Did you try re-installing the OS? The controller may use a different translation scheme than the adaptec.

Also, WB cache can cause problems with some SCSI controllers. You may want to try with WT instead as well.

I had a typo. I meant to say that I am using the Intel DX4-100 in WT mode. I have edited my original post.

Different translation scheme - I was kind of afraid of something like that. Unfortunately, this is my only [remaining] 50-pin narrow SCSI test drive and I didn't really want to reinstall DOS just for testing purposes. I'm not sure what else to do for testing the DTC. I really need some more IDE-to-SCSI solutions.

To show the jumper settings, I have also attached a quick PDF scan of the paper manual. I figure this may come in handy for someone who needs it.

The card contains two SRAM chips, which I confirmed are good using my SRAM tester.

The card defaults SW14 to ON, which is to disable Synchronous Negotiation. I have set this to enabled. This is the only switch I adjusted from the default configuration.

I believe the issue is related to the 1 GB HDD size limit of the DTC 3274 host controller, while my Adaptec 1542CP host controller has the BIOS option to allow for extended BIOS translation for HDD's larger than 1 GB. As my Seagate SCSI drive is filled beyond the 1 GB limit, the DTC was having issues.

My goal was to benchmark the DTC 3274 VLB, Adaptec 1542CP ISA, and Adaptec 2842A VLB. As I am out of 50-pin narrow HDD's and do not want to format my remaining test HDD, I pulled my ACARD IDE-to-SCSI adapter out of my 386 and attached an IDE HDD. Shown below.

However, I am concerned about the accuracy of the comparison when using a bridge adapter. When I connect my regular 50-pin SCSI drive to the DTC, I get this:

When I attach the IDE drive to the bridge adapter to the DTC, I get this:

With Adaptec adapters, the utilised transfer speed is shown after the HDD's model number. Following the Seagate SCSI drive, it shows 0484. I'm not sure what this is. Perhaps the HDD's firmware? Upon connected the ACARD bridge adapter to the DTC, I get 05.0 and this looks a lot to me like the drive is connected using the asynchronous transfer protocal, which is 5 MB/s. The fast synchronous mode allows for 10 MB/s. I have the DTC's dip switches set for Synchronous negotiation and fast synchronous transfer, so I'm not sure why the ACARD is getting put into async mode. This will invalidate the benchmark results.

The ACARD model is the 7720U, and according to the manual, supports synchronous rates up to 20 MB/s. So... I might need to pilage the 68-pin to 50-pin adapter from my 486 and use an Ultra160 SCSI drive for this test. For whatever reason, I cannot get the 80-pin SCA to 50-pin SCSI narrow adapter to work on these older SCSI controllers. Anyone have experience with these? Looks like this,

I clearly remember, the same SCSI HDD (a 2GB 50 pin Seagate) booting properly from a 1540CP w/o any additional hoppla, whereas needing an Adaptec driver to be installed tru config.sys to boot with a 1520B (even needed to be formatted and partitioned with Adaptecs own utilities).

The second card, 1520B was also a "BIOSless" card that need to be configured via a utility rather than entering the BIOS and changing the parameters there: https://storage.microsemi.com/en-us/downloads … =aha-1520b.html

There are DOS drivers for adaptec SCSI cards and you need to use them accordingly for "BIOSless" controllers, as afar as I can understand. This is from the Adaptec DOS driver package text file:

1If your new SCSI drive is not under BIOS control, you can follow
2   these steps:
3   
4   1.)  Install the appropriate driver for your controller in the Config.sys using
5        the following syntax (this is for all ASPI drivers)
6	Device=[drive:][path]Drivername /D [Parameter]
7
8        Example:  Device=A:\ASPI2DOS.SYS /D /P140 /Q10 
9
10   2.)  Install the driver for the removable disk device, ASPIDISK.SYS using the following
11        syntax
12        Device=[drive:][path]ASPIDISK.SYS
13
14   3.)  If the removable drive is to be installed using vendor software,
15	for example, IOMEGA TOOLS, do not use the Adaptec disk driver.  
16        It is only necessary to load the driver for the installed host adapter.
17
18   4.)  If the drive is unformatted, AND you are using the Adaptec driver, the disk
19        must be partitioned with the program AFDISK in order to have the OS assign
20        a drive letter.  Run AFDISK and follow the prompts to partition your drive.  The
21        drive is high level formatted for DOS at the same time, so the DOS FORMAT command
22        is not needed.
23

Adpatec DOS driver package:

https://storage.microsemi.com/en-us/speed/scs … dosdrvr_exe.php

This is what happens with Adaptec cards, I have no idea about DTC or any other brands. I hope this will help someway.

I used all kinds of SCSI socket comverter adapters (68 to 50, 50 to 68, 80 to 68, 80 to 50 & 68) by the way, and never ever had a case that they work with a given SCSI controller card and not the other one. As long as everything are already in working condition and the SCSI devde (HDD, optical, etc.) in question is compatible with the controller, of course. My experience is limited to Adaptec brand cards though.

Luckily, my Adaptec 1520B, 1540, 1542, and 2842 all have BIOSes. The DTC has a BIOS as well, just not an onscreen utility. The solution to my original problem is that I cannot use more than 1 GB with the DTC controller. My issue now is trying to get some non 50-pin narrow SCSI drives to work. The 68-pin to 50-pin adapter I took from one of my other machines works. So my plan now is to setup DOS/Win3.11 on a 1 GB drive, then compare the 3 controllers, as well as see if there is any performance hit with the ACARD+IDE vs. U160 on a 50-pin adapter. This is starting to seem like a lot more work than I had in mind. Half of me is wanting to put it all away now.

Those cheapo SCA adapters are improperly terminated (they don't terminate the upper bytes) thats why they usually don't work. How do you know the ACARD adapter is only running at 5MB/sec? The SCSI BIOS is almost certainly showing the drive's firmware revision which the ACARD pulls from the ATA ID command. Drive benchmarks should easily get over 5MB/sec from that Western Digital drive, even with the bridge. If unsure, test it with a fast CF card+adapter.

Concerning the SCA adapter, I was using a cable with terminator. The SCA adapter, from what I could discern, doesn't have any termination option. I wonder if the terminator on my 50-pin cable has gone bad.

I don't know that the ACARD is running at 5 MB/s, just that the 5.0 looks awfully suspicious. If it is just a coincidence that this HDD has a firmware version 5.0, then let me change the IDE HDD and see if the 5.0 changes to another number.

whereas needing an Adaptec driver to be installed tru config.sys to boot with a 1520B (even needed to be formatted and partitioned with Adaptecs own utilities).

How did you manage to do this? You can't boot without a driver installed in config.sys, but for the config.sys to be executed, you must have booted already..? That kind of can't really work. 🙄

You don't need drivers to boot from a SCSI card, drivers are used for different OS systems to know the card is there and for using stuff like removable media drives and optical drives in DOS.
There is no HD size limitation with SCSI cards, but you have OS issues (like 2GB fat partitions for DOS etc).
You need a 50 pin SCSI HD for that card with proper termination (single ended not LVD). The SCA adapters have no termination and neither do SCA hard drives so you need a proper terminator at the end of the cable.

You might need to reformat the drive when jumping from different manufacturers SCSI cards of that era.

The problem with the SCA to 50-pin adapters is the lack of "upper byte termination" (cheap 68 to 50pin adapters have this problem too). The upper 8-bits of the wide SCSI bus on the drive need to be terminated since they are not connected to anything on a narrow 8-bit controller.

Unknown_K wrote:

There is no HD size limitation with SCSI cards, but you have OS issues (like 2GB fat partitions for DOS etc).
You need a 50 pin SCSI HD for that card with proper termination (single ended not LVD).

When using the Adaptec SCSI controller w/extended BIOS translation, Dos 6.22 will see 2 GB. When using the DTC SCSI controller with the same HDD, DOS 6.22, that is, FDISK, will see 1 GB. I hypothesise that if I turn off extended BIOS translation on the Adaptec host controller that DOS will also only see 1 GB.

Concerning the SCA, I've given up on it with 50-pin narrow host controllers. They seem to work just fine with Ultra wide controllers. For 50-pin SCSI controllers, I can use the 68-pin version of SCSI drives just fine with passive 50pin-to-68pin adapters.

Not sure about "BIOS translation" but the SCSI card should see the whole HD (I have quite a few SCSI cards here in ISA and VLB). DOS will see the whole drive too (depending on the version of DOS) and then you just make 2GB or smaller partitions until you use the whole drive up (can only boot DOS from the first partition).

68 pin SCSI drives have native termination on them (you just need a mini jumper to enable it).

You could probably just get away with using a 50 pin to 68 pin SCSI adapter on the SCSI card and then use a 68 pin SE cable with terminator on the end and those SCA drives with generic SCA to 68 pin adapters.

I have already tried a 50-pin female to 68-pin female on the SCSI card end and it just don't work. I use an non-terminated 68-pin cable, and terminate on the HDD with jumpers. I don't understand exactly why because the 50-pin male to 68-pin male adapter, when placed on the HDD side, works just fine with a 50-pin cable going to the SCSI card. I even measured a few pins on 50-pin female to 68-pin female adapter to ensure that they went to the same places as the 50-pin male to 68-pin male adapter. The SCSI card shows some gobbly-gook for the HDD when using the female/female adapter. I only have one male/male adapter and had to take it out of a cased system. I have ordered 2 more. I have actually run into all this mess before but didn't remember the exact solution for each scenario.

NJRoadfan: I have connected a different brand IDE drive to the ACARD and it shows YAR4, so I guess the 05.0 display on the original IDE drive was just a coincidence and it is in fact the firmware revision. I have so many quite IDE HDDs, I just wish I had more ACARDs.

Unknown_K, when using the DTC, FDISK will only see 1022 MB as the size of the entire drive. I haven't yet tried to create multiple 1 GB partitions, but considering it can only see a 30GB drive as 1 GB, I'm guessing that I'd have to add another physical HDD to get more than 1 GB out of this controller. It may very well be that

I will proceed with testing. I find it interesting that the DTC has onboard DIP SRAM. It comes with 16K standard, but I can increase it to 64K. Any idea if this will increase HDD performance?

Here's the 50-pin male to 68-pin male SCSI adapter I am using.

Here is a screenshot of Fdisk when using the DTC 3274VL and a Seagate ST336705LW. Notice it reports the total disk size is 1022 MB. The partitions it is showing on there were created on another computer. I tested fdisk after loading the DTC driver, and when not loading the DTC driver. The result was still 1022 MB. Some kind of SCSI controller limitation?

Here is a screenshot of Fdisk when using the Adaptec AHA-1542CP and a Seagate ST336705LW. It reports 35003 MB, which is the size of the disk.

Here is a screenshot of the Adaptec's BIOS configuration utility. Notice the option for extended BIOS translation for drivers larger than 1 GB. Out of curiosity, I disabled this feature and checked FDISK. I was surprised when it still showed the HDD size of 35003 MB. These results lead me to believe there is some limitation of the DTC which is causing the HDD size limit.

My plan is to benchmark the DTC 3274VL, AHA-1542CP, and the AHA-2842VL using, both, the ST336705LW Ultra160 hard drive and the ACARD 7720U with a Western Digital WD800 Caviar drive. I haven't done a lot of HDD benchmarking. Is there a particular DOS and Win3.11 benchmark for HDD which everyone likes?

I was thinking of using Winbench96, the Disk Winmark benchmark for Win3.11. For DOS, maybe Sysinfo (disk), Checkit3 (disk bench), DrHard v3 (disk), and Speedsys (random access time, buffered read speed, linear verify speed, and linear read speed).

I created a 1 GB primary DOS partition on the DTC 3274 using the ST336705LW drive. I installed MS-DOS 6.22. To speed up the installation of Windows 3.11, I thought I'd remove the DTC 3274, put the AHA-1542 back in, connect both the ST336705LW and my 50-pin narrow SCSI drive, load Win3.11, then copy files over. Well, after some time of copying, the system hangs. I started to copy files over in DOS, but it took too long. So I put both HDD's in my dual PIII 850 with integrated Adaptec SCSI, booted W2K, but W2K will not see the HDD formatted with the DTC 3274. It will see my primary 50-pin narrow SCSI drive just fine though and can read from it. So I'm wondering if the DTC sets up the HDDs such that only the DTC can read/write to the drive reliably?

Frustrating. When I try to boot into Win3.11, I get these errors:

Not sure what this is about. I think I'll just bench the SCSI cards in DOS. I want this off my test bed.

Well, I took about all I could with this dumb DTC SCSI card. It has been put away. I recorded some benchmark values, but not nearly to the extent I had planned.

Results are with Speedsys only. Motherboard is the PC Chips M912 v1.7.

ISA AHA-1542CP w/ST3217N (50-pin narrow)

Random Access Time = 16.3 ms
Buffered Read Speed = 1745 KB/s
Linear Verify Speed = 2640 KB/s
Linear Read Speed = 2046 KB/s

VLB DTC 3274VL w/ST336705LW (ultra160)

Random Access Time = 6.9 ms
Buffered Read Speed = 4188 KB/s
Linear Verify Speed = 2464 KB/s
Linear Read Speed = 3514 KB/s

VLB DTC 3274VL w/ACARD 7720U + Western Digital WD800 Caviar IDE drive

Random Access Time = 9.7 ms
Buffered Read Speed = 4147 KB/s
Linear Verify Speed = 9261 KB/s
Linear Read Speed = 3121 KB/s

The results are generally unremarkable. VLB SCSI controller with U160 drive is obviously faster than an ISA SCSI card with an ordinary 50-pin narrow drive. Seek time is 2.5x faster for the VLB/U160, as generally could be expected. The transfer rates of the VLB SCSI were about double that with the ISA SCSI. As for the ACARD w/IDE, the seek time is less than the U160, but substantially better than with the ISA+narrow SCSI. The transfer rates with the ACARD were about the same as with the U160 drive on this system. The glaring difference with the ACARD setup is the "Linear Verify Speed", which was 4x greater than with the U160 drive. Why is that?

My primary objective was to compare the DTC 3274VL with the Adaptec 2842VL, but due to the software difficulties I am having with the DTC, I gave up. It could also be an issue of this particular PC Chips board and the DTC and not necessarily a fault of the DTC. The 1 GB limitation of the DTC is a substantial drawback though. I did connect up two HDD's to the DTC, so if you need 2 GB of HDD space, you would have to use multiple hard drives.

derSammler wrote:

whereas needing an Adaptec driver to be installed tru config.sys to boot with a 1520B (even needed to be formatted and partitioned with Adaptecs own utilities).

How did you manage to do this? You can't boot without a driver installed in config.sys, but for the config.sys to be executed, you must have booted already..? That kind of can't really work. 🙄

You are quite right, my mistake. 😊

The computer was booting from another drive (CF if I remember correctly) and SCSI disk was the secondary drive. The problem was DOS 6.22 was not aware of the SCSI drive (even it was only 2GB in size). After I installed some ASPI drivers as suggested by Adaptec in config.sys, and using Adaptecs own parititoning and formating utilities, the HDD was completely working ok.

tayyare wrote:

derSammler wrote:

whereas needing an Adaptec driver to be installed tru config.sys to boot with a 1520B (even needed to be formatted and partitioned with Adaptecs own utilities).

How did you manage to do this? You can't boot without a driver installed in config.sys, but for the config.sys to be executed, you must have booted already..? That kind of can't really work. 🙄

You are quite right, my mistake. 😊

The computer was booting from another drive (CF if I remember correctly) and SCSI disk was the secondary drive. The problem was DOS 6.22 was not aware of the SCSI drive (even it was only 2GB in size). After I installed some ASPI drivers as suggested by Adaptec in config.sys, and using Adaptecs own parititoning and formating utilities, the HDD was completely working ok.

That's completely normal behaviour. The 1520B BIOS is only there to boot from a SCSI drive when there's no IDE drive present.