I have not completed the custom pga-132 to PGA-168 SXL-only interposer, nor has anybody else to date. The design was shown to work, just need someone to produce the PCB.
Hmmm, now I'm intrigued. One of my brother's best friend inherited a small company that also does PCB manufacturing. I could ask him about the interposer thingy, as I'm would love to have this for my projects too.
@feipoa: any ideas how I should approach him about this? Just show him the pdf with the 3D stuff and ask him if he could produce it, or are there any specific documents that I could show him?
@Feipoa
Don't forget to show-off when you complete the setup.
Cannot remember if i ever tried EIDE2300Pro in Win95.
I keep VLB hardware on up to DOS + Win 3.1. The Windows 3.1 driver works fine as far as i can tell.
VLB drivers are problematic for Win95 and later versions of Windows.
Saving myself the trouble.
EIDE2300Pro DOS driver goes with the /T /M0:8 flags here.
@AndrewK2685
DTC 2278 is another badass VLB EIDE controller that just works.
So you are saying that on the same hardware DTC2278 is less problematic for you in Windows environment than Promise EIDE2300Pro, correct ?
DOS drivers are discarded in Windows i think.
@JohnBourno
It didn't sound like Feipoa had a complete implementation that was ready for PCBing.
But there are several projects going on around here that can benefit from access to manufacturer.
pshipkov, I didn't see /T as one of the options in the readme file for EIDE2300. What does this do?
JohnBourno, best to wait for someone with multi-layer CAD experience, free time, and personal motivation to come forward. So far, the layout and gerbers haven't been created in their entirety. If you know somebody at this company who has a keen interest in retro computers and wants to donate their time and expertise, I'd suggest they read this thread: Custom interposer module for TI486SXL2-66 PGA168 to PGA132 - HELP! and pickup where others have left off. All the information is there. To remain as small as possible, we're probably looking at a 4-6 layer design.
Plan your life wisely, you'll be dead before you know it.
I think jumpers can override the default policy, which can be overriden by driver flag.
But at some point i tested turbo/fast/normal speeds and ended up leaving it there.
For me yes, I had a bit less problems with the DTC.
It was not trouble free though. If I remember correctly
I had some issures with Win98SE and reverted to MS driver
don't remember the details though and haven't tested NT4.
For the promise one maybe there was something wrong
with my setup, can't really tell. But I'm gratefull its still working
DTC 2278 is another badass VLB EIDE controller that just works.
I've had one of these CIB sitting around for years. I think I tested it out once just to make sure it worked, but I didn't notice anything particularly interested about it. What makes it desireable?
I think I have another similar card from SIIG that I liked better for some reason, but can't remember why. Are those any good?
it is a good implementation of the EIDE standard.
fast and reliable.
covers wide range of storage devices.
from early mechanical HDDs to present day CF cards.
perf metrics are very consistent - not every VLB ide/scsi controller can offer that.
works on wide range of motherboards.
good drivers.
At first glance the DTC card really looks like nothing special. It uses a run of the mill Winbond multi I/O chispet. You would think that all cards using the Winbond chips would perform about the same, but it seems this is not the case. Is there something special inside that chip with the DTC sticker?
I'll have to dig out my SIIG VLB card one of these days. I'm kind of a SCSI guy, so I never bothered to benchmark it. I just remember it being pretty slick looking compared to some of the others. It seems you didn't have this model in your testbase. I remember it claiming support for something beyond PIO4, but maybe I'm just getting old. Pretty sure I still have the box and documentation.
This is the only photo I have of it (it was taken around 2009 or so). Maybe the only reason it looks slick is because of all those awesome stickers and yellow jumper caps. The drive connectors are also installed pointing upward, which is something rarely seen on VLB controllers. But even if you take all of that away, the PCB just has a nice green colour and looks like quality. I'm kind of a sucker for basing the performance of stuff on how it looks.
*edit*
I checked out a few old messages on usenet, and some people claim that the card is based on an adaptec chipset, while others claim it's a rebadged "Promise 2300+". Some other messages confirmed it uses a Promise BIOS. It seems to me it clearly uses a different PCB than the 2300PLUS.
Keep in mind that dtc 2278 models E/B and E (latest revision) are EIDE interfaces.
Previous versions are IDE implementations. Some of the early ones not even PIO 4 ready.
Maybe you had one of those and thats why the "meh" memory of it.
As for the SIIG controller - looks tidy. Somebody really put the effort back then.
Notice the 3 yellow capacitors in the lower right corner - that a is Promise EIDE2300Plus chipset, but with stickers.
SCSI guy yourself you say.
Nostalgic reasons ?
I kind of gave up on them.
Aren't nostalgic reasons the whole point of the hobby?
I think once all my SCSI drives die (so far all of tem still work), I might have to give up on the hobby. Either that or use a lot of SCSIDE bridges.
Can you explain how you can identify my SIIG card as having a promise chipset simply by looking at a few tantalum capacitors?
Aren't nostalgic reasons the whole point of the hobby?
Yes indeed, but the meaning of my question was slightly different. 😀
As for the SIIG controller.
The 3 caps configuration is typical for P EIDE2300Plus, but other controllers have them there too. So i had a second thought about my presumption.
Ran a search online and one obscure message says "just bought SIIG VLB EIDE controller with Promise bios", which reinforced back my initial stance.
There are several different versions of the Promise EIDE silicon. Unlikely to be PDC20230. Higher chance for PDC20630, judging by the level of integration.
Looking at PDC20630 based adapters - their layout widely varies.
Don't bet money on my P EIDE2300Plus presumption of course.
If you have the drivers - that will show immediately what's under the labels.
Running perf test will be very indicative too.
The card was made in early 1997, so whatever it is should be pretty late revision.
I really need to get myself a proper motherboard for testing VLB cards. I used to have a big pile of VLB 486 boards, but now all of my stuff is VL/EISA, and I have a Leopard board which of course is only 16-bit.
I ran some extensive tests with the Promise EIDE2300 Plus and the results caught me off guard.
1)
If you just try to load the original DOS driver in config.sys, e.g.
DEVICE=C:\EIDE2300.SYS /T /M0:8
Core tests yields around 4100 KB/s
I noticed the Promise installer is very particular if you want to use it. First, it insists on running the DOS/WIN31 installer from a diskette only and the diskette cannot be write protected. It also must run without anything loaded in config.sys/autoexec.bat. I complied. The installer runs some speed tests to determine the fastest and safest speed with your particular hard drive. The documentation which came with my EIDE2300 card states that it modifies the DOS/WIN31 driver files (EIDE2300.386 and EIDE2300.SYS) themselves after these tests, which took about 15 minutes for my system. I'm guessing this is why the diskette needs to not be write-protected, because it replaces the drivers on the diskette. This seemed like a really odd approach to me. The documentation states that once these files are modified, they cannot be simply used on other systems without running the lengthy tests again to customise them to a new system.
Once the installer finished its job, it says that my PIO and DMA speeds are 7 and the new EIDE2300.SYS file returned double results in Coretest, to approx. 8700 KB/s.
2)
In Windows 3.11, the customised Promise driver for Win31, EIDE2300.386, allows for use with 32-bit disk access instead of the default 16-bit disk access. HOWEVER, although the Promise documentation recommends connecting the CD-ROM and HDD to the Primary IDE connector, I am no longer able to use my CD-ROM in Windows 3.11. When I try to access the CD-ROM, I receive this beautiful blue screen error:
1Windows 2 3This program tried to access your hard disk in a way that is incompatible with Windows 32-bit disk-access feature (WDCTRL). This may cause your system to become unstable. 4 5Press any key to continue _
tao yen
If I comment out the EIDE2300.386 device in system.ini with a semicolon (;) and reboot, I am able to use the CD-ROM as usual, but disk access is 16-bit. Has anyone else run into this issue and if so how was it resolved?
3)
It is possible to use the CD-ROM on the Secondary IDE port (ISA) on the Promise Eide2300 Plus but this seems less than optimal. When I load Windows 95 with the CD-ROM port on the Secondary IDE port, Windows complains that the CD-ROM is being run in compatibility mode. Is there any way to resolve this? I remember having issues with the Promise Win95 driver, so I am using the default Microsoft ESDI/IDE driver.
Plan your life wisely, you'll be dead before you know it.
1. Never ran their installer. Just padded config.sys with their driver + args = ~14Mb/s in Coretest. This is a very strange and weird install process you described. Not sure what to say about it.
2. I don't use CD-ROMs in retro PCs, sorry - cannot provide any feedback. I bet controller+drivers were not tested against later/faster optical devices, so things going wrong is not surprising.
3. Good to know that second port works with optical drives. No idea if they can be connected to the EIDE interface.
As far as i remember from the distant past - CD-ROMS better stay on IDE.
Maybe there was wisdom in that.
I use CD-ROMs in every retro PC. For my BL3-75 system, I am using a CD-ROM from March 1997. It is a Hitachi CDR-7930, which is 8x. I doubt it being a year or two newer than the I/O card is the issue at hand. More likely is that Promise didn't care about compatibility proofing their Windows 3.1 drivers. I say this because the manual mentions if the user runs into issues with the CD-ROM on the primary port, try it on the secondary port. The booklet casually implemented foreshadowing.
I do not think the Secondary port is EIDE.
From my experience, I've had much better luck with hardware and driver compatibility with respect to SCSI CD-ROM drives. The controllers and devices were more expensive than IDE and most likely they put more time/money into their development.
By the way, I also have a ZIP-100 drive planned for this system. I'm not looking forward to what's in store given the fundamental errors with the CD-ROM.
Plan your life wisely, you'll be dead before you know it.
