SIDE3 has become a nightmare

[@bulldogcraft]

As a software developer I feel your pain and frustration. You have my respect and I appreciate everything you do for the Atari 8-bit community.

[@xBeetle]

what a rant! But you did a good job pointing out the problem. Will stick to my two Side2.... Keep up the good work, Jon

[@ChrisCromwellHP]

It would seem to me that “Lethargic” has long already fired you from working with him ever again. That would explain why he completely cut you off and moved on with someone else with newer projects. Lethargic would say that you were being extremely insubordinate with him. But yeah, Lavoric or Lethargic or whatever is name is, was only obsessed with profit, he didn’t want to lose alot of money he didn’t have with beta testers; so he just sent them all out as they were. Definitely not the best way to do things, but this is Poland we’re talking about here! 🤷‍♂️😂 I hope you will find another revenue source soon. I really enjoy your videos and British accent/slang talk! I subscribed! I’m a new owner of an Atari 800XL that I heavily modded and recapped myself. This thing is a blast to use with the entire library of games!

[@radovansabo9863]

Ďakujem

[@flashjazzcat]

Thank you!

[@buriedbits6027]

1:21:06 facts and evidence always prevail. Kudos mate.

[@qqryqq123]

All I can say - thank God I purchased two SIDE3's (and a batch of U1MBs) last year. And thanks Jon for your support to A8 community and excellent firmwares.

[@Madgitty2]

Hi Jon This is my baby i gather, can i say what a incredible customer service you have given, you have let me know every step of the way with all the faults that you have found and what you are trying to do to fix this. I feel so bad this has happened and the amount of time and money to solve this problem, i am eternally grateful.

[@flashjazzcat]

That means a lot, and frankly makes all the trouble worthwhile. Thank you. Sorry for the short hiatus, btw. It doesn't look like there's any interest in further analysis of your setup, so what say we get things tied up now so we can get your computer back to you? I'll get on that tomorrow. ;)

[@Madgitty2]

@@flashjazzcat thank you so much

[@atariandre5014]

Completely understand you and support you. Haven’t updated my U1MB and SIDE3 for some time as it just works fine for me on my 600XL with no chips changed….

[@OscarFowler]

It's really disappointing to hear that Lotharek has become so difficult to work with, FJC. As a life-long troubleshooter, programmer, and IT professional, you have all my sympathy in this situation which could've easily been avoided. I've always been impressed with the quality of your work on both the technical and artistic end of things, and I hope the Atari community continues to benefit from your output for decades to come.

[@byteoncomputerservices8054]

I'm glad I bought my SIDE3 3.1 upon release. When Candle mentioned to me on the AA forums that he doesn't believe in open source, alarm bells went off in my head regarding situations exactly like this. I understand that software and schematics should be kept close to one's chest, but JED files should be made available upon request in situations such as this - Better yet, 'situations such as this' shouldn't exist in the first place regarding the final product. Jon, your obvious passion regarding the product is exemplary. Every time I needed help regarding U1MB and SIDE3 on AA, you were always beyond helpful.

[@flashjazzcat]

Thanks so much for the positive comments. What can I say other than that I think customers are entitled to after-sales assistance if they need it, and it seemed to me a wise investment of time to help them end up with a working system. IIRC, U1MB was initially intended to be effectively 'build it yourself if you want', but the situation became altered when the IP changed hands. And after that, we had the 'Pancake 1MB' board, which was a reverse-engineered U1MB clone which would have been quite useless had the developers (who lacked the ability to write their own CPLD code) not had access to the freely available JED binaries floating around on the Internet. There was also a SIDE2 clone. That situation clearly informed future policies regarding JED files on other devices, the fear presumably being that someone could reverse-engineer SIDE3 with relative ease, stick the JED on the CPLD, and start selling them. In any case: I think users are a little more forgiving of firmware issues on 'build it yourself' open-source hobbyist devices than on closed-source, commercially marketed hardware which supposedly sells with a warranty, regardless of the fact that everything under discussion exists under the 'hobbyist' umbrella to some extent.

[@paulmetcalf4471]

Wow Jonathan! That was a long one for sure! Worth the watch to the end though. I was waiting to make a decision on which cart to buy, the Side2, Side3, AVG, the ultimate cart, and finally the AVGSubcart. This video reinforced my decision to ultimately (no pun intended) go with the AVGSubcart in the end. It just seemed to support every file type and also your Sideloader firmware. With that said, even the Subcart has its issues that I haven't quite figured out how to get past them. My machine is a perfectly stock 130XE and I'm hesitant to do any major mod upgrades to it, so the Subcart seemed the way to go, or so I thought. I also own a Fujunet v1.6, and to be honets, even though the commnuity has said great things about it, I'm not thrilled with it. There are too many firmware updates and constant development changes that make it more and more bug ridden. The only product I was ever completely satisfied with was the S-drive Max. It just works, however it doesn't support all file types. It's a shame that a lot of these projects that are supposed to be heavily tested before being released to the Atari community really aren't. I agree with you, the end purchaser/user is not and should not be a beta tester. The Karin minidrive is another one I purchased recently as it looked like a good idea, but once again in the end it only sort of works. Now the A8-picocart is the latest thing the community is raving about.......it clearly hasn't been tested enough either. So for now, I'm going to save my money and wait until something has been out that actually works and doesn't get changed all the time for the sake of cost or whatever. Matter of fact I'm most likely going to start parting ways with some of the new gear I bought thinking it would be great after seeing videos that other people have made stating this....the whole you have to get this as it's the latest, greatest thing for your Atari. I wonderhow many of those people were paid by the product developers to put such videos up? Hmm......anyway, I hope things can be resolved on the whole Side3 matter with you and Lotharek.

[@flashjazzcat]

Thanks for seeing it through to the end. :) Expectation management is super important in the realm of hobbyist equipment, I think, especially concerning forty year-old computers. Even AVG, SubCart, etc, are not going to be completely immune from issues (although they do seem remarkably robust, generally speaking), and it's only the well-established stuff like S-Drive max which is going to be feature-complete and more or less bulletproof. Of course it takes time for people to perfect hardware and software when it isn't necessarily their primary occupation, but this is all the more reason not to throw away five years of arduous progress with a single poorly-tested firmware revision. My most recent of slightly unscrupulous salesmanship was when I was commissioned to put together a Rapidus 800XL, the shopping list for which had been drawn up by the vendor (whom the client had approached directly). The client - as it turned out - didn't even know what most of the upgrades he'd been sold were for, let alone how to use them or why he might need them, and when we started to encounter the inevitable snags during the very ambitious build, it was down to me to enact the expectation management (it turns out this thing you're paying for doesn't work with another upgrade, this other thing just doesn't work period, etc). Somehow, at the end of it, and after paying a considerable amount of money, the client appears happy with what he received, despite the fact I absolutely could not warranty it as working perfectly. In any case, we have an opportunity to 'fix it and move on' here. All that remains to be seen is whether the opportunity will be taken.

[@paulmetcalf4471]

Owning a completely stock machine was important to me when I purchased a 130xe a couple of years ago. I had one back in the day, and I wanted to replicate that system. I went so far as to reverse a "big" RAM mod someone had done to this machine at some point in its life. I know it's amazing what can be done with a lot of these new mods, but it really doesn't interest me much. The only modern things that interest me has more to do with file storage and loading than anything else. And those files are pretty much just games I had back in the day, not the mega byte + demo software that seems to keep being developed. The s-drive just works, but because it's emulating real floppy drives, it's slow, just like the originals were. I never understood why Atari went the serial port route with most of its periferals when they had a perfectly good parallel buss available to them ......

[@tschak909]

New JED works with a significantly marked up phase2 delay shifter... Some would call this cartel behavior.

[@flashjazzcat]

I think it's completely by accident, but it might well be the case. 'Never attribute to malice what can be attributed to incompetence', although since Lotharek is now inferring that he is considering materially penalising me (via recovery of firmware commission) for customer-returned cartridges RMA'd because of a poorly tested JED about which I knew nothing until three months after it was put on every production cart, malice might actually explain it.

[@user-Atamigaputer]

very disappointed in lotharek's lack of taking responsibility

[@wieczor3000]

Ser his answer first

[@flashjazzcat]

The answer in which he fails to mention I told him of issues two months before I published the video, and fails to mention that Candle urged him to send me a copy of JED 2.3 and a means to flash it in February 2024, which would have circumvented the entire situation? Every single premise presented in my video - whose central complaints concern the lack of testing and the complete exclusion of yours truly - stands.

[@wieczor3000]

​@@flashjazzcat Thank you for your answer, I didn't know about that. I think gentlemen, this failure has been made, and regardless of who is to blame, as an experienced engineer, I advise you to hold a meeting together to establish clear and inevitable development, testing and release procedures. Point by point - easy to understand and follow. Also - minimal test environments. This will certainly help avoid misunderstandings in the future. 99% of them always come from the assumption that something is obvious - NOTHING is obvious should be the rule :)

[@user-Atamigaputer]

@@flashjazzcat yep i will no longer be buying from lotharek until I see an apology

[@flashjazzcat]

@@wieczor3000 Sensible advice. Let's see if there's any desire to move forward. Of course nothing should be taken for granted when changing firmware or hardware, or even when decreeing that such changes are needed.

[@jrherita]

I wish I could offer something to actually help here. I appreciate your passion for the Atari platform and community. The Ultimate 1MB / Incognito firmware is a work of art IMO. Thanks FJC.

[@flashjazzcat]

That's very kind - thank you!

[@TheWoj76]

Jon, the U1MB board you discuss in detail starting around minute 22 also seems to have some corrective hand soldering on the square chip underneath, at least under your filming light it occurs to me so.

[@flashjazzcat]

Agreed: it looks as if it has shorts on the legs, but it doesn't. Maybe a trick of the light.

[@colinellett]

Entertaining video! I've bought quite a few things off Lotharek over the years and had no problems. I was thinking a getting a Side 3 for my Atari 65XE (upgraded to 130XE with VBXE). I think I'll hold off for now and stick with my SIO2SD. That Ultimate 1MB + Side 3 addon would be amazing, but will it ever be ready to be released. Doubtful ....

[@flashjazzcat]

SIDE3 has been withdrawn anyway at the time of writing, which is the first responsible thing he's done concerning the entire issue.

[@mattsworld7943]

I’m already on my second Side 3 cart after it just died. You had kindly reflashed my first Side 3 after I had a failed JED update. Which turned out to not be compatible anyway. I won’t be buying another as most of the time it works. But only in half my Ataris, which is what it is. I can truly understand your anger Jon. Until this mess is sorted out I won’t be spending my money. Everything seemed rock solid a few years ago.

[@flashjazzcat]

You've definitely been very unlucky there, since I've never known of a SIDE3 to just drop dead. Despite the methods I employed in the video, I would definitely advocate simply RMAing a defective device, on the assumption that customer returns will be handled in good faith. In field JED update failures are not uncommon, meanwhile, and I believe this to be down to updates being attempted on cartridges which were exhibiting stability issues on the machine to begin with. The situation is hardly helped by something I didn't even mention in the video, which was the frustrating and complete lack of beta testing of the JED flashers, despite my repeated pleas to distribute the tool to people in possession of bad flash recovery methods (i.e. the Lattice platform cable and a copy of the JED). Any hope of this basically vanished once fear of hardware cloning prevented JED files from being placed on general release or even distributed to any but a handful of 'trusted' parties. And of course the whole matter of in-field JED updates is now moot, since the cost-reduced GoWin CPLD completely lacks the 'Wishbone' interface essential for software JED updates.

[@achaney]

Best rant on KZbin, Watched him unleash the fury 3 times now. Phew... he was HEATED!

[@flashjazzcat]

I can only say I learned from the master - Louis Rossmann. :)

[@MoparStephen]

What an awful situation. I agree - it is quickly tarnishing a reputation (not yours) that took years to build. The more time goes on, the less faith I am putting in a certain seller. Not to mention waiting for almost a decade for a new Simple Stereo device, since version 2 was a complete failure that never worked.

[@flashjazzcat]

Personally I thought discontinuing the last incarnation of the Lotharek Stereo board was a bit premature, regardless of whether POKEYs are like hen's teeth. The thing had the u-switch logic built into it and is exactly what I would purchase for customers just wanting stereo audio in the majority of cases (I have one of them here, and I'm holding on to it). Presumably the assumption was that Simple Stereo 4 (of which I received a pre-production sample nearly two years ago, and am no nearer installing it in any machine) was 'imminent', but - as you say - it's still nowhere to be seen. The same is true of External U1MB, Incognito Lite, Incognito 3, etc. Certainly I was pretty much out of the loop for about six months while moving house, but to produce firmware for any of these products, I first need a) confidence they will ever see release, and b) working boards with current and actively maintained CPLD code. I have neither.

[@jamesalexander9468]

@@flashjazzcat I remember your video on those and wondered for the longest timem why there was no progress on any of that. Same with rapidus, I wanted 1 but soon found out it doesn't work well with U1MB or other hardware.

[@TzOk]

Przemek doesn't seem to speak to anyone who dares to express any kind of criticism of his products. Also, he is a self-taught practitioner and is missing a theoretical background in electronics.

[@flashjazzcat]

I suppose one can position oneself as 'just a salesman' or someone who considers himself to have sufficient technical insight to mandate fundamental changes to the hardware. Both situations cannot exist simultaneously.

[@TzOk]

@@flashjazzcat Many people can't tell the difference between knowledge and experience, which leads to overconfidence.

[@JimWood28]

Jon, my god talk about doggedness. You're a true bloodhound mate. More power to you.

[@SimonWells-f8s]

So after those purple Incognito and u1mb are sold out will there be no more u1mb at all? That seems odd.

[@flashjazzcat]

U1MB is a steady seller, so I doubt that will ever be taken off the market. I see no likelihood of it being succeeded by it's external (PBI/ECI) counterpart, either - if that ever materialises.

[@buriedbits6027]

1:02:51 what is JED?

[@flashjazzcat]

Customer's cart and its replacement both shipped with JED 2.3.

[@angrydove4067]

Wow, I had no idea about this problem, I have the same one as that customer got and it works perfectly without swapping the chip in the 800XL. Whew! Thanks for this video. I came here after watching Lotharec's video, wondering WTF is going on.

[@flashjazzcat]

Yep - as mentioned, it can randomly work if you're lucky. No wonder you were left bemused after watching Lotharek's video, mind you. :)

[@troyquigg4411]

@flashjazzcat - Great video as always! Excellent troubleshooting! Just wondering why there is no source control on this so that you would be able to see what changed and roll back easily (without worrying about finding an older version file that may no longer exist). No need to make it open to the public, but just my $.02.

[@flashjazzcat]

Well, that would make sense and is exactly how I work. I have a hard enough time unpicking diffs run on prior loader versions when I lose track of something, but without those revision histories, I'd be done for. Quite honestly, detail concerning what was changed in the JED is vague, but I hope (and assume) Candle knows exactly what was changed and can unroll or amend things as necessary. If not, we're in trouble.

[@TrimeshSZ]

I have worked on a couple of projects porting stuff to those Gowin devices and I think I have an idea what's happened here. The Gowin chips have a somewhat unusual architecture because their logic blocks (which they call "CFUs") contain both logic and routing resources, so depending on the placement the fitter might have to route signals through multiple CFUs. This makes setting a long default maximum propagation delay and just assuming that it's always going to be much quicker than that (ok, it's bad practice - but it works pretty well on the Altera and Xilinx parts) a major problem because if you tell the fitter that this signal can tolerate lots of delay then it may well take you up on it and route it halfway across the chip. My guess is that the HDL had timing constraints that were either missing or too loose but worked anyway with the original FPGA.

[@flashjazzcat]

These kinds of concerns definitely ring true concerning my prior conversations with Candle concerning the subtelties of CPLD signal routing and resources (not that I have much knowledge about it: I just asked because I was interested to know why the same VHDL can function differently depending on compilaton conditions, etc), but what's still puzzling here is how and why the older Lattice JED ported to the GoWin CPLD works well here while the newer JED tweaked specifically for the GoWin CPLD does not.

[@TrimeshSZ]

@@flashjazzcat It's hard to know - but my experience has been that even small changes can have drastic effects on the final routing, including significant changes to the timing parameters of signals that you haven't touched at all. This is generally true on FPGAs, but the distributed routing on the Gowin parts makes it even more apparent. Basically, if a signal has a timing constraint then you have to tell the fitter about it and it's unsafe to make any "it should be fast enough" assumptions, especially as you use more of the available logic capacity. The good thing is that in my experience the software does exactly what you tell it to do and if you have a fully constrained design it will perform as required.

[@flashjazzcat]

@@TrimeshSZ That sounds absolutely right. I remember Candle would send me a Lattice JED, I'd test it, find it had issues, and then Candle would simply make some compile-time adjustment to the routing or resource usage and everything would magically work. It's exactly the nightmare I referenced in the video, though: if the hardware behaves indeterminately but the developer swears up and down that it 'must work', the software naturally becomes a suspect and it's possible to enter a death-spiral of software and hardware changes which actually end up making the situation worse. This is the exact reason that despite the known imperfections in Loader 0.69, core functionality has been exhaustively proven to work over the course of two years and it provides a baseline constant on the software side which is the bare minimum requirement while we diagnose these JED issues.

[@gyorka]

@@TrimeshSZ I was suspecting there is something that goes along with how fast GoWin software was able to compile this code. When it came to constrains, I always looked for worst propagation paths, but ON original design (lattice machxo3), never on GoWin parts just because there were other issues regarding Lattice/GoWin swapability, and GoWin chips not starting at all (they were fine if programmed to SRAM, but never ever booted up from embeded flash after power was applied ). Anyways, thanks for the tip - i'll certainly check this out WHEN I find enough motivation right now - whole thing brought me down to the point I don't want to do anything more than I must stay alive (don't be too worried though)

[@gyorka]

well, maximum clock for the design is 84.2MHz, (I actually use 57.something - just phi2 going on 32x pll chip) and then synchronised with one-hot event scheluder based on a counter running off that pll output and resynced to phi2 falling edge every cycle - when it works, i have 34ns of stable window of operations i can move freely within, but when it doesn't - no changes i can make to the code will change this - simple ls08 swap (even for another ls08 chip) will and this puzzles me the most. no negative slack present within the design, worst clock path has 4.8ns of slack, and all concurent stages are pipelined and use semaphores between stages just in case. i'll probably refactor this code one day and make it more modular, but for now, if nothing new emerges i think i'm done here

[@AdamKlobukowski]

I've watched this and Lotharek's answer. I'd like to start with that I love and apriciate what all three of You (Lotharek, Candle, FJC) do for the community. That said, I think you seem to have big communication and project management issue here. What we see here is a hobby project that grown and is not hobby anymore. What is obvious for me is that you need a project manager person to hold it all together. You need project management, release management and maybe more. I hope that all this problems will soon come to pass, and we'll soon see more incredible releases :D

[@flashjazzcat]

Lotharek's response video (which I will rebut in the fullness of time) seems to completely miss the point. Not only that, but it contradicts Candle's point of view on a number of issues. Lotharek fails to mention that back in February (according to Candle), Candle repeatedly asked Lotharek to send me a copy of the new JED and a means of flashing it, so that I (as the author of the Loader and drivers) could test it prior to release. Not only did I receive nothing, but even after requesting the GoWin programming interface three months later when I first learned of JED 2.3's existence, I received nothing at all. So it seems to me that due process was advised by the author of the JED 2.3 (who was solicited into creating it) but simply not enacted. This negligence comprises the central complaint in my video. He also indicates that he 'heard nothing' for four months until my video was published, when in fact I raised the alarm bells concerning JED 2.3 on 17 May, two months before the video was released. While I agree that this is a project management issue, simply not excluding 1/3 of the three individuals involved does not require class-leading management skills or titanic intellectual capacity. Two or three Whatsapp messages and potentially one £25 GoWin dongle sent in the post would certainly have prevented the whole situation from occurring. Instead I was left in ignorance for three months and then left hanging for a further two months while attempting to sort the situation out myself. I suppose - given the fact Lotharek hilariously entertains the notion of recovering from me firmware commission paid on devices RMA'd because a poorly tested JED was released without my knowldge - I can whistle for the £130 I spent on a replacement cartridge and a GoWin platform dongle. Having spent two hours talking the matter through with Lotharek last night, and actually believing we understood one another and that he was determined to fix the issue, it's disappointing to say the least that he produced a 'statement' thereafter in which he lays out his plans to materially penalise me (and Candle) yet further.

[@AdamKlobukowski]

@@flashjazzcat that is exactly why I think that some project management is needed here. Clear lines of communication. Clear separation of responsibilities. Release management. You (and that is a plural You) outgrew hobby level of doing things.

[@flashjazzcat]

@@AdamKlobukowski Sure - I'm not disagreeing with you there.

[@Doug_in_NC]

I gave up with my (original version) Side 3 years ago after it failed to work reliably even with a 74F08 in my Ultimate 1MB-upgraded 800XL.

[@flashjazzcat]

That's very unfortunate. I must admit it was - prior to this - unprecedented for the F08 fix not to work on machines which have been through my hands. I have heard of a few cases where it didn't work for other users, however.

[@michaelwessel4953]

Totally understand your frustration. Some cart pushers trying to maximize their profits whereas other like you doing the bulk work behind these extensions with little to no prospects of ever getting their fair share of the cake. Dismayed how it turned out. They are complete idiots by not keeping you tightly coupled into the loop. For the hardware I developed and distributed, I always treated my contributors and developers like kings. I.e., they got everything they needed for doing their work for free. But, I also never managed to make a dime from my products, really 😅

[@flashjazzcat]

Thanks for the considered comments. And yes: it's not as if I wasn't an integral part of the testing process when the original VHDL was being developed (indeed, it was a case of Candle sending me an update, and me testing it), so Lotharek enacting changes without my involvement was completely retarded. I think it was at that point that I realised I was dealing with people who a) don't understand what testing is, b) don't understand software, and c) can't carry out basic communication. The situation was therefore entirely untanable. Even Lotharek himself retrospectively blames a 'lack of communication', although he fails to acknowledge the efforts I made to resolve the matter privately during the six weeks between my discovering the problem and releasing the video. The firmware has remained free to download for end-users since the day it was first released, but obviously commercial vendors of hardware with a high markup which is functionally useless without said firmware are expected to cough up money. Lotharek has told me himself that he was getting tired of paying me my commission anyway, so this miserable outcome had been on the cards for quite some time. It wasn't fun the whole time it lasted, but it was fun for a while. ;)

[@edbrown1166]

Thanks for sharing. I have had issues with a different product (Rapidus for XL) and installed it myself (I'm no stranger to installing/soldering) and found that it just doesn't work. But, now after watching this, I'll try swapping out the LS08 for a F08 and see if the Rapidus starts to work. I've tried reaching out to Lotharek about the board not working but, got no reply - as far as I'm concerned that's piss poor customer service.

[@flashjazzcat]

Heh - Rapidus. Don't get me started on that. There is quite a backlog of material on this channel concerning Rapidus and the issues it has with U1MB, VBXE, etc. Watch this if you haven't already: kzbin.info/www/bejne/nnnJfKGcmKiNY5I

[@philiphoeffer7442]

I guess I won't be upgrading my SIDE2 for a SIDE3.

[@fragalot]

I'm not sure I'll buy a Side3 just yet... though im not in the position to buy much these days. Tho i'd be happy to test a Side3 cart on my stock 800, and 800XL with a RAMBO XL (256k) upgrade.

[@flashjazzcat]

That's exactly what's needed (testers), but how to equip them with the ability to test different JED files when no-one is allowed access to the JED files? :)

[@glenndoiron9317]

@@flashjazzcat The problem with distributing the .jed files is two fold. 1. People might flash the wrong .jed file for their PCB revision and brick the board (applies if there are different board revisions with the same FPGA). 2. Chinese clone companies will build the boards and sell them for $5 since they didn't have to pay anyone to reverse engineer the FPGA or write the software (also you are going to be bothered about supporting a clone board which you made no money from and had no quality control over).

[@flashjazzcat]

@@glenndoiron9317 Quite - these were among the concerns already considered. However, in the case of (1), the GoWin CPLD lacks the 'Wishbone' software programming interface, and therefore any field update requires the GoWin platform cable (£28 including P&P), and if the tester/user has the cable, a bricked board can by definition be just as easily un-bricked. The cloning issue, meanwhile, was the driving inhibitor of any wider field testing of any kind, everyone apparently being a potential pirate. I already experienced the 'Pancake Ultimate' board which - being subject to no formal licensing arrangements - may be hidden in any number of 8-bit Ataris without having generated a penny in IP revenue for myself or Candle O'Sin (notwithstanding the fact I received several unsolicited arbitrary payments which I suspect pertained to instances of the 'Pancake').

[@fragalot]

Wait 13:10 how are they programed in the first place? Do you use some other device with the chip NOT in the cartridge? I guess when they are mass produced they are programmed before they are soldered to the PCB and made into cartridges?

[@flashjazzcat]

Your guess is as good as mine. I know the loader needed to be signed off before this production run started, but that's on the ROM chip. It looks like JED 2.3 was applied to the cartridges after the fact with a cable, however (they came from the factory with JED 1.3 on them).

[@GadgetUK164]

What a nightmare lol! It's also worth pointing out that the JED would need to be based on the Gowin chip! ie. I believe a Lattice .jed file won't work on the Gowin chip - unless it has been designed to be more or less a clone. You can get those Gowin programmers much cheaper (like £20 ish or less). There needs to be better communication between the hardware guys and yourself! Edit: I realise you worked that out - just getting to the end of the video now! What a load of work you had to do - you gave me a bunch of laughs at various points lol! I was about to buy one of these, I might postpone that until the jed issue has been worked out... The obvious question at the end of this - what the hell is 2.3 supposed to fix?!? PS: Have you tried turning it off and on again?

[@flashjazzcat]

Well, all I can say is watch till the end. ;)

[@GadgetUK164]

@@flashjazzcat I have done =D I posted my message a bit prematurely lol!!! Great work - I had to program one of those Gowin chips recently (using same programmer I think) - I didn't enjoy installing the software etc lol. I hope they can work out what's wrong with 2.3, and realise they need to keep you in the loop!!!

[@flashjazzcat]

Yeah: I don't think it's unreasonable to expect to be kept in the loop with hardware I install very frequently and for which I maintain the firmware, and get paid for so doing. On that subject, I was paid June's commission this evening without having to chase it up, but there was no covering email or any reference to the original programming cable (which was simply never sent out), nor any enquiry as to whether I was able to overcome the issues I reported two months ago almost to the day, whether any other expenses were incurred (which there certainly were), etc. No apparent interest in the outcome whatsoever. I'm glad it was amusing in places, anyway. I even get a few wry laughs out of it, watching it back. 'Is it a Mac or PC?' 'Yes'. LOL.

[@flashjazzcat]

@@GadgetUK164 As for what JED 2.3 was meant to address: without going into a lot of detail, I remember the conversation which presumably (without my knowing) led to its creation, and I do think that it was a case of red-herring issues caused by the configuration of the system or systems used for last-minute testing of SIDE 3.2 stock before it shipped to the vendors. Premature to say the least.

[@c1ph3rpunk]

I’m going to have to check mine, was ordered in early December from Lorharek, received mid-January in the US. I just put the U1M I ordered the same time in, so far it appears to work but I haven’t pushed it much yet. Would you happen to have a video on the 74LS chip replacement? While my 800XL is open, I’ll just swap it out.

[@flashjazzcat]

You'll be fine if the cart was recived in January. JED 2.3 dates from February 2024. U1MB should also be fine in every configuration; fear of complication with the Alliance SRAM 'appears' to have been unfounded. I don't have a video concerning 74F08 installation, but it does crop up in several videos on broader topics. If 74LS08 is socketed (it's right at the front of the middle row of small horizontally-aligned ICs), you're golden: it's a straight swap. If not, you'll need chip extraction tools and a 14-pin socket.

[@c1ph3rpunk]

@@flashjazzcat did some reading late yesterday, is the one that’s usually swapped U18?

[@flashjazzcat]

@@c1ph3rpunk That's the one, yes.

[@jamesalexander9468]

Jon, You've done more than most with helping people with their 8bit gear. Even in instances where you didn't get paid for it like when you helped me via email last year with my 800XL (which did get fixed at least in part with your advice). What you've shown here is conclusive proof of what the problem is and how to fix it and just as importantly that it's not your fault despite what the unprofessional buffoon says. The situation smacks of somebody (who isn't you!) looking to cash out with 1 last whack at the pinata with little regard for others but it will backfire on him going on the reactions I'm seeing in the comments here. You were 1000% right calling this out for what it is. I'm 1 of the people glad you made this video showing the reality of the situation. Despite the fact I got a U1MB and a Side 3.1. Both still work fine. But I've since added a 130XE and an XF551 that needs a bit of work. Frankly I'm gonna be looking for something different for upgrades on that system (not decided which yet but won't be another side 3).

[@flashjazzcat]

Thanks so much for the kind appraisal, and while I'm pleased to hear your existing U1MB/SIDE3 setup is working well, obviously it's disappointing that you're turned off from purchasing these upgrades again in the future. It's not my intention to denigrate U1MB and SIDE 3 (indeed, that would be like cutting off one's nose to spite one's face in one sense), but I do think people basically deserve to know what it is they're thinking of buying. I'm sure the new Ultimate 1MB boards will work fine (well, I hope so), and I'm hopeful that the SIDE 3.2 issues will eventually be worked out, but frankly it doesn't take much to lose the trust and confidence of the customer and user base.

[@jamesalexander9468]

@@flashjazzcat Well I been thinking of what to add to the 130XE since I got it. Less urgency on ram expansion as it already has an extra 64k which lets it run nearly all of the software for these machines & thinking of an Sdrive max. The problem that you've pointed out here is its likely that as of now that the U1MB will probably work the Side3 most likely will not and its not likely to be anytime soon since somebodies ego (not yours) is involved. What makes it worse is the jed file needed to get the newer cpld working isn't publicly available and the cable to update it is something I'd only use once. In the mean time my electrolytic cap set for the 130XE just arrived yesterday & I've still gotta fix the XF551 (I still like having at least 1 floppy with a system) which is giving me the slow boot error problem despite the fact I've already replace the notorius 7.4 version of its firmware with 7.7 (tried the hyperxf but kept getting a strange prompt & it didn't help the problem either).

[@Clancydaenlightened]

What about altera cpld? Epm570t and larger fpgas Downside is 3.3v level shifting needed I could throw and altera max ii, esp32, some sram, and SD card Don't really need a ROM, just an os handler (driver) and menu, can load that via Sio...

[@Clancydaenlightened]

That way I can use wifi, and also use Bluetooth to connect to a computer or phone load some ROMs, or wirelessly print on lan printer

[@Clancydaenlightened]

Ask lotharek how feasible it is to replace the ram in Atari with flash memory If the R/W cycles are under 100ns The Atari shouldn't tell it's non volatile Just need different boot init for the base 64k, and 512k page on top Then use the megabits above in place of the floppy drive Even modern flash chips still have parallel data modes And you don't need battery backup

[@Clancydaenlightened]

How much faster can my Atari "800xe" run without waiting on dram refresh

[@buriedbits6027]

The root cause appears to be hardware revisions made without communicating these changes to the software developer. To make matters worse, it seems the device is put on the market, leaving users to catch the bugs. That said, the transparency shown in addressing these issues is commendable and incredibly gracious.

[@flashjazzcat]

Thank you. It certainly seemed to me - after days of careful editing and repeated full viewings prior to release - that unless the piece was viewed through a fog of indignation and annoyance, it would be impossible to reasonably argue with any of the points presented. The response which appeared shortly thereafter adopted the only possible strategy other than simply saying sorry and undertaking to fix the problem: namely, blame-shifting and choosing to ignore the fact that nothing was done about the problem for two months after I privately raised the alarm and provided a detailed diagnosis of the issue in mid-May.

[@FalconProductionsus]

I hope that these differences/disputes/problems can be worked out. I hate to see this. My Atari hobby gives me something to look forward to. I totally understand the frustration expressed, but we need the combination of the hardware availability, your excellent firmware, and support. We will all be in a mess if there is not a supplier of the hardware, support, and firmware. I hope that this does not become a wall that puts and end to great combinations of hardware and firmware for the end users.

[@flashjazzcat]

I spoke to Lotharek in private for two hours on Wednesday evening and genuinely believed by the end that he was intent on making things right. Unfortunately in his recently published statement on KZbin, Lotharek infers that he is considering penalising me yet further (over and above time, money and resources already invested) in a material sense. So it's hard to know whether I will be involved in any of these products at all going forward.

[@glenndoiron9317]

A .jed file (in this case) refers to a bitstream which is used to program an FPGA. The old .jed file for the Lattice fpga has absolutely no chance of working on the Gowin (unless the people at GoWin stole the Lattice IC chip masks and are making identical clones.) If you were somehow able to send up the old .jed file, at best the FPGA is bricked until the correct configuration is written ... but more than likely, there will be internal gates configured incorrectly, resulting in outputs connected to other outputs, and/or fpga pins which should be inputs, configured as outputs. In either case the FPGA probably self-destructs.

[@flashjazzcat]

Yes, this became apparent as soon as I set the programming software up, which is fortunate since no-one thought to tell me beforehand that GoWin didn't use JED files and that I'd need the '.fs' bitstream, previously (in the case of version 1.3) produced from the Lattice VHDL sources compiled specifically for the GoWin CPLD. Then a few tense moments when it appeared that Candle had consigned GoWin JED 1.3 to the trash, before finally the file was produced, flashed, and the problem determinately fixed.

[@FelipeBalbi]

I was considering ordering u1mb and side3 for two of my machines. Guess I’m not ordering them anymore.

[@flashjazzcat]

You certainly won't be ordering SIDE3 from Lotharek now, since he's withdrawn it from sale, and done the same with the FJC-firmware U1MB for reasons completely unknown (stock version is still available).

[@pfcrow]

If I'm following this correctly, the bad JED code is only an issue with DMA. Is there any way to work around this and not use DMA with a firmware update? Obviously that would cause a performance hit, but at least the cartridges would work. Or maybe a better solution would be to blacklist the JED version in the firmware and have it report a warning message on boot. I haven't followed all the details of the problem carefully enough to know, and it's likely not worth the effort on your end.

[@flashjazzcat]

It initially seemed like a DMA issue, but I speculate later that it could simply be a CPU SRAM access issue. I think it's reasonable to entrust the designer with the task of figuring out what the problem actually is, however. I am disinclined to remove DMA SRAM to SRAM copies from the directory reader since reliance on DMA elsewhere in the software is not optional, and if the hardware cannot work sufficiently well to build an array of filenames, it will surely fail elsewhere and in similarly problematic ways. Older, more primitive versions of the loader did not leverage DMA when building the directory, however, and that fact simply encouraged lo-fi, high-noise commentators to conclude that the newer loader was buggy when in fact more extensive reliance on DMA simply exposed the instability of the system. It's also worth remembering that JED 2.3 caused - on video - the two identical SIDE 3.2 cartridges to not even get as far as running the loader and instead send the OS direct to the memory test. That last idea (blacklisting JED 2.3) is very appealing, however, and something I had already considered. :) But as you say... what is the bloody point? Thanks for the comment!

[@pfcrow]

@@flashjazzcat A self-test might be generally useful if you can identify an easy way to do it and the most likely types of faults to look for. But likely not a top priority.

[@flashjazzcat]

@@pfcrow It's surprisingly difficult to simulate the exact operating conditions present in the loader using a separate stand-alone stability testing tool. I have tried; I might even have such a tool lying around for SIDE3 which I produced during the pre-release phase. The problem is that the loader is simultaneously banking ROM code, calling the DMA engine, writing information to the SRAM on the cart, etc. It was explained to me at one point by Candle that a problematic development JED was exeriencing problems remaining synchronised to the bus during these kinds of interleaved operations (note: I summarise here and speak without tremendous technical insight). As you can imagine, this presents a very difficult environment when it comes to debugging, and I have spent (wasted) a disproportionalely large amount of time over the past five years being sent down blind-alleys searching for software bugs which - for the most part - did not exist and having to justify and prove the functionality of code which somehow works or doesn't work depending on the JED revision or Phi2 signal integrity on the host machine. Therefore the onus is always on me to 'prove' my code, which is especially ironic given how lax the testing has become on hardware and JED revisions. Meanwhile, my code long passed the forty-thousand lines of assembler mark (and is due for another big refactoring pass), while the VHDL represents one or two thousand lines of code, and once the board layout is done, it's done.

[@ChrisPatti]

Sorry for all the trouble you’re having. Communication is impossible.

[@flashjazzcat]

One can speak, but one cannot make others listen. :)

[@darrenhunter2324]

Watching this Video has put me off getting any thing of lotharek for my ZX Spectrum, seen there is issues with Atari Hardware.

[@flashjazzcat]

Imagine what fun it will be for me if I have to purchase more of the stuff for customers who want it in the future. I will be braced for issues every time the postman arrives.

[@darrenhunter2324]

@@flashjazzcat I feel your pain, Been a Dev my self. (Ok it C#, HTML5, CSS and Javascript) with a bit of DB thown in there. I agree with other comments made on the Video about needing a Project Manager to help sort this mess out.

[@flashjazzcat]

@@darrenhunter2324 I'm coming round to that way of thinking more and more as well. The fact is, the 'coming soon' products I advertised in late 2022 have no development strategy that I am aware of, assuming they ever see the light of day at all. And here I am now with the two products which generate the bulk of my firmware revenue withdrawn from sale on an open-ended basis (and I wasn't even personally notified of their withdrawal), and no way of planning ahead regarding use of my time when it comes to firmware development (since I have no idea whether it will be needed and whether it will generate any further income).

[@josch1710]

Oh man, I understand, you are mad. I remember chatting with you about similar stuff on AtariAge, when I was still there

[@flashjazzcat]

Yes. This is just the latest development in a long-running saga of what was once an enjoyable and uncomplicated way of earning a modest living turning into a high-stress ordeal.

[@robertplestenjak]

Watching this got me thinking. My SIDE3 3.0 is collecting dust for a while now, since it won't work with 0.69 driver on any of my machines. I downgraded JED from 1.3 to 1.1 and now it looks like it's working. Thanks! Lotharek stopped selling SIDE3!

[@flashjazzcat]

That's great: I'm so glad a downgrade (from 1.3 o 1.1 in this case) worked for you, and thanks for reporting the outcome here. :) Doesn't thus just underline that there is already plenty of variance between the behaviour of the two prior (and reasonably well tested) JED revisions, without adding yet a third (poorly tested) JED (and on a different CPLD) into the mix! As for SIDE3 being withdrawn: while I think that's a sensible move in the short-term, I'd have liked to see it happen in mid-May when I first reported the issues to Lotharek in private rather than two months later when I had to report the issues in public. More concerning again is that Lotharek has withdrawn FJC-firmware U1MB boards from sale without bothering to give me advance notice that this was happening.

[@klaso]

I tried upgrading my 3.0 (having JED 1.0) to SIDE3 loader 0.69 and it behaves badly, so I went ahead and upgraded the JED to 1.3 and the SIDE3 no longer behaves reliably. Any hints on how I can downgrade the JED?

[@flashjazzcat]

@@klaso You can use the 1.1 JED flasher included in the loader package on my website. My SIDE 3.0 wouldn't work well on the 1200XL I'm selling in the subsequent video, but after downgrading it from JED 1.3 to JED 1.1, it worked well.

[@klaso]

@@flashjazzcat Thank you, I’ll try that.

[@klaso]

@@flashjazzcat Seems I put myself in a bad position. The S3JED11.xex says SIDE3 not present for the SIDE3 I ran the JED update on. To make matters worse, I earlier updated the SIDE3 loader on a second SIDE3 and 0.69 does not behave well either but on this I haven’t (thanks for small wonders) updated the JED so it’s still 1.1. Is there an earlier SIDE3 loader available to downgrade the still working SIDE3 bar the loader to? Any recommendations for the seemingly bricked SIDE3? (This should teach me a lesson)

[@joeventura1]

A beater test?

[@w1katari]

doesnt work with u1mb correctly, doesnt work with usb power adapter, latest demo rewind 2 doesnt work - on other computers works.. somewhere something works, somewhere not.. :/

[@flashjazzcat]

USB power connectors are agents of chaos. As mentioned on the forum, Rewind 2 is working fine here, so it CAN work. Cold comfort - I realise that.

[@przemekkobel4874]

Glad after all the odyssey it didn't end in finding something like SD card going bad.

[@flashjazzcat]

Quite. That's why it's best not to jump to conclusions when testing. But this is the recommended SanDisk branded SD media, shown to work in all other cart revisions. And in point of fact, even a scrambled directory could not result in the kinds of errors seen here.

[@RowLikeaPro]

I bought a new side3 cartridge from Lotharek a couple of months ago and it has JED 2.3 and Hardware Version 3.1 and it works fine on my 600 XL

[@flashjazzcat]

That's great news. I'm really glad you avoided these problems.

[@RowLikeaPro]

@@flashjazzcat I also have the 64 kb upgrade from Lotharek, perhaps thats why it works? What will happen if I install 1MB uppgrade......

[@flashjazzcat]

@@RowLikeaPro Possibly. I made an educated guess in the video that his test machines would have 64K SRAM and/or 'O2 Fixer' installed in them. I'll be interested to know if U1MB upsets anything. Maybe all one has to do is spend a further £30-50 to make these things work, LOL. :)

[@RowLikeaPro]

@@flashjazzcat Also I have not installed the O2 Fixer, so that's probably not needed to fix the problem :-)

[@flashjazzcat]

@@RowLikeaPro That's reassuring, at least. :)

[@richardkelsch3640]

No hardware revision nor firmware revision, nor JED revision should be released to the wild until they have been thoroughly tested by all parties involved in its development. Sure, it takes longer, but it avoids BS like this happening. Lotharek totally screwed the pooch on this dumpster fire. My life experience as a software engineer and sysadmin has been "if it ain't broke, then don't fix it".

[@flashjazzcat]

At the very least. I don't consider something truly fit for release until it's also been tested by some end-users, at least if we're talking hardware changes which can't (easily) be undone in the field. I completely agree that it would have been better to just 'leave things' be, as well, but clearly something completely spooked the vendor at the last minute.

[@richcsst]

@@flashjazzcat End users are the last step for testing. The primary steps must be hardware, firmware and configuration before the users get to test it. He did cursory hardware checking, no firmware checking and likely nobody else checked it. I have seen this happen many times in professional projects and this always resulted in a major FUBAR chew out and sometimes someone losing their job. This is the biggest difference between "amateur" and "professional" and also why consumer releases can take a while. QA is massively important. As a sysadmin, if a project or code hasn't been checked by at least the dev team, testing in stage, and three members of the QA team, the request to push code to production gets a hard "NO". In this situation, all four aspects (hardware, firmware, JED,. users) are the "QA" team.

[@fragalot]

37:07 Wow i thought you wrote the firmware on an actual Atari 8bit machine.

[@flashjazzcat]

I don't think it would be possible at this point. The source code for the FAT driver is 100KB long, and that's just one of over twenty source files.

[@skywiseca]

@@flashjazzcat Did you at one point write drivers N such on an Atari? What was that setup like?

[@flashjazzcat]

@@skywiseca I did start a rewrite a while back and commenced work on a 'NET' driver for SpartaDOS X, but simply got side-tracked with other things for the moment.

[@buriedbits6027]

56:43 oh bother.

[@flashjazzcat]

Yep. And that's before the software even got a look-in.

[@waltciii3]

Man, I have 3 Atari 400s in various state of MOD ready for an Incognito Lite!!! Just sitting... 48K Gaming every other month... Waiting for an innocuous capability expansion... I probably need to sell two of them at this point...

[@flashjazzcat]

I somehow have two or three 400s as well, one of which was donated for the specific purpose of Incognito Lite development. I doubt I'll actually use those machines without Incognito Lite and an aftermarket keyboard, for sure.

[@DarrenCoull]

Feel for you, I really do... feels like the vendor (Rhymes with Shipwreck) is interested in selling new shiny things and has no interest in supporting the hardware previously released. Of course, the market size for these sort of products is limited, so I suspect this is why, to keep a fresh revenue stream coming in. Not keeping you in the loop with revisions, and the technical reasons why is unforgivable, in my opinion.

[@flashjazzcat]

Thanks for the kind comments.

[@shoutyshouty]

Interesting Lotharek has suspended sales of the Side 3.2 with FJC firmware: SALES OF SIDE 3 is unfortunatelly supsended. in light of recent events and FJC video I AM NOT ABLE TO SUPPLY DEVICE with right FIRMWARE .... Awaits answer from Candle.....

[@flashjazzcat]

That's the first competent thing he's done, albeit two months or five months late depending on how you look at it. Clearly there's a communication issue, since Candle believes Lotharek should test the JED while Lotharek believes it should have already been tested and proven fit for release by Candle. Or by me? But I'd have to be told it existed and sent a copy first, which Lotharek didn't seem to consider worth doing. I can't make it make sense no matter how I try to add it up.

[@mjmr914]

Rough times sounds like they are trying to move away from your firmware

[@flashjazzcat]

I suspected for some months that I was being goaded into a rage-quit. But it actually feels more like boiling frog syndrome at the moment, as one FJC-firmware device is withdrawn from sale, then two days later, another...

[@darthverminates9708]

​@@flashjazzcatmaybe you and candle can find another hardware assembler to work with. I bought a memory board for my atari 800 from reifsnyderb recently and it's clean.

[@buriedbits6027]

59:28 sorry but, you ‘noticed’?! Why weren’t you made aware of this by Lotharek? You have to discover this? Uff 1:01:40 oh my god. Not an ideal situation. Lotharek WTF bro!

[@buriedbits6027]

1:03:41 lotherak comes across as a guy who likes to sail friction free without any regard for the reliance he has on the guy who built the software. My opinion mate but based on what I see with my two eyes that wide open. Come on Lotharek, take responsibility. Start talking with your software developer. Be humble man.

[@jonathanbandelier7901]

After watching Lotharek's very long and emotional response, I gather that he sees his role as being the PCBWay of Poland, leaving the design and test to you and Candle O'Sin. You made it absolutely clear that you did your part, but why is it that you send all the blame to Lotharek, while Candle, who - if I understand correctly - released the botched JED version, only gets kind words from you? Shouldn't he also issue a video to explain his side of the story?

[@flashjazzcat]

Lotharek's assertion - in his response - that he leaves design and testing to me and Candle O'Sin is wholly inconsistent with his actions, though, isn't it? If you watched my video properly, you'll understand that Lotharek commissioned JED 2.3 from Candle after making an executive decision that JED 1.3 was unstable and timing adjustments were required. I also made it clear that Candle O'Sin asserts that: 1. Circumstances prevented her from carrying out anything other than perfunctory first-hand testing of JED 2.3 (something she reiterates to this day). 2. She stipulated to Lotharek that JED 2.3 should be sent to me along with a means of flashing and testing it prior to deployment on production cartridges. However, I was sent nothing and not informed of any changes, so the only way to send blame in any direction other than Lotharek's is if you choose to doubt the veracity of Candle's statements. While it might not have gone amiss if Candle had independently warned me of what was going on at the time, it's rather easy to say this in hindsight. Now that Candle has formally ended cooperation with Lotharek, I very much doubt she'll be interested in pontificating further on the situation, but regardless: that's a matter for her.

[@buriedbits6027]

21:50 I am up to here on your video but if the hardware (the 1MB) is being sold, somewhere by someone then it would be unreasonable for you to contend with those non out of the box configurations. I purchase all my Atari accessories and the like from Lotharek in Poland and never had an issue. I have an 800XL and a side3 and it works great. You need testers. It’s a basic requirement and you need users to test. Read the ISTQB syllabus it will tell you why. If the software developers name is on the device, or not, he must stay in the loop. This product 33:42 is defective. I won’t even get into EU consumer rights law that cover EU consumers for two years. On to other topics, waiting for external 1MB expansion for 800XL I have a lot of respect for the pain. If hardware is being modded the software developer must must must be made aware of it. All changes must be documented. Test before release.

[@flashjazzcat]

Thanks for the balanced remarks. I think I made the illustrious reputation enjoyed by the Candle/Lotharek products absolutely clear, and that what we're observing here isn't calculated negligence, but simply the cumulative effect of cutting corners here and there and making a couple of unilateral YOLO decisions. Anyone can make a mistake in isolation, but the trend of perfunctory testing and the complete disregard of the lead software developer has represented an emerging pattern of behaviour since I first learned of the SRAM changes long after they had been enacted. In the context of consumer rights and what the customer is entitled to expect: the vendor has benefitted from the kind of expectation management I have voluntarily undertaken for the past four years concerning the mitigation of SIDE3 stability issues, such that it was more or less accepted that the host machine might require a 74F08 chip in order for the cartridge to work. But that's already a tenuous situation, and changing things such that the first-aid measures are no longer guaranteed to work means those previously carefully managed expectations are thrown straight out of the window. Test before release, as you say.

[@buriedbits6027]

@@flashjazzcat Absolutely, I completely agree with you. As a software tester, I understand the challenges you faced. Despite not being the best among my peers, I can see that you did everything right under the circumstances. You worked hard for your customer, safeguarding their interests and your reputation, and made this issue public, which is commendable. The big lesson here is that even the smallest change can turn into a defect if not properly tested. Additionally, software and hardware developers must communicate any new requirements to each other in a documented manner. This ensures risks are assessed, and the necessary testing is conducted to prevent defects in user and functional requirements. Including a few external expert users can help identify issues early, allowing for prompt analysis and resolution. I sympathize with the customer whose hardware was affected. You diligently tried to resolve the issue and suspected a faulty chip, replacing it in the process. Unfortunately, this rendered the device non-functional. While this effort unintentionally damaged the product, your decision to cover the cost was incredibly noble. Although it felt somewhat unfair since the customer was not at fault, your commitment to resolving the issue and taking responsibility for the outcome was commendable. I plan to watch your video again to ensure I fully understand the situation. An honest conversation between hardware and software developers is crucial. This incident can provide valuable lessons to prevent similar occurrences in the future. From my perspective, experienced and hardworking developers who build fantastic products often face unexpected and disappointing results. These can cause customer dissatisfaction, have a negative economic impact, and necessitate expensive fixes. Your transparency in handling this situation builds trust, and I can't thank you enough for your candor.

[@flashjazzcat]

@@buriedbits6027 Thank you for that excellent commentary. You clearly watched the video and understood it. The only point I need clarify is that replacement of the GoWin CPLD with a Lattice part was intended to facilitate JED flashing using existing tools (I didn't suspect the GoWin chip itself was faulty - I simply had no way of programming a GoWin CPLD). In any case, I suppose it was unrealistic for the vendor to watch the video and reach the same conclusions you did, but nevetheless I don't think a strong sense of obligation to the customer and to the product and development team needs to be in any way mututally exclusive, and if it is, something is seriously wrong and that's a situation I cannot possibly inhabit long-term.

[@PetrBelohoubek-ot5ok]

That gowin FPGA is proably able to Run whole your computer inside of it... I dont get the point of this "cart" at all, it would me more practical, and proably even cheaper that this "card"... My 25 USD Tang nano 20K can run whole Atari ST with no problem at all (and you can run it on even cheaper 9K version)

[@flashjazzcat]

You'd be surprised. The cartridge emulation logic consumes a large amount of the CPLD resources and there is little room left for anything else. The point of it, anyway, is to allow easy mounting of cartridge images of all conceivable types on a real computer, along with executable loading, hard disk functionality, etc. At the end of the day, there is no point in having any physical hardware whatsoever if you look at it that way, since I can emulate everything on the free-to-download Altirra emulator and experience everything in cycle-exact accuracy on my PC's monitor.

[@OldManBadly]

bee-ta?

[@michaelstoliker971]

The customer isn't a beta tester...you have seen how Microsoft operates haven't you? Sorry about your frustration. I'm almost glad I can't find one of these things in the US.

[@flashjazzcat]

LOL. Well, you make a good point, but I think there's a difference between the hardware and software side of things. At least Microsoft can issue fixes in the form of Windows Updates, and likewise I can release updates to the Loader which end users can apply without any specialised equipment whatsoever. And that's important, since the Loader is such a complex piece of software that I still have not fully implemented every feature I want to include (that's why the current release loader is versioned at 0.69). And I feel a strong obligation to 'finish' it, but how can I further improve the user experience if the hardware threatens to deny the user any experience at all? What we have here is more along the lines of an Intel CPU with an execution bug which can't be patched with an in-field microcode update.

[@michaelstoliker971]

@@flashjazzcat I learned how to use Windows Recovery thanks to Windows Updates...just saying.

[@flashjazzcat]

@@michaelstoliker971 Haha. I managed to mangle the boot loader on a Windows 11 install the other month, but I usually have no problems. I have a dual-boot system, though, and I had somehow messed up the EFI folder.

[@bobbob-xf3yg]

You need a stiff drink

[@flashjazzcat]

LOL. Had a few at the weekend. Helped for a while. :)

[@kranibal]

I am unfortunately a new owner of a SIDE3 with JED 2.3....

[@flashjazzcat]

The question is: does it work for you? I'm not maintaining that JED 2.3 cannot possibly work on any machine, nor that JED 1.3 was immune to issues. If you want to get in touch with me via the contact page of my website, I would very much appreciate any detailed feedback on the behaviour of your JED 2.3 cartridge. :)

[@kranibal]

@@flashjazzcat I will contact you to provide you some of the problems I have encountered so far. Most common one is scrolling down a long list of files making the SIDE3 cartridge to freeze.

[@flashjazzcat]

@@kranibal Received with thanks!

[@user-Atamigaputer]

The cart retailers should be paying you a big bonus for solving their stuff ups. And if lotharek has cracked the shits he should apologise

[@flashjazzcat]

Acting on the information provided would be a start, but let's see what happens.

[@marcins6071]

try fujinet

[@darthverminates9708]

It doesn't emulate cartridges. Love my fujinet but someday I'll probably buy another product that can.

[@marcins6071]

@@darthverminates9708 do you like (s)nesctrl interface? you can plug in snes nes ps1 etc controllers even wireless controllers

[@mynameisleeyesitis]

So i should get the SUB cart then?

[@flashjazzcat]

It's a good opition if you prefer not to trouble yourself with stability issues and don't care much about what the firmware is like or whether the device integrates with U1MB.

[@mynameisleeyesitis]

@@flashjazzcat to be honest im struggling to wrap my head around the advantage of a side3 compared to a subcart. Have a fujinet and 10502pc coming from retrolemon. I was wondering if I buy the side3 from there will it come with the good firmware still. Same goes for the 1mb upgrade. As I understand it the side3 with u1mb should cover anything the subcart can do in theory, but also provides hard disk support (which 1mb plus fujinet does anyway) so still struggling to decide haha. Thanks for the videos by the way you have provided me hours of interesting viewing

[@mynameisleeyesitis]

@@flashjazzcat bearing in mind vbxe and 1mb are definite future purchases after the fujinet and then after the cart (buying too much at once a: will set the wife off and b: wont give me enough chance to play with each iteration on my atari setup)

[@flashjazzcat]

@@mynameisleeyesitis Well, Candle is now offering SIDE 3.3 at PCBWay and foretunately the bill of materials specifies a Lattice CPLD instead of the problematic GoWin chip, so if you buy via PCBWay you'll avoid JED 2.3 entirely and get a SIDE3 which presumably works at least as well as SIDE 3.0/3.1.

[@flashjazzcat]

@@mynameisleeyesitis SIDE3's loader is nicer and the cartridge fully integrates with Ultimate 1MB to provide a DOS-usable, OS-compliant hard disk on the SD card, which AVG/Sub cart completely lack. You can also start the SIDE3 loader with the 'L' key directly from the U1MB setup menu. SIDE3's lack of SIO connectivity should be a non-issue if you have FujiNet anyway (and SIDE3 will handle ATRs when used with U1MB), and you won't know yourself when the hard disk has 64KB/s throughput instead of 12KB/s maximum. :)

[@nasty_niff]

O2 fixer! Wasted my cash there! Total con

[@flashjazzcat]

Its inventor still seems convinced the device had some positive impact on your system, despite the fact I demonstrated on video that O2 Fixer was not needed and accomplished nothing at all. And I think the promotion of and reliance on that device is proving extremely damaging to the credibility of the other products it is supposed to 'fix'.

[@petertheatarian]

@@flashjazzcat ​ From a purely engineering point of view, O2/Fixer improves at least one class of problems, namely the sensitivity of SIDE 3.1 to the clock line capacity. It was tested on about 30-40 Atari computers in mid-2023. The available hardware at that time included SIDE 3.1, JAD 1.3 (black PCB without PLL), firmware 0.47 (you might say it's old, but it was the only one available then). Unfortunately, an inventor cannot predict the future or eliminate future errors. It is possible that if there had been no nonsense talking and no search for other solutions, this small circuit would have become a fixed point regarding capacity, timings, and other parameters, and the author of SIDE 3.x would have adjusted his works to these new constants (which are variable in every Atari due to NMOS technology). The inventor never claimed it was a breakthrough solution saving everything. O2/Fixer consists of two 74F08 circuits implemented as AHCT 1-gate gates. Finally, I would like to add that all computers serviced by the innovator have these circuits, including the machines of top Atari developers in Poland. Lotharek also installs them by default because then we have a reference point for other constructions. In my opinion, there is a financial war in the background between the author of SIDE3 and the producer of SIDE3. As you know, the topic is interesting (for an engineer, something that doesn’t work is interesting), so it will be checked, but it’s probably not a priority because there are other projects. Thanks for mentioning Jon, and you really don't have to keep attacking the only person here who cares that SIDE3 works because everyone else in Poland has long been using AVG. Despite your persistent attitude, I still believe your software is functionally better. Greetings from Warsaw! PS: Let me remind you at the end how it turned out that I got involved with this at all. It wasn't assigned to me. I bought SIDE 3.0 and SIDE 3.1, which didn't work stably on my Atari. Here, the Polish Atari scene is only capable of cleaning the edge connector with IPA and pressing the processor into the socket. I am also a user who bought a product twice, the one "worked on" by Candle and you. Both didn't work stably. SIDE 3.0 doesn't start at all on the Atari XEGS, where the PHI0 signal isn't buffered. The problem isn't the inventor but the fact that the designers don't test their projects on machines other than their own with expansions other than their own. Simply put, everyone here hates other creators. Maybe the issue would be easier to solve if people started talking to each other instead of hating each other. Don't you think, Jon?