Troubleshooting on LED matrix.

0da6a0f95b3c9e1a

Hallo Latigid on. Murphys law schlägt mal wieder voll zu…
Jedoch erst einmal Grundsätzliches: ist es gewünscht / erlaubt sich hier in deutsch zu verständigen? Ode sollen doch besser alle “Ausländischsprachler” mitlesen können. Oder vielleicht sogar per PN, wenn es zu tief in eure Konstruktionsgeheimnisse eintauchen sollte?
Also: 2 Tage lang, nachdem ich alles komplett verschraubt hatte, lief Matrix ohne Probleme. Doch jetzt kam sie beim Booten nur bis zur Anzeige “Disconnected”, obwohl Loopa angeschlossen war. . Keine einzige LED leuchtete auf.
Ich also alles wieder aufgeschraubt, etwas an den Kabeln gewackelt, Funktion wieder da! Testmodus: alles wie gewünscht. Strom ist auf jedem Board, Kabel soweit auch alle durchgängig.
Kabel wieder sorgfältig verlegt und befestigt, zugeschraubt: NEIN!!! Wieder nur Bootscreen, keine LED, “Disconnected”. Ab jetzt konnte ich sie nicht mehr zum Leben erwecken. Im Testmodus (ohne SD-Card) reagieren alle Slider und der Joystick, aber keine LED, ob die Tasten überhaupt ansprechen, habe ich auch noch nicht herausgefunden.
Jetz benötige ich erst einmal einen Plan, wo ich welches Signal abgreifen kann, und wie ich diese Signale verfolgen kann, um den Ort Unterbrechung einzukreisen…

0fc5608d995b44ff

English would be good 🙂 So that other builders with potentially the same problem can better understand what’s going on and how the problem was finally solved! 🙂

If you have access to a scope, Andy should have more informations to trace the pins going into J44 and J89 of the Neo boards.

But it probably makes sense to start with first things first - i’d recommend to disassemble again and test with a single connected neo board first - my suspicion is that the power wiring might not be perfect (“Wackelkontakt”) - measure the TRACO switchers on the neo boards to see if they have proper 12V input and 5V output.

Best regards and good luck!
Peter

17703227d9536da1

I think English would be better just as Peter says to potentially help out others.

I also agree that it is probably something to do with cabling. Were the molex connectors assembled with the specified parts? Were they (incorrectly) soldered together? Or correctly crimped with a proper tool? Are there any collisions especially under the first neo board?
If you build the wirescanner you can also test those cables. If there are disconnections with a bit of wiggling, it can point to assembly issues.

Best,
Andy

0da6a0f95b3c9e1a

Ok. I took some measurements so far. Isn’t J44 responsible for LED action? I have no signal there, except some weak but steady pulses on pin 3. No single Neo is working. Also annoying is, that the program runs until it shows: 1 disconnected, 2 disconnected and then stops. No way to get into the menue. No matter if any board is connected.

17703227d9536da1

(please help me to help you… there are unanswered questions…)
J44 sends the serial LED data on pin 6. It must connect to Core R9 and R8 is a +5V pullup to PB6.

You could just be super unlucky and have the first LED in the chain dead. The THT resistors carry the data signal to/from the mm8 connectors.

0da6a0f95b3c9e1a

Unanswered questions:
Incorrectly soldered molex connectors: i think no, it worked for 2 days properly. 12V on every board.
No need to crimp the contacts, got some prewired connectors. Collisions: couldn‘t find any. Wiggling on any connector doesn‘t show any effect any more.
I‘m working now for more than 20 years in our company, developing, programming, manufacturing and maintaining meanwhile thousands of microcontroller circuits. (Working on agricultural machines.) So I thought it would be a walkover to get this thing running. But something went wrong here completely…. 😅
Further measurements following during the day.

0da6a0f95b3c9e1a

In letzter Zeit falle ich wohl einfach zu leicht in Panik.
No power on the 5V line of Neo 1. Ordering parts…..

17703227d9536da1

Interesting, so a deceased voltage regulator or diode maybe? Or something with the cables?
The Core board actually does supply +5V on pins ¾ of J89. I don’t recommend it as a permanent fix, but you could consider to bridge the J89 pins to the THT via:

0V is not connected through the ribbon but you could consider to bridge D70 pins to supply the 0V from J89 into the neo board 0V net. (It would also return through the Molex cable).

I haven’t tried this, but it is just an idea to get power to your first neo board. The current capacity of the ribbon might not be enough to light a lot of LEDs.

You mentioned pre-wired cables: are the lengths optimised to fit in the case? Sorry for the tight spacing! Peter insisted on a second BLM port! 😆

0da6a0f95b3c9e1a

Desoldered UREG 1(and c105 also) but still got a short circuit between +5V and GND. One of the other capacitors may be faulty? Let‘s find out… any suggestions how?
2nd strange observation: only on neo1, not on the other ones, I have also a connection between the shortcutted lines (5V and GND) to the switch plate / metal spacers. Where could there be a connection?

17703227d9536da1

Synthfox Desoldered UREG 1(and c105 also) but still got a short circuit between +5V and GND. One of the other capacitors may be faulty? Let‘s find out… any suggestions how?

You could go over the whole board with hot air reflow and touch up all caps?
Have a look also on 165 chip pins 15/16 (15 is 0V, 16 is +5V).

17703227d9536da1

The only THT pins that carry +5V are on the voltage regulator and J89A. The micromatch header shouldn’t have long pins, so maybe the Vreg?

0da6a0f95b3c9e1a

Last try, before I declare this mission as a complete failure…
I managed to get the LED matrix working again, but now most of the buttons are no longer responding. Only when I press one of the four buttons in the middle (9th) column on the first neo board in the row (I also tested them separately) I will get a strange lightshow. (See photo) I swaped all cables, replaced the ICs 1 and 2, same result. May the problem now be on the axis board?

17703227d9536da1

Are you running in test mode with no uSD card inserted?

What does it mean “managed to get the LED matrix working again”? So the LEDs will light up properly now? LEDs are not in a “matrix” but are serially chained through the 8-pin headers. Can you control the LEDs with the sliders?

Slider action could also be visualised on the OLED test mode, even if no LEDs light up. So you can verify if the sliders send the correct data.

So the buttons are not giving the expected responses? In test mode you should see the LEDs light up in a cross shape. You can try connecting the Core cables to another neo board first to see if it makes a difference. The strange behaviour could be explained by a backwards IC or bridged pins on the shift registers, or the IC types being mixed up.

Do you get proper +5V (measure across 100n caps)? What have you done with the BAT54 diodes?

What power supply are you using and does it give proper +12V?

17703227d9536da1

You can check the integrity of the serial clock located on pin 2 of 165 registers with an oscilloscope.
Serial inputs come in from the previous register on pin 9
What do you see on pin 14 of the 595 registers? The same clock is on pin 11 of this chip.

0da6a0f95b3c9e1a

Yes, I am testing without SD so far.
Sorry, the LED are of course sequential. The problemso far was in deed a pin touching the button grid when I fastened the screws. LED working means, boot animation runs, I can control the lighting with the axis board sliders, sending correct data.
No cross shape, see photo. Only the mentioned 4 Buttons do a response at all, on each board which is connected as the first in the row. No BAT54 istalled. Should I?
IC issues: I did not touch the ICs since they were working properly last week. Then I exchanged IC1 and 2 whith the same result.
Power Supply is this one, I can see no irregularities measuring the output on the oscilloscope.

I will do the suggested measurements this evening…

P.S. Wäre es möglich, einen etwas größeren Ausschnitt des Schaltplanes zu erhalten, um die Signalwege unter den verschiedenen Boards besser verfolgen zu können? Ich schwöre, ich werde ihn nach erfolgreicher Anwendung gut zerkaut runterschlucken, damit er nicht in “falsche” Hände gerät. 😆 Aber so gehe ich bald am Stock mit dem Teil!

17703227d9536da1

Maybe the +5V short circuit damaged some of the ICs? I can’t really explain what you see otherwise.
Do still try to connect to the 2nd/3rd/4th neo board first up, also try disconnecting subsequent boards to try to localise the fault.

If you can measure a significant difference in +5V voltage between neo boards, the BAT54S could help to bring them back together. Another option is to bridge the BAT54S pads to connect 0V from board to board rather than back to the PSU through the Molex connectors. Yet another option is to bond the 0V lines from board to board with thicker gauge wire, rather than relying on the ribbons or Molex. This could reduce ringing or over/undershoot in the clock lines. But if you say that the signals look good in the scope trace, then I don’t think it’s that.

0da6a0f95b3c9e1a

Update: I have clock signal through all neo boards. But when I connect J89 on the axis, clock collapses. Next chip to swap: neo, ic3?

17703227d9536da1

The clock is wired in parallel as you see, so maybe there is a bad cable or backwards/wrong IC2/3 on axes?
Any button response if axes is disconnected?
The basic layout/routing of the shift registers on axes is the same as above, just it is directly wired to the buttons/LEDs

0da6a0f95b3c9e1a

Finally…It’s done again. After replacing 2 more 165’s and one 595 and loosing a lot of nerves, I could reintegrate it into my setup. But by the way I learned a lot about adressable LED chain and how to realise a switch matrix…

17703227d9536da1

Gratuliere! Weißt du genau woran das Problem mit den Chips liegt? Überspannung oder Überhitzung beim löten bzw. gebrückte Pins?

Das du was gelernt hast ist in der Tat etwas gutes und definitiv ein Teil von DIY.

0da6a0f95b3c9e1a

latigid on Beim Zusammenschrauben hatte sich das Metallraster der Taster verbogen (vergessene Spacer unter dem 1. Neo-bord! ), genau auf die Lötpins der Steckverbinder durchgebogen, wahllos Signal- und Stromkontakte überbrückt und dadurch die diversen ICs zerschossen. Ein / zwei schlecht gelötete Dioden fanden sich auch noch. Drückt mir die Daumen, bald kommt der Seq dran…