Power delivery issues with JST vs dupont cables

[graydetroit]

Hi, I'm building a music sequencer device using the Teensy 4.1, and I'm really confused about an issue I'm encountering involving power delivery (I think) when using JST cables vs. typical Dupont cables.

Project info:
- The project uses various components: a 2.42 SSD1309 OLED display (SPI), a TLC5947 powering 22 LEDs, 28 switches (matrixed), capacitive sensing contacts, 4 dual-channel DACs.
- The project is a multi-board project that requires some sort of wired connection between the main (bigger) board and the I/O (smaller) board.
- The I/O board is the one which the Teensy is mounted to.
- The Teensy is powered externally by 5V via a Recom 78E5.0-0.5 dc/dc converter, which is powered by a Recom 78E12-0.5 dc/dc converter, which is in turn powered by a mains wall wart. This is so I can have +12V, +5V, and +3.3V signals for various components I'm using.
- For now, since I'm prototyping, I'm just using 90-degree pin headers and JST and/or dupont connectors to test the connections between the main and I/O boards.

Problem:
I put together a simple test program which cycles through all the LEDs and then runs the Adafruit example OLED graphics example code. When connecting the I/O board power and ground signals to the main board using a 4-pin JST connector, it seems as though the Teensy gets stuck in a restart loop. The OLED and LEDs turn on and the Teensy's LED all hang for a second, and then everything shuts off and the cycle repeats.

However, when connecting the I/O board power and ground signals to the main board using typical Dupont connectors, the program runs fine. The JST connector is oriented correctly, so it's not a matter of it being reversed or anything like that.

Here are two videos showing this issue:

First video -- using the JST connector (data sheet here).
Second video -- using the Dupont connectors.

Here is a picture of the I/O board 4-pin section (top view):


Here is a picture of the main board 4-pin section (top view):


Since this is a multi-board layout, I'm hesitant to share the actual schematics. I can if need be. But I'm wondering if the problem is something easily identifiable just by observing the behavior. I was thinking maybe it's due to the JST connector somehow allowing _too much_ current to be consumed causing the Teensy to restart, while the Dupont connectors are possible limiting the current a bit more? Maybe I need to limit the current that the LEDs consume via the current-limiting resistor used by the TLC5947 (currently using a 5.1k resistor)?

Thanks so much in advance for any help.

 

[PaulS]

Perhaps it's not the connector type but the thickness of the wires used, that may be the issue?
From the video, it looks like the wires used with the Dupont connectors are thicker.

Paul

 

[MarkT]

Try to source cables or wires with a definite wire gauge spefified - some of the cheap ones have homeopathic amounts of copper in them and are only good for signal.

 

[KenHahn]

Definitely sounds like a wire gauge issue causing excessive voltage droop when the LEDs kick on.

If the JST wires come from China, which include most sold on Amazon, it is common practice for them to be labeled with a larger AWG than they actually are. I have JST pig-tail assembles here that are labeled as 22AWG on the insulation and sold as 22AWG wire assemblies but are actually only 26-27AWG when measured with calipers.

 

[graydetroit]

Thank you all. I will probably just try and use a couple terminal blocks with some guaranteed 20 AWG wires instead of using pre-crimped cables. This was a very frustrating issue, thank again!

 

[graydetroit]

Plot twist, when I added thicker individual wires ~20-22 AWG, then I saw the same problem as before. I checked one of the dupont wire gauges and they are more like 26-28 AWG, much thinner. I then swapped out each individual signal wire and found that when either the ground or 3.3V wires are thicker, then the problem occurs. I'm thinking it's _not_ the LEDs necessarily, because I put a much higher resistor (24K) for the IREF pin on the TLC5947 and I didn't see the problem go away.

I almost think it's the OLED but I wasn't able to find the current consumption specs for the module I'm using (DIYMore 2.42" SSD1309 OLED module). I guess next steps is that I could try putting a resistor between the 3.3V line and the VCC pin on the OLED module. Not really sure, but I'm thinking the thinner gauge wire reduces the current bandwidth on the display enough to allow the Teensy to not be overwhelmed.

 

[MarkT]

Ah that may simply be overloading the 5V DC-DC converter which is only rated at 0.5A. I would recommend passing 12V to the sub module and adding point-of-load DC-DC coverter on that board. The 12V converter can deliver more power, with less current, and voltage droop isn't an issue if you then convert back to 5V

Another possible problem is having too much electrolytic decoupling capacitance for the converter(s), they have usually modest maximum capacitive load ratings.