Notes on a recent kit assembly and an issue using the Pi 5

[Steve Platt]

Hello all,

I recently got a pidp-11 kit.  A quick report on my experience assembling it.

1. Super-easy to assemble.  I've been soldering for many many years, so any tight situations, well, weren't.  The instructions were clear and well written. The videos were clear and well-presented.  Even though the ordering was different, both made the assembly very straightforward.
2. The extra parts thrown in were greatly appreciated.  Even if I didn't use them, aside from one diode (broke it, don't ask) and one nut that wanted to be on the floor.  Somewhere.
   
3. The standoffs going from the circuit board to the front panel were nylon, not brass.  The threads aren't particularly good (short nylon threads are like that), and they wouldn't grip the (brass) threaded inserts on the front panel.  I got some longer nylon standoffs from my parts box, cut them down to size, and they worked well.
   Still, I'm not overly happy with them.  When I removed the board, some of the screws holding the board came out (as they should have), but some of the standoffs came out of the front panel.  An annoyance.
   I ordered a set of brass standoffs and will replace the nylon with brass.  I'll probably use some lock-tite to hold them in place, and use some silicone spray in the top threads to make the screws easier to insert (and remove).
4. I installed the simple/base software on a Pi 5.  Simh ran nicely (RSX-11M, haven't been on that in decades!  And Unix version 3 and 5, my first unix experiences!).  But.. I ran the various switch tests on the switch panel.
   I got to the one (on the video) where you turn S21 through S0 on in sequence and do a Load-Addr.  Worked well until S21-S1 were on.  Turned S0 on, hit load-addr, and around 4 LEDs in the addr bar went dark.
   Reproducible, sort of.  Sometimes the same LEDs, sometimes different ones.  The console (tty) showed that the led-displayed values were being read; later memory examination showed that the led-displayed values were also being written.
   All of the diodes were in the right direction.  All of the LEDs worked individually. All of the soldering rock-solid.
   Isolated it to S11-S0.  One of the strobe groups on the switch-reads.  If *any* of the switches S11-S0 was set down, the correct LEDs glowed.  Put all of them up, and some of the LEDs went dark.
   Sounded like a fanout issue.  Too many switches up, voltage drops, some of the 1s read as 0s.
   Swapped power supplies (put a larger one on the system).  Same problem.
   Swapped the Pi-5 for a Pi-4.  Issue solved.  Pi 4 worked well.  So, for now, I'm sticking with the Pi 4.
   
5. No real complaints outside of that.  A great kit, and I'm enjoying looking at the blinking right now.  Next up, probably play with RSX-11M, get the LISPF4 source up and running on it (I used it a bit in the early 80s), and see what it can do.
   

So, anyone else see this LED issue with a Pi-5?

Next up: refreshing my Linux, in particular, this version of Linux and its desktop (right now, I'm running the full 32-bit Linux with pidp11 installed on the pi user).  And on it, how can I set up the pi user to, after login and desktop creation, start a single terminal running pidp11.  I have a shell script that checks for pidp11 already running; if not, it'll start up a new version.  So just starting lxterminal on this script once, after the desktop has been created, will work well.

Steve

[Tom Lake]

There was an issue a long time ago where some LEDs would go out when too many were lit. I believe a change in resistor values fixed it. The Pi 5 draws more current than the older versions so maybe that's the problem.

[oscarv]

Steve,

On Saturday, April 27, 2024 at 12:09:03 PM UTC+2 steve...@comcast.net wrote:

1. I got to the one (on the video) where you turn S21 through S0 on in sequence and do a Load-Addr.  Worked well until S21-S1 were on.  Turned S0 on, hit load-addr, and around 4 LEDs in the addr bar went dark.
   Reproducible, sort of.  Sometimes the same LEDs, sometimes different ones.  The console (tty) showed that the led-displayed values were being read; later memory examination showed that the led-displayed values were also being written.
   All of the diodes were in the right direction.  All of the LEDs worked individually. All of the soldering rock-solid.
   Isolated it to S11-S0.  One of the strobe groups on the switch-reads.  If *any* of the switches S11-S0 was set down, the correct LEDs glowed.  Put all of them up, and some of the LEDs went dark.
   Sounded like a fanout issue.  Too many switches up, voltage drops, some of the 1s read as 0s.
   

Hmm! I have almost never heard of this problem again, since we moved to using 820 ohm resistors for the switch groups. There's 3 of them just above the switches on the PiDP-11 PCB. They sink current for a group of switches. They used to be 1K resistors, and that was too high a value sometimes when all switches in a group were set.

Switching to 820 ohm resistors fixed that problem, except... I hear of this problem about once every 18 months. And there are two solutions:

1. replace 820 ohm with something between 400 and 800 ohms, it is not critical, it just allows more current to be sunk.

2. the underlying cause, as far as I could determine, is small current leaks around the GPIO connector pins from flux. Reflowing the solder on those pins and cleaning up with alcohol has been the remedy in all 4 cases so far. This was the fix back in the days of the 1K resistors too.

Let me know if #2 solves your problem? And otherwise #1 will solve it too.

Kind regards,

Oscar

oscar.v...@hotmail.com for quick reply!

[Steve Platt]

Yeah, imagining something along that line.

All of the LEDs work.  LED test switch, all on, bright and solid.  No flicker.

I'm imagining a similar issue with the switches.  I'm new to the Pi world, but have around a decade or so working with Arduinos.  On these, I configure switches on an OC line with an internal pullup.  I've designed and built switch arrays (custom-built electric piano keyboard encoders, soldered a LOT of diodes...); on these, the resistors are all internal to the Arduino units.  Just a lot of strobe and sense lines.

Is this how the GPIO lines work on the Pi?  Are the individual lines configurable as Input, Output, or Input-Pullup in the same manner?  Or do they require an external pullup?  (Don't have the PiDP schematic handy; can't remember seeing it, but I did trace out some of the circuitry while assembling it.)  So I'm not sure if S0-S11 are open or closed when in the down position.  But I could see how, when all in the up position, there might be a drop on the lines.  Typical of fanout problems -- driving too many lines at once from a single source.

Now that it's up and running with the Pi 4, I'm somewhat loth to desolder resistors and solder in new ones.  Ah well.  But, if I were so inclined, any suggestions on which resistors and what new value might solve the issue?

--
You received this message because you are subscribed to a topic in the Google Groups "[PiDP-11]" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/pidp-11/CxzRC7x6S7Y/unsubscribe.
To unsubscribe from this group and all its topics, send an email to pidp-11+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/pidp-11/7a3cd02f-0d6c-421d-bbea-380197f17c2fn%40googlegroups.com.

[oscarv]

Steve,

For the technical details (yes, they are as you expected), see here: https://obsolescence.wixsite.com/obsolescence/pidp-11-technical-details

On Wednesday, May 1, 2024 at 5:18:57 PM UTC+2 steve...@comcast.net wrote:

Ah well.  But, if I were so inclined, any suggestions on which resistors and what new value might solve the issue?

So my first suggestions would be to reflow the solder on the GPIO pins and clean around the pins with alcohol (vodka has been reported to be fine too). Because this is a tiny current leak somewhere, nothing problematic but that is the cause.

But a resistor around 500-600 ohms would also do the trick.

Kind regards,

Oscar.

[Steve Platt]

Oscar,

Thanks -- if I switch back to the Pi 5 I'll swap out the resistors.  I know which ones they are.  I'll start with swapping just the resistor on the S0-S11 bank since that's the bank that has the problem.

Oh, I already reflowed the solder, and I use a board cleaning spray (flux solvent) to clean my boards as needed.  They're basically flux-free -- two full baths to be sure after the first bath didn't fix the issue.  Still didn't work, sigh.

Steve

--
You received this message because you are subscribed to the Google Groups "[PiDP-11]" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pidp-11+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/pidp-11/95110c38-5654-453b-b185-49ae975834b1n%40googlegroups.com.

[oscarv]

Steve,

On Thursday, May 2, 2024 at 3:14:04 AM UTC+2 steve...@comcast.net wrote:

Oh, I already reflowed the solder, and I use a board cleaning spray (flux solvent) to clean my boards as needed.  They're basically flux-free -- two full baths to be sure after the first bath didn't fix the issue.

Thank you for that information - yes, swapping the resistor will fix this. The underlying cause, now that the GPIO is taken out of consideration, should then really be an out-of-spec diode (if that is a known thing at all? I am not an EE). Or messy solder point elsewhere causing a current leak, but it does not sound like you're prone to soldering sloppiness.

Hmm. Although easily fixable I keep thinking of the root cause.  

If you or anyone else has an idea or suspicion, do let me know. I might learn something. Schematics attached for anyone who wants to think along.

Kind regards,

Oscar.