Hi there! Did you configure the XTIDE ROM file using the official DOS utility before programming it into an EPROM?
Otherwise, it won't work.

XTIDE ROM - 3COM 3C509B-C

If you don't have a DOS PC at hand, you can try those:
Re: IDE disk-on-module + 386

Good luck! 🙂🤞

As above, the rom file MUST be configured first. The documentation used to be pretty terrible for not making this immediately obvious. Not sure if that is still the case.

Thanks Jo
No I Didn't ...
I just assumed the premade binaries were set to run at a default address of D000...
obviously not the case..
I'll have to get one of my motherboards set up and run dos on it.... I guess.
Thanks also for the links.
regards
VS

You're welcome. 🙂
It's not your fault, you did everything fine. The pristine binary simply lacks the needed checksum.
As megatron-uk wrote, the need for the utility wasn't really documented.
Not sure if that has changed since, either. 🤷‍♂️
In fact, I've struggled with XTIDE the same as you when I tried it first.
Found out about the DOS utility by checking the tutorial for the original XTIDE card for 8-Bit PC slot.
Then it all made sense to me, when I realized that the possibility to install XTIDE Universal BIOS on an ethernet card
is essentially just a nice by-product, a free gift so to say.

Edit: Typos fixed.

Vipersan wrote on 2023-03-28, 18:37:

I'll have to get one of my motherboards set up and run dos on it.... I guess.

You can run that util using DOSBox too. I haven't tried but I suppose it should also run on 32-bit NT systems via CMD.EXE - just don't try to use any auto-detect functions, pick your options manually.
The resulting file (after writing the config) should be power-of-2 in size, for example 8192 bytes long, that's the easiest way to tell if it's ready for EPROM or not yet.

Broke out the Toshiba luggable T3200SX..
ran up DOS.
erased the 27C128....ran up xtidecfg ...but this time used IDE_XT.bin ...which is more in keeping with the Compaq 286.
loaded up the checksumed id_ext binary with the TL866 ...clear buffer with 0x00 ...to fill up the unused eprom space with zeros.
Apparently very important.
flashed the eprom and re-fitted in the network card set as previously shown.
Now I get this at boot :-
So...getting there

Some progress obviously...but not sure what that 1h error is ?
The boot order appears wrong C: before A: ?
how do you change the boot order ?
Also ..
If I connect up the original C: drive it is still seen ...and will boot although it isn't "identified"...
Once in the OS I can see a directory of the A: drive which is a high density 1.2mb 5.25 inch floppy....but cannot boot from it.
This has me very confused as obviously XTIDE is running ...sort of.

It may well be that XTIDE cannot handle the hard drive and floppy drive interfaces of the Compaq correctly ?
rgds

Now this is curious...
It still has the boot order C then A
of course C is not physically connected...so it does try and boot from A ...but then eventually times out and re-boots the compaq entirely...
Which allows another go at booting from A after getting no response from C
I left it in this loop whilst I went for a cup of tea.
...when I returned it had finally booted from A:
see picture..
I still need to reverse the boot order..but it did eventually boot the compaq setup floppy in A:

Hi again, I'm not familiar with those portables, but I do assume they're AT class machines.
So chances are high they're already using some sort of AT-Bus/IDE interface, rather than XT-IDE.

If so, the AT buils of XUB might be better suited perhaps. The backups I made back then are configured for auto-detection
and should scan for HDDs on primary/secondary IDE ports.

Speaking under correction, though, it's been quite a few years. 😅

From what I learned when installing my XT CF Lite card,
the XT-IDE Interfaces can be very weird/special at times. The 8-Bit cards exist in different models, also.

For the XTIDE card that I got online, only the supplied ROM image would work (which was using 8086 code, but I wanted an V30 build) .
Other, more recent XT BIOSes didn't work for it. So I had to go back to the original BIOS.

Definately debateable Jo..
These machines are a bit of an enigma to me also..
Certainly at a crossroads in development I think...they could I suppose actually benefit from using an AT version.
The existing drive interface is odd indeed...and proprietory to Compaq.
For example this is circa 1987- ish..and maximum hard drive size would have been around 40mb.
Anything larger was inconcievable at that time....and certainly not in the hard drive list.
Hard drives at this time were set by Type ...and geometry ..rather than auto detect.
You would have to tell it what the drive was.
No way for example can you hook one of these hard drives up to an ide to usb interface ...and pull data off that way.
So yes the drive would indeed have more in common with AT I suppose.
In actual fact I have seen such hard drives described as AT/XT.
A very early version of XT...and at the very end of AT.
regards

To change the boot order you can configure it using the same tool which sets the ROM address.

You should also be able to override it on each boot by pressing on of the Function keys listed at the top of the Xt-ide splash screen - you can then change to floppy, ROM, or one of the detected ide drives.

Vipersan wrote on 2023-03-29, 00:14:

Definately debateable Jo..

Definitely. 😁

But as I said, I don't know these PC models, so I can only guess. 🤷‍♂️

Vipersan wrote on 2023-03-29, 00:14:

The existing drive interface is odd indeed...and proprietory to Compaq.
For example this is circa 1987- ish..and maximum hard drive size would have been around 40mb.
Anything larger was inconcievable at that time....and certainly not in the hard drive list.

It's likely not the norm, but the Commodore Setup of the time did list a 112MB type.

Re: Emulation on MS Windows 3.1x ?

My old 286 using a Schneider BIOS (Phoenix derivative), didn't by contrast.. So it depends.

Vipersan wrote on 2023-03-29, 00:14:

Hard drives at this time were set by Type ...and geometry ..rather than auto detect.
You would have to tell it what the drive was.

I know, I literally grew up with a 286 PC.. 🙂
My father and me installed an 80 MB Conner drive in it, but there was no Type 47 (user definable).
So my father did choose a 40 MB model from list.
The drive didn't complain, since the drive geometry wasn't exceeding overall drive capacity.

That being said, AT-Bus or ESDI drives of the late 80s do implement some form of geometry translation.
They can use L-CHS (logical CHS) as opposed to P-CHS (physical CHS).

For older HDDs (P-CHS), not exceeding drive capacity wasn't the only factor to use fake geometry.
To my understanding, not exceeding the actual numbers of physical heads and sectors was also required.

Last but not least, some 386 BIOSes frpm '92/' 93 can auto-detect CHS values.
They have an option for this in CMOS Setup.
XTIDE Universal BIOS has the same ability, I think.
It will try to figure out drive geometry in auto-detection mode.
Also works for old, pre-LBA drives, thus.

PCem/86Box rely on XTIDE Universal BIOS, too, for their PC/XT emulation.

As far as I know, auto-detection itself was no problem for IDE/AT-Bus HDDs of the late 80s/early 90s.
It rather was the PC BIOS that was so limited.

Here's an interesting site with information about early HDDs.
Drive firmware from the 1980s onwards wasn't simple, at all.
It had "sector translation" already.

http://redhill.net.au/o/glos2.php#sectrans
https://www.redhill.net.au/d/2.php

Vipersan wrote on 2023-03-29, 00:14:

In actual fact I have seen such hard drives described as AT/XT.
[..]
A very early version of XT...and at the very end of AT.

I believe you. Those early HDDs supported both 8-Bit and 16-Bit versions of IDE.
Like Compact Flash cards still do today.
Some of those Commodore PCs had XT-IDE on-board, AFAIK.

On the PC side, however, XT-IDE and IDE are differently implemented.
They occupy different memory addresses, different I/O ports.
They're not on the standard addresses the 90s era IDE channels are.

In general, XT BIOS itself differs a bit from AT BIOS in how it uses memory space to store its parameters in RAM.

Also, there's a difference in how IDE/XT-IDE communicate.

XT-IDE is using 8-Bit transfers, while IDE uses 16-Bit transfers.
On an XT class system, 16-Bit transfers aren't possible. Not directly, at least.
Multiplexing is required (converting to 2x 8-Bit transfers).

Some of the XTIDE cards may do this, yes, but it's not the old XT-IDE anymore, but normal IDE. The change was likely made to enhance compatibility with IDE hard disks or non-standard CF cards.

The old XT-IDE interface was intended to easily connect HDDs using 8-Bit I/O, because 16-Bit I/O wasn't available on PC/XT bus.

AT class systems with a 16-Bit ISA bus usually don't physically need XT-IDE anymore, since they can use regular IDE/ATA interface.

Not looking favourable for the IDE_AT.bin
Went digging for another 27C128...but seems the one I'm currently using is the only one I have.
Note to self...must get a few more.
Anyhoo...20 minutes to erase it ..then exactly the same process...wrote a checksummed IDE_AT bios to the chip and replaced.
Yes it does run...but cannot boot from either the 40mb hard drive ...or the 1.2mb Floppy.
I guess this system shares more in common with XT than AT.
There are other versions in the list https://www.xtideuniversalbios.org/binaries/r625/ ...but non specific to a 286 system.
Not an easy process to try each one...with a single chip as there is a 20 minute erase time for the chip each time.
definately must get a few more 27C128
If anyone has used XTIDE with this exact system...or indeed early 286 systems...perhaps they would like to share their experiences ?
rgds

also ..thanks for the heads up megatron.
I'll look into changing the boot order next time.

Update. I did remember an old thread about the 3200SX.

Toshiba T3200SX BIOS?

The HDD of one 3200SX user was a Conner CP30104..

Re: Toshiba T3200SX BIOS?

https://stason.org/TULARC/pc/hard-drives-hdd/ … -SL-IDE-AT.html

Edit: Judging by this video, the Compaq Portable III 286 used an WDA-L80 HDD at some point.

https://www.youtube.com/watch?v=ys018PQ4fYI

https://stason.org/TULARC/pc/hard-drives-hdd/ … SSL-IDE-AT.html

Since the Portable seems to have ISA slots, it might be possible to install a soundcard with IDE connector.
The early CT17xx Sound Blaster 16 cards or some ES688F cards have Secondary IDE channel that's configurable via jumper.

The AT BIOS can only boot from primary master, but XTIDE Universal BIOS can boot from secondary master, too. I tried this myself.

Yes ..I remember the thread well referencing the Toshiba 3200
I got a lot from that thread and the 5200 thread.
Ian_B had mastered these lovely old machines...and in fact modded the bios for the 5200 if not both.
I still have my highly modded 5200....and the 3200 with colour lcd screen.
I had no choice in fitting this...on account I bought it with the plasma screen smashed.
I guess this is why I fell in love with orange gas plasma period relics...and why I am now renovating Compaq Lunchbox machines.
Once an addict ...always an addict...🤣

So..
I have tried IDE_AT.bin
IDE_ATL.bin
and IDE_XT.bin
All now run versions of XTIDE at boot...but only IDE_XT.bin has been successful in booting the original hard drive ...and sometimes the floppy drive.
I must conclude that the Compaq III 286 lunchbox has more in common with XT than AT.
..but having said that..it isn't a perfect fit for this PC ....at least when using it as a replacement for the Compaqs own bios using a modified network card in an ISA slot.
The Irony is that there are 2 vacant 28 pin bases on the main board which can be used for XTIDE ...but requires the original compaq bios chips modding in order to 'map' these vacant sockets into the system.
The XTIDE bios then has to be split...odd/even and flashed to 2x 27C256 which are then fitted into these 2 vacant slots.
All this is beyond my skill set of course.
but it has been done already by LeFlash...but sadly not for my 'early' version of the Compaq III 286.
see this thread..
Compaq Portable III modded BIOS / XTIDE
regards
VS

megatron-uk wrote on 2023-03-28, 18:33:

As above, the rom file MUST be configured first. The documentation used to be pretty terrible for not making this immediately obvious. Not sure if that is still the case.

Twice now I've forgotten that and spent an hour debugging a problem that didn't exist. I wish the default ROM just included a default setup/checksum, hardly anyone needs to change the defaults anyway...

Just erasing another 4x 27C256 to give the LeFlash modded bios and XTIDE .bins another go..
Just in case I FFF'd up in the flashing.
This time I will be extra carefull to fill the unused sections of the 256 chips with 00 ...in case this creates a checksum problem as by default when the chips are UV erased...they are filled automatically with FF.
worth a punt I guess.

So...
This is interesting.
I managed to get the LeFlash modded bios to actually boot.
Sadly the internal original hard drive is not recognised and will not boot..
However something is identified as master - at 1F0H
Obviously LeFlash is using the AT version of UBXTIDE.
I say this because of the port reference 1F0H and his modded bios and additional XTIDE chips split odd/even are behaving exactly as my attempt to run IDE_AT.bin din via the network card.
So ..maybe I should now prepare a more modern hard drive to try ..preferably with DOS already on it ?
How did I make it work on my older MKl compaq II 286 ?
I prepared two more chips U72 and U89 (split modded bios) with the LeFlash bin files...but this time on 27C256 chips..
and made sure to fill the unused bios space with 00.
Having realised a mistake last time I tried in that my older 286 was fitted with 27C128 (16k) chips...I found a jumper on the mainboard E10 and moved it from linking pins 1 and 2 (default factory setting) to link pins 2 and 3
This then accomodates a larger capacity chip 27C256 (32k)
That isn't all though I now have to tell the bios to map the 2 vacant sockets into the system..
This is done by linking jumper E12 1 and 2
factory default being E12 2 and 3.
This action now maps in the empty sockets...where I can place the 2 x 27C256 chips I flashed with Xeven and Xodd....again split by LeFlash.
Many thanks for these .bin files btw...if you read this LeFlash.
So there it is ...
XTIDE is now installed on the mainboard without need for ISA slots or cards....but does not boot the original 40mb hard drive.
No surprise there as IDE_AT.bin does not see it either.
rgds
VS