brs-tester improvements (DEC negative logic)

[djg]

My goal is to get a chip tester to help restore one of System Source's pdp-9's. https://lists.vcfed.org/archives/li...org/message/SM4MZN3X3CPMRGFBLVSEPIO2GS5AI5DO/. Looks like some minor changes would be good before ordering boards. Contacted author and should have design files this weekend. Assuming the eagle imports ok into kicad I will do a some changes and order boards. Will also want to use it for my linc-8.

I would like to get comments from users on what changes would be good to roll in hardware before ordering boards. I've started looking and this summarizes what I know so far. There are a bunch of things marked incomplete. If those aren't just software it would be good to know.

I saw todo here. Need to dig more to see which need hardware changes. If you know let me know.
Also found possible updates.
Also changes in vectors.

Changes
1) I thought might as well make the +10 and -15 variable so can do margin testing as part of the board test. Did margin test on my straight 8 when it was working and found one bad card. So not critical but could help. Will look at switchers so only needs one input voltage. If not will implement the change to single power connector to prevent swapping from possible updates.
2) Renames from possible updates
3) Not sure advantage moving measurement circuit to plug in board from possible updates. Thoughts?
4) Change to BA & BB pins. Currently they are connected to +10 and -15. For R series cards only dual I know of are r210 and r211. For those they should be connected to test signals instead. Looks like BC is ground on r210 and r211. r210 has signals on BA and BB. Some cards like G209 do need +10 and -15 on BA, BB. If I understand correctly the relays on level shifter isolate the pins so if they are off the G adapter boards can put what they want on the board pins so can put the +10 and -15 on BA and BB.
Looks like we aren't testing BB/DB input to card so should add another level shifter so all pins can be tested.
5) Hardware for pulse measurement. Assume we need a small FPGA or raspberry pico (only 23 GPIO boards, RP2350 chip 48) to capture timing and changes. Have people had failures do to slow outputs that we should add capability to capture propagation delay for all tests while we are at it? I don't remember any slow output problems in my troubleshooting.


For the vectors showing incomplete tests
b105 - File doesn't say what is incomplete
b132 - File doesn't say what is incomplete
b204 - Test resistor inputs. Not sure if that needs hardware changes.
b213 - Need sw support for pulse. Assume hardware doesn't need changes.
g208 - Hardware mod needed. Details in file. This seems to need another board like G219 uses. Board does not seem to be included in design files.
g209 - Need sw and hw for testing. Does not provide information on add on board.
g219 - Need sw and hw for testing. Add on board is in design files. Assume vector file not finished to use it.
r111 - Unclear what is incomplete
r203 - Work in progress, vectors need to be finished
r210 - Needs changes to BA & BB pins.
s203 - Doesn't say what is incomplete.
s205 - Doesn't say what is incomplete.
s206 - Doesn't say what is incomplete.
s220 - Needs '0' output load tests. Doesn't have vector file at all. Does have r220 file so assume using that.
s602 -Need hw for pulse measurement
s603 - Need hw for pulse measurement

 

[m_thompson]

The Pi-5 on my tester is powered from a little 5.0V switcher. The Pi periodically reports a low input voltage. The wall wart that you can buy from Pi puts out more than 5.0V so no low voltage messages. I need to see if I can raise the output voltage.

I did the first pass at a User's Manual. I would be happy to continue that work based on comments from other users.

I had plans to write a GUI tool to control the tester, but haven't done much. Having a tool where you could manually control the output/input pins would be really useful for creating vector files and debugging a broken flip-chip.

 

[djg]

Files are eagle. I have an old 7.2.0 that I was able to convert them from binary to xml format that kicad can import. Will need some cleanup. If someone has more modern eagle/fusion and wants to try conversion PM me and I'll send you a link to the files. Don't know if newer will import better but may be worth a shot.

 

[tingo]

Perhaps it would be simply easier to just redraw the schematics and redo the layout in KiCad?

 

[djg]

Perhaps it would be simply easier to just redraw the schematics and redo the layout in KiCad?

Import issues weren't that significant so less time than redrawing. All done. Now on to the desired changes.

 

[thunter0512]

I wonder if there is a viable alternative to the many relays in the design. The relays are relatively expensive, have a limited life and create electrical noise when they switch.
Also Pulse amplifiers (R602, R603, etc) are currently not well supported. One work around is to attach an oscilloscope and monitor the pulses manually. Maybe someone can think of a better solution.

 

[djg]

I wonder if there is a viable alternative to the many relays in the design. The relays are relatively expensive, have a limited life and create electrical noise when they switch.

Haven't had a much time as expecting so I'll leave this alone for now to finish it. I also don't have one to test changes so more chance breaking something. If I get bored after I have it working may look at other options.

Also Pulse amplifiers (R602, R603, etc) are currently not well supported. One work around is to attach an oscilloscope and monitor the pulses manually. Maybe someone can think of a better solution.

That was the change to add a FPGA so it can capture timing info the PI can check. Still looking at it.