I just wanted to share my success in building the Xi 8088 for those who are hesitant about taking this project on. I did have some stability issues as first, but after building SMBaker's active termination board, it has been soak tested on turbo (8 mhz) for hours without hanging once. A few observations or things which I think helped:

I noticed in another forum thread about the project that Sergey emphasised using F series logic for U15 to U23. Although ACT are mentoned on the build page and can work, it seems like F is preferred.

When building the termination card, I followed appendix B of this document to determine which lines should be terminated. Basically, that means isolating legs on the resistor packs for all IRQ and DRQ lines, plus I/O CH RDY. I used 220 ohm packs.

There is a newer version of the BIOS on Sergey's github which you can assemble. While most of the improvements relate to the Micro 8088, there are some which affect the Xi 8088 so it is worth doing and you can inject the latest XT IDE universal bios version while you're at it.

I don't have a ton of cards plugged in, so it's possible I could start to have issues with all slots populated. I have no use for upper memory or adlib support though.

So the termination is simply 220R to +5V on all IRQ/DRQ lines and IOCHRDY?

maxtherabbit wrote on 2022-05-14, 14:23:

So the termination is simply 220R to +5V on all IRQ/DRQ lines and IOCHRDY?

No, the other way around - I terminated all lines on SMBaker's pcb except those. This is the active terminator board regulated to 2.7v, using 220 ohm packs. Anything that said NC on the table in the linked pdf, I cut the leg off the resistor pack and covered it with electrical tape.

SanguineBrah wrote on 2022-05-14, 15:37:

maxtherabbit wrote on 2022-05-14, 14:23:

So the termination is simply 220R to +5V on all IRQ/DRQ lines and IOCHRDY?

No, the other way around - I terminated all lines on SMBaker's pcb except those. This is the active terminator board regulated to 2.7v, using 220 ohm packs. Anything that said NC on the table in the linked pdf, I cut the leg off the resistor pack and covered it with electrical tape.

I have no idea what "smbaker's pcb" is or where to look for it so I'm blindly guessing here

maxtherabbit wrote on 2022-05-15, 03:00:

I have no idea what "smbaker's pcb" is or where to look for it so I'm blindly guessing here

It's mentioned earlier in the thread but I guess it's a lot of pages now. Link

My Xi 8088 is alive! (again) My Xi 8088 build worked like a champ, but then after a year or so it wouldn't even post - no activity on the bus at all. So it sat for years. Recently I took another shot at it... The soldered joints "looked ok", but for kicks I redid them ALL. Plugged it back in, and booted right up. Go figure, I knew my soldering skills were lacking. If it helps anyone struggling - double and triple check your solder connections and don't trust "it looks ok"!

Anyone have a creative idea for a case for your build? Right now it's in a beat up 30+ year old beige case but I'd prefer something to show it off. I was thinking maybe one of the open motherboard concepts. Anyway, just curious what others are using and if there are any active builds of the Xi 8088 going on. It's quite a challenging build, my next step is to get off using the floppy and install an isa-cf card.

I just updated the BIOS version from 0.9.0 to 0.9.9 and there is now 16 second delay before the machine posts. Is this normal? EGA card being used. None of the dip switches or presence/absence of cards seem to affect this behavior.

Hello guys! can anyone recommend a case to mount micro8088?

bazingaa wrote on 2023-06-26, 10:37:

Hello guys! can anyone recommend a case to mount micro8088?

What type of backplane are you gonna get/use? this determines the size of said case. or you could even make your own 😁
I got a original IBM 5160 case (sold for "sleeper" builds). current backplane only has 3 isa slots and no power, kinda have to hack it anyway. miniATX towers can be cute 😁

CodeAsm wrote on 2023-06-26, 13:16:

bazingaa wrote on 2023-06-26, 10:37:

Hello guys! can anyone recommend a case to mount micro8088?

What type of backplane are you gonna get/use? this determines the size of said case. or you could even make your own 😁
I got a original IBM 5160 case (sold for "sleeper" builds). current backplane only has 3 isa slots and no power, kinda have to hack it anyway. miniATX towers can be cute 😁

I have Segey's backplane, https://github.com/skiselev/isa8_backplane
I was thinking about getting a microATX case but my worry is that only got 4 expansion slots and it would be full for external interfaces.
May be full ATX with 7 slots would be useful ?

I plan to use below cards.

CPU - micro8088
VGA - Trident 8900CL
Sound - PICOGUS OPL2/GUS
NIC - I have few ISA 16bit ones, need to check compatibility
UART (+USB mouse) - may be this https://github.com/hkzlab/ISA_DualUARTParport + https://github.com/LimeProgramming/USB-serial-mouse-adapter
XTIDE - May be this https://github.com/glitchwrks/xt_ide/tree/rev_4 (I have PCB for this, but not sure about soldering CF slot https://github.com/monotech/monotech_xt-cf-mini)

I only found one IBM 51xx cases on ebay and it is $200, I am not from USA and they are pretty rare here.

If anybody is building a Xi 8088 and has been having an issue dialing in the frequency of the OSC crystal, I have a suggestion. On both of the boards I have built, adjusting C54 would give a much higher frequency than required, making it impossible to get to 14.31818mhz. My assessment is that this is a result of the load capacitance being too low on one side. While the values of ~33pf on either side of the crystal (C52, C53) are calculated correctly, the addition of the trimmer in series with C53 significantly drops the capacitance on that side of the crystal. I found that swapping C53 with a 100pf part, and C54 with a 60pf trimmer together yeild something in the ballpark of the 33pf and make it possible to hone in on the required frequency.

Moogle! wrote on 2023-05-06, 03:41:

I just updated the BIOS version from 0.9.0 to 0.9.9 and there is now 16 second delay before the machine posts. Is this normal? EGA card being used. None of the dip switches or presence/absence of cards seem to affect this behavior.

It might be it looking for a PS/2 mouse, I have a similar problem with my xi8088 derived XTjr 8088 based machine. Try plugging in a ps/2 mouse and see if it boots faster. This pause used to trip me up a lot on my custom build when testing, I kept thinking I'd broken something! I mean to edit my custom bios to see if I can reduce that delay...

Having completed a second Xi 8088 build, I have more to add. I encountered a lot more stability issues this time. It seems like there was a big jump in instability when plugging in older ISA cards using lots of TTL chips. For example, stability was fine using a Trident VGA card but swapping to an original IBM CGA card caused a lot of faults.

I also found that the choice of CPU made a big difference in stability. I have had major issues with 3 different Oki 80C88-2 from a variety of sources (some likely legit, some questionable) but then switching over to a V20 or what I am using now, which is a Fujitsu MBL8088-1, caused a significant stability improvement. It is tempting to chalk this up to supply issues but I am not so sure.

When experimenting with SMBaker's active termination board, I found that stability was better with the voltage set a little lower (2.5v vs the 2.7v mentioned in the blog post). It is possible that more tweaking or different termination resistors could improve things further.

Curiously, I have found that clock frequency plays no role in the likelyhood of errors - all tests are the same in turbo mode (8mhz) as they are at 4.77mhz. I am at the stage now where stability is rock solid with the CGA card, XT IDE and Sergey's floppy/serial card but if I add a vintage parallel/serial card, Checkit's RAM test will throw up errors after an hour or so.

Good day !
First, I'm showing off a little: I finally assembled Kh8088, which had been working for a long time, into a case. For some reason, I love 19" cases)))
In the photo: Xi8088 by skiselev, CGA Redux by hkzlab, OPL2 and FDC by skiselev, VCF XT-IDE Rev 2 and ISA Backplane by Alexandru Groza.
However, there were not enough free slots to add a network card and and EMS card. I will have to do a little more metalwork to move the COM2 and CF card to a separate panel and free up two slots. But that is an optional option.
But there were several shortcomings that I (perhaps) did not pay attention to earlier, or they did not appear earlier.
1) During a "cold" start, the Xi8088 "hangs" at POST 08h for 20 to 40 seconds. Then it boots normally and during "hot" restarts this problem does not occur. I already found information about this problem in other discussions. There, as a reason, they indicated the need to use 82c59 of specific manufacturers, namely AMD. Unfortunately, I only have a Toshiba 82c59 at the moment. I have already ordered several sets of different 82c59, even including the ex-USSR KR1810ВН59А - so I will experiment. But maybe someone has already found other ways to solve this problem.
2) Periodically (quite often), 1.2 MB and 1.44 MB disk drives do not work well. That is, they work - write, read, copy from one to another. But when trying to format the diskette, the most various errors are issued. It is sometimes possible to make a system diskette by chance in one case out of ten. And then after that it is not always possible to boot from it. This is definitely not a disk drive problem or a controller problem - everything works fine on any other motherboard. Alexandru Groza wrote about the same problem with his 386SBMC project - and yes, I also faced the same problem with his 386SBMC. Alexandru Groza believes that the matter lies in the timings on the DMA. I have Xi8088 almost completely assembled on K1533 (74ALS), and maybe it is worth trying to replace Page Registers and X-Bus Transceivers with 74F? - simply, apart from the incorrect operation of disk drives, no other problems arise...

mogwaay wrote on 2023-07-27, 21:42:

Moogle! wrote on 2023-05-06, 03:41:

I just updated the BIOS version from 0.9.0 to 0.9.9 and there is now 16 second delay before the machine posts. Is this normal? EGA card being used. None of the dip switches or presence/absence of cards seem to affect this behavior.

It might be it looking for a PS/2 mouse, I have a similar problem with my xi8088 derived XTjr 8088 based machine. Try plugging in a ps/2 mouse and see if it boots faster. This pause used to trip me up a lot on my custom build when testing, I kept thinking I'd broken something! I mean to edit my custom bios to see if I can reduce that delay...

I did something and it randomly stopped doing this. Don't remember what though. 😒

Hi everyone,

I've recently completed my build of the Xi8088 and it's been a great experience overall. After some initial issues I managed to get the system to be very stable under most conditions - CheckIt passes all tests (aside from the known DMA channel 0 issue), and extended memory checks also run without any problems. However, I've run into a strange issue with digitized audio playback and I'm wondering if anyone here has ever encountered something similar.

Using either WSS or Sound Blaster digitized playback (via DMA) causes immediate system instability. Sound Blaster playback produces screen artifacts, followed by a freeze. This is especially noticeable in Prince of Persia, where some random pixels appear on the screen during the playback. WSS playback causes an instant freeze or reboot, sometimes preceded by a corruption of the screen contents (lots of random colorful blinking characters appear). I'm testing the WSS playback using my own C program to play WAV files, it works perfectly on my other, 386 machine. The source code is available here if someone would like to take a look: https://github.com/Lefucjusz/WAVWSS

What I've already tried:
- Adjusted wait states — no effect.
- Replaced glue logic chips with various families: 74S, 74LS, 74ALS, 74AHCT - no improvement.
- Changed 8288 from MBL8288 to SAB8288 — no change.
- Replaced DMA controller from M5M82C37AP to NEC D8237 — no improvement.
- Tried running in both 4.77 MHz and Turbo mode — issue persists in both.
- Tested four different sound cards: CX4235-based, YMF719S-based, OPTi931-based and CYF S2801 - all produce the same result, instability and sooner or later crash.

One more thing that I noticed is that floppy transfers (which use DMA too) appear to function correctly, but if I move the mouse in Norton Commander during a floppy transfer, I occasionally see artifacts (random characters) on the screen (though the system does not crash in that case).

I wanted to ask whether anyone here has ever successfully used digitized DMA playback (either WSS or SB) on a Xi 8088? I'm still fairly new to retrocomputing, so it's entirely possible I’ve missed something obvious or am unaware of a well-known limitation.

After many long nights with the logic analyzer and oscilloscope, I finally found the root cause of the issue - and to my surprise, it turned out to be the graphics card.

I was using a generic Trident TVGA9000i-based card that I’ve had for years and used in other machines without problems. Since the PC booted normally, displayed video correctly, and all software appeared to work fine, I didn’t initially suspect it at all. Plus, the same combination of graphics and sound cards worked flawlessly in my 386 system.

The issue is specific to the PC/XT and is caused by a known quirk with the DMA implementation in such systems: the ISA ALE signal is incorrectly pulsed during DMA cycles, which can confuse some cards that will start behaving unexpectedly. In my case, the graphics card seemed to respond to the ALE signal during a DMA transfer, placing data on the bus at the same time the RAM was supplying audio sample data to the sound card. This caused bus contention - effectively, the graphics card and RAM were "fighting" over the data lines. One chip would drive a line high while the other tried to pull it low, creating invalid, non-TTL logic levels and, likely, corrupting data in RAM, which then caused system freezes.

Once I spotted this, I compared my card to Sergey’s ISA Super VGA design (which uses the same Trident chip), and noticed that in his design, the ALE signal isn’t connected to the ISA bus at all, it’s tied to +5V instead.
So I tried a crude fix: I masked the ALE pin on the graphics card’s ISA connector with Kapton tape. I booted the PC, ran some software… and it all worked! No more crashes, no more graphical artifacts - problem completely gone. Apparently, this is a known issue with some ISA VGA cards in XT-class systems, and even Sergey has mentioned this fix in some forum post, which I unfortunately hadn’t seen earlier.

So, to anyone wondering: yes, digitized audio playback using DMA does work perfectly fine on the Xi 8088. It is even possible to play CD quality WAV files through a WSS-compatible card, which I initially doubted!

very interesting. I think I too was using a trident 9000... I dont think I had any other vga card to try at the time. I might have to go back to it and see what I can find out.

Simple solution: Use a CGA card! 😁