For the NEAT chipset there's a DOS tool to set the various fancy chipset registers, even if the BIOS does not expose them. Is there such a tool for SCAT as well? That would make me less dependent on the BIOS I'm currently using. There's a DMA-related setting the chipset registers that isn't exposed in most BIOSes, but it needs to be toggled from its default for it not to eat floppy disks for all meals of the day.

If such a thing existed, I'd be willing to play more with alternate BIOSes..

douglar wrote on 2023-12-14, 01:27:

anyone get their board working at 16Mhz or 20Mhz?

anyone try the mr bios for 386sx scat or the generic 286 bios?

In case anyone is still playing with these...

I'm running mine with an 8MB SIMM & Harris CPU at 16MHz. C&T BIOS, XTIDE ROM on an SMC ethernet card and an 8GB Apacer CF in a no-name IDE-CF adapter on a no-name Winbond I/O-everything card.

It runs stable at 25MHz as long as I don't touch the IDE subsystem, so I *think* it would run at 20MHz if I had a 40MHz can.

I couldn't get it to recognize *any* CF without the XTIDE ROM, running either the C&T or the AMI BIOS.

Doc

DrDoc wrote on 2024-07-31, 07:48:

In case anyone is still playing with these... […]
Show full quote

douglar wrote on 2023-12-14, 01:27:

anyone get their board working at 16Mhz or 20Mhz?

anyone try the mr bios for 386sx scat or the generic 286 bios?

In case anyone is still playing with these...

I'm running mine with an 8MB SIMM & Harris CPU at 16MHz. C&T BIOS, XTIDE ROM on an SMC ethernet card and an 8GB Apacer CF in a no-name IDE-CF adapter on a no-name Winbond I/O-everything card.

It runs stable at 25MHz as long as I don't touch the IDE subsystem, so I *think* it would run at 20MHz if I had a 40MHz can.

I couldn't get it to recognize *any* CF without the XTIDE ROM, running either the C&T or the AMI BIOS.

Doc

I have one of these with the AMI BIOS and it works fine with CF, but I'm instead using an IDE SSD. With the C&T BIOS, it won't boot from IDE at all using only the BIOS.

So you must have changed the oscillator to change the CPU speed? I didn't think the chipset allows this on its own.

eesz34 wrote on 2024-08-15, 18:26:

DrDoc wrote on 2024-07-31, 07:48:

It runs stable at 25MHz as long as I don't touch the IDE subsystem, so I *think* it would run at 20MHz if I had a 40MHz can.

Doc

So you must have changed the oscillator to change the CPU speed? I didn't think the chipset allows this on its own.

I changed both oscillator and CPU. I haven't tried to crank it up with the original CPU.

It's actually pretty snappy for a 286 even at 16MHz

Doc

DrDoc wrote on 2024-08-16, 02:15:

It's actually pretty snappy for a 286 even at 16MHz

You don't say.. I have a pair that came as 16Mhz, they run DesqView with httpserv, ftpsrv, ircd and sbbs (synchronet bbs node). And a novell client (for the bbs). Apart from the BBS - which is kinda sluggish because SBBS really isn't for a 286 served over IPX - http, ftp and ircds are nicer and snappier than I had dared guess. Even survived being mentioned on hackernews..

These JCS286 boards are awesome.

/Eirik

In the previous post I was going to compare the JCS board's performance to my Teknor TEK-AT1, and when I looked at the TEK board I found... an F82C835D chip.

The TEK BIOS is a 1Mb OTP but looking at it in the hex editor I found the same string as starts the C&T color BIOS at 0x010000, so I split it and burned one...
When I first booted it I got a "Parity Check 1" error. I had to install all the original DRAMs, boot, and disable parity. As far as I can tell if you've been running the C&T color BIOS and disable parity check before shutting down, this won't be a problem.
The GQ programmer app says the checksum is "00941400"
Ctrl-plus and Ctrl-minus toggle the CPU clock between 16MHz and 8MHz. Switch and LED are non-functional.

I ran all 3 BIOS's through "Phil's benchmark collection" on floppy. All 3 had all the performance settings turned on but in the TEK BIOS I had to go into the last register page to twiddle the RAM wait state to zero:

AMI 286-BIOS (c)1990 -
didn't play nicely with my 128MB CF, with or without XUB
6.7 in Superscape
Landmark 24.55 CPU 5285.16 chr/ms
TOPBENCH 46
System Info"Computing Index" 12.1

C&T "awsum" color BIOS (c)1988-1990 -
I did NOT set RAM WS to zero on this one
5.3 Superscape
Landmark 18.20 CPU 4636.98 c/ms
TOPBENCH 36
SI 8.9

TEK C&T v2.41 1993 -
Before tweaking the RAM WS it got identical results to the color BIOS
It's the only BIOS that offers a choice of boot devices
6.5 Superscape
Landmark 24.54 CPU 4636.98 c/ms
TOPBENCH 44
SI 12.1

And finally, if you want to you can break this out yourself. The Teknor TEK-AT1 BIOS file is on theretroweb on that board's page.

Doc

DrDoc wrote on 2024-08-16, 10:07:

Ctrl-plus and Ctrl-minus toggle the CPU clock between 16MHz and 8MHz. Switch and LED are non-functional.

Too late and too brain-burned - Obviously I meant Ctrl-Alt-plus and Ctrl-Alt-minus

Doc

Now I'm curious what the C&T BIOS would look like at 0WS .. Also I wonder what the AMI BIOS does differently. I may have to test that at some point..

ltning wrote on 2024-08-16, 10:27:

Now I'm curious what the C&T BIOS would look like at 0WS .. Also I wonder what the AMI BIOS does differently. I may have to test that at some point..

I suspect that it has something to do with the other wait state settings in the AMI BIOS. DMA and bus, IIRC.

The problem with the CF cards kind of kills the speed boost for me though.

Ok, so what kind of 72-pin simm seems to work? Because I've tried two of those new-ish production simms you can get on ebay and both give me error 0B post error, which means "first 64k failed." Has anyone else tried these things? I do remember a period correct Kingston 4 meg stick worked for sure.

Ok, so the 4 meg Kingston stick does work. The odd thing is it doesn't work (I get a 0B post error) unless both DIP ram banks are populated and the bios even seems to count both the DIP ram and the 4 meg SIMM. The initial ram test outputs 640K conventional, 384K shadow, and 4096K extended. I've ordered a IBM OEM 8 meg 70ns stick, so hopefully it works. Even if it doesn't, 5 megs total ain't bad. Also, I tested the MR BIOS for the 386 SCAT chipset in this board, it didn't work unsurprisingly.

Ok, I got the IBM OEM SIMM in. It works. So it seems like there's a compatibility issue with slightly later 8 bit parity SIMMS. Basically it seems to me if it's an 8 meg parity SIMM with 9 chips on either side (8 memory, 1 parity I think) then it may not be compatible with this board. If it has 12 on either side, though, it may be compatible. So make sure you get a good look at the SIMM you're buying if you want to go to 8 megs. If you plan to stick with the onboard 1 meg and add a 4 meg SIMM, the ones that worked for me were 60ns Kingston parity SIMMS with 8 chips on the front and 4 on the back. I suspect you could run 4 megs on the board (if you could find any of the 1 meg x 4 bit 20 pin DIPs, they seem to be unobtanium) with a 4 meg SIMM to max out the memory, that'd have an aesthetic advantage as well as a cooling one if you were going to go for some faster 286 speeds.

Jackal1983 wrote on 2024-08-26, 17:25:

Ok, I got the IBM OEM SIMM in. It works. So it seems like there's a compatibility issue with slightly later 8 bit parity SIMMS. Basically it seems to me if it's an 8 meg parity SIMM with 9 chips on either side (8 memory, 1 parity I think) then it may not be compatible with this board. If it has 12 on either side, though, it may be compatible. So make sure you get a good look at the SIMM you're buying if you want to go to 8 megs. If you plan to stick with the onboard 1 meg and add a 4 meg SIMM, the ones that worked for me were 60ns Kingston parity SIMMS with 8 chips on the front and 4 on the back. I suspect you could run 4 megs on the board (if you could find any of the 1 meg x 4 bit 20 pin DIPs, they seem to be unobtanium) with a 4 meg SIMM to max out the memory, that'd have an aesthetic advantage as well as a cooling one if you were going to go for some faster 286 speeds.

For what it's worth, I got eBay item 163052598756 and it works perfectly with the DIP memory removed.

eesz34 wrote on 2024-09-03, 22:05:

Jackal1983 wrote on 2024-08-26, 17:25:

Ok, I got the IBM OEM SIMM in. It works. So it seems like there's a compatibility issue with slightly later 8 bit parity SIMMS. Basically it seems to me if it's an 8 meg parity SIMM with 9 chips on either side (8 memory, 1 parity I think) then it may not be compatible with this board. If it has 12 on either side, though, it may be compatible. So make sure you get a good look at the SIMM you're buying if you want to go to 8 megs. If you plan to stick with the onboard 1 meg and add a 4 meg SIMM, the ones that worked for me were 60ns Kingston parity SIMMS with 8 chips on the front and 4 on the back. I suspect you could run 4 megs on the board (if you could find any of the 1 meg x 4 bit 20 pin DIPs, they seem to be unobtanium) with a 4 meg SIMM to max out the memory, that'd have an aesthetic advantage as well as a cooling one if you were going to go for some faster 286 speeds.

For what it's worth, I got eBay item 163052598756 and it works perfectly with the DIP memory removed.

Huh, that's weird. Both of the sticks I have that didn't work are similar. There might be something wonky with my board or maybe I didn't go through the right procedure. The one that worked is the IBM one I posted below. The other one is one that's on the way that has chips that might work well at fast 286 speeds with 0 wait states per this post: Re: 3 (+3 more) retro battle stations

Also, here's the 82c235 databook/datasheet. I think someone posted it on another thread, but I figure it might be good to have it in this thread. Regarding the clock: on page 52 it shows how a regular battery should be interfaced with the chip. The really important thing is that it calls for a 6 volt battery backup, not 3.0 to 3.5V like most other clock chips. So a 4 AA or AAA battery box would be bang on for this thing. However, if you're planning to use a solder in Varta replacement like Wiretap's "Not-A-Varta" or Scrap Computing's Barrel Battery Blaster, you might need to substitute a different battery holder or mod the design. Digikey PN 36-126-ND looks to be a direct swap for the regular single cell holder and should work with the Barrel Battery Blaster. It also appears to have enough tension that a single cell is still usable, so you can use it in both single and double cell applications. It is quite chunky though, so you might need to trim a bit of plastic off.

Well, it looks like IBM certainly binned the chips they got from OEMs heavily. The 60ns Siemens SIMM runs into a wall before 20mhz- the board boots with a 38mhz crystal (19mhz) but gives bad memory errors when I try a 40mhz crystal (20mhz). The 70ns IBM SIMM Boots at 25mhz and does a full memory count without any errors on the JCS board! Looking forward to getting this thing in a case. I'm betting the real limitation will be the lack of a /3 divider for the ISA bus, 12.5mhz is right on the ragged edge as far as sound and I/O cards go.

So for some reason, I cannot use my Promise DC200M on (at least one of) my JCS286/SCAT boards. I know I've tested this in the past, but for some reason on this one it fails in a weird fashion:
I configure the board as I usually do; HDD type 1. This works in *every* other 2/3/486 board I have, a high two-figure number.
In the Promise BIOS, when I try to fix the geometry (which is all zeroes, for some reason, despite the "drive info" page showing correct geometry), the only input I can access is the LBA enable/disable - and it won't let me change it. The Promise BIOS claims the system BIOS has already enabled LBA! This is obviously wrong.

I've tried AMI, C&T ("awesome") and the TEK BIOSes, they all behave the same way.

Next step is to try XTIDE, but that will take a bit more work.

Perhaps this can be helpful info in figuring out why these boards apparently struggle with some CF/IDE drives?