Let's upgrade an early Socket 7 board beyond the limit (Part 4): K6-2 test and VRM announcement

[@tiagosolano9728]

Viewer from Brazil here. I teach electronics and a group of my students that are very interested in retro computing will try and build your project on proof boards. If you're interested I can share the results and benchmarks after we're done.

[@hgbugalou]

I love this idea of young people being excited about retro computing. You can learn so much going back to the roots of the tech and that knowledge can still be applied to modern systems.

[@necro_ware]

Sounds great! :)

[@MidnightHunters]

Brazil viewer here too ❤

[@juniorbcm5375]

@@MidnightHunters One more over here :)

[@Alex.Adametz]

Your VRM works better than many factory VRMs of those times! 🙂

[@p_mouse8676]

About the change in ripple; I don't know exactly this DC-DC converter, but many of them go into some kind of energy save mode (lower frequency) or even hickup mode for light loads. This also looks like showing on the little scope you have. I am one of those electronics engineers, doing this during breakfast, hahaha. I will have a look at your design to see if there are some improvements possible. Using a shielded inductor is already one of them btw. Really sad to hear that people already ripping of your idea. Seems to be very common these days unfortunately. It's so lame. I often build in some kind of "mistake" in the design I share on the web. People who want to use my design, have to contact me first.

[@matusekpetr7806]

Just a tip, regarding space occupied by capacitors - you can put silkscrenn on both sides of PCB and populate capacitors from opposite side in case they are facing towards CPU cooler. 🙂

[@simontay4851]

Yeah, exactly. I was going to post a comment saying that but you already did.

[@absalomdraconis]

Good strategy there.

[@drPeidos]

License is fair. Great job.

[@--fishiiki-]

Seeing the board get repaired would be cool! Absolutely love your work and all the love you put into. You're doing great!

[@magnum333]

Amazing work! Bravo! You should prosecute those who infringe the GPL. Contact the FSF or the EFF, they will give you legal advice. Not that I know how that goes about but I've heard they help with such cases.

[@matthewday7565]

If the capacitors are on plated through hole vias, maybe they could be mounted on the reverse if front clearance is a problem. Also, you may want to add extra vias or wire throughs if using tracks both sides to lower resistance

[@BrassicGamer]

Came here to say this.

[@cburgess5294]

Always happy to see a new video drop! Great content as usual!

[@ReinaldoGonzalezreix2x]

that 0.4 board is just beautiful!

[@xXBeefyDjXx]

It is awesome to see this project blossom. It is a very interesting design and your performance is great. I look forward to maybe building one someday.

[@Choralone422]

Excellent work! So glad to see projects like this work successfully!

[@precisionxt]

You deserve the credit and the fact you are sharing this free means more people can contribute and help improve the design. I hope this new licensing helps keep your contributions from being stripped out and sold under some other branding. I will be buying an RTC module from your store soon for my Compaq 386. Thank you for all your hard work!

[@necro_ware]

Thank you for your kind words. Unfortunately, there is one problem with your plan, it's just that I have no store :)

[@precisionxt]

@@necro_ware i was just about to come back here and edit my comment after I checked, I was mistaken. Regardless, thank you for your hard work on this.

[@lorenzo.c]

Impressive work! Extra points for making your own toroidal inductors 😁 In case you are interested, I'll leave a note on the relationship between characteristics and dimensions of electrolytic capacitors. For a given series, the ESR is associated with the dimensions that is to say all capacitors with a certain Length and Diameter will have the same ESR. The ESR decreases with increasing volume. The product of Rated Voltage x Capacitance is also (roughly) proportional to the volume so, for example, in a given L x D shape, if VoltageA = 0.5*VoltageB then CapacitanceA = 2*CapacitanceB (roughly). Different series differs for endurance and max working temperature. It doesn't seem to me that, in your case, you would benefit much from pushing either so you may instead prioritise lower ESR or higher capacitance in the given volume but it's your call. You may observe higher ripple when the CPU goes into that "power saving" mode because the buck converter goes from Continuous Current Mode (CCM) to Discontinuous Current Mode (DCM, the current in the inductor goes to 0 in each cycle). The ripple voltage is typically higher in DCM. The current threshold between the two modes is 1/2 the inductor (peak-to-peak) ripple current which, in turn, is lower for higher inductance. One of the benefits of higher inductance is that the buck regulator operates in CCM even at lower output current. As you found out, you had to trade nominal inductance for size & resistance. Larger capacitors would also lower the output voltage ripple but, again, you are constrained by the physical envelope of the design. A way to lower the noise while staying in the same physical envelope is increasing the switching frequency (then, with the same output current, the inductor current ripple would be lower). This would require reviewing the choice of inductor and MOSFETs. Which MOSFETs have you used? Can you please add the voltage rating of the capacitors? Again, great job!

[@necro_ware]

Thank you very much. Yea those are constraints, which I was messing with. The volume proportions are which I was trying to aim, due to the sizes, that didn't work completely due to limited size :) In regards of CCM/DCM that's very interesting information, thank you very much! I'll have to evaluate that in depth for better understanding. As for higher switching frequency, I considered first faster controllers, but most if them are modern parts, which are harder to solder and they are more expensive. I tried to keep that cheap and easy to solder, so everyone can build it. I'm currently AFK, so can't tell which transistor model it was, but there's a BOM on the project site. Thank you for the info once again!

[@lorenzo.c]

@@necro_ware I agree with the emphasis you place on the design being feasible for as many people as possible and it's tricky to find a good solution (Renesas ISL6545A, the 600kHz version, is now obsolete 😭). Seen the BOM: I hadn't understood that one column is LCSC part numbers. My fault! As others have written, you could use both faces of the PCB for components, for example SMD on 'top', (some of) TH on 'bottom'. This can help with the layout and make it tighter. Can anyone point me to the pinout of the VRM socket (or 'header')? PS: I don't have KiCad 😓

[@necro_ware]

@@lorenzo.c there is a link to the specification on the project site, right in the end

[@blubaustin1]

Could always lay the capacitors on their side if they are to tall. Also am wondering if polymer caps would be better?

[@brzeczyszczykiewiczm]

I really waited for it. Before you presented your first video about that VRM extension, I knew what it is for, but I didn't know how to make reaplacement or what is the pinout. But since first part when you explained so many things, I was watching any progress in that case. I hope one day I can repay you. The sad is there are people who don't appreciate someone's work and do so stupid thing like to erase any sings of your work and they claim it as their own. I really don't understand it :(.

[@comedyflu]

Awesome video ❤ I thoroughly enjoyed 11/10, will watch again!

[@projectartichoke]

Very cool. It may be possible to make the board work in either orientation by putting a second connector at the top of the board. Understandably there are wide high-current traces, but they could be made more narrow by soldering solid wire along the critical ones permitting the addition of a second connector. Alternatively, you could place the tall capacitors on a second small board and have it plug into either side of the main board with polarity-keyed plugs.

[@jantestowy123]

Very cool work :)

[@georgeaiaskaridis5812]

Wow! That VRM module is very nice! Great work! About that other mainboard.... angled pins should do the trick and be more flexible selecting capacitors. Looking forward for more on this topic and K5 CPUs 😀

[@SianaGearz]

The ripple tuning is actually spectacular, great success. The routing is kinda ridiculously overbuilt for the 15A target though, like, you don't even have to worry or really spec through hole inductors, just via stitch the area and you'll be good i think. Not that i mind handwound at all.

[@rkurbatov]

Such a cliffhanger. Eagerly waiting for the next episode.

[@minombredepila1580]

It is a shame that people take the credit from you (by removing your name). Thank you so much for sharing. Will make my own as long as I could get a board that supports it !!!

[@julianbarron5293]

id love to see board comparison tests!

[@TheRasteri]

awesome project :)

[@Yrouel86]

I've been binge watching your videos these past few days and it has been really enjoyable seeing all those motherboard and other cards being repaired and restored. Since I noticed you use one of those IDE to CF card adaptors, have you considered connecting pin 20 of the IDE connector to 5V to power directly such adapters? I did it on 2 of my motherboard and makes those adapters even more convenient so I was just wondering if you thought about it as well

[@necro_ware]

Hi, glad you like it! :) Yes, if you look closely on one of my IDE controllers, you will see, that I use it with CF card and without external power.

[@Yrouel86]

@@necro_ware Oh yeah right. In my case I did it directly on the motherboards, bodge wire underneath. It's not that bad and it's worth not having the extra power cable running to the adapter while using it

[@phillycheesetake]

17:31 Naaaawwwww what's in the baaaaawwwwx? What's in the box?! Thanks for the update, looking forward to how this project develops!

[@necro_ware]

LOL

[@vswitchzero]

Great video and project as always! I love the enhancements you've made and its pretty amazing that it doesn't even get warm at this point! I'd love to try this on my 430FX Full Yes 82430 board, which has the proper VRM socket... and I'm very curious whats in that box now! 😄

[@Thomsonicus]

Love it!

[@tomaszkoszela8433]

Thank You so much for shearing your knowledge. Unfortunatelly I destroy 1 mainboard trying to do it my selve. Withaut exact knowledga how this look in details i made it mirrored :-). now i wonder if i didnt burn also CPU. Is there any way to check if the CPU is burned ?

[@necro_ware]

Thank you! I'm sorry to hear, that your experiment went wrong :( It's a bit hard to predict, what would happen, if you turn around the VRM, but I'm afraid, that it instantly will kill everything on the board, including the expansion cards. As far as I see, the +12V would be connected with the ground on the VRM and sent back to +5V, +3.3V, VCORE and VIO.

[@3dfxvoodoocards6]

Interesting video, like!

[@adrian_sp6def]

You can also solder capacitors and inductor on second side of board in case of cooler space conflict

[@andrewhamop6665]

Fantastic!!

[@Inject0r]

@Necroware could you also add SMD capacitor pads on top of those through hole pads, just so anyone’s able to use either one of those? Whoah 🤯 this project is so damn awesome! I currently have an Intel Advanced/EV Socket 5 motherboard with the same header. I really want to be able to use a 166MHz Pentium, as I ran that on my board back then. My board back then was a Socket 7 board with some extra jumpers for setting the multiplier higher than 2x, but I should be able to run it at 166 anyways! :) Thanks for the great content and projects. You’re doing awesome work, and you’ve really created a big impact in the retro community. Much appreciated!

[@Inject0r]

Necroware for president!

[@n.shiina8798]

i second this SMD caps idea. they're thin so there's no more space constraint between the module with CPU cooler or other parts on the motherboard. on top of being thin, now we're able to use both PCB plane to be populated for the caps if necessary

[@therealjammit]

The two tall input capacitors can be installed from the opposite side of the board.

[@jb2590]

Great video, lets try that VIA chipset board

[@travismoore7849]

Great job on getting the k6 running. Can you use small super capacitors to replace the big capacitors at 3.5v and 0.47F ? Or the 5.5v coin super caps at 0.22F. Just mainly for smoothing out power for DC?

[@AG-jj3lx]

Really appreciate your passion and sharing your adventures!

[@SudosFTW]

As someone battling a removable VRM that gets hot enough to cook a steak, I'd love to see this sort of project come up for the motherboard found in the Dell Dimension 8100, OptiPlex GX400, and Precision 330. It's all the same board, some are missing onboard audio etc. but they're still all the same platform, which is Socket 423. The reason I'm battling it is heat generation with higher-wattage CPUs... and given I'm about to finally do an upgrade on it to a much faster Pentium 4 with a Socket 478 adapter, I'd love to be able to make the heatsink not run at a million degrees. but I just don't have the skills needed to make it happen... PCB layout and EE at that level is just too much for me.

[@TradieTrev]

Used to love to overclock these suckers back in the day! You've gone full tilt modding this one!

[@Constantin314]

what you do is so awesome! totally agree with you on the license modification. some people are just disgusting. why don't you make them yourself and then sell them? for sure i would by it from you rather than an unknown person. make a store with prestashop since it's free and post them there for selling, have one or two models preped for an order and that's it, you don't have to go industrial level :)

[@rkirilkin]

Great idea. I try to do this mod!

[@AiOinc1]

Haha, I needed one of these VRMs really bad about 2 years ago. Wish they had the notch in them for the little clips to grab them.

[@necro_ware]

They do have in revision 0.4 as I explained in the video. But you don't really need them.

[@awd42]

I think your proposed voltage selection revisions make sense, with one exception: wasn't there a K6 which ran at 3.2V? I'm not aware of any CPUs ever using 3.1V, maybe that one could be dropped instead? I've never seen any of the mobile CPUs in a desktop board, but I guess supporting them makes sense. Here's an idea: make the mobile (1.x V) voltages an optional section, with the appropriate resistors and a jumper to select "mobile" mode, which would substitute some voltages. In "standard" mode, you'd have the common 4-bit binary setup of 2.0 - 3.5V seen on many Socket 7 boards. Maybe that's too complex, just a thought...

[@necro_ware]

Yeah, that missing 3.2V is the result of resistor network distribution. I'm trying to get equally distributed voltage over the range, this is still work in progress, but K6/233 runs stable at 3.1V if the ripple is low. I even was able to use it at 2.9V flawlessly. So, I guess, if 3.2V setting is missing in the end, it wouldn't be critical.

[@Stjaernljus]

a hacky solution to the tall caps problem would be adding an option to mount them on the back of the pcb either standing up or laying down.

[@DevilbyMoonlight]

This is top work! -

[@Kedvespatikus]

I can visualize Necro's 20 years later project: running a Ryzen 7950X3D in an Intel Z370 mainboard. :)

[@memadmax69]

I wonder if it would be possible to connect a secondary power supply thru that VRM card to more directly power the CPU?

[@darthtripedacus1]

I have no use for this but I thank you for your additions to the community. I'm saddened to hear of.people stealing your board design and removing your name from it. That's low and dirty.

[@B24Fox]

For better ESR on electrolytic caps, you should use caps with higher voltage ratings. It's very common, for example, to see 16v rated caps on 3.3v lines, just to improve ESR on the cheap. (and higher voltage spec, makes the caps thicker, not taller; so it would also help with the space clearance problem)

[@pipschannel1222]

Very cool project 👌 I would love to try one out on one of my P54 socket 7 mainboards but I'm not sure if it will work on them..

[@necro_ware]

You can compare the pinout of the vrm connector before you try.

[@pipschannel1222]

@@necro_ware I believe the pinout matches on one of them but there's little to no info available on that board, an early Gemlight GMB-P54SPV rev. 0 so even setting its jumpers may be a bit of a challenge.. It has a 100MHz P54C on it.. I guess I'll have to leave this one alone as I don't want to blow it up experimenting without any technical info ;-) The other one is a later model version 4 Biostar MB-8500TAC-A which doesn't have the VREG-board-header but 2 separate through hole regulators soldered directly on the board. Later versions had 3 regulators for lower voltages and supported MMX CPUs.. I may be able to add another regulator to match the version 6 config of this board, hopefully without blowing things up 🙂

[@crashmatrix]

Since you're asking, I wanted to run the S7-VRM on a IBM PC 350 board (6587-9BT). It comes with an unpopulated VRM header with the usual layout. I was looking for a Semtech MP55C-3.3-2.8 VRM on ebay-like sites, but got nowhere. Then I found your board, and figured I might as well order a few off pcbway including assembly. Then I got it in my head that I could probably do it for cheaper if I was willing to hack the board. So that's the current plan, I've ordered a XL4016 based DC-DC converter which I plan to set-and-forget at 2.8V for the MMX conversion of the board. The 12 V power rail on the supply is barely in spec to supply the max 17 watts I expect to see. I did mess up, because I assumed I could feed it off the 5V rail. Oh well. I'll update the comment and let you know when I've blown up the board, the CPU, or anything else when it inevitably goes wrong somewhere because I know I'm not that comfortable with making and hacking electronics yet.

[@necro_ware]

For a Pentium MMX that might work, but for a K6-2 that would be too weak. XL4016 is specified for up to 8A, but k6-2 can pull almost twice as much in the worst case. Furthermore Intel VRM specs are defined for 5V as main current supply, but xl4016 needs 8V or more. There is 12V pin, but that one is only for low current application. It will burn the board if you try to get the requited power from there. You can add a molex connector though to get the 12V.

[@jonatas464]

Yeaaaaaah! Here we go!

[@simontay4851]

What switchmode controller IC are you using and is there a DIP version. Some people might want to redesign/built it on prototyping/stripboard using through hole resistors, capacitors and MOSFETs they already have in stock if they can't afford a PCB. You did say you had difficulty sourcing some of the SMD resistors and caps.

[@necro_ware]

Finding that parts as TH is even harder today. Also, a lot of work went into the layout, probably every change will influence the ripple behavior. But if s.o. wants to do that, the exact parts are all mentioned in the project.

[@K10driver]

Wow! You want to know what i think? Insane! Great! Overwhelming! I need to talk with you about my clockgenerator Problem ❤ And yes, i want to see a Benchmarathon! With all the chips you have and i can send to you (if you want :) )

[@xenoxaos1]

Adding an inductor between the input capacitors could lower ripple if you're seeing ripple on the input.

[@necro_ware]

That's true, an additional LC filter would be a good option. But so far I'm trying to avoid additional inductor, I would like to reduce the size of the module instead.

[@simontay4851]

You don't need to reduce it. I would rather have a properly designed board that works well with low ripple than a board built down to a size/price. If additional LC filters on the input and output require a taller board, so be it.

[@necro_ware]

@@simontay4851 of course, if it needs that, I'll do.

[@horusfalcon]

It kinda bites that people took advantage of your generosity. The GPL has provisions in it that require any contributions to a GPL'd project to be contributed back to it, and to not remove the GPL licensing from the project, but the Creative Commons license is a good one, and I really hope you have better luck with it.

[@nickwallette6201]

Some thoughts: 1) Capacitor height -- Lots of ways to solve this. As you mentioned, parallel smaller caps, but also you might think about laying them down like axial caps, and if they're fat enough, maybe even route out a section of the board so part of the cap body can go through a cutout in the PCB. Obviously you'll want to place the body cutout somewhere that isn't a power plane (like the edge), which might be challenging. You said you had some trouble with decreasing the input filter capacity, which was one thing I was going to recommend trying -- generally, you don't need a ton of capacity on the input side, since it's already regulated upstream at the PSU. Just enough to handle peak switching current. That will of course depend on how well filtered the motherboard's power planes are. 2) Inductors -- Don't shy away from SMD parts due to concerns over current capacity. The data sheet tells you what it can handle. Provided you have minimal resistance / inductance on the traces to and from the inductor, you don't need to second-guess what the pads are or are not capable of. The part vendor's engineers have taken care of that for you. :-) Trust in them. 3) Layout -- You might get better performance by optimizing your layout. Keep the switching parts as close as you possibly can. The signal from the controller to the transistors, and the in and out traces of the Q, L, and C components. Fat traces or planes are good, but really try to minimize the total loop from the in cap to the out cap as well. That path should drive your layout, with everything tucked in around the main loop, wherever it will fit. 4) Licensing -- This is obviously a personal decision, but I just want to encourage you to consider your mindset on this. Anything of value you put out into the world is going to get taken. But, you can choose how you feel about that. IMO, if I publish something, more power to you if you want to run with it and sell it, or whatever. I see that as a net win for the community, because I'm going to do a terrible job at making them available as a product, so if someone else wants to handle that, you have my whole-hearted blessing. I don't particularly care about credit -- nobody knows who I am, or would care, anyway, so why fret about it just to feed my ego? You made a reasonable point about wanting feedback to find its way to you, but IMO, you'll probably never get as much feedback from a product in the wild as you do from these videos. Chances are, if it looks like a complete solution, it will be consumed as one, and very few people will give it any consideration beyond that. Here, you have an enthusiastic community that is invested in the saga, so you're going to get lots of armchair hobbyist engineers throwing in their 2 cents. (guilty!) Once it can be downloaded or bought, it'll just be a black box what takes one thing and produces another, and you'll probably never hear from an end-user that wants to help you make it better. (And if you did, you'll probably have moved on to your next project anyway.) Not trying to tell you how to handle your own intellectual property, of course, just gentle encouragement to contemplate a more "que sera, sera" attitude toward idea theft, since it's largely inevitable, and you hadn't much intent to profit off of it, so why create stress about it? If you _see_ it as a good thing, then it _is_ a good thing.

[@necro_ware]

Thanks Nick! Thanks a lot for the detailed comment, I appreciate it a lot. Here are my 2 cents ;) 1) This doesn't seem to be a big problem. Most mainboards are designed in a way, that the module looks away from the CPU, so caps have enough space. Others have usually enough space for 16mm caps as far as I can see. It's a rare case, where caps wouldn't fit and then they can be soldered from the opposite side. 2) Of course I trust the specs, but I want to be safe with my PCB design as well :) My initial design was with SMD, but I had problems to obtain what I needed. Either they were out of stock, or too big, or too expensive, or too weak. Eventually I decided, that it is much easier to make my own coil and call it a day. 3) I invested many hours into the layout. It was checked by two of my friends, which are electronics enginers. Every trace is calculated and everything is layed out strinctly as suggested by the datasheets. Distance to capacitors, distance to the coil, distance between the parts, which can influence each other, traces and planes widths etc. Since I'm not an electronics engineer and miss the required experience, I decided to do everything strictly by the specification :) 4) Yeah, personal thing of course. Eventually, I have to release it under some license and wich one is basically a free choice. I want, that people are able to find the origin, schematics, additional information and whatever they want to know about the project. This project has no firmware or software, but others do and may be someone wants to make an upgrade, or start to experiment and contribute, who knows? I don't do my projects to be black boxes, I'm a friend of "by community for community" approach, otherwise I can stop all this instantly. Once again, thanks a lot! I always make notes and try to keep all the suggestions and opinions in mind, when developing new projects or improving the old ones. Not only technical questions, but even such exciting thoughts about the license. This is the first time, that I'm trying the CC-BY-SA, currently I'm not quite sure what to think about it. Time will tell ;)

[@nickwallette6201]

@@necro_ware Yeah man! I appreciate everyone who contributes to the options we have to keep old machines alive. I feel bad sometimes, because I've seen a lot of product developers get bent out of shape about clones, and at the end of the day, it doesn't seem like anyone has ever found a solution that prevents it. Particularly not the terms of a contract. ;-) I often feel like so much energy is wasted worrying about "what if it gets cloned?" (it will), and "what can I do to stop it?" (contain the wind and rivers maybe?) that would be much better spent remembering what's most important. Family, friends, good health. Everything else? Small potatoes. Hopefully, the community, particularly fellow KZbinrs, help direct people to the origin of such projects, so everyone has the opportunity to benefit from the resources you (and others) have carefully designed and curated. And for those that just buy from the cheapest source, regardless of authenticity? Well, good luck to them. Anyway, whatever you do, cheers. You put a lot of effort into this, and it definitely shows. :-)

[@glitchwrks]

Very cool project! I was wondering last week if anyone had made a modern VRM for Socket 7!

[@tigheklory]

Why not use two axial caps for the input voltage? You can have the same capacity but they lay lower.

[@DxDeksor]

Awesome ! I loved this series. I don't have many vrm capable boards anymore, but I may still try to build a few for myself or other people. Would you like a K6-2+ 570 for your benchmarks (unless this already is the surprise you're talking about :D ) ? This reminds me when I was using a Shuttle HOT-557 as a retro PC and I wanted to upgrade it to a K6-2 450. When I set the CPU to the proper speed, it crashed while booting windows 98. After fiddling with the jumpers I noticed that running it at 300MHz was working. I didn't understand why immediately, until I tried to mess with the jumpers more and accidentally touched the MOSFET's cooler ... It burned my finger. I took a thermometer and even though I'm not sure my measurement was 100% accurate, I saw that it was running at at least 80°C. Definitely too much. I noticed also that even with a normal pentium that MOSFET ran really really hot (burning hot). So I screwed a fan in that MOSFET's cooler and then I tested the K6-2 450 again and it ran without any issues. Later I read the motherboard's manual and saw it wasn't designed for CPUs above 200MHz😅 This board has a spot for a VRM though, so I guess it could benefit from your project :D By the way you said the link was in the description, but it's not :( Probably a mistake😄

[@necro_ware]

Oh my... of course I forgot the most important thing, the link :) Thank you! Fixed!

[@ArtemAleksashkin]

Ааа еще одна серия!!! Класс!

[@adrian_sp6def]

Do You consider that Your ground clip from oscilloscope proble are long and form a one loop coil that could pick up rf nocie that cpu produce ?

[@necro_ware]

Yes, I tried also with short pin to the side of the tip as well. There was no difference.

[@FrancisFjordCupola]

I might be mistaken; but I think CPU's around that era used to run full-steam ahead all the time. There was some utility on Win95/98 to put the CPU into idle mode. Then CPU's (that started to consume more, so the increased consumption 100% became more of an issue) started to incorporate idle modes.

[@SchkuenteQoostewin]

I would like to have one of these crafted for me. Physically incapable of building this myself with only one working hand. So who can I buy one of these through for my M520 PC Chips board? (yes it has the proper slot)

[@andycristea]

Nice work! I wonder if the VRMs for the Pentium Pro are also standard across manufacturers...

[@desolder75]

The Pentium Pro VRMs are also standardized. I've built a simple adapter to use them in my old Socket 7 board.

[@andycristea]

@@desolder75 Thanks!

[@ArtemAleksashkin]

15:36 Тут можно легко исправить - отпаять высокие кондеи и на другую сторону их кинуть сохранив полярность - будет работать.

[@necro_ware]

Это точно, но тот второй с кондёрами пониже тоже нормально тянет, я там фильтры получше впаял. Так что должно сработать. Увидим потом, если тот борд вообще жив :)

[@OuTannu]

cool project.... and yes love to see the VIA chipset board running something else than the "official" supported CPU's 🙂

[@Stefan_Payne]

Have you tried the VRM with either polymer capacitors or a mix of Polymer and normal Electrolytics? Or some Panasonic FM? Reichelt migh thave some, as well as some Kemet A750...

[@Einimas]

You can also just solder the normal caps on the othe side of the board if there are space constraints.

[@googleevil]

Please more AMD K5. It was my first CPU I saw in real and tried to overclock. It runs great on 110Mhz and probably above.

[@Stefan_Payne]

maybe add some SMD Solder Pads for the Filter Caps?

[@rhuwyn]

Is the pinout for VRMs largely standard? Xurious if it would work on boards with Pentium Pros or Dual slot1 boards that need vrms.

[@necro_ware]

It is standard for the mainboards, which support Pentium MMX and compatible CPUs with dual voltage.

[@rickardandersson5840]

So sad to here that someone else would take credit for all your work!!👿 They should be a shame for themselves!

[@horusfalcon]

For a historical take on this, see Sir Rudyard Kipling's poem "A General Summary". It is the most succinct and yet humorous explanation of a human activity that is no joke.

[@2dfx]

Again, a taller board would allow the use of through-hole components for those SMD passives!

[@Jackpkmn]

Excitement!

[@SnipE_mS]

Sorry if this was explained elsewhere but being a noob electronics tinkerer I’m very interested in the tools you use specifically the component tester and the oscilloscope. Do you have those listed anywhere?

[@T3hBeowulf]

I have a 437FX mainboard that is similar to that FIC board you mentioned with only a linear regulator. During some initial inventory and status testing, I confirmed that regulator gets crazy hot. The board was fitted with a P75, but due to a jumper misconfiguration (i.e. some were missing), the system booted that chip at 133MHz and I burned my fingers brushing against the heatsink fins. Fixing the jumpers helped some but adding a dedicated 30mm fan to the regulator "fixed" the burn concern issue. I absolutely love the thought, engineering and testing that went into this VRM. Great project!

[@NuffMan_]

Switching psu is cool, but is it quiet? Like if you would use cheap soundcard, will you hear the psu in headphones? Also be sure it's not produsing any outside noise, like noise noise that one with good hearing can hear. this may sound really anal but im one of those people that can hear every modern device whining away, i cant even use normal phone chargers because i can hear them whine annoingly lmao. Weirdly never heard any whining from computers, unless you count the coilwhining

[@necro_ware]

Well, it's not more noisy, than any other hardware which uses switching regulator. It is definitely worse, than a linear regulator, but you can't use a linear regulator for any desktop CPU, which was made in the last ~25 years. They simply pull too much current.

[@kmnl926]

Mount input caps on other side of PCB if they block CPU.

[@felixokeefe]

You could save a little space on the pcb if you used the molex 90152-2130 connector

[@fhunter1test]

Dos and low power could be 'dosidle' driver.

[@jerry1333]

Test more mainboards! MORE!

[@labibleatarilesitedesatari6724]

You are a great hacker!

[@L0wcash]

There's a protective sticker on your scope's screen.. remove it for a picture with much better sharpness and contrast... Love your video's!

[@necro_ware]

I added that sticker, because the screen is glaring like crazy and blinding the cam. I played around with light, but decided to put that sticker on screen. It's slightly blurry, but therefore I don't need to search the angle all the time :)

[@L0wcash]

@@necro_ware Whatever works for you is fine with me.. The screen is readable enough on video.. Thank you for the great content!

[@clintthompson4100]

Hey Necroware I really am loving this project. I am not meaning to be a pain I was wondering if you have any new Info for supporting the Cyrix MII CPU. It does use more wattage than the K6 CPU. Not sure if you saw my latest comment on your last video on the VRM. Granted I know I could use the K6-2 but to me I want to have something different for once since I just have to be stubborn and hard headed. Not trying to be mean in anyway about this project. I am just getting anxious for the final release and if I can use that CPU. I hope you have a good one and again I am in no way trying to be a jerk.

[@necro_ware]

Hi Clint, strange, I though I answered your question back then. MII consumes more power compared to K6-2/400, but it needs less current at higher voltage, so that should run with this VRM. I have no MII, so I can't tell definitely though.

[@clintthompson4100]

Only partly. Another person was trying to give more of an answer but yea you mentioned you don't have a Cyrix MII CPU but I hope if its possible for you to get one of the MII 333 or 300 to use a test out. Have a good one and I look forward to your next revision when your parts fully come in. Thank you for this project.

[@Dutch-linux]

Why don't you place the large caps on the otherside of the pcp ? that would give you all the clearance you need !! ?

[@necro_ware]

Yeah, one way to workaround it on the boards, where it is needed.

[@doc_emmettbrown]

Hello, what should I do with an intel advanced \ml motherboard?

[@shadowwarrior3521]

Hello, my VRM sometimes doesn't start from the first time but from the second or third try. Tried to remove C8 but it is still happens time to time. I can't determine the pattern of such issues. Sometime it can happen on cold boot, but sometimes if I turn off the PC and turn on at once. Tried different PSU also. I have also Asus motherboard with Triton 1 chipset. Asus PCI/I-AP55T. Also it seems doesn't happen on lowest voltage (1.6v - 1.7v) but on voltages starting from 1.8.

[@Permudos]

If Im right, I know, whats in the Box 😁 And if I remember right, this CPU worked in an Asus P5A with ALI Aladin V Chipset before.

[@necro_ware]

Of course it is! Or may be not? ;)

[@Permudos]

@@necro_ware Im very excited to see it in the next video. And if it is: especially the difference in performance - or at least, if there is one 😉 (Im expecting the limited bus speed as a real show stopper for most benches)

[@Inkvisitor]

As far as I'm concerned, all Socket 7 motherboards should come with a built-in Necroware VRM. Can we go back in time and fix this ?

[@AmstradExin]

Time to dig up like 20 motherboards then. What Motherboards I found: Intel ADVANCED/ATX (430FX66) (No pins), Chaintech 5SEM M103 (SIS 551X) (No Pins), Chaintech 5IDM2 M105 (430FX). Too bad I don't have the money to just fuck around and find out. Who knows if Installing the pin header would even work.

[@AmstradExin]

Also....Aaaaaaamiiiigaaaaa

[@JamieBainbridge]

In my experience licence violators do not care. GPL effectively has the same copyleft restriction as CC-BY-SA in this situation. It seems if one does not want to be disappointed in licence, the only choice is to be okay with no-restriction CC0 (or Public Domain, WTFPL, etc).

[@djpirtu2]

A little offtopic: but i've been struggling with many Asus P5A mobos which all does the same thing (with k6-3+), just reboots in Windows while idling. I can run 3DMark for hours but idling is bad. Voltage ripple could be the reason and that is maybe caused by bad caps... I really need to buy an oscilloscope.

[@n.shiina8798]

be careful with low ESR caps. too low ESR can also destabilize your VRM

[@amirpourghoureiyan1637]

some hobbyists in the community are just selling their boards on pcb stores since people tend to ignore any license you put on the work

[@necro_ware]

From Germany it's hard to sell those things legally.

[@DavoShed]

That sucks that someone else would claim your work as their own I don’t fully understand what you do but it’s fun to watch. Name and shame them. The next scum bag may think twice. 🤠

[@necro_ware]

Some people just don't think too much about what they are doing.

[@DavoShed]

@@necro_ware some people are also only think about themselves. I like your videos. I have worked with all these PC’s since the 80’s. (Aka the dawn of time) Never understood the actual electronics though. Glad these sneaky people have not detoured you.

[@Ironclad17]

Sadly I don't think people who ignore gpl, are going to respect cc license.

[@necro_ware]

Actually GPL doesn't prohibit to remove names, however you are still right, because after they removed the names, they have to open their changes under GPL again, but they don't :( Let's just hope.

[@ddognine]

It's really sad that pirates ripped off your design for their own aggrandizement, but at the same time, no surprise. We live in a post-modern, secularized world where morals, good vs evil, right or wrong, etc. are irrelevant to the almighty dollar. Furthermore, there are regimes that actually encourage this sort of behavior. The whole point of sharing your design so freely is to garner feedback from similarly inclined individuals to build an even better vrm. Just look at what happened with Stand Up Math's code for words using the letters of the alphabet. However, given the amount of e-waste from Asia, I am not surprised they took your early design and ran with it because "good enough" is often too high a bar. I really hope though you can generate the sort of collaboration and feedback that Stand Up Maths did. You definitely deserve it!

[@necro_ware]

I wouldn't consider those people as pirates, they just don't think, that what they are doing could be harmful. Also many viewers seem to assume, that those "ripped copies" were from China. That's not the case, Chinese vendors are not interested in retro hardware, all those issues were by Europeans.

[@pvc988]

DOS is pretty dumb about how it implements its idle state. Instead of using HLT instruction (and lowering power consumption significantly) which was desinged into CPU specifically for that purpose it uses tight loop. It was ok for pre-CMOS devices, but made absolutely no sense for CMOS ones. There is a tool called DOSIDLE which takes care of that. Occasionally there are some sideeffects of using it but most of the time its pretty damn useful.

[@hgbugalou]

I suspect some Chinese vendors are doing the copying of your design. They are really ruthless with this stuff. 😔

[@necro_ware]

No Chinese vendors seem not to be interested a lot in retro tech. Those were mostly Europeans.