x86 card revival thread

taddy said:

Anyway, I configured it with Q610 PDS + Houdini II + Cyrix 5x86 100GP.
Cyrix 5x86 100GP is an option from reply.

Click to expand...

Is it working? I couldn't get my AM586 working in my 630.

Phipli said:

Is it working? I couldn't get my AM586 working in my 630.

Click to expand...

The standard configuration of 6100+Houdini II works,
but I have never been able to get it to work with the Cyrix 5x86 100GP configuration.
So I don't think it will work.

However, even if it is not working, the panel is getting very hot,
so there is still a possibility that it can work if the cooling is strengthened.

Does it work with the DX4-100Mhz-ODP configuration?
If it doesn't work, I'll revert to standard.

taddy said:

Does it work with the DX4-100Mhz-ODP configuration?

Click to expand...

Yeah, my Intel DX4 ODP works fine, although the speed increase is small. Shame the boards don't have some cache in them.

Hi everybody, I am kind of crossposting this from another thread, so sorry if I am breaking some rule - I got one of the excellent Mac286 replica cards made by @max1zzz and I am trying to figure out how to make it work. Max was so kind to add in the package the Mac software disks and even the 5.25" DOS floppies, which I can use connecting the Apple PC 5.25 drive to the card(s).

I installed the cards in my Mac IIfx, turned off 32bit addressing, configured the control panel extension for managing the serial interfaces as "compatible" and started the Mac286 v. 3.02 program. There I set external floppy to PC drive, and D: root directory (this is the shared dir between System 7 and DOS) to a folder that contains up-to-date DOS commands, coming with the program.

Everything seems fine, I can boot dos from the floppy, then I run FDISK and partition the virtual C: drive (which is a file on my Mac hd), then I have to remember to activate the partition (damn DOS...) , then I format C:, copy DOS files from floppy 1 and 2... and then when I boot, the autoexec run commands DSTEP1 - 2 - 3 (which should activate the shared dir) ... and then it hangs there. Any chance to solve this, or find a manual anywhere for the card (it's two days of googling and I can't find anything)? what would be your hints? also @max1zzz if you shared your C: drive file online somewhere, could you please point it out to me / us here?

Thanks all in advance!

Answering to myself but it might be useful for other users:

on Mac Garden there are images of the MS-DOS C: drive file (10 or 20 megs) already set up, so that solved the drive issue -
but what really did the trick was using System 6 (6.07 to be precise, the last localized version AFAIK) because the card seems to work only every now and then under System 7 even with 32bit mode off and serial ports on compatible mode.

Now let's install GEM and / or Win 2.03

jeremywork said:

I also took some better quality photos of these Orange386 and OrangePC 290 cards.
View attachment 45833

Click to expand...

@jeremywork , these photos have been a great help, but I still can't read some of the chip labels and resistor values. I'm trying to see if I can populate the missing components on a lower tier 200 series card (210 in my case) and turn it into a full blown 290 model. My card is missing both the cache and PCMCIA components. The PCMCIA controller chip at U93 I can read fine in your existing photo, but it also appears that Q3, Q7, and Q8(?) to the left of the NuBus connector (facing down) are also related to PCMCIA and are unpopulated on my card. The components that are missing for the cache appear to be: the cache chips themselves at U2, U3, U4, U5, and U6; some logic gates at U7 and U10 at the top left of the memory slot; as well as U8 and U9 which are located under the voltage regulator heatsink near the OPTi chip.

There are also some components on the reverse side of the card which seem to differ. The important looking ones are: R20, R21, R22, R108, R109, R110, R111, R112, R113, Q1 and Q2. R20, R21, and R22 are near the back of the card with the rest of the components in question being located to the right of the NuBus connector (facing down).

I believe I've seen that @micheledipaola also has one of the fully populated 290 cards? Any photos where I can clearly read the component text in question would be greatly helpful in this endeavor. Thanks in advance if anyone is able to assist!

evanboonie said:

@jeremywork , these photos have been a great help, but I still can't read some of the chip labels and resistor values. I'm trying to see if I can populate the missing components on a lower tier 200 series card (210 in my case) and turn it into a full blown 290 model. My card is missing both the cache and PCMCIA components. The PCMCIA controller chip at U93 I can read fine in your existing photo, but it also appears that Q3, Q7, and Q8(?) to the left of the NuBus connector (facing down) are also related to PCMCIA and are unpopulated on my card. The components that are missing for the cache appear to be: the cache chips themselves at U2, U3, U4, U5, and U6; some logic gates at U7 and U10 at the top left of the memory slot; as well as U8 and U9 which are located under the voltage regulator heatsink near the OPTi chip.

There are also some components on the reverse side of the card which seem to differ. The important looking ones are: R20, R21, R22, R108, R109, R110, R111, R112, R113, Q1 and Q2. R20, R21, and R22 are near the back of the card with the rest of the components in question being located to the right of the NuBus connector (facing down).

I believe I've seen that @micheledipaola also has one of the fully populated 290 cards? Any photos where I can clearly read the component text in question would be greatly helpful in this endeavor. Thanks in advance if anyone is able to assist!

Click to expand...

I'll try to take some more detailed photos of this card next time I have the machine it lives in on the bench. Can't promise that'll be tremendously soon, but I'll push it up the priority list a bit.

evanboonie said:

@jeremywork , these photos have been a great help, but I still can't read some of the chip labels and resistor values. I'm trying to see if I can populate the missing components on a lower tier 200 series card (210 in my case) and turn it into a full blown 290 model. My card is missing both the cache and PCMCIA components. The PCMCIA controller chip at U93 I can read fine in your existing photo, but it also appears that Q3, Q7, and Q8(?) to the left of the NuBus connector (facing down) are also related to PCMCIA and are unpopulated on my card. The components that are missing for the cache appear to be: the cache chips themselves at U2, U3, U4, U5, and U6; some logic gates at U7 and U10 at the top left of the memory slot; as well as U8 and U9 which are located under the voltage regulator heatsink near the OPTi chip.

There are also some components on the reverse side of the card which seem to differ. The important looking ones are: R20, R21, R22, R108, R109, R110, R111, R112, R113, Q1 and Q2. R20, R21, and R22 are near the back of the card with the rest of the components in question being located to the right of the NuBus connector (facing down).

I believe I've seen that @micheledipaola also has one of the fully populated 290 cards? Any photos where I can clearly read the component text in question would be greatly helpful in this endeavor. Thanks in advance if anyone is able to assist!

Click to expand...

I will do my best to take these pics, but as @jeremywork just wrote, I think it will take a bit to find the time.

Using a combination of more detailed photos and some excellent schematics of an OPTi 82C495XLC implementation, I believe that I was able to reverse engineer all of the optional components on 2xx series OrangePC cards. This effort should allow for upgrading any lower tier 2xx series card into the fully loaded 290 model.

The biggest gap in my current effort is for cards that are missing serial/parallel functionality. I only saw evidence of one example in photos and it had the PCMCIA cage, which obscured many components in the serial/parallel area. The 210 card that I currently have in my possession has serial/parallel fully populated. Therefore, my current upgrade is focused on adding PCMCIA and/or L2 cache to cards that currently lack that functionality.

I have created a DigiKey list containing all but one of the components needed for the upgrade. I have placed location and notes in the "Customer Reference" field. The component values I am most unsure about are R22 and U8, but I'll update my list as I get things working and/or receive more detailed photos of a fully populated 290 card. The one component that cannot be purchased from DigiKey/Mouser/etc. for this project is the PCMCIA controller chip, which is a CL-PD6710/CL-PD6720 depending on your PCB revision. Cards accepting a 10/12 model controller have the 144-pin QFP footprint while cards accepting the 20/22 model controller will have a 208-pin QFP footprint. There seem to be a decent amount of the requisite PCMCIA controller chips floating around on eBay, but you could also harvest it from a scrap board if you can find one.

If you are interested in expanding upon this to add serial/parallel components, I found some good photos of that section of the card here.

@evanboonie Here's some photos of my 250 in case they're of help https://68kmla.org/bb/index.php?media/albums/orangepc-250.24/

I got the cache working! After installing all of the components, initially the card completely failed to boot. After replacing the Am5x86 that I had installed in the board with an Intel 486 DX4, it would boot, but it threw an error message that the L2 cache was "BAD" and was being disabled. I had originally used a 4.7k ohm resistor at R22; as when running a simplified SPICE simulation of the circuit in question, anything smaller than that had a decent amount of ringing on the falling edge of a signal given the capacitance specified for the pins involved. However, such a large resistor also delayed the rise time, so R22 was my first thought to troubleshoot. After replacing R22 with a 1k ohm resistor, the cache was recognized and actually shows up correctly! I don't have a Windows installation for this version of OrangePC to test with yet, so I'm not sure how stable it is; but I do know that it at least functions in basic DOS programs and the card does "feel" faster, for what that is worth. I'm still not positive what R22 should be for optimum performance/stability, so any photos or reports of R22 from a card with factory cache installed would still be greatly appreciated. Here are photos of evidence and the mods involved:

nickpunt said:

@evanboonie Here's some photos of my 250 in case they're of help https://68kmla.org/bb/index.php?media/albums/orangepc-250.24/

Click to expand...

Thanks! This will definitely be helpful when trying to add the PCMCIA to my card. It also looks like most of the logic chips are populated under the PCMCIA card cage on your card? If so, serial/parallel really would just be a matter of adding the GoldStar Prime 2C (or compatible), the 24MHz crystal, and the headers.

@evanboonie great work, what Mac are you using the OrangePC card in? I've a 290 in a Mac Iici, it runs well however full screen video speed is slow, only OK at quarter resolution on screen for games. Not sure if a limitation of Iici video or the card itself. Be keen to hear if you have similar peformance on your card.

Byrd said:

I've a 290 in a Mac Iici, it runs well however full screen video speed is slow, only OK at quarter resolution on screen for games. Not sure if a limitation of Iici video or the card itself. Be keen to hear if you have similar peformance on your card.

Click to expand...

I am running my tests in a Quadra 650 overclocked to 40MHz. I can confirm that full screen video is kind of bad in most cases. Sometimes, when launching an application, the color gets stuck in grayscale. Switching out of, then back in to fullscreen will make some applications appear in color and run with acceptable performance. Other applications (mostly more advanced games or 3D benchmarks) have slide show level performance on fullscreen, but are acceptable in a window.

From what I understand, the lack of video pass-through performance is why all of these PC cards ended up moving to video switching hardware with external dongles. Outputting video directly from the card to the display solves a bunch of these performance issues.

I have some more updates on my quest transform any 2xx series card into a 290:

1. Thanks to @nickpunt , I was able to determine which resistors and capacitors are needed for adding serial/parallel. I have since broken up my DigiKey list into separate ones for each piece of potentially missing functionality:
   • Cache
   • PCMCIA. Additionally, requires a PCMCIA card cage and a CL-PD6710 for Rev01-Rev03 boards or CL-PD6720 for Rev04 boards. The aforementioned components can be found on eBay, AliExpress, or harvested from existing boards. There are some PC/104 PCMCIA boards out there that have a CL-PD6710 and a compatible card cage on them.
   • Serial/Parallel. Additionally requires a GoldStar Prime 2C chip. I cannot find the GoldStar on eBay or AliExpress, so harvesting from existing boards may be the only option.
2. While my Am5x86 used to work fine, it simply will not POST in my OrangePC after adding L2 cache. It is one of those PNY "surface mount on an interposer" models with DIP switches. I have tried forcing it to WriteBack/WriteThrough with its DIP switches, but neither option seems to work. I thought this might have something to do with the bodge wire I have seen on the back of the CPU socket in several photos of 290 cards. The bodge ties the clock multiplier and WB/WT cache selection together (jumper 2 on the OPC). I did get my Am5x86 to boot once after installing the bodge and setting J2 to the AB position (GND for 2x/4x mult and write through cache). However, after trying to overclock the bus to 40MHz, I was never able to get the Am5x86 to boot again, even back at 33MHz bus. The location of the bodge wire makes me highly suspicius that Orange Micro found issues with certain caches on clock quadroupled cards, and/or had trouble enabling write back cache on newer 486 DX4 cards that included it. Either way, my OrangePC still works great with the 486 DX4 with or without the bodge installed; but I am very curious what others experience with L2 cache and a Am5x86 CPU have been.
3. I'm still not positive what the value of R22 is. After studying the two blurry photos of 290 cards and comparing the look of R22 to other resistors blurry shapes, my best guess is that it is a 220 Ohm resistor. 220 Ohm does work and I was able to pass many benchmarks that heavily utilize the cache with 100% stability at a 33MHz bus. I'd still love to have confirmation of what was put here from the factory though.

I'm actually starting to think that my Am5x86 is somehow broken. I tried putting it back into a Houdini II DOS card that it was previously known working in and I can't get it to POST there either, regardless of cache settings. I think I'll try to get a 3.3V Am5x86 and test that once I install the regulator on my OrangePC.

Speaking of the Apple DOS card, holy cow is its memory speed faster than the OrangePC 290! Without L2 cache, the Houdini feels light years ahead of the 290 in benchmarks. Disk speed is also almost doubled on the Houdini, but that makes sense given PDS vs NuBus. I feel like the OrangePC 290 should have faster memory performance though given its much more advanced chipset. My best guess right now is that the default BIOS settings are extremely conservative with timings/wait states. I'm not sure if the BIOS settings are passed in by the OrangePC application or are hard-coded on an on-card EEPROM, but either way, changing them won't be straightforward. I'll probably look into this at some point though.

I dumped the EPROMs on both my 210/220 (card has 220 written on it, so that is what I'm calling it) and my card simply labeled "OrangePC" (Cx486SLC with one ISA slot) which I'm calling the "200". There really isn't anything in them besides some basic identification info, but I'll upload them here in the event that anyone needs them for some reason in the future.

After digging into the OrangePC 3.8.2 application with ResEdit, it is clear to me now that the app itself is injecting the BIOS config. I'm not really a hardcore BIOS hacking type, so I'll probably just leave this for now. But if anyone out there is more familiar with classic Macintosh programming and/or PC BIOS editing, I'd definitely accept some help in that department. It would be quite fun if we could get into the BIOS settings for these cards and actually make changes to memory timings and such.

Thanks for posting about your travels with your OrangePC card @evanboonie Interesting about the video speed, the Windows driver gives OK performance but full screen 320 x 200 DOS video modes are not stellar, perhaps some adjustments with Scitech Display Doctor might help out.

Can confirm a drop in Hypertech 5x86 did not work on my OrangePC 290 (same CPU does not work well on the Houdini, either - presume CPU microcode is lacking, card can boot to DOS on Houdini but not much else). I installed the 5x86 not long after getting my card and the VRM blew on it not long after. Since repair it's been fine and perhaps the CPU was a red herring the card had not been used for several years before I obtained it.

Am also using an Intel Overdrive DX4-100 in mine which is a good match but does get toasty therefore recommend a fan nearby. I recently found an Evergreen 5x86 upgrade with jumper settings for WB/WT cache voltage and multiplier but the pins are fragile and prone to snapping will keep in the tower 486 I have where it is happy.

Would love to see a more optimized BIOS! I'm guessing some of the RAM and cache timings have wait states. Having restored an early 486 VLB motherboard with said 5x86 above, the minor tweaks to timing can double RAM throughput and give the CPU chosen a big boost.

Question about the Joystick ports on the 610, 630, 6100 vs. the PCI PC cards. I was trying to decide if/how I could do the MIDI modification on my 630 system, and I was considering modifying the Joystick Port to be compatible with the Sound Blaster MIDI adapter. But I'm not able to find any information on the 630's Joystick pinout. Looking at the traces, it appears that it might be the same as the 610, and that the 6100 might also be the same as the 610. But, I'm not sure. I've written up what I've found out so far at: https://www.savagetaylor.com/2024/12/13/macintosh-lc-630-performa-640dos-midi-modification/. Does anyone have the Joystick pinout for the 630 or 6100 from an official source? I've found the TIL articles for the 610 and PCI PC card pinouts.

More updates on OrangePC shenanigans:

Naming clarity:
After digging around more in the resource fork of the OrangePC 3.8 application and looking at the decl ROM contents on my NuBus OrangePC cards, I have determined that the card with a 386sx/Cx486SLC and single ISA slot is just called the "OrangePC" with no number/suffix. I also have the manual for the original card and it does not refer to any number/suffix on the end of the name. 210-290 cards with the PGA-169 socket for full 486 CPUs is referred to in the code as the "OrangePC 2".

Original OrangePC CPU upgrade:
I desoldered the original Cx486SLC my card came with and replaced it with a TI486SXLC2-G50 (3.3/5v version) and it works great! After reading on some Amiga forums that the TI chip can safely run at full 5v (some users report years of stability), I tried it on my OrangePC and was completely successful. I haven't managed to get the clock doubling working yet, but application load times are already helped by the extra L1 cache. Most strangely, the memory bandwitth as reported in benchmarks is actually higher than my OrangePC 2 with L2 cache... I'm really starting to think that either the benchmarking utilites are inaccurate on the OrangePC 2, or the BIOS settings really are optimized horribly.

OrangePC 2 PCMCIA addition:
The CL-PD6710 I ordered from China is not a CL-PD6710. I'm not sure what it actually is, but the VCC/GND pins are not in the correct locations and the silkscreen is rotated 90 degress and looks sloppy. I have another one on the way to test from another seller, but if that one doesn't work out either, the only option for sourcing these may become desoldering them from existing boards. I also have a PC104 card with a CL-PD6710 on it, so I'm confident I can get it working in the end. It would be nice if it could be done without sacrificing anther piece of hardware to get it working though.