Yes, the adapter also supports a mouse. However, you can only use one device per adapter. E.g. I cannot use my Logitech combo via the receiver because it only connects to the keyboard port. Additionally, it can deal with XT and AT systems.
@@bitsundbolts If it had a second plug for a ps/2 mouse I'd probably order right away! I also happen to have a Logitech combo reciever that I use with my PIII setup, and also with my laptop. Just plugging the receiver into the machine I use is so nice, and I don't have to have separate keyboards. It worked OK in Win98SE, but for other software Bios Legacy emulation needs to be turned on, and it doesn't work all that well, not to mention the mouse not working in DOS at all. Funnily enough I have a wireless Logitech USB and PS/2 keyboard and mouse combo as well, but the mouse is missing and the keyboard part uses USB. TLDR.: if you can add a separate plug for mouses as well, TAKE MY MONEY
@@bitsundbolts that would be my question, if the adapter supports the different XT keyboard protocol as well, because in that case I'd find it very handy. Also, how does it handle the cursor keys in XT mode?
I don't have an XT system nor do I have experience with it. The creator however did implement the XT protocol. Maybe we have to have a few rounds of further development done with the help of the community.
Always a pleasure to help out if I can but on this occasion you did all the work! Thanks for mentioning me! And great idea about using sunshine to cure solder mask, I also have trouble with the UV lamp. I shall do the same… oh wait… 🇬🇧 🇬🇧 🇬🇧
Haha, it's definitely a benefit to have 330+ days of sunshine per year! You just have to get used to the heat in the summer - or go back to a better climate during the hot season! Thank you Tony for taking the time. It was very helpful to exchange thoughts over this issue!
I would try a few more things before call it a day. It could be a poor implementation of the FOREX chipset, maybe making some investigation on this. Some issue with cache on the board, try remove it and observe relevant changes. It could be as well some electrical issue, like a pin with bad joints or some bad cap causing ripple and therefore instability. Please let us posted about it! 😃
3:08 Oh that is just so satisfying to see the braid soak up the solder. Great background music. I love your videos, and similar troubleshooting and repair channels.
Thank you for the mention. Regarding low performance... it is probably mostly due to the ISA bus. The performance difference between ISA and PCI is huge. Especially noticeable with benches / games that have higher frame rates. For a 486 DX2 66MHz with ISA 8.3MHz, 15fps doom max details looks reasonable. You can improve the performance significantly by changing ISA clk to 1/3 or 1/2. Also some ISA cards have a jumper that is for 0WS, this also improves performance. Now the sysinfo performance issue, I have no idea a very spooky behavior.
I agree, I can live with the poor performance in games, but the weird behavior of SI is unexplainable so far. I need to tinker a bit more with this board, but there are so many other things I need to take care of 🙂. You made very interesting videos. Do you consider coming back with new content on TH-cam?
Thank you. I am on a long hiatus for sure. But all this time I was still working on my retro stuff. I have accumulated a ton of retro hardware now + made many new projects and mods. It is just that I do not have much time now + I am trying to put some order to my stuff here first. So I will come back to videos for sure :). I just very busy now T-T. Most of my videos where on COVID.
I had a Forex MB with 3 sockets for 386, 486 and 387. Basically the bigger(and older all ISA) brother of this one. I used it with AMD 386 @ 40MHz and later upgraded it to Intel 486 SX/2 @ 66MHz. That involved Quarz change and EEPROM change. And I also upgraded the SRAM.
Nothing wrong with old solder. I'm 55 and still have some solder that is older than me that was given to me by uncles that have past. I find the old stuff flows better for me.
That slot is an "OPTi Bus" and there were at least TSENG video cards for it. There was other OPTi Bus like that but with what looks like a regular 8 BIT ISA slot behind it, I guess the single slot variant was mostly for video cards because it's missing some ISA signals and the dual slot variant can have enough pins for all VLB signals, but I never seen a card for it.
I remember seeing something at the scrapyard that might have been compatible, but I didn't pay attention to that item because that connector looked nothing like a PC connector.
A good source of UV for things like erasing EPROMs, curing solder mask etc - are UV germicidal bulbs. Essentially like fluorescent bulbs but without the phosphor coating, and with quartz envelopes instead of plain glass. A small 8W T5 tube will fit into standard fluorescent light fixtures, and is powerful enough to erase an EPROM chip in 5-6 minutes. That is what I use. However I agree, when you live in Dubai you have your UV source up in the sky and it's free. :)
These are entirely different use cases and call for different light sources. EPROM have their sensitivity peak that corresponds exactly to the germicidal lamp, these were made for each other. You should also protect yourself against this, this light is very hazardous. Resin photoinitiations don't actually work at wavelengths like that and you can cause degradation instead. They want UVA ideally. 330 to 420nm should about work. These are made for UVA CCFL. You don't need special protection against it. Trying to cure resin with an LED is of course a little frustrating, since the LEDs that they sell all contain acrylic enclosures and cannot work at sub 400nm, so you're only reaching the tail end of the sensitivity. But the sun is alright. In turn it's remarkably difficult to erase an EPROM with the sun since a lot of UVC gets lost in the atmosphere!
@@SianaGearz I did a test this spring, I took a new EPROM with recently burned and tested bios software and placed it in my window facing the sun. We had sun almost everyday this spring. After 3 weeks a small percentage of the bits was erased. After aprox 5 weeks only a few bits remained. The last bit was reset after 6-7 weeks, it was a stubborn one 😆 Just for the science of it, didn't document it very specific though.
Great to hear. Let me just put more details here that others might find helpful. The adapter supports AT keyboard, or a mouse to PS/2. It can also be used for XT systems to connect a USB keyboard.
As my supply of 9x & AT compatible mice is rapidly drying up (I have 2 left that aren’t Serial ball mice), I’m especially looking to convert later usb mice to PS/2 interface for testing. This would 100% solve my specific problem. I have several PS/2 keyboards that are still AT compatible & curently don’t use XT hardware. Having XT compatibility would be a godsend for many others in the retro community.
We may work on two different adapters - one for AT and the other one for XT, but it's not decided yet. This adapter will solve all those issues you mentioned. The creator developed the firmware for this adapter and I even flashed it a couple of times. So, it can be upgraded in the future with more features.
You could try higher L2 cache amount than 128Kb and make sure the cache ships are 15ns or faster. Also try changing cache timings to improve performance.
19:45 - I have 3 versions of that exact Logitech wireless keyboard! They're so great for just a quick key/mouse combo, and the batteries last forever! I love mine!
Yes, you're right. I have seen reports where people use a BIOS from similar boards (similar chipset and features). It could improve performance. I'll keep this in mind. Thanks!
Rule out the socket, and bios by swapping to a 386 CPU and seeing how that compares to your previous 386 benchmarks? Dual socket boards could be iffy at the best of times and usually had some kind of compromise to allow using one type over the other. The chipset maybe more 386 with 486 extension thrown in to be compatible with both. Certainly an interesting one. also did you use a jumper for the turbo, or the bios setting? I didn't really pick up on that bit.
Also, not sure if he still does work with BIOS's but if you can get in touch with a chap called Jan Steinbeck he used to unlock hidden functions and patch bios's for all sorts of things back in the day.
I think I've solved it. You should have used green solder mask. Audiophiles knew that using a green marker around the edge of CDs made them sound better, so it stands to reason that green solder mask would make a motherboard faster. I amaze even myself sometimes...
Turbo - this board spams Cache Flush pin with ~HKz signal to slow down. Constantly flushing whole cache is slower than no cache at all. Speed - 10MB/s ram speed in Cachechk is really bad, its very slow 386 system bad. Bios ram/cache timing settings need tuning.
Will definitely be in the market for a keyboard adapter! I want one of the new Model F keyboards, but they are USB only, and my entire retro setup works via a PS/2 KVM switch.
Great video. Thought I had this motherboard and would have tested it to compare results, but turns out I have a Forex based 386 only motherboard, and I have a dual CPU 386/486 2VLB USA Chipset (SC-9204-a 4L08F1061 Japan) and my memory combined the 2, oh well. The dual CPU board looks similar, minus the OPTi chip and my has 2 legit VLB slots.
Perhaps there was an oxidation issue and you should just wait for the microcode update in mid-August 😛 Seriously, this is quite odd behavior, but something I think you will be able to find out and make another interesting video about 🙌
I hated OPTi chipsets during 386/486 times. They were significantly slower on the ISA bus than any other chipset. They were also very unstable and liked to crash for no reason.
I don't know about the stability, but I am really surprised by the poor performance. I will compare it to other ISA only boards and see if those OPTi chipsets should be avoided for 486 retro machines.
Can you also try without the USB keyboard adapter? Just use an older one with a PS2 connector and a DIN-5 adapter. It's probably not related, but I've seen very weird behavior with 386/486 systems in the past (such as sound blasters clicking and popping loudly or actual video artifacts, because of... SD card to IDE adapters).
I think the USB/ps2 adapter would very useful. I saw a video that I can't seem to find now that actually tested using a USB keyboard on older computers that supported them slowed the performance down around 3-8%, due to the constant interrupt polling of the bus. I'm paraphrasing because I wish I could find it and my memory is spotty. Edit: Ha, it was your video I was thinking of! I feel silly. I'm not in the market for one myself, I don't have the room (or budget) for many hobbies, so I'm enjoying retro PC hardware vicariously through youtube.
When I read the first part of your comment, I was certain you saw my video 😅 - well, you notices it then. Yes, this adapter will solve that problem with USB keyboards on older systems. It's a nice little device and I'm using it these days all the time!
Hello Bits und Bolts. What a nice channel you have! By mistake i visit your channel and immedatly i subscribe. I have a strong UV lamp for the solder mask. It works great. Nice to see your videos. Greetings from Steven from the Netherlands
That USB to DIN Adapter certainly is a useful product. Luckily I am the proud owner of a Model M and I saved some keyboards with Mini DIN / PS/2 plugs that are in great shape, but I suppose the day has come for many where that isn't the case anymore and then such an adapter is invaluable. An affordable USB to RS232 / PS/2 adapter for mice would be great too! Regarding the performance, I concur with Atheatos that ISA bus tweaking is the way to go. Maybe you can also get a hand on an actual OPTi local bus graphics card some time for comparison? Would make for a nice rarity. They might be quite the unicorns though, I've personally never come across one!
As mentioned in the video this is very likely NOT EISA, it's an "OPTI local bus", it's label "local bus" in both the PDF and the BIOS. And yes, this is a known problem, as he mentioned using an EISA card in this motherboard will result in a bad day.
I tell you what would be really useful (for me) in a keyboard adapter, the ability to upload custom keyboard layouts to convert a QWERTY keyboard to something else and switch between them with a keyboard shortcut. I use the COLEMAK layout, and while I do have a custom keyboard with that layout, it's an old Unicomp that I've never been too happy with. It would be better if I could just use my spare modern board, and being able to switch back to QWERTY when I need WASD would be nice too. Obviously it's probably not worth putting the effort into making this feature, since I imagine I'm pretty much the only person who actually needs it. Everyone else has a keyboard that sends the right scancodes for their layout, so I can't imagine anyone else out there needs this.
Nice, simple repair. Glad it works, just a shame about the performance, or lack thereof. It's possible this is an issue with the chipset. I always got the feeling these 386/486 chipset boards were actually a 386 chipset kludged to work with 486 chips. If it were me, I'd be tempted to slip in a 386 chip and see if you get the normal expected performance from that. Otherwise, there may be some issue with the jumper settings, since we really have no idea if those instructions are for your exact revision of the board, or just something pretty close. And even if it is, I've seen boards and cards shipped with a supplemental note showing the correct jumper settings. Evidently the manufacturer got it wrong in the manual, and decided it was cheaper to just print a note with the correction rather than re-print all the manuals.
Good points! Trying the board with a 386 might give some clues. It could very well be that the chipset just doesn't take advantage of 486 CPUs and just somehow works. I'm sure, a 386 will never reach 7+ fps in PC player or 15 fps in Doom - so, it definitely benefits from the higher frequency, but it just can't compete with a proper 486 implementation. We shall see. Maybe it will just be a good 386 system 🙂
About the Norton SI weird behaviour, I would try another version of the program. The version Phil used in the benchmark is from Norton Utilities ; I would try a earlier version just to see if the issue is there (win world pc have them all). Also could try another sysinfo software like aida32, troubleshooter, astra (from Hiren´s boot CD early versions) just to see if the hardware identify routines can replicate the weird behaviour. I never saw something like that, but I believe that the issue is the routine that Norton SI implements to identify the installed hardware (poking I/O adresses?) may alter some bytes and so diabling the cache or something similar. It might be a bug in the BIOS. Try to auto config the BIOS for performance and then for stability to see if there are any difference. Also, may try to see if with an 386 CPU this motherboard would behave the same.
I also have a 386 / 486 board (Edom MV008) with an OPTi chipset and it suffers from similar slow performance, although mine has VLB and I am using a VLB graphics card. Even using /3 divider (11 MHz) on the ISA bus only helps a bit, but not much. From what I can tell, the problem seems to be related to memory write performance as that is really slow (8MB/s so even slower than yours). If I disable the external cache, write performance gets significantly faster (almost 2x) so it seems the cache implementation is the culprit. It might be set up as a simple write through cache and I have no BIOS option to change that. While disabling the cache will improve the memory speed, it doesn't compensate for the loss of cache in the general system performance so I leave it on and just accepted that this platform is probably just not good for 486.
You could use a LIR2032 battery (same size as a CR2032) which is rechargeable. I don't know if you can charge a LI battery with a circuit that was intended to charge a NiCad.
I don't know either. It's a pretty simple circuit. One diode and one resistor. There is no overcharge protection anywhere. I think I'll be ok with a non-chargeable CR2032.
@@bitsundbolts That explains why those old Varta NiCads leaked back in the day. If you left the computer on all the time is continuously charging the battery and at some point it is going to star off-gassing. I have no proof of this, but it seems logical.
@@tigheklory Nah i think they just degrade chemically whether you use them or not, just from ambient heat. Most of these machines were disconnected from power before they leaked, and never turned on again for 25 years, and in the interim there's a leak. Just a little gas should have just diffused out of there no problem.
@@tigheklory Nah, they were inefficient enough that throwing 5V at them wasn't enough to get them to 100% so you ended up with some equilibrium point short of full charge. Given the very slow discharge rate, they don't need to play games for extra battery capacity.
IIRC the UV curing solder mask needs UVA light, you can use the sun but it's much easier to use "nail gel coat" UVA led lights.. Don't ask me why the ones actually intended for UV solder mask is so crappy but I agree they are. You can buy reasonable nail gel curing lights for next to nothing and say a 10W one should cure it in something like 5-30 seconds depending on thickness. This is not surprising given that it basically IS nail gel coat you're applying, just different label. Regular nail polish also works and dries on it's own but is much less "hardy" than either of the UV variants. Those lamps won't work with EPROMs since they need a much shorter wavelength (253.7nm, near UVC) which is also why you have to be far more careful with EPROM eraser lamps.
Thanks! Well, I found it in a case. Except for the battery damage, the board is in very good condition. But you never know. With old hardware, you never know...
I think I have a use case for that adapter - I have a Model F XT converted to USB, and with it I would be able to use the same board on practically every PC ever made :^)
I have an AM4 mobo with a ps/2 port I'd like to use for interrupt driven keyboard input. An adapter like that would be interesting if it supported high polling rates (1kHz+). :)
Very nice video! Your RAM bandwidth is 22.7 MB (shown at 24:27). Other people get 37.4MB which is more than 50% faster. I think that there might be something wrong about 30pin RAM interface implementation. (does it matter which four slots you use? Upper or lower?)
Hm, then I should be able to power on this board with only two memory modules. But the CPU has a 32-bit data bus. Could there be some magic happening in the memory controller on the chipset?
@@bitsundbolts Well there is a chance they used each slot individually in order to support 16 bit and 32 bit memory configuration? Does this motherboard support 386 SX CPUs? It worth a try to play around with memory slots. Even to populate all.
The 387 extension is not the same, as the 0807 extension. The 32 kb 87em. Exe loads into the pc, and resides in there once it is run, even if it can’t run. Heretic game also has it inbuilt, but only enables the maths coprocessor if it is forced to. To force it, the cpu cycles have to be set to very high. It is important for laptops to draw the frames in TH-cam when it is in energy saving mode. Tricky, a good laptop has the feature turned off by default.
This only works if you have a keyboard that implements the PS/2 protocol. When USB got more popular, keyboards supported both and you could use it with USB or ps/2 with a simple adapter. That's no longer possible with modern keyboards that use a wireless dongle
@@bitsundbolts Ah, my 'modern' USB keyboard(s) still work, although they are not Wireless, I don't personally like Wireless keyboards (and have never found myself thinking 'I wish this keyboard was wireless' :D) so maybe it's just wireless ones? I'll have to test all the KB's I have and see if any don't work (some basic USB ones, some more 'advanced' with RGB and Screens etc) :)
I would expect the more modern keyboards to not work with a simple USB to PS/2 adapter. I think the older models have a way to detect if they're connected to a USB port or PS/2 and transmit with the correct protocol.
@@bitsundbolts Have so far tested my Asus ROG Strix Flare - that worked.. (no RGB obviously, separate USB for the LED's) Logitech G510s - This also worked, but only as a basic keyboard (again, the display etc doesn't work, as requires a 2nd USB) Basic Dell "e-waste" Keyboard - This also worked, no problems here... Basic 'unknown' e-waste 'thin' keyboard - This one DIDN'T work.... I have a couple more to test when I dig them out of my storage tomorrow :)
Those passive adapters only work if the USB keyboard/mouse is PS/2 compatible as well. This was fairly common in earlier devices but nowadays you'd probably have to find one specifically that states PS/2 support.
I have had the same problem with Norton SI some years ago and stopped using it. I think it disables cache on some systems because of some faulty CPU detection code. Try testing the system without L1 and/or L2 cache and check against the scores you got after running Norton SI.
Interesting. I'll definitely investigate further. This behavior of System Information is just so odd. But at least I know that I have to watch out if I use it in the future
I wouldnt say 'dont plug anything in'. just verify the pinout, vogons has a thread on how to tell them apart. I think OLB expansion cards always(?) has "Local Bus" written on them near the edge connector.
I mean, if you do find a compatible expansion card, then I guess it's possible to use it. However, as I understand, you need specific drivers to make the card work - sometimes, those drivers are even board specific.
I have to say the use of a connector _that already has one use_ should require that they somehow distinguish -- maybe by installing the socket 180 degrees around so the notch in the edge connector is in a different place. This was a very bad design decision and I'm glad it fell out of favor even harder than EISA itself did.
I would love to have one of the adapters, better yet two (if mouse can be added) to run my DEC Alpha Multia. Supposedly some older USB keyboards can be switched but I have never found one.
The adapter has a firmware developed by the creator. There is a possibility to support models that may not yet work properly in the future. And yes, a second adapter makes sense for a mouse.
I have worked on a similar adapter using a Raspberry Pi Pico called ps2x2pico which does support keyboard and mouse simultaneously, with a hub or a logitech unifying receiver
@@bitsundbolts Cool! It also supports AT and XT (through the ps2pico firmware), two PS/2 passthru ports and its part of the PiKVM pico hid for emulating PS/2 devices with a PiKVM.
This had to be a budget motherboard for the time, and IMHO mostly meant for 386 CPUs, providing an update path to early 486DX and SX. People thought they wouldn't be missing out on the performance when upgrading just the CPU.
@@phljns that's why there's only one, and even that is not standard EISA, so I'd consider it more of a gimmick than a useful thing. This approach was common for budget motherboards for a long time.
Jack-of-all-trades boards like this tended to do lots of things poorly. The 486 support is probably bugged as you found, but good luck finding a BIOS to run 486 CPUs properly. I wonder if a 386 DX 40 would give a better experience all around. In reality, that is what probably nearly all boards like this were actually used with as they would have been sold to people that couldn't actually afford a 486 system at the time. That reuse of the EISA physical socket that is even labeled EISA but will literally destroy expensive and rare EISA cards (both then and now!) is just terrible design.
I suspect your problems are related to the fact that it's a multi CPU board and the chipset is neither fish nor fowl nor good red meat. I'd set that the implementation for the cache is screwy/borked in some fashion. Try it without cache with the 486 and with and without cahce wth a 386.
Good point. I have a feeling that the frequency changes in certain situations. But i am still confused why system information would change the systems speed
I don't like ml2032. Yes it's safe and it won't leak. But also it will die in a few years. You'd have to plug in your computer regularly to charge the battery. If you use the machine too much or too little, the cell dies in a couple years. cr2032 you put it in and forget about it for a decade. For all classic NiCD batteries, you can get NiMH replacements of stacked button cell type. NiMH is safe against leakage if you want a machine which works like original with regard to battery. Which is to say a little annoying, if you put the machine away for a while, the battery will be flat, but at least it will fully self recover when you actually use it. The endurance and self discharge of NiMH is much better. You can even cheat and install a connector on the board and mount 3xAAA cells off board somewhere in computer enclosure, the Eneloop like type, this works very well actually.
I would like something that would allow using at keyboard with motherboard that needs xt keyboard. Well if you could use usb keyboard with motherboard that needs xt keyboard it would work too
the memory would be a bottle neck! but not likely ! I would test it in 386 mode! if you get the same results! the bottle neck is the cpus to "North Bridge" the number of lanes of data thru put! the ps2 mouse port and key board ports are same eclectically the same just different positions in the ps/2 connector... they both use the same clock circuit same transmit and receive and Vcc and Gnd its just a matter of flipping a couple of wires to making the USB to ps/2 function as a mouse .
I have a similar 386/486 board, the FIC 4386-VC-HD. Paired it with an i486 DX2-66, CL GD5429 ISA graphics-card, 256KB cache and 20MB 32-pin SIMM RAM. It's also a mixed bag, getting a Topbench score of 110, Doom max 13.7 fps, PCPlayer 8.6, speedsys 24.92. Sysinfo works though with a score of 130.7. Your main memory and L2 cache look way too slow. My board has 18.1 MB/s memory and 46MB/s L2 cache in cachechk. The main bottleneck is ISA bus in both our cases I believe. Sadly my Award BIOS doesn't let me change the divider and I don’t know which clock speed it even runs at.
That's very good information! Thanks for sharing. So, we do score about identical in many benchmarks. Except for system information - which seems to work for you.
@bitsundbolts Thank you for the great videos! Glad I could help and provide info. Sysinfo bugging out might be due to different chipsets, my board has a VIA one. Wish I had an EPROM programmer, retroweb lately got an AMI BIOS for my board that might allow to change ISA dividers. If you want to compare any other benchmarks, let me know and I'll try my best to report them back.
Thanks for your offer. I might come back to it. But I also have so many other 486 boards that I need to look at - I think I should have around 7 boards with ISA only.
I wonder if the Forex chipset makes compromises to support both the 386 and 486 bus. I.e. you get 386 bus performance even when you have a 486 installed.
Surely you tried with a coin cell battery plugged in? Otherwise worth a shot, it's been proven in the past that some boards behave weirdly without a battery
I've got a similar motherboard (same chipset soldered AM386DX40) , found that switching from a 80386 to a 486 CPU required the turbo switch to be physically jumpered (JP14 block). My board is the MECER Corp B800. Could very well apply to yours :)
I'm using a jumper on the pin headers. You're right, it is the easiest way to get such a board to its optimal performance. Edit: does your board score also low on system information (if you would have that data)?
I have recently tested all new solder available in our electronics supplier here and comparing it with my solder that I have bought around 2000 the new ones are garbage.
I have tested only the leaded solder available at the shop. I will not name brands but I can say that there were also german and dutch "made" solder. Even the new batch of the same brand of solder as the one that I have is signifficantly worse. I suspect that it has to do with the metal purity and content of the rosin flux inside.
Bürklin has it but doesn't want to sell it to you by mail. Maybe if you drop by their facility in person, i didn't try. Or if you have a registered business, then they sell it to you no problem. TME might be easier, it's hidden if you're browsing from Germany but if you have a link i heard you can buy it. They carry Broquetas and Cynel. Eleshop will just sell the same Broquetas leaded solders to you from the Netherlands.
Id love to have a wireless keyboard on the bench. One less cable . But i would want the dongle small without a cable. If you know what i mean. AT keyboard to usb dongle. Love the videos , keep up the good work.
just fresh from watching this video and my first thought is chipset. some bug or limitation in the chipset. or it could just be a really crap board. they did make a ton a crap back then
Haha, maybe... Unfortunate because it would allow testing 386 and 486 on the same board. Maybe the 386 performance is ok, but they dropped the ball for 486.
Did you check to make sure the keyboard adapter wasn't the cause of the weird performance issues? Probably not but best to rule it out. My guess is it's probably the bios itself, and if there even is an updated bios, it's probably unobtanium.
He covered that. No, it's not, and it was the worst connector that could have been chosen for this _proprietary_ interface. At the very least they could have spun the socket around 180 degrees so the notch falls in a different place.
One should be able to colour match the mask by mixing these Mechanic UV mask resins, but i haven't tried. The application looks thick, but i know it's difficult to apply thinner. I have had to apply the mask much thinner for it to actually successfully UV cure. But then i don't do sun :D Oh also i noticed recently Relife has a clear solder mask which they promise is much quicker and easier to cure. Haven't gotten any myself yet. Most of my keyboards (not the mechanical one i use day to day on my computer) and quite a few mice have internal PS/2 functionality and work with a passive USB to PS2 adapter, and i thought everyone had it like this, because they're extremely cheap and widely available. But i guess if you want to use your favourite input devices, wireless ones and the like...
I had this multiple times that the solder mask cures outside, but is still liquid under the surface when using the UV light. When it's under the sun, it cures properly.
I am wondering if the 386 BIOS may be causing/creating an issue. Perhaps you could configure it with a 386-33, if you have one, and run some tests on that. If the performance is as expected, for the selected 386 CPU, perhaps a new BIOS will be in order, should you choose to reinstall the 486.
I take it this era of systems can't actually use more than one socketed CPU at a time? Would be pretty interesting if somebody could build a system that could somehow utilize multi CPU.
Yes, those boards will only work with one CPU at a time. Think of the memory access for instance. Those memory controllers cannot accommodate two CPUs requesting access to the memory, or ISA slots. Most of the traces go directly to the CPU socket. I have another 386 board I need to work on. There are almost no other components on the board except for the CPU.
I tried building a ps2x2pico twice and failed both times. I threw in the towel and I'd love to buy a prebuilt solution that accepts USB mouse and keyboard. I use a USB switch and would love to hook it up to one of these if it supported both mouse and keyboard.
The adapter supports mice and keyboards. At the Moment, you would need two devices. That is a problem if you're using a dongle that acts as a USB hub to transmit both, mouse and keyboard signals.
Both wrong, it's a proprietary slot that was (stupidly) built from an EISA slot physically, and the second socket is for a 386 as covered in the previous video on this board.
Yes, the adapter takes its power from the port on the PC port. That port also needs to power the keyboard - so, there is enough juice for the little adapter as well.
Did you put a heatsink on the CPU in the end? How important are the heatsink instructions on the chip? Do you usually get away without a heatsink even on a chip that says it needs one?
Yes, I used a cooler. I just put a socket 7 cooler on the CPU - not clipped anywhere since this socket doesn't allow for it. The CPU gets hot and the heatsink warms up quite a bit. I think you should follow the instructions. Those CPUs should have a heatsink and airflow
I’m interested in the usb/ps2 adapter. It would make much easier and flexible using modern mice and keyboards. I have some stocked, the keyboards are ok, mice are from Aliexpress, I would rather use some Logitech. It’s getting more difficult to get PS/2 hardware and eventually is gonna get expensive. I would buy a couple of this adapters.
I'll make a dedicated video about it and show off the capabilities of this adapter. The creator is still ironing out a few small things, but the prototype is working pretty well!
A bit surprised that you are based in The United Arab Emirates. You sound like one from Germany. Didn't know that retro computers are easily available in UAE. Is there a community for old electronics and computers in UAE?
I moved to Dubai almost 20 years ago, but I grew up in Germany, went to school, and left right after that. The retro community in the UAE is quite small. I just know a few people who are into old retro PCs. If I wouldn't have been introduced to the scrapyard that receives old hardware from time to time, I wouldn't be able to make those videos.
Opti-chipsets are just soo sloow... I have 386/486 hybrid board with VLB and it's fast with 486DLC CPU but super slow with 486. Memory bandwith is miserable, whatever timings are used. So, try a 386 CPU 😅
@@bitsundbolts It might be a good match for the AMD 386/33, which most likely will happily run at 40. That should offset some of the shortcomings of the board.
If the adapter also was a mouse I think it'd be a very appealing product!
Yes, the adapter also supports a mouse. However, you can only use one device per adapter. E.g. I cannot use my Logitech combo via the receiver because it only connects to the keyboard port. Additionally, it can deal with XT and AT systems.
@@bitsundbolts If it had a second plug for a ps/2 mouse I'd probably order right away!
I also happen to have a Logitech combo reciever that I use with my PIII setup, and also with my laptop. Just plugging the receiver into the machine I use is so nice, and I don't have to have separate keyboards.
It worked OK in Win98SE, but for other software Bios Legacy emulation needs to be turned on, and it doesn't work all that well, not to mention the mouse not working in DOS at all.
Funnily enough I have a wireless Logitech USB and PS/2 keyboard and mouse combo as well, but the mouse is missing and the keyboard part uses USB.
TLDR.: if you can add a separate plug for mouses as well, TAKE MY MONEY
@@bitsundbolts that would be my question, if the adapter supports the different XT keyboard protocol as well, because in that case I'd find it very handy. Also, how does it handle the cursor keys in XT mode?
I'll check if something like this is possible. Might be an update to a later version.
I don't have an XT system nor do I have experience with it. The creator however did implement the XT protocol. Maybe we have to have a few rounds of further development done with the help of the community.
What's crazy is that the CPU in that wireless USB receiver is probably almost or more powerful than the 486 it's attached to!
What's even crazier is that a lot of these USB to PS/2 adapters (and I'm not sure if this one is an exception) basically run on a 6502 core.
Always a pleasure to help out if I can but on this occasion you did all the work! Thanks for mentioning me! And great idea about using sunshine to cure solder mask, I also have trouble with the UV lamp. I shall do the same… oh wait… 🇬🇧 🇬🇧 🇬🇧
Haha, it's definitely a benefit to have 330+ days of sunshine per year! You just have to get used to the heat in the summer - or go back to a better climate during the hot season!
Thank you Tony for taking the time. It was very helpful to exchange thoughts over this issue!
@@bitsundbolts We have 330 days of clouds and rain here :) I have 35 days to retrobright and apply solder masks to all my projects! :)
That requires very good planning if you have projects like that! 😜
@@tony359 but your country has Silverstone Circuit, it pays back for all inconvienances :) rainy race - it is awesome!
@@tennickjestzajety69 I knew it was Silverstone's fault! :)
I would try a few more things before call it a day. It could be a poor implementation of the FOREX chipset, maybe making some investigation on this. Some issue with cache on the board, try remove it and observe relevant changes. It could be as well some electrical issue, like a pin with bad joints or some bad cap causing ripple and therefore instability. Please let us posted about it! 😃
3:08 Oh that is just so satisfying to see the braid soak up the solder. Great background music. I love your videos, and similar troubleshooting and repair channels.
Thanks 😊
Constantly see stuff on your channel that I never knew even existed. Y'all have the coolest scrapyards!
Haha, great to hear! Thanks 👍
i like the new natural, laid back style! greetings from germany :-D
oh yes forgot to add that, I also like the new "no script" style! :)
@@tony359your channel is awesome, too :-D thx for the good stuff
I really like the change in tone of your last few videos. You keep making them, and we'll keep watching :)
I think the keyboard adapter is great, would have saved me a lot of time.
We're working on the prototype and get it ready as soon as possible!
Thank you for the mention.
Regarding low performance... it is probably mostly due to the ISA bus.
The performance difference between ISA and PCI is huge.
Especially noticeable with benches / games that have higher frame rates.
For a 486 DX2 66MHz with ISA 8.3MHz, 15fps doom max details looks reasonable.
You can improve the performance significantly by changing ISA clk to 1/3 or 1/2.
Also some ISA cards have a jumper that is for 0WS, this also improves performance.
Now the sysinfo performance issue, I have no idea a very spooky behavior.
I agree, I can live with the poor performance in games, but the weird behavior of SI is unexplainable so far. I need to tinker a bit more with this board, but there are so many other things I need to take care of 🙂. You made very interesting videos. Do you consider coming back with new content on TH-cam?
Thank you.
I am on a long hiatus for sure. But all this time I was still working on my retro stuff.
I have accumulated a ton of retro hardware now + made many new projects and mods.
It is just that I do not have much time now + I am trying to put some order to my stuff here first.
So I will come back to videos for sure :). I just very busy now T-T. Most of my videos where on COVID.
I had a Forex MB with 3 sockets for 386, 486 and 387. Basically the bigger(and older all ISA) brother of this one. I used it with AMD 386 @ 40MHz and later upgraded it to Intel 486 SX/2 @ 66MHz.
That involved Quarz change and EEPROM change. And I also upgraded the SRAM.
Nothing wrong with old solder. I'm 55 and still have some solder that is older than me that was given to me by uncles that have past. I find the old stuff flows better for me.
That slot is an "OPTi Bus" and there were at least TSENG video cards for it.
There was other OPTi Bus like that but with what looks like a regular 8 BIT ISA slot behind it, I guess the single slot variant was mostly for video cards because it's missing some ISA signals and the dual slot variant can have enough pins for all VLB signals, but I never seen a card for it.
I remember seeing something at the scrapyard that might have been compatible, but I didn't pay attention to that item because that connector looked nothing like a PC connector.
@@bitsundbolts Nothing worse than missed opportunities.
Unfortunately 😔
A good source of UV for things like erasing EPROMs, curing solder mask etc - are UV germicidal bulbs. Essentially like fluorescent bulbs but without the phosphor coating, and with quartz envelopes instead of plain glass. A small 8W T5 tube will fit into standard fluorescent light fixtures, and is powerful enough to erase an EPROM chip in 5-6 minutes. That is what I use. However I agree, when you live in Dubai you have your UV source up in the sky and it's free. :)
These are entirely different use cases and call for different light sources.
EPROM have their sensitivity peak that corresponds exactly to the germicidal lamp, these were made for each other. You should also protect yourself against this, this light is very hazardous.
Resin photoinitiations don't actually work at wavelengths like that and you can cause degradation instead. They want UVA ideally. 330 to 420nm should about work. These are made for UVA CCFL. You don't need special protection against it.
Trying to cure resin with an LED is of course a little frustrating, since the LEDs that they sell all contain acrylic enclosures and cannot work at sub 400nm, so you're only reaching the tail end of the sensitivity. But the sun is alright.
In turn it's remarkably difficult to erase an EPROM with the sun since a lot of UVC gets lost in the atmosphere!
@@SianaGearz I did a test this spring, I took a new EPROM with recently burned and tested bios software and placed it in my window facing the sun. We had sun almost everyday this spring. After 3 weeks a small percentage of the bits was erased. After aprox 5 weeks only a few bits remained. The last bit was reset after 6-7 weeks, it was a stubborn one 😆 Just for the science of it, didn't document it very specific though.
@@tomchristian The glass window blocks most of the UV.
This adaptor is sorely needed.
Great to hear. Let me just put more details here that others might find helpful. The adapter supports AT keyboard, or a mouse to PS/2. It can also be used for XT systems to connect a USB keyboard.
As my supply of 9x & AT compatible mice is rapidly drying up (I have 2 left that aren’t Serial ball mice), I’m especially looking to convert later usb mice to PS/2 interface for testing. This would 100% solve my specific problem. I have several PS/2 keyboards that are still AT compatible & curently don’t use XT hardware.
Having XT compatibility would be a godsend for many others in the retro community.
We may work on two different adapters - one for AT and the other one for XT, but it's not decided yet. This adapter will solve all those issues you mentioned. The creator developed the firmware for this adapter and I even flashed it a couple of times. So, it can be upgraded in the future with more features.
I can't express how amazing this new natural style of videos is! Totally different from the other "heavily scripted" videos 😅. Keep them coming!!!
You could try higher L2 cache amount than 128Kb and make sure the cache ships are 15ns or faster. Also try changing cache timings to improve performance.
19:45 - I have 3 versions of that exact Logitech wireless keyboard! They're so great for just a quick key/mouse combo, and the batteries last forever! I love mine!
Stannol makes a similar alloy with great flux. It's still available to me here!
20ns cache chips seem slow - if I recall I think 12ns was the norm for the 486 era
Love Tony's contents. Really friendly in the comments section, too.
For testing purpose You could try find bios for motherboard with similar specs, especially chipset. It may run.
Yes, you're right. I have seen reports where people use a BIOS from similar boards (similar chipset and features). It could improve performance. I'll keep this in mind. Thanks!
Rule out the socket, and bios by swapping to a 386 CPU and seeing how that compares to your previous 386 benchmarks?
Dual socket boards could be iffy at the best of times and usually had some kind of compromise to allow using one type over the other. The chipset maybe more 386 with 486 extension thrown in to be compatible with both.
Certainly an interesting one. also did you use a jumper for the turbo, or the bios setting? I didn't really pick up on that bit.
Also, not sure if he still does work with BIOS's but if you can get in touch with a chap called Jan Steinbeck he used to unlock hidden functions and patch bios's for all sorts of things back in the day.
Na época eu tive essa placa, tive um 386... 40mhz ela queimou... depois comprei 486 dx2 66... numa placa melhor...
eu tambem acho que o problema principal é a placa mae... depois por ser um AMD...
@@netowork3d the fact it is an AMD has nothing to do with it. clock for clock the 386's were great CPU's.
I think I've solved it. You should have used green solder mask. Audiophiles knew that using a green marker around the edge of CDs made them sound better, so it stands to reason that green solder mask would make a motherboard faster. I amaze even myself sometimes...
Turbo - this board spams Cache Flush pin with ~HKz signal to slow down. Constantly flushing whole cache is slower than no cache at all.
Speed - 10MB/s ram speed in Cachechk is really bad, its very slow 386 system bad. Bios ram/cache timing settings need tuning.
I'll try to run the board without cache and see if it makes a difference.
Will definitely be in the market for a keyboard adapter!
I want one of the new Model F keyboards, but they are USB only, and my entire retro setup works via a PS/2 KVM switch.
This usb to ps/2 adapter will be very usefull.
that brown socket is a 32bit socket, I had a HDD controller for that socket in my 486 PC
They still make your Solder wire. Balve Zinn is in my neighboring town. 😊👍
Oh wow! Nice! I wonder if they would be surprised to know that there is still one of their solder wires out there from 1998.
I would try a different bios from another board with the same chipset, and see if it makes any difference
Good luck finding a board with that chipset.
Great video. Thought I had this motherboard and would have tested it to compare results, but turns out I have a Forex based 386 only motherboard, and I have a dual CPU 386/486 2VLB USA Chipset (SC-9204-a 4L08F1061 Japan) and my memory combined the 2, oh well. The dual CPU board looks similar, minus the OPTi chip and my has 2 legit VLB slots.
Perhaps there was an oxidation issue and you should just wait for the microcode update in mid-August 😛
Seriously, this is quite odd behavior, but something I think you will be able to find out and make another interesting video about 🙌
I was going to say the CPU is defective, but you ruled that out. lol
Yes, that was my thinking. But the Intel DX2 scored identical. Maybe it's just the board/chipset :(
I love that keyboard adapter!
Yea the adapter is cool, i want it!
I hated OPTi chipsets during 386/486 times. They were significantly slower on the ISA bus than any other chipset. They were also very unstable and liked to crash for no reason.
I don't know about the stability, but I am really surprised by the poor performance. I will compare it to other ISA only boards and see if those OPTi chipsets should be avoided for 486 retro machines.
Can you also try without the USB keyboard adapter? Just use an older one with a PS2 connector and a DIN-5 adapter. It's probably not related, but I've seen very weird behavior with 386/486 systems in the past (such as sound blasters clicking and popping loudly or actual video artifacts, because of... SD card to IDE adapters).
I think the USB/ps2 adapter would very useful. I saw a video that I can't seem to find now that actually tested using a USB keyboard on older computers that supported them slowed the performance down around 3-8%, due to the constant interrupt polling of the bus. I'm paraphrasing because I wish I could find it and my memory is spotty.
Edit: Ha, it was your video I was thinking of! I feel silly.
I'm not in the market for one myself, I don't have the room (or budget) for many hobbies, so I'm enjoying retro PC hardware vicariously through youtube.
When I read the first part of your comment, I was certain you saw my video 😅 - well, you notices it then.
Yes, this adapter will solve that problem with USB keyboards on older systems. It's a nice little device and I'm using it these days all the time!
Hello Bits und Bolts. What a nice channel you have! By mistake i visit your channel and immedatly i subscribe. I have a strong UV lamp for the solder mask. It works great. Nice to see your videos. Greetings from Steven from the Netherlands
Thank you Steven! Glad to hear that you like the videos!
I'm interested in that USB to PS/2 adapter!
That USB to DIN Adapter certainly is a useful product. Luckily I am the proud owner of a Model M and I saved some keyboards with Mini DIN / PS/2 plugs that are in great shape, but I suppose the day has come for many where that isn't the case anymore and then such an adapter is invaluable. An affordable USB to RS232 / PS/2 adapter for mice would be great too!
Regarding the performance, I concur with Atheatos that ISA bus tweaking is the way to go. Maybe you can also get a hand on an actual OPTi local bus graphics card some time for comparison? Would make for a nice rarity. They might be quite the unicorns though, I've personally never come across one!
One EISA slot, nice It has features compared to microchannel arquitecture from IBM ps/2 line
As mentioned in the video this is very likely NOT EISA, it's an "OPTI local bus", it's label "local bus" in both the PDF and the BIOS. And yes, this is a known problem, as he mentioned using an EISA card in this motherboard will result in a bad day.
I tell you what would be really useful (for me) in a keyboard adapter, the ability to upload custom keyboard layouts to convert a QWERTY keyboard to something else and switch between them with a keyboard shortcut. I use the COLEMAK layout, and while I do have a custom keyboard with that layout, it's an old Unicomp that I've never been too happy with. It would be better if I could just use my spare modern board, and being able to switch back to QWERTY when I need WASD would be nice too.
Obviously it's probably not worth putting the effort into making this feature, since I imagine I'm pretty much the only person who actually needs it. Everyone else has a keyboard that sends the right scancodes for their layout, so I can't imagine anyone else out there needs this.
I work in an IT Refurbishment/recycling place and we have absolutely loads of PS/2 keyboards lying around lol.
7:30 there is Pb / Lead in the solder wire, be careful tho, you wont ever get this kind of solder wire nowadays...
They already make PS2-USB adapters. Have done so for 20+ years.
Nice, simple repair. Glad it works, just a shame about the performance, or lack thereof. It's possible this is an issue with the chipset. I always got the feeling these 386/486 chipset boards were actually a 386 chipset kludged to work with 486 chips. If it were me, I'd be tempted to slip in a 386 chip and see if you get the normal expected performance from that. Otherwise, there may be some issue with the jumper settings, since we really have no idea if those instructions are for your exact revision of the board, or just something pretty close. And even if it is, I've seen boards and cards shipped with a supplemental note showing the correct jumper settings. Evidently the manufacturer got it wrong in the manual, and decided it was cheaper to just print a note with the correction rather than re-print all the manuals.
Good points! Trying the board with a 386 might give some clues. It could very well be that the chipset just doesn't take advantage of 486 CPUs and just somehow works. I'm sure, a 386 will never reach 7+ fps in PC player or 15 fps in Doom - so, it definitely benefits from the higher frequency, but it just can't compete with a proper 486 implementation. We shall see. Maybe it will just be a good 386 system 🙂
@@bitsundbolts That's what I was thinking, if it performs like a normal 386, just find a better board for the 486 CPU and keep this one as a 386.
About the Norton SI weird behaviour, I would try another version of the program. The version Phil used in the benchmark is from Norton Utilities ; I would try a earlier version just to see if the issue is there (win world pc have them all). Also could try another sysinfo software like aida32, troubleshooter, astra (from Hiren´s boot CD early versions) just to see if the hardware identify routines can replicate the weird behaviour. I never saw something like that, but I believe that the issue is the routine that Norton SI implements to identify the installed hardware (poking I/O adresses?) may alter some bytes and so diabling the cache or something similar. It might be a bug in the BIOS. Try to auto config the BIOS for performance and then for stability to see if there are any difference. Also, may try to see if with an 386 CPU this motherboard would behave the same.
I also have a 386 / 486 board (Edom MV008) with an OPTi chipset and it suffers from similar slow performance, although mine has VLB and I am using a VLB graphics card.
Even using /3 divider (11 MHz) on the ISA bus only helps a bit, but not much.
From what I can tell, the problem seems to be related to memory write performance as that is really slow (8MB/s so even slower than yours). If I disable the external cache, write performance gets significantly faster (almost 2x) so it seems the cache implementation is the culprit. It might be set up as a simple write through cache and I have no BIOS option to change that.
While disabling the cache will improve the memory speed, it doesn't compensate for the loss of cache in the general system performance so I leave it on and just accepted that this platform is probably just not good for 486.
You could use a LIR2032 battery (same size as a CR2032) which is rechargeable. I don't know if you can charge a LI battery with a circuit that was intended to charge a NiCad.
I don't know either. It's a pretty simple circuit. One diode and one resistor. There is no overcharge protection anywhere. I think I'll be ok with a non-chargeable CR2032.
@@bitsundbolts That explains why those old Varta NiCads leaked back in the day. If you left the computer on all the time is continuously charging the battery and at some point it is going to star off-gassing. I have no proof of this, but it seems logical.
@@tigheklory Nah i think they just degrade chemically whether you use them or not, just from ambient heat. Most of these machines were disconnected from power before they leaked, and never turned on again for 25 years, and in the interim there's a leak. Just a little gas should have just diffused out of there no problem.
@@tigheklory Nah, they were inefficient enough that throwing 5V at them wasn't enough to get them to 100% so you ended up with some equilibrium point short of full charge. Given the very slow discharge rate, they don't need to play games for extra battery capacity.
definitely interested in the wireless PS2 Adapter
I had this board back in the day.. got it cheap because it ran so slow.
IIRC the UV curing solder mask needs UVA light, you can use the sun but it's much easier to use "nail gel coat" UVA led lights.. Don't ask me why the ones actually intended for UV solder mask is so crappy but I agree they are. You can buy reasonable nail gel curing lights for next to nothing and say a 10W one should cure it in something like 5-30 seconds depending on thickness. This is not surprising given that it basically IS nail gel coat you're applying, just different label. Regular nail polish also works and dries on it's own but is much less "hardy" than either of the UV variants. Those lamps won't work with EPROMs since they need a much shorter wavelength (253.7nm, near UVC) which is also why you have to be far more careful with EPROM eraser lamps.
Thanks for this great information!
i'll be honest...gave it 0 chance of working! awesome video and repair, BuB
Thanks! Well, I found it in a case. Except for the battery damage, the board is in very good condition. But you never know. With old hardware, you never know...
I think I have a use case for that adapter - I have a Model F XT converted to USB, and with it I would be able to use the same board on practically every PC ever made :^)
The adapter also supports the XT protocol ;)
But for you, the adapter will take care of AT systems and you're good to go 👍
if you ever have need of decent 60/40 solder with activated rosin flux that's 1mm diameter, MG Chemicals 4896-227G is excellent.
Nice!
yes the adapter would be useful
I have an AM4 mobo with a ps/2 port I'd like to use for interrupt driven keyboard input. An adapter like that would be interesting if it supported high polling rates (1kHz+). :)
Very nice video! Your RAM bandwidth is 22.7 MB (shown at 24:27). Other people get 37.4MB which is more than 50% faster. I think that there might be something wrong about 30pin RAM interface implementation. (does it matter which four slots you use? Upper or lower?)
Good point. Maybe there is a difference between the memory banks. I'll check it. Thanks for suggestion
It might be cheaped out and the company went with a 16bit bus. Would also explain the results.
Hm, then I should be able to power on this board with only two memory modules. But the CPU has a 32-bit data bus. Could there be some magic happening in the memory controller on the chipset?
10MB/s in Cachechk, this is _really_ bad. I would start by lowering all possible tuning timing knobs in bios.
@@bitsundbolts Well there is a chance they used each slot individually in order to support 16 bit and 32 bit memory configuration? Does this motherboard support 386 SX CPUs? It worth a try to play around with memory slots. Even to populate all.
Please would you share the soldering iron temperature range to use while working on the traces? 🙏🏻
I'm usually at 330 degrees Celsius
@@bitsundbolts Thank you! 🔝
The 387 extension is not the same, as the 0807 extension. The 32 kb 87em. Exe loads into the pc, and resides in there once it is run, even if it can’t run. Heretic game also has it inbuilt, but only enables the maths coprocessor if it is forced to. To force it, the cpu cycles have to be set to very high. It is important for laptops to draw the frames in TH-cam when it is in energy saving mode. Tricky, a good laptop has the feature turned off by default.
will that USB to PS/2 adapter split out the mouse signal if you use a PS/2 spitter? and potentially a PS/2 to serial adapter?
At the moment, it only translates one signal at a time. It supports AT mouse or keyboard, but not simultaneously. It also supports USB to XT keyboard
I just use a USB>PS/2 that I got with a mouse years back, and a PS2>AT (5-pin DIN) adaptor and seems to work fine! :)
This only works if you have a keyboard that implements the PS/2 protocol. When USB got more popular, keyboards supported both and you could use it with USB or ps/2 with a simple adapter. That's no longer possible with modern keyboards that use a wireless dongle
@@bitsundbolts Ah, my 'modern' USB keyboard(s) still work, although they are not Wireless, I don't personally like Wireless keyboards (and have never found myself thinking 'I wish this keyboard was wireless' :D) so maybe it's just wireless ones?
I'll have to test all the KB's I have and see if any don't work (some basic USB ones, some more 'advanced' with RGB and Screens etc) :)
I would expect the more modern keyboards to not work with a simple USB to PS/2 adapter. I think the older models have a way to detect if they're connected to a USB port or PS/2 and transmit with the correct protocol.
@@bitsundbolts Have so far tested my Asus ROG Strix Flare - that worked.. (no RGB obviously, separate USB for the LED's)
Logitech G510s - This also worked, but only as a basic keyboard (again, the display etc doesn't work, as requires a 2nd USB)
Basic Dell "e-waste" Keyboard - This also worked, no problems here...
Basic 'unknown' e-waste 'thin' keyboard - This one DIDN'T work....
I have a couple more to test when I dig them out of my storage tomorrow :)
Those passive adapters only work if the USB keyboard/mouse is PS/2 compatible as well. This was fairly common in earlier devices but nowadays you'd probably have to find one specifically that states PS/2 support.
Is that "extended local bus" the same thing that used to be called "VESA local bus"?
I think my 486 used it for the video card.
I have had the same problem with Norton SI some years ago and stopped using it. I think it disables cache on some systems because of some faulty CPU detection code. Try testing the system without L1 and/or L2 cache and check against the scores you got after running Norton SI.
Interesting. I'll definitely investigate further. This behavior of System Information is just so odd. But at least I know that I have to watch out if I use it in the future
I wouldnt say 'dont plug anything in'. just verify the pinout, vogons has a thread on how to tell them apart. I think OLB expansion cards always(?) has "Local Bus" written on them near the edge connector.
I mean, if you do find a compatible expansion card, then I guess it's possible to use it. However, as I understand, you need specific drivers to make the card work - sometimes, those drivers are even board specific.
@@bitsundbolts vga cards come with the drivers in ROM. personally I do have an OLB motherboard + an OLB gfx card.
I have to say the use of a connector _that already has one use_ should require that they somehow distinguish -- maybe by installing the socket 180 degrees around so the notch in the edge connector is in a different place. This was a very bad design decision and I'm glad it fell out of favor even harder than EISA itself did.
I would love to have one of the adapters, better yet two (if mouse can be added) to run my DEC Alpha Multia. Supposedly some older USB keyboards can be switched but I have never found one.
The adapter has a firmware developed by the creator. There is a possibility to support models that may not yet work properly in the future. And yes, a second adapter makes sense for a mouse.
I have worked on a similar adapter using a Raspberry Pi Pico called ps2x2pico which does support keyboard and mouse simultaneously, with a hub or a logitech unifying receiver
Nice. I'll have a look
@@bitsundbolts Cool! It also supports AT and XT (through the ps2pico firmware), two PS/2 passthru ports and its part of the PiKVM pico hid for emulating PS/2 devices with a PiKVM.
The unknown bus connector looks like a EISA Slot 32bit 16MHz.
Its easy to test ... If a ISA Card Work in it, its EISA!!
As Artheos warned him - if its Opti Local Bus then plugging wrong card will result in escaping magic smoke
@@rasz crazy Shit ... Its the Same connector But electronic incompatible. Also cards Look simulare WTF
This had to be a budget motherboard for the time, and IMHO mostly meant for 386 CPUs, providing an update path to early 486DX and SX. People thought they wouldn't be missing out on the performance when upgrading just the CPU.
The 32bit slot is bit out of place for a budget board
@@phljns that's why there's only one, and even that is not standard EISA, so I'd consider it more of a gimmick than a useful thing. This approach was common for budget motherboards for a long time.
Jack-of-all-trades boards like this tended to do lots of things poorly. The 486 support is probably bugged as you found, but good luck finding a BIOS to run 486 CPUs properly. I wonder if a 386 DX 40 would give a better experience all around. In reality, that is what probably nearly all boards like this were actually used with as they would have been sold to people that couldn't actually afford a 486 system at the time.
That reuse of the EISA physical socket that is even labeled EISA but will literally destroy expensive and rare EISA cards (both then and now!) is just terrible design.
I suspect your problems are related to the fact that it's a multi CPU board and the chipset is neither fish nor fowl nor good red meat. I'd set that the implementation for the cache is screwy/borked in some fashion. Try it without cache with the 486 and with and without cahce wth a 386.
Try measuring ISA clock. It should always stay at 8.3 mhz
Good point. I have a feeling that the frequency changes in certain situations. But i am still confused why system information would change the systems speed
@@bitsundbolts Yes, that's quite odd behavior indeed, but also very interesting at the same time. 🙂 So please let us know of any new findings!
I think you could use a ml2032 inplace of the the usual button cell it's rechargeable and fits the standard holder.
Yes, I have seen those. I never got one of those though. Do you think this would be a better option? Wouldn't they start leaking again in the future?
I don't like ml2032. Yes it's safe and it won't leak.
But also it will die in a few years. You'd have to plug in your computer regularly to charge the battery. If you use the machine too much or too little, the cell dies in a couple years. cr2032 you put it in and forget about it for a decade.
For all classic NiCD batteries, you can get NiMH replacements of stacked button cell type. NiMH is safe against leakage if you want a machine which works like original with regard to battery. Which is to say a little annoying, if you put the machine away for a while, the battery will be flat, but at least it will fully self recover when you actually use it. The endurance and self discharge of NiMH is much better. You can even cheat and install a connector on the board and mount 3xAAA cells off board somewhere in computer enclosure, the Eneloop like type, this works very well actually.
I would like something that would allow using at keyboard with motherboard that needs xt keyboard. Well if you could use usb keyboard with motherboard that needs xt keyboard it would work too
The adapter does support XT as well
the memory would be a bottle neck! but not likely ! I would test it in 386 mode! if you get the same results! the bottle neck is the cpus to "North Bridge" the number of lanes of data thru put!
the ps2 mouse port and key board ports are same eclectically the same just different positions in the ps/2 connector... they both use the same clock circuit same transmit and receive and Vcc and Gnd its just a matter of flipping a couple of wires to making the USB to ps/2 function as a mouse .
I have a similar 386/486 board, the FIC 4386-VC-HD. Paired it with an i486 DX2-66, CL GD5429 ISA graphics-card, 256KB cache and 20MB 32-pin SIMM RAM. It's also a mixed bag, getting a Topbench score of 110, Doom max 13.7 fps, PCPlayer 8.6, speedsys 24.92. Sysinfo works though with a score of 130.7. Your main memory and L2 cache look way too slow. My board has 18.1 MB/s memory and 46MB/s L2 cache in cachechk. The main bottleneck is ISA bus in both our cases I believe. Sadly my Award BIOS doesn't let me change the divider and I don’t know which clock speed it even runs at.
That's very good information! Thanks for sharing. So, we do score about identical in many benchmarks. Except for system information - which seems to work for you.
@bitsundbolts Thank you for the great videos! Glad I could help and provide info. Sysinfo bugging out might be due to different chipsets, my board has a VIA one. Wish I had an EPROM programmer, retroweb lately got an AMI BIOS for my board that might allow to change ISA dividers. If you want to compare any other benchmarks, let me know and I'll try my best to report them back.
Thanks for your offer. I might come back to it. But I also have so many other 486 boards that I need to look at - I think I should have around 7 boards with ISA only.
I wonder if the Forex chipset makes compromises to support both the 386 and 486 bus. I.e. you get 386 bus performance even when you have a 486 installed.
That seems to be the most common suspicion among many commenters.
Surely you tried with a coin cell battery plugged in? Otherwise worth a shot, it's been proven in the past that some boards behave weirdly without a battery
I had an external battery hooked up. I don't have a button cell holder that fits at the moment
I've got a similar motherboard (same chipset soldered AM386DX40) , found that switching from a 80386 to a 486 CPU required the turbo switch to be physically jumpered (JP14 block). My board is the MECER Corp B800. Could very well apply to yours :)
I'm using a jumper on the pin headers. You're right, it is the easiest way to get such a board to its optimal performance.
Edit: does your board score also low on system information (if you would have that data)?
I have recently tested all new solder available in our electronics supplier here and comparing it with my solder that I have bought around 2000 the new ones are garbage.
I absolutely love my solder wire from 1998. , but I must admit, I haven't tested anything new. Maybe it's better that way
What have you tested? Have you tried leaded solders from TME? They carry Broquetas and Cynel.
I have tested only the leaded solder available at the shop. I will not name brands but I can say that there were also german and dutch "made" solder. Even the new batch of the same brand of solder as the one that I have is signifficantly worse. I suspect that it has to do with the metal purity and content of the rosin flux inside.
The solder is 38% Pb. I guess, it is not possible to buy that in the EU anymore.
Bürklin has it but doesn't want to sell it to you by mail. Maybe if you drop by their facility in person, i didn't try. Or if you have a registered business, then they sell it to you no problem.
TME might be easier, it's hidden if you're browsing from Germany but if you have a link i heard you can buy it. They carry Broquetas and Cynel.
Eleshop will just sell the same Broquetas leaded solders to you from the Netherlands.
Id love to have a wireless keyboard on the bench. One less cable . But i would want the dongle small without a cable. If you know what i mean. AT keyboard to usb dongle. Love the videos , keep up the good work.
We could change the adapter to plug directly into the PS/2 port without a cable. But I have to check with the creator. All suggestions are welcome!
just fresh from watching this video and my first thought is chipset. some bug or limitation in the chipset. or it could just be a really crap board. they did make a ton a crap back then
Haha, maybe... Unfortunate because it would allow testing 386 and 486 on the same board. Maybe the 386 performance is ok, but they dropped the ball for 486.
Did you check to make sure the keyboard adapter wasn't the cause of the weird performance issues? Probably not but best to rule it out. My guess is it's probably the bios itself, and if there even is an updated bios, it's probably unobtanium.
Good point, although unlikely, I will try run the SI benchmark with a different keyboard.
Can you sttill buy leaded solder?
That's an EISA slot.
He covered that. No, it's not, and it was the worst connector that could have been chosen for this _proprietary_ interface. At the very least they could have spun the socket around 180 degrees so the notch falls in a different place.
@@mal2ksc I had an 80386DX motherboard with EISA slots, and the connector seems identical.
One should be able to colour match the mask by mixing these Mechanic UV mask resins, but i haven't tried.
The application looks thick, but i know it's difficult to apply thinner. I have had to apply the mask much thinner for it to actually successfully UV cure. But then i don't do sun :D
Oh also i noticed recently Relife has a clear solder mask which they promise is much quicker and easier to cure. Haven't gotten any myself yet.
Most of my keyboards (not the mechanical one i use day to day on my computer) and quite a few mice have internal PS/2 functionality and work with a passive USB to PS2 adapter, and i thought everyone had it like this, because they're extremely cheap and widely available. But i guess if you want to use your favourite input devices, wireless ones and the like...
I had this multiple times that the solder mask cures outside, but is still liquid under the surface when using the UV light. When it's under the sun, it cures properly.
I am wondering if the 386 BIOS may be causing/creating an issue. Perhaps you could configure it with a 386-33, if you have one, and run some tests on that. If the performance is as expected, for the selected 386 CPU, perhaps a new BIOS will be in order, should you choose to reinstall the 486.
I take it this era of systems can't actually use more than one socketed CPU at a time? Would be pretty interesting if somebody could build a system that could somehow utilize multi CPU.
Yes, those boards will only work with one CPU at a time. Think of the memory access for instance. Those memory controllers cannot accommodate two CPUs requesting access to the memory, or ISA slots. Most of the traces go directly to the CPU socket. I have another 386 board I need to work on. There are almost no other components on the board except for the CPU.
Try it with a real din keyboard. Maybe it has something to do with it. Also try a 386 maybe? Does it underperform with it as well?
I tried building a ps2x2pico twice and failed both times. I threw in the towel and I'd love to buy a prebuilt solution that accepts USB mouse and keyboard. I use a USB switch and would love to hook it up to one of these if it supported both mouse and keyboard.
The adapter supports mice and keyboards. At the Moment, you would need two devices. That is a problem if you're using a dongle that acts as a USB hub to transmit both, mouse and keyboard signals.
That is a VESA slot, and I believe the second slot is for a math co processor.
Both wrong, it's a proprietary slot that was (stupidly) built from an EISA slot physically, and the second socket is for a 386 as covered in the previous video on this board.
I am a little surprised there is enough current to drive everything off of the keyboard circuit.
Yes, the adapter takes its power from the port on the PC port. That port also needs to power the keyboard - so, there is enough juice for the little adapter as well.
Did you put a heatsink on the CPU in the end? How important are the heatsink instructions on the chip? Do you usually get away without a heatsink even on a chip that says it needs one?
Yes, I used a cooler. I just put a socket 7 cooler on the CPU - not clipped anywhere since this socket doesn't allow for it. The CPU gets hot and the heatsink warms up quite a bit. I think you should follow the instructions. Those CPUs should have a heatsink and airflow
Is System Information somehow staying in the memory or does it have an unwanted passenger in the code?
I’m interested in the usb/ps2 adapter. It would make much easier and flexible using modern mice and keyboards. I have some stocked, the keyboards are ok, mice are from Aliexpress, I would rather use some Logitech. It’s getting more difficult to get PS/2 hardware and eventually is gonna get expensive. I would buy a couple of this adapters.
I'll make a dedicated video about it and show off the capabilities of this adapter. The creator is still ironing out a few small things, but the prototype is working pretty well!
Looking forward to the review. It will be a great addition to the retropc community!
A bit surprised that you are based in The United Arab Emirates. You sound like one from Germany. Didn't know that retro computers are easily available in UAE. Is there a community for old electronics and computers in UAE?
I moved to Dubai almost 20 years ago, but I grew up in Germany, went to school, and left right after that. The retro community in the UAE is quite small. I just know a few people who are into old retro PCs. If I wouldn't have been introduced to the scrapyard that receives old hardware from time to time, I wouldn't be able to make those videos.
Opti-chipsets are just soo sloow... I have 386/486 hybrid board with VLB and it's fast with 486DLC CPU but super slow with 486. Memory bandwith is miserable, whatever timings are used. So, try a 386 CPU 😅
Interesting. So, it might just be an issue with that board and it may be better to just use it with 386 CPUs
@@bitsundbolts It might be a good match for the AMD 386/33, which most likely will happily run at 40. That should offset some of the shortcomings of the board.
Opti chip here is a 206 = peripherals, nothing to do with CPU/RAM communication thus no influence on system speed.