Strange output during testing

[Scott Atkinson]

Hi All,

I've just completed the soldering of all the LED's and other components, except for the switches. When I run the test program, I get all the lights working correctly. But I also get some random led activity as well. I can live with it, but was wondering if anyone else has had the same problem.

Before I solder the switches, I also want to know if this is normal behaviour. When the test program gets to the switch test. I get random keypress detection. Is this due to floating inputs on the pi (i.e. no switches connected), or is it another problem. When I short out the switch contacts, I get the press detected properly.

Am I being overly cautious, or is there a possible problem? 

Cheers,

Scott

[Heinz-Bernd Eggenstein]

I doubt very much that the result will be different after soldering in the switches, so you probably want to debug this before doing more soldering.

There is a troubleshooting guide for download here: https://groups.google.com/g/pidp-8/c/QdPFqUrTXO0/m/e_RmvAOABAAJ some details are not up-to-date but it will still help.

It explains that the switches are read out in groups, and that the GPIOs used for reading whatever group of switches is selected at a given time use the internal (so nothing you soldered) pull-up resistors, so an open switch or a missing switch will be sensed consistently as a HIGH level on the input GPIO, there is no floating input even if the switch is missing. The only way that the signal can go to LOW (and the software thus sensing a closed switched) without a switch installed  is if something on your PCB is pulling the input GPIO down. The switch sensing GPIOs are also involved in turning the LEDs on and off, so I guess the sporadic " random led activity" you mentioned and the sporadic switch activity are actually related!

Forgive me the silly question: are you 100% sure about the polarity of all your LEDs ??? An LED that is soldered in the wrong way is about the only thing I could think of, other than an error with the PCB traces (unlikely), to cause this.

Oscar does provide some spare parts for the kit which is nice of course but it doesn't allow to check the soldering work with the IKEA-Test (Any parts left after assembling?). Then again, I don't see how any forgotten element could pull those GPIOs down.

If all that doesn't help you probably want to check whether your PI is in an undervoltage state at the time you see the failures: when using the HDMI output, a little yellow lightning symbol in top right ? corner would indicate that, if you are using the Pi headless via ssh etc, the

dmesg -T

command  van be used to see if undervoltage events happened and when.

Good luck hunting down the problem.

HBE

[Heinz-Bernd Eggenstein]

Oh, and while you still haven't soldered in the switches, you still have time to decide whether you want to solder in the deposit switch the way the instructions tell you to do it or in the way that is more authentic, but that would then require a tiny change on the PCB or in the software, see https://groups.google.com/g/pidp-8/c/1s7g54Y2-Y8/m/H1fwzaSRFwAJ

I wish I had known that detail before, I'm now to scred to damage something and will probably not desolder my deposit switch to turn it around.

HB

[Scott Atkinson]

HI HB,

All of the LED's are ok, and light up correctly.

I think it may be the connector for the Pi, as looking at the soldering using a magnifying glass, it looks like some of the soldering could be dry joints. I'm also going to go through the circuit as you suggested using the guide, and see what happens. I'll let you know what happens when i do.

Thanks for you reply to my questions.

Cheers, Scott

[Rick Murphy]

I had a problem with my PiDP8i where it worked fine outside of the case, but when mounted the top row of LEDs (the PC) wouldn't light up.

Pressing on the PC board could bring back to life. It turned out to be one of the pins for the Pi connector wasn't soldered at all. Took a bit of head scratching to figure out where the problem was.

    -Rick

--
You received this message because you are subscribed to the Google Groups "PiDP-8" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pidp-8+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/pidp-8/73121feb-ccc7-49df-a16e-b8e2934a3ef5n%40googlegroups.com.

--

Rick Murphy, D. Sc., CISSP-ISSAP, K1MU/4, Annandale VA USA

[Heinz-Bernd Eggenstein]

THe funny thing tho is that I cannot see how a cold solder joint could cause the problem in this thread: that the Pi occasionally sees a closed switch when there aren't any switches installed yet.

Consider the schematics on Oscar's site that expains how the switch sensing (and LED blining) is managed:

https://static.wixstatic.com/media/ce503a_f36338fce4234da9a7a6bc7be3ca8e75.png/v1/fill/w_630,h_341,al_c,q_85,usm_0.66_1.00_0.01/ce503a_f36338fce4234da9a7a6bc7be3ca8e75.webp

For the Pi to think that the switch is closed, it has to read a "zero" at the GPIO connected to "col1" in this diagram when that GPOI is acting as an input with internal pull-up resistors activated.

So for this to happen, something has to drain charge from "col1".  Without the switches installed, the only thing that col1 is connected to , via a resistor, is an LED (several of them, in the full schematics), but because of their polarity, they won't let thru any significant charge coming from "col1". Now if you introduce a cold soldering joint here and there in the schematics, it won't help to drain that charge as well...

HB 

[Obsolescence]

Hi,

Apologies! I've been absent for quite some time due to family health reasons.

But - my proposal for fixing this would indeed be:

1- reflow all the pins on the Pi connector

2 - then clean them with alcohol (vodka, cotton tip is fine) to get rid of flux. Amazingly, this has solved about a dozen cases of spurious keypress problems! I'd never have believed it. Excess flux can be a problem.

3 - check if the metal connectors on the Pi do not pierce through the insulation tape shielding it from the PiDP board.

Scott,

No, there should be zero random LED activity in the test program. Email me if the above does not fix the issue. Worst case, I send you a new PCB plus small electronic parts...

Kind regards,

Oscar.

[Neil Higgins]

Hang on a minute!!! You want us to use Vodka?? No way! Isopropyl alcohol (painter’s variety).

[Warren Young]

On Wednesday, December 1, 2021 at 2:20:44 AM UTC-7 1955...@gmail.com wrote:

You want us to use Vodka??

In countries where ethanol isn't taxed to the sky as a morality play, it's an excellent option, provided it's pure enough and the "impurities" are nothing but water. So, vodka, 190 proof Everclear, etc.

I've been informed by people with organic chemistry knowledge that it does a better job on rosin fluxes than IPA, but if IPA is cheaper where you live, it's also fine, provided it's high enough in purity. Low-purity IPA will leave a white haze on your board which is possibly conductive, and there's a lot of water in the less pure preparations of IPA, which is a danger of shorting contacts if you don't dry it properly after cleaning.

 

Isopropyl alcohol (painter’s variety).

I was under the impression that denatured alcohol as sold for paint thinning is methanol, the "denaturing" was meant to make it undrinkable. IPA doesn't need that step, which is one reason it may be cheaper in any given locale.