Comparing the ARMSID, Nano SwinSID & MOS 6581 Sound Chips for the C64

In this video I'm going to try out my new ARMSID (modern Commodore C64 SID replacement), and compare it to the Nano SwinSID. At the end of the video I'll do some comparisons with those 2 chips and the original MOS 6581
Links:
Nano SwinSID - www.tolaemon.com/nss/
ARMSID - retrocomp.cz/produkt?id=2
Matt Gray - / mattgrayc64
High Voltage SID Collection (HVSC) - www.hvsc.c64.org/
Chapters:
0:00 Intro
2:15 Nano SwinSID Listen
9:34 Chip swap
11:11 ARMSID Listen
22:10 Cybernoid II
24:32 Monty On The Run
26:40 The Last Ninja II
29:56 SIDBurners 7 Intro
32:48 Ghosts 'n Goblins
35:50 Bubble Bobble
37:53 Thats All Folks (Flimbo's Quest)

[@lochmarnegoat9812]

The fact that you can get replacement SID chips is amazing, regardless of minor discrepancies in the quality.

[@DavePoo2]

Yep, I think both of these replacements are excellent. And they both run about 40ºC cooler.

[@turrican4d599]

Minor?

[@danielmantione]

I have reconstructed the source code of the SwinSID and I think I can improve it. There is a few things I still want do with the reconstructed source code, before I start improving it, so don't expect anything on short notice, but I think the SwinSID can still do better than it currently does.

[@DavePoo2]

That is good news. I was under the impression that the SwinSID was open-source? is it not? It would be great if it could be improved. It's not far off being a decent replacement as it is, but unlike in video images, where the human brain can kind of filter out a blip here and there, it can be quite jarring when there are glitches and bleeps in the sound sometime. If you need any help then hit me up, there is a contact email on my channel page.

[@danielmantione]

@@DavePoo2 Swinkels has released source code for the SwinSID SE, which was an early SwinSID. The SwinSID Nano is completely different and the firmware has few similarities. The source code for the SwinSID Nano was never released and the improved "Lazy Jones Fix" firmware by Codekiller that everyone uses was binary only as well. Both Swinkels and Codekiller have disappeared, and the SwinSID website went down. It's been 10 years since the Lazy Jones firmware was released. As your video underlines, the ARMSID is fantastic and I am a happy user of the ARMSID, but the SwinSID still makes a lot of sense and is very popular. So I tought it was a good time to act and check if improvements were possible and now I am able to confirm this. The code is public, can't link but you will find it. I will drop you a notice when I have something ready.

[@GadgetUK164]

This is fantastic news! One thing it does lack is OSC3 and ENV3 read support - not sure if that can be supported or whether the MCU used just doesn't have enough cycles to accommodate?

[@danielmantione]

@@GadgetUK164 The current hardware can't do it and the reason is pretty crazy: The chip select needs hardware interrupt 0, which shares a pin with port D bit 2. The rest of port D is connected to the data bus. Because bit 2 is unavailable, it is connected to port C and you need bit manipulations inside the interrupt handler. This costs so much time that it is impossible to get data on the bus fast enough to support reads. However, now that I have source code, I am no longer restricted to the Atmega88. I have been already doing research and two chips are attractive: The AVR32DB48 and the Chinese LGT8F328P. Both allow the full data bus to be read/written in one cycle and should allow read access. They have timers with input capture capability, which makes it possible to support POTX/POTY without involvement of the CPU. This is future work, I will focus myself first on better firmware for the current hardware.

[@DavePoo2]

github.com/dmantione/swinsid

[@sbmicro1896]

ARMSID is very good !

[@DavePoo2]

Yep, I'm very pleased with it.

[@GadgetUK164]

Excellent video Dave =D I wasn't aware of the Rodland issue - that may relate to the OSC3 and ENV3 (lack of read support). The Last Ninja 2 problem (and maybe others you noticed) I think may relate to CS timing. I had to mod one of mine with a cap to sort, and I think maybe that's what's going on with the C64 board you have there. I will re-test that game in my upcoming SID related video, and give you a mention there too =D

[@DavePoo2]

Thanks, I've seen Daniel's comments about the filter read support. Maybe if I could disassemble the music routine for it, then I could verify if it tries to read it. Do you want those D64's I made or any of the raw footage I recorded from the games (for your video)?

[@dans.8198]

Thanks for the comparison! It would have been even more useful to play the *same* music segments with all three chips instead of switching during playback.

[@timetorelaxfocus9642]

I miss Commodore computers.

[@DavePoo2]

Projects like the nano swinsid and the armsid are keeping them alive.

[@totoonthemoon3593]

Thank you for this great video! The ARMSID is good but still not close enough to the original 6581.

[@DavePoo2]

I think it's pretty darn close

[@Wikcentral]

I have several sid replacements. The Arm2Sid being probably the one I like the most. FPGA sid is really good as well. but a bit more messy with wiring. The backsid is good, but for the money I think the armsid is a better buy. The Arm2sid and an Ultimate64 with the sid tap into an audio mixer makes for a great way to change up your sid combinations.

[@DavePoo2]

What is the BackSID like? the problem with that one is it is out of stock currently. I'm a big fan of my BackBit even though it isn't super compatible with everything.

[@Wikcentral]

@@DavePoo2 I need to try out the last firmware but I do not think anything new relates to why it sounds. It does have a simple gui program to set the filters. It has been a while since I listened to the backsid (I am not even sure which machine I have it in at the moment) but it sounded really good the last time I tried it. Maybe a tad bit thick compared to what I am used to. This is not a bad thing, just a bit different. In a world with more time I would like to put all of my SID options through a bunch of sound testing. Arm2Sid is what is currently in my Ultimate64 which is what I use the most I would say.

[@Svante]

My old MOS 6581, started to fail on me, so I replaced it with the ARMSID, and I'm really happy with it. I was first thinking about buying another MOS, but since they are starting to die now, this was a much better choice. Definitely worth the money.

[@DavePoo2]

I think that's very wise. Even if the 6581 does work, it really runs hot. I've added a heatsink to the one I have.

[@danielmantione]

The only original chips that you can still find are sold by butchers who slaughter C64s for their chips. It is a dubious business that IMO best isn't sponsored, we want as many C64s are possible to be preserved and this business only makes the opposite happen.

[@DavePoo2]

@@danielmantione In the past, I've seen eBay sellers who have entirely working C64's being sold as individual parts.

[@bitoxic]

Dave, you should try using a TapeCart and load all your SIDs in the SD card... there is Sidplay64 TapeCart Edition (ARMSID version) that makes it easy to browser hundreds of SID without changing disks etc. The ARMSID version switches between 6581 and 8580 depending what the SID tune was made for. Actually it would be interesting if you do a review of the TapeCart... its a neat device!

[@DavePoo2]

The BackBit actually had a built in SID player. So you can just go to any .sid tune on the SDCard and play it, however I found that the player likes to crash or just not play some of the tunes from the HVSC properly. The same tunes can play correctly though the Sidplay64 (but unfortunately I had to compile them into a .D64 to get it to work on the BackBit). I've never heard of TapeCart, thanks for the tip.

[@rawberg_se]

Great video again, ARMSID is really really good, to me it sounds better than the original SID sometimes. And i like that you included Ode to C64, I love that tune.

[@DavePoo2]

Thanks, it is a good tune. I added it because Adrian's digital basement plays it a lot.

[@e8root]

SIDs are hissy, clicky and horrible - BUT they handle certain parts in some songs much better than any emulation/simulation. You can feel real SID have oscillators which react with some inertia and each sound is dependent on sounds before whereas all emulated SIDs are at times too precise for their own good. Imho SID emulation will improve and in time we will get all the bugs/features emulated to the point the only tell sign of real SID will be reacting to bus/electrical signals in the computer the chip shouldn't see/hear... which could also be recreated in hardware. Even if that happens I will still like to have real analogue SIDs. For now getting real chips is "gotta get this whole C64..." affair because a) often less expensive and b) if we pay ridiculous prices for SIDs people will buy perfectly fine C64's to make them organ donors. For my U64 I got MOS8580 for cheap (~$25) and for now scooped 6581R3 from my 128D... but already got broken breadbin C64 (which was case/keyboard donor for U64) with broken PLA chip - I will repair the motherboard it and the intention is to put some SID inside. Maybe some SID replacement? I wouldn't put SwinSID really because of all the issues it have but ArmSID might be a good pick... maybe some day, we will see.

[@TRONMAGNUM2099]

Listening to both I think the Armsid does sound better, but I have been pretty happy with my SwinSid. I grew up with a 64C and I think the Swinsid replicates the 8580 pretty closely. If you have the money I'd say get the Armsid but if you are on a budget the SwinSid will probably be fine for 99% of people.

[@DavePoo2]

I agree, I think the SwinSID is pretty cool, and it does seem to replicate the 8580 better (even when it's in 6581 mode).

[@wimwiddershins]

I'm "lucky" in this way to, coming from C64c the humble SwinSid sounds fine for the most part.

[@DavePoo2]

@@wimwiddershins I had the bread bin when I was a kid. I didn't actually know that the C64C was actually any different, I thought they had just updated the case.

[@HifiCentret]

There is a method of playing (relatively) very high quality samples on the SID. I have a by now almost vintage MMC64 cartridge (ground breaking when it was released) with a plugin to play if I remember correctly unsigned WAV files (need to be mono, 8 bit correct samplerate) and it totally trumps all the samples in vintage game. Info can be found in the Individual Computers wiki section. The downside though is it's extremely CPU intensive so it can't do anything else. If I remember correctly it needs to update a register every 20 cpu cycles which makes programming extremely tricky. I wonder if the armSID can handle that ? Had they been able to do that back then it would have been jaw dropping. Of course storage would be the huge issue. A typical song is 3 MB of data. And by the way that way of playing samples have a 15 kHz (or there about) whine to it.

[@DavePoo2]

I haven't had a go at any of the SID tunes that did "sampled sound". But it's worth noting that some of the later C64's with the new SID also wouldn't be able to do that due to changes to the hardware, so if you put the ARMSID in the mode to emulate those chips, then it should behave the same. Pretty amazing that people figured how to play samples on the SID back in the day, but it was proper abuse of the SID, taking advantage of undefined behavior.

[@HifiCentret]

@@DavePoo2 isn't true. First of all the usual volume hack is just lower on the 8580. Second the method I'm talking about isn't the volume hack. It's a PWM style hack where the side plays very high frequency triangle waves which is frequency switched very fast in a PWM style. It plays just as well on the HMOS if not even better than the NMOS. Downside is that it is very CPU intensive. You can research the older MMC64 cartridge. There was a wav plugin for it. Very impressive.

[@C64sidVSsid]

J ai acheté un armsid et je trouve cette alternative fabuleuse. Fan de demoscene, certaines musiques sonnent même mieux que l original. J ai réglé l armsid en 6581 et 8580, avec les réglages appropriés le son est époustouflant. J ai hate de voir la prochaine mise a jour :) par contre une petite augmentation de volume sonore de base serait top ! Sur mon c64ii l avantage avec l armsid est que l on peut modifier en software le digifix meme si on en a un en hardware. J ai rangé mon 8580 dans une petite boite en sécurité :) le problème c est qu avec mon digifix j avais des distorsions sur certains samples. Il est réglé à environ 30 pour cent avec la résistance. L armsid est vraiment génial, merci pour votre video. Just pour le swinsid, il est mieux adapté pour remplacer un 8580, pour un c64c c est viable mais pour un breadbin il vaut mieux investir dans un armsid, vous auriez pu parler des réglages de l armsid, bien calibré c est une tuerie, difficile de différencier l original de la copie. A savoir aussi que j ai eu de nombreux sid 6581 et aucun ne sonne de la même manière. Le meilleur était la révision r4, mais même avec les même révision comme les r3, le son pouvait être différent de l un a l'autre... Pour finir j ai aussi un fpga sid, il est excellent en dual, en mono il faut l utilser en dual sinon le son manque de puissance. Par contre j ai quand même rencontrer des bugs sur des musiques plus important que l armsid. Par exemple morphing de dane craque avec le fpga.

[@greenfoam]

I really like the Arm2sid, the sidkick is good also if you want something that can do a bit more

[@dougjohnson4266]

The only issue with the Nano SwinSID is that it has no paddle or mouse support, but I am still very happy with my SwinSID with the games I play.

[@HifiCentret]

Just ordered the armSID. Can't wait to try it out - especially I'm curious how it handles "high quality" waveform playback. Wether it sounds like a real SID, worse or better - or even fail to do anything which sound in the right direction. I'm collecting parts for a SixtyClone which I as far as possible won't use any original CBM parts - even if they're actually a little cheaper than the modern replacements (Like the MOS8701) or I have them already (as original CBM parts). But one thing I don't really agree on. The price of the SwinSID. Yes cheaper but not by as much as it makes sense to get a SwinSID unless you're on a really tight budget and even then it's easy finding original chips just as cheap with a little patience. Do you have to pay VAT on importing the armSID? where as the SwinSids are made by many so finding a local is just about always an option. Beside unless it's the SwinSID ultimate you loose mouse and paddle support - playing Arkanoid with joystick or keyboard just isn't the same!! And hey no wonder armSID sounds better. Just look at the specs of the chips. The arm core of the armSID by far wins the arm wrestle here.

[@DavePoo2]

I don't remember paying any extra fees to get the ArmSID in the UK, but they did take a while to arrive. The SwinSID yes is not as high a quality product as the ArmSID, but yes it's cheaper, and now the source is available on GitHub, so the possiblity of it being improved is good. The ArmSID is closed source, so it could just dissapear at any point.

[@HifiCentret]

@@DavePoo2 Sorry I tought you was based in the US :D I don't know the custom rules in UK. In EU everything has to get at least VAT on it. In the old days you could be lucky but not anymore. But with the new EU initiative where shipper can pay the VAT up front so you don't have to pay the "fine" for the shipping company to handle the VAT then it's OK - but regarding US shipping to Europe is still quite expensive. But not all sellers use it or are aware of it. Chinese sellers are typically but US' aren't. Fearing the creator of ARMSID stops is IMHO pure hypothetical. Might be closed source, yes. But I'm sure another guy will make firmware for it or his own hardware design. It's not some obscure IC that is used so really anobody who want can just sit down and write their own firmware. I'm pretty sure the actual SID emulation he didn't write completely himself but uses some of the open source algorithms available. The ARMSID with its greather power and precision and more inputs can better do a good job than a chip with lower specs and have support for paddles/mouse too. I don't know about the SwinSID ultimate. I've thought of trying that one - if nothing else for the fancy lightshow. But there's also the BackSID which supports paddles/mouse. Don't know if that uses the same chip as ARMSID. On the downside (for us in Europe) is I believe it's a US product (haven't double checked that). So really no fear if the guy behind ARMSID stops and don't open his code :)

[@donaldblakley6796]

Thanks for the comparison video. I have 2 armsids on the way . However, the swin nano is ok, but in some games like donald duck's playground (favorite when I was a kid) makes everything in the jobs in game not work properly.

[@danielmantione]

I recommend to compare the SwinSID sound to the 8580. The SwinSID simply doesn't have the compute power to accurately reproduce the 6581 filter and the only thing the 6581 jumper on the SwinSID does, is to adjust the filter cutoff setting. The 8580 filter needs much less computer power to reproduce and therefore running the SwinSID in 8580 mode and comparing with the 8580 makes more sense.

[@DavePoo2]

I noticed when doing the comparison that it seemed to be behaving more like the 8580. The messed up Rodland music seems to match the with the way an older version of VICE, which used to do the same thing. I also noticed that when you play Ghouls 'n Ghosts on an 8580, some of the instruments go missing, and the same thing happens on the SwinSID.

[@skeggjoldgunnr3167]

That MOS 6581 just stole the show.

[@DavePoo2]

It steals the show if you can find a working one. They get rarer every day.

[@skeggjoldgunnr3167]

@@DavePoo2 1983 When my C-64 was made. I was in the Army. Excuse me, but computers bleeped and blooped in 1983 IF YOU WERE LUCKY! Wow.

[@DavePoo2]

@@skeggjoldgunnr3167 The "bleep" of the BBC computer still gives me a nostalgia hit. All PC's used to bleep when you turned them on (via the internal speaker), but they don't seem to do that anymore.

[@skeggjoldgunnr3167]

@@DavePoo2 HA! When I do builds for my customers I track down my little bag of internal PC speakers...You may get a post single short beep as usual but I don't think beep codes are a thing any more. There's an entire I2C SPI-ROM diagnostic system in there now just like an automotive canbus / OBD I & II ECU bus. Beep!

[@penguinfromdeep]

There's certain nice grunge in the original Sid chip that neither of the replacements really capture from your sound comparisons. I think Nano Swinsid sounds very clean and lifeless with lack of real oomph. The Armsid is closer but again sounds cleaner than the original sid without the organic feeling. But of course it is also matter of taste and not everybody will really care about those differences, it is good that we have these replacement chips and they sound good for what they are.

[@DavePoo2]

Yeah, the original sid has lots of distortion, which gives it, it's sound. Even Commodores own later chip is a much cleaner sounding device. The nano does sound quite clinical, but when it's in use I don't notice (unless it does something glitchy and wrong). The armsid though is certainly very good, and I think I actually prefer it to the original sid.

[@stephenwhite506]

I wonder how these SID replacements will go with a SuperCPU. The 1Mhz 6510 can, at best, write to the SIDs registers every 3 cycles where as a SuperCPU can hit it on every cycle. These replacements may not like that. I'm creating a SuperCPU clone so I will have to get some of these replacements and see how they handle it. Cool video, keep up the good work!

[@DavePoo2]

I suspect that they can't handle the higher clock of the CPU at all.

[@danielmantione]

The SwinSID generates 1 sample in 24 6510 cycles. You can do multiple writes to the SID per sample, but this does mean that after writing to the SwinSID, it may take 24 6510 cycles before your write has an effect. The ARMSID has a faster CPU, so may generate more samples, but even the ARMSID isn't fast enough (not by far) to generate one sample per 6510 Hz.

[@AbAb-th5qe]

What about the FPGASID? What do you think of that?

[@DavePoo2]

I don't have one.

[@evansdm2008]

Swinsid had a pitch difference. Tiny amount higher frequency. It’s also doing something a bit wrong with ADSR… but still good effort

[@DavePoo2]

Yeah seems so. It could just be that the SwinSID is actually just emulating the 8580 chip, which might have similar properties. Either way, if you listen to the SwinSID on it's own, the pitch isn't enough to be noticeable.

[@danielmantione]

@@DavePoo2 The SwinSID emulates a SID running on a perfect 1.000 MHz clock frequency, completely ignoring wether it runs in a PAL or NTSC system. Indeed, the difference in pitch is small enough that it won't be noticeable, but combining the SwinSID with a real SID in a dual SID stereo setup wouldn't work very well.

[@DavePoo2]

@@danielmantione Ah, that explains it so it's about 1.5% out. Initially I suspected that my real 6581 might be out of spec, but I noticed the noticed the ARMSID seems to do ok and was in tune with the real thing.

[@C64sidVSsid]

If you want keep vic more cold you must put a new thermal past, it s conductor

[@DavePoo2]

I've already done that to mine, and I've kept the metal can that doubles as a heat sink. I also have a heat sink on my real SID

[@elmariachi5133]

I think it's less the perception than more the evalution, what is making you think of the SwinSID Nano better than other. An objective rating system just wouldn't evaluate having audible errors in most songs as 'really, really good' ;) ('Really good would be about 90% of songs playing perfectly, because from the overall choice of existing english words that describe a rating, there are just about 10% better, than "Really good".). Also there's the issue in regards od the definition of 'sounding good;, because many people, like me, just want the emulation to sound as close to the original as possible. So there's no 'better than a SID', that's just not possible. It's not about the perfect clearness, volume, base and whatever, it's about not noticing the difference compared t a real SID, when playing a game (maybe even without knowing there's a fake SID inside). And of course some strange sound issues that obviously don't stem from the way a real SID generated a sound, but only can be emulation aritifacts, destroys the illusion. Of course the Nano has improved much since it's eralier days, eapecially with the Lazy Jones fix, which made most of the sounds at least kind of simliar, opposed to the painful screeching sounds it generated before. So I can see it as an emergency solution, but not sufficient for playing and enjoying the games undisturbed. The ArmSID for sure is the best overall solution currently, considering it's 'just good enough' sound emulation, feature completeness and price. If I had enough money, I would still prefer the FPGA SID, because there still is a clearly noteable advantage vs. the ArmSID there. But IMHO the SwinSID Nano has the potential of cathing up to the ArmSID so far, that it could ne the better price/performance solution with an upcoming firmware update.

[@DavePoo2]

I don't think the SwinSID is better than the others, I think it's really good, for what it is. Now the source has been reconstructed, it's also open source now, github.com/dmantione/swinsid - so that goes a long way to it being improved further down the line. It's also very cheap compared to the others, so it's not trying to be the best, it's trying to be the best in it's price range. The real SID is always going to be the best, but ultimately, it's not in production so it's numbers will dwindle away to nothing. The ARMSID is better than the nano but more expensive and not open source. So I stand by my words, that the SwinSID is really good (but it could be better, and maybe it will be)

[@JestersDeadUK]

Wow, the Nano really struggled with LN2

[@DavePoo2]

It did. I checked a video showing LN2 on the 8580, I thought that might do the same, but the real 8580 seems to be able to handle it.