RQDX1 and RQDX2 woes..

[MattisLind]

Hello RQDX1 and RQDX2 experts!

I have a BA23 chassis with a 11/23 CPU some memory and then a choice of a RQDX1 and RQDX2 controller. I have a ST412/RD51 disk to go with it.

When starting off I didn't know neither of the RQDXn boards nor the drive or the cables were OK, so there were plenty of unknown variables.

Formatting the ST412/RD51 with the RQDX2 (Firmware seems to be 10.0D according to the labels on the EPROMs) resulted in the stepper motor going from track 0 to track 305 back and forth a couple of times then it just hung. The drive has been used in some DEC system but not necessarily with any of these controllers.

Then testing with the RQDX1 which gave a completely different result since after it finished going back and forth it wrote that it had ran out of revectoring tracks (or something similar message). The RQDX1 firmware is a bit fishy since the EPROMs are marked 23-238E4 and 23-239E4. But someone has removed the UV window cover at some point and put som ink there instead. They might have been reprogrammed. The number on the EPROMs would imply version 7 firmware according to this doc.

The same doc mentions that the RQDX1 and RQDX2 is essentially the same board:

The RQDX1 later underwent another ECO that allowed it to be resident in
any Q-bus slot after the CPU. This ECO involved changes in etch work,
micro-code and nomenclature. The controller name and module number
changed from the RQDX1 with module number M8639-YA stamped on the
handles, to the RQDX2 with module number M8639-YB stamped on the handle.
The firmware revision became known as version 9.4E, and was compatible
with the RX50 and RD5n (n = 1, 2, 3) series devices.

As mentioned above, the RQDX1 controller became the RQDX2 with an ECO to
the board itself and a change to the firmware allowing it to support the
latest mass storage devices on MicroPDP-11s and MicroVAX I's (RD52s).
The announcement day (MicroVAX II) ROMs (version 10.0D) were available
to be used with the RD53s, and have been upgraded once, to the version
10.0E micro-code. These two versions of the firmware will boot any of
the RD devices available for the MicroPDP-11s or the MicroVAXs.

Does this mean that it is possible to put a RQDX2 firmware on a RQDX1 board? As far as I understand the problem was that the RQDX1 has to be the last board since it wasn't handling NPG / Bus grant as it was supposed to.

How can one deduce the actual version of the firmware? Aside from the label on the EPROMs. Is there a version string somewhere to be retrieved?

To rule out hard disk and cabling problems I attached the drive to a AT style MFM controller and ran Speed Stor from a floppy (it must have been more than 25 years ago I did this last time). The drive was flawless! Not a single bad track after half an hour of surface analysis.

I then got back to the RQDX1 again which now didn't even started to format. It immediately complained, not even trying to format the drive.

On the other hand with the RQDX2, this time it didn't hang! It formatted it successfully this time. I then was able to initialize it under RT-11 and copy all RT-11 files and then boot of DU0!

But still the problems remain that the RQDX1 is not working. Of course this could be a faulty board but since it was working and moved back and forth the formatting process seemed to work.

So what is the experience with RQDX1 and RQDX2 boards? Is certain firmware better than others. How can I know what firmware is on the board? Is the RQDX1 and RQDX2 firmware interchangeable? Why was the RQDX2 able to format the drive after it has been low level formatted on an AT clone? What is the process the RQDXn does to figure out what drive is connected (this process seems to confuse the mfm emulator that David Gesswein has designed)?

 

[MattisLind]

So I did some more research into this myself.

What are the differences between a RQDX1 and RQDX2.

Hardware

The etch has changed but not so much. They didn't even bother changing the E15 designator to E16...

The PROM chips in the red area are the microcode for the DMA sequencer. They all have been updated between RQDX1 and RQDX2. RQDX1 the chips are marked 193F3 to 201F3. On the RQDX2 these are marked 224F3 to 232F3.

The green marked PAL chips are one PAL16L8 which in the RQDX1 is marked 098J5 while the RQDX2 uses 207J5. The other is a PAL16R4 which on the RQDX1 is marked 074K3 while the RQDX2 uses a 115K3.

Finally there are two 74F74 added to the RQDX2 compared to the RQDX1 board. These are the blue marked chips.

These updates probably solves the problem with the RQDX1 not propagating the BIAKI and BDMGI signals. Thus there can be no interrupt or DMA devices after the RQDX1 board in the bus.

Firmware

There are various revisions of the T11 software for both the RQDX1 and RQDX2. As it turns out the RQDX2 firmware CAN be run on the RQDX1. This has been indicated by old posts in various news groups. I can confirm this as well. I tried a RQDX2 10.0D firmware in the RQDX1 and it not just started up correctly it also was able to read a disk that was formatted by the old RQDX2 controller (which was the donator of the firmware). Copying files in RT-11 worked just fine. But if works equally well with other operating systems like RSX-11, VMS etc I have no idea about. But it probably worth a try?

So why didn't the RQDX1 work in the first place? The usual embarrassing story about changing to many variables at the same time. The root cause was probably that the disk had some peculiar format that made both controllers spit. Formatting the disk with a ST class MFM controller cleared this. But meanwhile I had also changed the line driver chip for the write signal (since it was socketed by previous owner of the board). Well. Now reading those letters isn't getting easier. I mixed up a 9638 with a 9639 inserting the wrong chip. So when using the RQDX1 with new firmware and a RQDX2 formatted disk it worked to read the directory once. But when I tried to bootstrap from it subsequent reads from the directory failed. The RQDX1 had probably wiped the directory somehow. Installing a 9638 driver instead and reformatting the drive and everything worked happily...

Different version of the firmware can be downloaded here. Please note that the RQDX1 firmware is 23-238E4 and 23-239E4. Apparently there is a overlap here with a KDF11-B firmware which has the same ID. The RQDX1 version 7 image identities are confirmed by this document and also by the marking on the chips in my RQDX1.

MFM emulator

I have been trying to hook up the MFM emulator that David Gesswein has designed with the RQDX1 and RQDX2. And it doesn't work well at all. Even though I have an known good image from the newly formatted ST412 drive. The RQDXn are simply incapable of detecting what kind of drive is connected. They probably do all sorts of magic detecting this and since the MFM emulator isn't aware of this it responds wrong.

I will connect a logic analyzer on the cable to see if it is possible to get a better understanding what magic the RQDXn is actually doing.

 

[MattisLind]

RQDX1 and RQDX2 woes - now working with MFM emulator

RQDX1 and RQDX2 woes - now working with MFM emulator

MFM emulator updates

I have been corresponding with David on the subject of not being able to use the MFM emulator with a RQDX1 or RQDX2. I put a logic analyzer over most of the signals and got a trace on how the RQDX1 behaves when formatting a ST412 drive. David then modified the MFM emulator code slightly and I tested it. I was then able to not only use the image read by the mfm emulator under RT-11 but also to format a brand new ST412 emulation drive. Now I have tried to format a RD53 which has 1024 cylinders and 8 heads and it passed ZRQBC1 and ZRQAH0 XXDP utilities just fine. It just took ages for it to format the drive. Operation under RT-11 works equally well.

One slight problem is that with the first SD card I got occasional SEEK ERRORs when formatting. David told me that this probably because of the buffering in the kernel so that finally it has to write and then if the SD card is slow it could take time. I saw seek times in range 100-200ms which the controller didn't accept. Using a faster flash or as per Davids suggestion - to use a ramdisk solved this problem.

So for all RQDX1 or RQDX2 owners: upgrade to at least firmware 10.0D and then use David's MFM emulator if you cannot find a working RD53, RD52 etc drive.