Thanks Rik,

This is my first time ordering a PCB from a file (pretty non-commercial if I'd say so). I'm having some trouble finding the size of the board/number of layers etc from the file. Any help would be appreciated.
The design looks great, I'll get that ordered once I know more details on dimensions.

New filter caps and NPN coming in today, I'll see if that can remedy whatever issue was happening. Will update.

Rhys

Peter,

I've had a chance to do a brief search for my orphaned 9825A system ROMs with no luck so far. I'll continue to look next week, but it sounds like you're way beyond needing them at this point. Still, it will be good to locate them.
--
Jack
www.computerarium.org

I hate the French postal service. They put the package on hold because THEY had mislaid the invoice. There was no mention of that in the tracking info and now they tell the seller that it might take 40 days to correct it. Presumably in addition to the fiwe weeks I have waited already. Heck I would probably BUILD memory faster than that if I had a schematic...

On Fri, Mar 5, 2021 at 11:12 AM, Anders wrote:

anyone

someone

Would be cool if anyone did that, just for the heck of it :)

I have now stopped the ERROR 1 issue, but I still need to do some further testing to make sure that it is reliable.
Dyke suggested a ground shield for the ribbon cable, and this was easy to try quickly I tested this and it made no difference, a least I was able to rule this out as a cause as in the board design has no ribbon cable.
One thing that has been nagging at me for a while is that the output buffers can only drive to a maximum of 5V and as I'm using 74LS and in the final design 74ALS the logic highs won't be +5V. Also the output buffers are driving the lines all the way back to the processor not locally to the BIBs on the ROM interface board. Further, I noticed that on the Option ROM side of the BIBs on the ROM interface the address/data lines have 2k2 pullups to +5V. So, I thought I'd try adding some 2k2 pullups
to my prototype to give the output buffers a bit of help and bingo now it seems to work.

Thanks to everyone for their input.

Regards, Peter

Thanks Dyke, this is one of the things I had considered and is on my list try. I have some adhesive backed copper tape here in my lab, so it will be easy to test. The only reason the board is connected by a ribbon cable is ease of access while I'm debugging the hardware. Now that I have seen the the 9831 code work once I'm more confident now that I have a signal integrity issue rather than some corrupt bits in the binary image.
Regards, Peter.

Peter,  something easy to try as an experiment, not necessarily a fix is to emulate a ground shielded ribbon cable.  Turns out significant relief is available when 'shared area coupling' is encountered ( per unit length degradation in signal integrity in ribbon cables with sparse ground returns ) when  a conductive surface is placed ( copper clad pcb for instance ) in as close proximity to the plane of the ribbon cable suspected of being a root cause.  It's a long winded description of an observation made long ago that an 'equipotential surface' merely need be present ( closer the better the geometric planes formed by the ribbon cable and the surrogate ground return path ), that is not otherwise connected, just properly positioned.  Similar thing is accomplished by making the ribbon cable shorter in that both reduce the 'shared area' between loop currents ( di / dt phenomena )

If signal integrity is the issue, the long ribbon cable is more likely to succumb to 'shared area' coupling than 'shared impedance'.  Series termination is one way to slow down the edges, however, it seems to take a lot of R to begin to see improvement in TTL interfaces which can begin to produce DC offsets ( if I remember all this correctly )  good luck with testing,  Dyke

Paul, till a few minutes ago I thought I'd fixed the issue. I made one change to my board and the 9831 code started working without the ERROR 1 after switching the machine off changing the jumper for the 9825A code and testing that, then changing back to the 9831 it is giving the error again. As I have an additional 200mm of ribbon cable between my board and the system ROM socket I may look at terminating the address/data lines or as the input and output buffers are separate devices try powering the output buffers from 6V so I'm not running them at the absolute maximum of 7V. Currently I'm using 74LS240 inverting buffers and unlike the 98228 clone which drives a pair of BIBs on the ROM interface board I'm driving quite a way back to the processor. For now I think the issue is with my design not the 9831 binary image.

Regards,
Peter

_._,_._,_

---

Well I am in the process of building my own, I would be done but a piece on the light over my work table broke and I had to take time out to turn a replacement.  I will probably finish it up tomorrow as it it get very late here. 

You do know that there are TTL chips that are functionally equivalent to the BIBs, except that the direction function is reversed and the pinout is not the same I am going to use 74LS640s, they are just like an inverting 74LS245.  I have not decided what I will use for address latches yet but I have a good stock of TTL to choose from.

Yes you long cable could be an issue however I encountered problems trying to extend the memory bus on my 9835 when I was designing a RAM card for it.  When I was building prototypes for ROM modules I made an extender that just barely bring a ROM slot outside the machine.

Paul.

Paul, till a few minutes ago I thought I'd fixed the issue. I made one change to my board and the 9831 code started working without the ERROR 1 after switching the machine off changing the jumper for the 9825A code and testing that, then changing back to the 9831 it is giving the error again. As I have an additional 200mm of ribbon cable between my board and the system ROM socket I may look at terminating the address/data lines or as the input and output buffers are separate devices try powering the output buffers from 6V so I'm not running them at the absolute maximum of 7V. Currently I'm using 74LS240 inverting buffers and unlike the 98228 clone which drives a pair of BIBs on the ROM interface board I'm driving quite a way back to the processor. For now I think the issue is with my design not the 9831 binary image.

Regards,
Peter

Hey Rik, that's interesting. I had a close look at your photos and apart from the three 74368 inverting buffers, the 1820-0068 is a 7410 and the 1820-1112 is a 74LS74, so no complex logic there. The PCA number is 09831-66529 so the board is specific to the 9831. My 9825A has the same connector plugged into a piece of PCB material with no traces, I assume to stop it flapping in the breeze. This looks like a way of identifying that the hardware is a 9831 rather than a 9825A. Possibly, this would. explain the ERROR 1. Also if the binary image was corrupt I would expect the software to crash rather than consistently giving error messages on certain operations and then being able to get the system prompt back by pressing CLEAR or RESET.

Hope you get a chance to take some clear images of both sides of the PCB.

Regards,
Peter

_._,_._,_

---

I just fired up my 9825A, which does not have that card, with a known good 9831A OS ROM and I don't get those errors that you saw.

Paul.

On 3/4/21 6:13 PM, Achim Buerger wrote:

Possibly the 9000 300/400 is to old for any web browser software. Netscape 1.0 was released end of 1994 when the 700 series was alread state-of-art.

People compiled those browsers on lots of different platforms, whether they were released for them precompiled or not. It's just a matter of finding the binaries or, well, compiling it..

The predecessor of Netscape was NCSA Mosaic; I built that on several 68K UNIX platforms at the time, but not HP, as we didn't have any of those systems.

-Dave

--
Dave McGuire, AK4HZ
New Kensington, PA

Take a look here:
https://www.webdesignmuseum.org/old-software/web-browsers

Hi all,

I found Netscape 2, 3 and 4 installation files, but for PA-Risc (9000 700/800) only and uploaded them to https://groups.io/g/VintHPcom/files/PA-Risc
Possibly the 9000 300/400 is to old for any web browser software. Netscape 1.0 was released end of 1994 when the 700 series was alread state-of-art.

Achim

The following files have been uploaded to the Files area of the VintHPcom@groups.io group.

• /PA-Risc/netscape-v202-export_hppa1_1-hp-hpux_tar.Z
• /PA-Risc/netscape-v301-export_hppa1_1-hp-hpux_tar.gz
• /PA-Risc/netscape-v30-export_hppa1_1-hp-hpux_tar.gz

By: Achim Buerger <achim.r.buerger@...>

Description:
Netscape Navigator install files for HP PA-Risc (9000 700/800).

Hey Rik, that's interesting. I had a close look at your photos and apart from the three 74368 inverting buffers, the 1820-0068 is a 7410 and the 1820-1112 is a 74LS74, so no complex logic there. The PCA number is 09831-66529 so the board is specific to the 9831. My 9825A has the same connector plugged into a piece of PCB material with no traces, I assume to stop it flapping in the breeze. This looks like a way of identifying that the hardware is a 9831 rather than a 9825A. Possibly, this would. explain the ERROR 1. Also if the binary image was corrupt I would expect the software to crash rather than consistently giving error messages on certain operations and then being able to get the system prompt back by pressing CLEAR or RESET.

Hope you get a chance to take some clear images of both sides of the PCB.

Regards,
Peter

Roland Dietiker <roland.dietiker@...> added folder /INTEREX

The following files have been uploaded to the Files area of the VintHPcom@groups.io group.

• /HP 9825/Crowbar_gerber.rar

By: Rik Bos <hp-fix@...>

Description:
The HP 98X5 crowbar Gerber files.

---

The following files have been uploaded to the Files area of the VintHPcom@groups.io group.

• /HP 9825/crowbar.pdf

By: Rik Bos <hp-fix@...>

Description:
HP 98X5 Crowbar circuit diagram