Too bad I don't have an ODPR75 to compare with. They have 8k of cache. Was there ever a DX4-100 with 8k?

@@Epictronics1 most AMD dx4 have 8kb, yes

@@Dxceor2486 Perfect, I'll see if I can find one and do a comparison

@@Epictronics1 beware though, even though this is very rare, you can sometimes get a 5x86 rebranded as a dx4-100 with 8k ... Which has 16k It happened to me once. In this (very rare) case of putting a dx4 with "8k" of cache and you see 16k, test it in another system just in case

@@Dxceor2486 Ok, thanks

I think you're on to something there

A DMA channel conflict was exactly what I was thinking as well. Wave playback on the SB cards need their own DMA and IRQ while the base address is primarily used for the Adlib/OPL music playback or for addressing the MIDI/controller port. A DMA channel conflict can also cause the system to hang or act a little screwy in some cases.

@@Choralone422 Thanks, I'll make a fresh install on another system and do some more tests with this card

I was thinking the same thing. Easy test is to go into the BIOS and disable all serial and parallel ports.

By the sound of it - it is not a DMA conflict. More like an IRQ priority problem. IIRC back at that times there was such a thing as an IRQ cascading with IRQ #0 being assigned to timer, IRQ #1 to keyboard and IRQ #2 assigned to another PIC handling extra 8 IRQ channels from #8 till #15. If you had any device sitting in this 8..15 range doing intense bus transfers it would cause issues with that IRQ priority being higher than sound card IRQ channel, no matter it being 5 or 7, and as a consequence sounding like SB does sound in this video. Problems like these were most painful in early PCI era when your PCI A/B channel for the slot your PCI SB card was sitting was PNP auto-assigned to IRQs channel having lower prio compared to SCSI or IDE controller. Any burst of disk activity would result in sound artifacts like these. Thus what I'd do to troubleshoot the issue here is to disable onboard disk and video controllers if possible and try using external IDE and SVGA cards to check if it helps. Can't make any promises though as I'm writing all of this from the memories I've got aging back to 90s-00s and it was, you know, quite a long time ago.

Yes, that is correct. The next step is to figure out the clock circuit. Thanks

@@CobraTheSpacePirate Thank you :) I think you are on to something here... There is no J37 on my board. However, while looking for it online, I found a very similar board that was sold with a 33MHz CPU... I'm waiting for parts right now, but there will definitely be a follow up on this project :)

@@Epictronics1 25mhz might be why your sound card isn't working properly. The ISA slot might be on 25/2=12.5mhz - which many cards could handle. 33.3/3=11.1mhz and that might be all the difference needed.

​@@Epictronics1 looking forward to your follow up. Your PS/1 videos are my favorite!

Thanks

I couldn't resist : )

Thank you : )

he does have nice camera work and editing.. but his vocals and audio fullness like narrating performance needs work. At least it isn't CoreTek which I use for white noise to fall asleep too.

@@naikjoy Thanks for the feedback. Could you be more specific? I'm still learning

@@Epictronics1 uhmm Idk how to describe it. You know like if you watch an interview with Chris Williamsom and Andrew Huberman about some neurobiology and their voices have a bit more umph to them..its not like its base and low frequencies much. I mean it is but its subtle and not all in your face / ears.. but like its pleasing to listen and more into you..more forward... and yea its nice.. I think the same is with LGR vids where his background lounge soundtrack are nicely in the back and his cheerful voice which its output its probably a bit more louder talking as if he was in a bar but at 7 pm... and like.. you can hear him nice loud and clear and cheerful... very good presentation and vocals and the sound quality its nice too. I think this now comes more to comparing how other podcasts or youtubers do it and how to replicate their quality on the cheap.

@@Epictronics1 but yes another thing is.. this is why PhilsComputerLab because of how his voice and how he narates and how pleasing is to listen too and his accent isn't as great as LGR.. this is why he is only at 133k subscribers.. plus very blend sense of style old fashioned thumbnails n shit like that. I think the best dude to watch and learn from..like by watching and recreating his style..reverse engineer his style is OptimumTech videos. Now that youngster does a good job. His camera work and editing is nice.. his sense of style is nice.. he is quite a cute looking sturdy man.. BUT his voice could be better more manly with a better presence.

Totally agree :)

Thanks :)

Thanks : )

It's an awesome game :) Thanks

That's very helpful, thanks. I'll give it a try

IRQ 2 is cascading to the second PIC that handled IRS 8..15. I.e. IRQ 10 signals itself through IRQ line #2 to the CPU. If you've got a choice between IRQ 5 and 10 - 10 is better as it is higher priority.

I had a look at his channel but didn't see any videos about it. Hopefully someone who reads this and knows what software that is will let us know

Thanks. I guess we need to find out if it can do 33MHz

Doom is capped at 35 fps.

@@gameblabla Thanks

yeah, I went on a cleaning spree and found that nasty virus on many of my old diskettes and hard drives. I think I spent two full days getting rid of that damn thing :)

​@@Epictronics1 you should put in the fastest math co-processor in this. DOOM & other programs run faster with a math co-processor.

@@andrewsadach3194 There is NO floating point math in DOOM - QUAKE was the game that pioneered use of the Floating Point Co-Processor. The reason DOOM runs better on some DX chips is because of Cache Size, Cache Policy [write-through or write back], & clock speed. Almost no Software used a Math Co-Pro except Lotus 123 & AutoCAD until QUAKE.

@@sionfiction9566 DOOM doesn't use the QUAKE engine. It was made using a variant of AutoCAD. All 3D DOS games used some form of AutoCAD back then, and combined them with BASIC programming language. That's what the DOOM game engine was compromised of.

@@sionfiction9566 before Blender, Unreal, and Unity... Everyone used a AutoCAD variant for 3D games.

You were lucky then, very nice machines : )

haha, I guess I should check the solder joints

We'll solder some more on the MOBO this week :)

Thanks, I'll do some more tests with different DMA settings

Or there is some trouble with the PCM parts of the board.

Thank you : )

I'll check, thanks

Totally agree

I'll make a comparison ones we have 3v on that board. That will give me more CPUs to choose from for tests

Thanks :) The caps looked a bit funny on this board. I better check

Thank you : )

@@Epictronics1 you are welcome

And other machines would have flourished more. Looking at Amiga machines...

That is a very good question. I feel bad for IBM for having lost it all. At the same time, the clones made computers accessible to more people and that is always a good thing

@@josedias5514 Yes, I can't quite understand why Commodore lost the game having the Amiga. looking at the market today, Apple and PCs coexist quite nicely on the market. Why wasn't there room for the Amiga?

@@Epictronics1 Comodore lost out because they were comodore. Ibm didn't loose out even making the same mistakes but they're Ibm. Apple didn't lose out because of the will of Steve Jobs but comodore didn't have any aces up their sleeves. The amiga's were ahead of everyone else but couldn't shake the low value ghost. A shame it was.

@@josedias5514 yeah, It's a shame really. I grew up with a c64 and I'm still a big fan

It sure is : )

Sounds like a fun project. Do you know if they are pin compatible?

I'll do a test

That's a possibility. Hopefully someone who reads this and have an original set will let us know

I found the "disable onboard CPU Hack" on Vogons years ago. Otherwise, I don't think there is one.

I am not aware of any tricks to solve this yet. I'll do some more reading and I'll figure it out. The ideal solution would be to adjust the refresh rate on the CRT with software to match a camera setting. Thanks

@@Epictronics1 except you can't adjust the refresh rate in the BIOS screen. That's most likely hardcoded to be 640x400 @ 70Hz. Good luck getting a camera to sync up with that. Easy solution. Not period correct, but cheap and easy. Use an LCD. Preferably with a period-correct aspect ratio of 4:3 or 5:4. Those should be cheap and somewhat easy to source on the used market still. I just saw one (17" SXGA) going for $5. That will eliminate any camera-based flicker.

@@SeeJayPlayGames I've got a couple of those but I try to keep things period correct as much as possible. I'd be happy if I could just get rid of the flickering in DOS and Windows

@@Epictronics1 unless you can adjust your camera's shutter speed to not flicker at 70Hz... nope. LCD = easiest. Some modes are 60Hz and wouldn't flicker, but a lot of them are 70. Mostly the 200/400 line modes such as those used by the BIOS and command prompt. Windows, you can get running at 60Hz most of the time. Although there are some modes that are like 56Hz or 72Hz or even 87Hz interlaced. This is really only a big problem with IBM compatbles and Macs, not earlier home computers, as most of those could be connected to a television.

@@SeeJayPlayGames Yes, I have noticed that my older machines don't produce much flicker. There are cameras that can be set to odd settings. I will probably get one of those when it's time to upgrade

Thanks :)

There is a missing chip very close to the ZIP socket. Maybe that is causing the problem? Unfortunately, there is no documentation about the socket to be found online

Cê tá bem identificado irmão kakakakak

Hmm... do you mean that the system might require that extra soldered chip to put the ZIP ram in use? We can try that

@@Epictronics1 from what I know from having one of those VLB cards long ago now that it has the other half of a megabyte it should be able to add the last ZIF chip with 512Kb or 1024Kb, though this will hinge on the BIOS being the same

@@wskinnyodden Thanks

Thanks Tony. I was hoping to find out with this project, but I don't seem to have the correct ZIP chips to do the test. Unfortunately, I can't find any documentation. Also, there is a chip missing near the ZIP. Maybe that's the problem?

Max video memory I ever found on a ISA video card was 1M but on a VLB 2M were supported (mostly upgradable or on high end cards). 2M+ were on PCI bus (2M or 4M on high end cards). So, my tought is that the soldered chips are meant to be 512k (low end) + 512k (low end factory upgrade) + 1M socketed memory chip (optional upgrade for high end). Maybe Epictronics check a chip in that socket now, after soldering that memory chip? (for missing pinout, just identify Vcc and ground pins to find the correct orientation)

@@sebastian19745 I'll try it out, thanks

@@Epictronics1 Note that socketed ZIP RAM chip only takes 40-pin 80ns sticks, and that seems to be what you used. It has a datasheet that gives a -1 to 7 volt CC and in/out rating. If either stick works in anything else, I say maybe test the socket for voltage, if the test Sebas suggested fails.

Ok, I think the manual just said to put an extra chip in that socket. I'll double check, thanks

@@Epictronics1 While I've never seen it myself, there may be some motherboard designs that have the TAG circuitry on the motherboard, rather than requiring a special TAG SRAM chip, and it just uses a normal SRAM to store the cache entry table. As you said, you'll need to consult the motherboard manual. Hopefully it's clearly stated whether it has on-board TAG circuitry or not. But it's more common than not that the motherboard needs the special TAG chip in the dedicated TAG socket, and the correct sized chip.

@@GGigabiteM I'll check, thanks

@@GGigabiteM To my knowledge, I've never run across a motherboard that requires anything special in the tag socket. It's just another SRAM chip. It may not be the same capacity, like you said, depending on the amount of cache installed. But, I've always treated it as interchangeable when they all call for 32K (e.g.) and that's never been a problem. It never even occurred to me that it would be any different.

​@@nickwallette6201 TAG SRAM is NOT another SRAM chip, it has comparator circuitry for hitting cache entries. YT shadowbans posts with links in them, so you'll have to look these parts up yourself. Here are just a few TAG SRAM chips from different manufacturers. SGS-Thompson MK48S80-15 Integrated Device Technology IDT 7174S Logic Devices L7C173TC-15 You can have all of the cache you want, but it's not going to work properly without a TAG chip, because the TAG chip tells the system when there's a cache hit. If no TAG is present, the cache will fill up, but it will either never be used, or have erratic behavior from OE2 toggling at inappropriate times. On TAG SRAMs, pin 1 and 26 are different from normal SRAMs. 1 is RESET/CLEAR (clearing cache entries.) Pin 26 is MATCH, which is used to determine when a cache hit is detected. On normal SRAMs, pin 1 is normally N/C (not connected) and pin 26 is OE2 (output enable 2.)

I'll check, thanks

It shouldn't. The digital I/O is all down at 0x220, with the end of that range supporting the CD-ROM interface. The OPL chip can be accessed in the 220 range, or at 0x388. Neither will conflict with the parallel port at 0x378 or 0x3bc. IIRC, 0x3bc was the original address for LPT1, used back in the PC / XT days. 0x378 became normal later.

Thanks! Yes, I used a TL866 plus. Good luck with the project

Thanks, D48 on order so going to take a backup of the original rom and take grub there.

@@Epictronics1 Thanks again for your help with the name of your programmer. I took a dump of my 2168's BIOS ROM (12/05/93 / 52G2974), but alas so far not been able to get PODP5V83 to post even with adjustment to j23-j25. I think my model just must be too old compared to the release date of the overdrive

I tied this, I also have the blue overdrive socket on my 2168 and installed the 83MHz pentium OD. Yes, it will not POST. We need BIOS FRU 76H6938.

Thanks, that's an easy test. I'll give it a try

I was thinking the same thing, and the parallel port is normally the culprit. For extra safety I would also disable both serial ports. I've had a sound card share a parallel port before without any issues except I couldn't print and play music at the same time. I'm not certain but the PS/2 might be sending bus data to the parallel port even though the port isn't selected. Many early machines did this because actually disabling the parallel port required extra chips. Setting the busy signal on a parallel port would tell the printer to ignore anything you "see" being sent.

@@therealjammit I'll give it a try, thanks

It's not uploaded anywhere. Any suggestions for a suitable place to upload? Does your board support the 3.45v required for the 5x86? If not, you may want to watch one of the upcoming videos :)

That's what I'm going to try next :)

Sounds like a bad chip to me. I'll see if I can figure it out

I'll see if I can find it, thanks

It's not available online. I have sent an email to the IBM PS/1 Resource Page to see if they want to host it. If they do, I'll post a link in the description of this video.

I would also like to get a copy of that BIOS for my ps/1

@@natasegde I'll put a link in the description ones it's available somewhere for download.

I'll double check

awesome, good luck

@@Epictronics1 any chance you can direct me on where to buy one of these flashed ROMs or how to flash my current one?

@@rjwelchiiiable Sure. All you need is a programmer. I use a TL866 but there are others. The file to the program is linked in one of the PS/1 videos.

@@Epictronics1 Thanks for the info. That may be out of my scope for now. It's beyond what I know how to do currently. Any chance you have any for sale?

​@@rjwelchiiiable Unfortunately no. You could check with your local vintage computer club. Someone there might have a programmer.

We better get that bus up to speed then : ) Thanks

Hence the system.ini file, which is a windows config file. I was thinking the same thing.

It got lost in the edit. I did the settings in Win unfortunately

Yep. Certain (most?) versions of the DOS configuration utility would write to system dot ini as well, if it had been told Windows was installed. But it's possible that some don't. I don't remember having to edit the Windows configs by hand much for Sound Blasters. Certainly for other cards, though. I think the Drivers control panel applet had a settings page for the Wave device driver that exposed those settings in the GUI.

Yes, that is my next goal, to figure out how the clock chip works. Maybe, it's hackable too

@@Epictronics1 I'm thinking there's a 25MHz oscillator on the board that needs to be swapped out with a 33.333333Mhz version. There may also be a divider circuit to keep the ISA bus frequency in spec, so you may need to find a 1/3 divider and replace it with a 1/4 divider, may be a jumper, IDK.

My memory of this stuff is all a little... not good, but here's some things I THINK I remember that might help... 486 bus speeds were 20, 25 & 33Mhz [There may have been a few 40 MHz Motherboards later in the 486's life]. As I recall Clock was typically set either by Jumper or in the Bios - you shouldn't need to fiddle with crystals or anything. The DX2/4 & SX2/4s were the first 8x86 CPUs with a disconnect between Internal & External Clock speeds. DX2s were clocked doubled & DX4s were actually clock TRIPLED - So a DX4 100 runs internally at 99Mhz while the Bus runs at 33. I could be wrong but I do seem to remember also that changing the Bus Speed also changes the ISA slot Clock speeds & anything plugged into them. So if your SoundBlaster 16 was an early model it might not like what is essentially an overclock. Interesting project Epictronics, brings back memories from when I used to work on this kind of stuff.

@@sionfiction9566 Thanks. I hope they are good memories :) Yes, we might get into some compatibility issues when we start clocking the board

ok, thanks, I suspected something like that. This will narrow it down when I do the troubleshooting

@@Epictronics1 yeah it's like 220 is one piece of hardware and 330 is another piece of hardware both on the same board. (I think those numbers are right. lol)

@@Epictronics1 In fact the early soundblasters (probably this one) has a yahama chip (I think it's yamaha) that is the same chip on an adlib card and a separate chip that's for wave. I believe later on they combined them, but inside the silicon it would be 2 completely different sections of the silicon.

@@awilliams1701 That is correct. This SB has a real Yamaha OPL-3 (edit)

@@Epictronics1 Pardon? Adlib is OPL2. SB16 is OPL3, which is software backwards compatible but not electrically compatible. One potential trouble is that the Soundblaster chips still do address decoding for the OPL, making them not super super separate; but also given Adlib mode works fine, that is something we can disregard for the time being. Also OPL subsystem doesn't need any IRQ or DMA, it's purly IOP, which makes it much simpler.

It's possible. I have another system that I bodged about a year ago. I haven't seen any issues yet. If it starts misbehaving, I know what to check, thanks

Things got really weird back then : ) Silly mistakes were made

@@Epictronics1 Oh I know, I remember when IBM lost billions due to losing their monopoly in the PC market. Once the BIOS was hacked/copied/reverse engineered (take you pick), the gates were opened for 'IBM-compatibles'. BTW, hi Mac. 😁

@@JenniferinIllinois Bad-mannered Mac Harvard says, Hi : )

I damaged quite a few chips and traces with chip pullers before I finally found this wonderful tool : )

​@@Epictronics1 I have quite an opposite experience. When you leverage that "tool" against that square chip it's just a matter of time when its 30+ year old brittle plastic will crack. Well, it's your life but this kind of "repair" videos give chills to many retro enthusiasts for sure.

@@fffUUUUUU I thought you were joking? I use to use chip pullers, but they damaged pins and traces. Then I found this tool and I haven't damaged a chip since then?

I'll make some new :)

@@Epictronics1 Thanks, I got one recently and I was going to use your videos as reference to upgrade it. :)

@@KhalidYousif87 Ok, you can use the 2168 videos. They have the same MOBO :) Have fun!

@@Epictronics1 thanks! :)

I was wondering the same...the ps/1 video playlist was my favorite.

I got a memory error when I tried to start Doom with 4MB. I didn't have time to check the details.

@Mr Guru I had 1.9. While I agree there are different versions and ems and xms memory was a huge struggle during the time it's possible to run on 4. I can also appreciate that he didn't want to fool with it and just and just more ram. At $50 a mb for ram in 1993 it wasn't as simple at the time and you made it work when you either run what you brung or didn't play.

Thanks for the offer. That project is too big for me. Have you tried to post a thread on the EEV blog forum?

@@Epictronics1 repairing just the motherboard I mentioned is too big? Okay, thank you for taking the time to reply

How about a 5x86-100?

@Epictronics1 also good2go

@@Phil-D83 We might try it soon :)