Some CF adapters don't ground pin 34 (PDIAG) - making it fail UDMA66 detect.
Connect pin 34 to pin 30 (or other ground pin) + cut trace that goes from 34 pin to compactflash card.

Credit : 0xCats

Yes, from Addonics (Model ADIDECF). If you can find one, they're often expensive. There's a few on eBay for $50 + shipping.

There are probably other vendors out there, but I have one of these:
https://www.addonics.com/product/intro/20

Alternatively, a SATA to IDE Adapter is another route to pursue, which could also be more cost effective and better performing.

well damn, $50. i'm pretty sure i paid less than $3 per cf adapter. maybe someday i'll make friends with someone who can solder.

The second one, the HX-2118P with Transcend 4GB/8GB/32GB cards makes it into UDMA 5 as reported by the bios on a 430TX Socket-7 system.

agent_x007 wrote on 2023-05-19, 18:56:

Some CF adapters don't ground pin 34 (PDIAG) - making it fail UDMA66 detect.
Connect pin 34 to pin 30 (or other ground pin) + cut trace that goes from 34 pin to compactflash card.

Credit : 0xCats

Do not do this for an adapter designed to connect to a cable like the second picture in OP's post. The PDIAG line is not meant to be connected to the host at all for 80-pin cables (but grounding is enough to satisfy most), but it is still required for communication between the master and slave devices.

vstrakh wrote on 2023-05-20, 05:23:

The second one, the HX-2118P with Transcend 4GB/8GB/32GB cards makes it into UDMA 5 as reported by the bios on a 430TX Socket-7 system.

thanks for the verification. i tried another 2 connector (single device) cable, that's never given me problems before, and got the same result, but when i swapped in a gnarled 24in 3 connector cable--bam, it detected the 80 wire conductors. (the cable only has to go 3in AT MOST for the adapter and now i got 24in of intestines.)

i'm a little perplexed on why this is happening. what is going on inside the connectors? i previously assumed blue, grey and black were all internally identical and the location of a cut wire determined the cable select positions. (so maybe without the cut wire only the blue/gray connector is different from the other 2...after all you can plug an 80 wire cable in backwards and only the cable select is affected.) i've tried to find internal diagrams online, but there are so many junk search results to wade thru

vstrakh wrote on 2023-05-20, 05:23:

The second one, the HX-2118P with Transcend 4GB/8GB/32GB cards makes it into UDMA 5 as reported by the bios on a 430TX Socket-7 system.
UDMA5 430TX Transcend CF 32GB.jpg

Would be impressive considering TX only supports DMA Mode 3 😀 That text is meaningless, its Drive capabilities reported by "identify drive" command.

stanwebber wrote on 2023-05-20, 19:16:

i'm a little perplexed on why this is happening. what is going on inside the connectors? i previously assumed blue, grey and black were all internally identical and the location of a cut wire determined the cable select positions. (so maybe without the cut wire only the blue/gray connector is different from the other 2...after all you can plug an 80 wire cable in backwards and only the cable select is affected.) i've tried to find internal diagrams online, but there are so many junk search results to wade thru

That's true for the traditional 40 pin type (with all black connectors), where barring any missing pins/wires for cable select it's a very plain 40 pin cable you can use just fine for your Raspberry's expansion port, but indeed the high speed 80 pin connectors are different not just in color!

Looking for this? 😀

https://courstechinfo.be/Hard/Images/Cable_at … _66-100-idc.gif

how authoritative is that diagram? if i read it correctly there's no difference between the blue and black connectors--at least in the cables that employ the cut wire method.

the essential difference is in the gray connector as every wire is connected in the black connector.

so my question is why would a single device cable with the blue and black connectors fail the 80 wire conductor test? the gray connector is practically transparent if nothing is plugged into it. how would the motherboard know if it's there or not?

stanwebber wrote on 2023-05-21, 01:21:

how authoritative is that diagram? if i read it correctly there's no difference between the blue and black connectors--at least in the cables that employ the cut wire method.

In addition to the PDIAG signal being cut, it's also probably connected to ground inside the blue connector (which is a good reason to not use the blue connector for drives - they may try to drive PDIAG high).

rasz_pl wrote on 2023-05-20, 21:49:

Would be impressive considering TX only supports DMA Mode 3 😀

Didn't know 😀 I've jumped over socket 7 back in the day, going from 486 into P2, missed lots of fine details...

i don't know what the deal is with my single device 80 conductor cables. i took a couple standard 2 device cables and trimmed them down to 5 and 2in from the blue connector, popped off the master black connector and crimped it back on the shortened cable with a portable table vise. i did this twice and both cables pass the 80 conductor motherboard test on the cf adapter with flying colors.

vstrakh wrote on 2023-05-21, 05:01:

rasz_pl wrote on 2023-05-20, 21:49:

Would be impressive considering TX only supports DMA Mode 3 😀

Didn't know 😀 I've jumped over socket 7 back in the day, going from 486 into P2, missed lots of fine details...

@rasz_pl, actually the i430TX chipset supports maximum UDMA 2 mode. But you are correct, this BIOS shows the maximum supported UDMA mode from the drive, not the actual UDMA mode used by the board/BIOS.

@vstrakh, this UDMA 5 display from the Award BIOS on your TX board is a BIOS bug. This bug is the source of the Windows 98 DMA issue.
I found that Windows 98 doesn’t like the BIOS reporting higher UDMA modes on older hardware and refuses to use any DMA mode when set in Device Manager. This results in a fall-back to PIO mode 4 with reduced Harddisk performance.
I’ve tested this on my i430TX board with an 11/1998 Award BIOS and a clean Win98 install, so I was using the standard Microsoft Bus Master IDE drivers.
Indeed the DMA mode would not stick and the atto disk benchmark measured a poor 9MB/s.
Luckily most Award BIOSes from 03/1999 or later have a fix for this UDMA bug so you may need a BIOS update for your board. I’ve tested an updated BIOS with this fix on my TX board and the BIOS now correctly showed UDMA 2 on UDMA 5 or 6 drives. In addition, Win98 allowed to select DMA in device manager and after a reboot I got 30MB/s in atto disk benchmark! 😀

Jan

Chkcpu wrote on 2023-05-29, 12:20:

Luckily most Award BIOSes from 03/1999 or later have a fix for this UDMA bug so you may need a BIOS update for your board.

It's a Lucky Star 5I-TX2A, and the the bios is already the latest one (O2-006 from 06/11/1998) as seen on The Retro Web 🙁

vstrakh wrote on 2023-05-30, 16:18:

Chkcpu wrote on 2023-05-29, 12:20:

Luckily most Award BIOSes from 03/1999 or later have a fix for this UDMA bug so you may need a BIOS update for your board.

It's a Lucky Star 5I-TX2A, and the the bios is already the latest one (O2-006 from 06/11/1998) as seen on The Retro Web 🙁

To prevent cluttering this thread about CF adapter UDMA issues, I send you a PM about a BIOS update for your board.

Cheers, Jan

Can I join this thread for a (relevant) question: how do I know what speed my CF should be seen by the system? Both my Lexar and Sandisk say UDMA 33, with 40 or 80 cables. The MoBo connector BTW has all the pins, the 80 wire cable has a blank on it, does that mean the MoBo (Shuttle 591P, Apollo MVP3) does not support higher speeds?

I'm using one of those CF adaptors which use a card opening at the back for the CF.

Thanks!

tony359 wrote on 2023-12-28, 20:37:

Can I join this thread for a (relevant) question: how do I know what speed my CF should be seen by the system? Both my Lexar and Sandisk say UDMA 33, with 40 or 80 cables. The MoBo connector BTW has all the pins, the 80 wire cable has a blank on it, does that mean the MoBo (Shuttle 591P, Apollo MVP3) does not support higher speeds?

I'm using one of those CF adaptors which use a card opening at the back for the CF.

Thanks!

from your mainboard manual...

Onboard PCI Bus Master IDE Controller Two Ultra DMA/33 Bus Master IDE Ports supports four IDE devices up to 33 MB/sec IDE transf […]
Show full quote

Onboard PCI Bus Master IDE Controller
Two Ultra DMA/33 Bus Master IDE Ports supports four IDE devices up to 33 MB/sec
IDE transfers and supports Enhanced PIO Modes 3 and 4 and Bus Master IDE DMA
Mode 2 devices

Ah, fair enough. So I shall be happy with what I have for now 😀

Thank you - sorry for not doing that myself!