#pcb design and interference

1 messages · Page 1 of 1 (latest)

hot kettle
#

Hey, I am learning more about this. And we have quite a few people with insight on this stuff, so I wanted to start a thread about good practices to create pcb designs that are robust against interference and inductive currents and different devices causing interference with each other.

MobiFlight also is enabling more people to create even more impressive and sophizticated devices, so I think it will be even more important to understand these things, because the more stuff there is in close proximity, the more important this topic becomes.

We had many people with issues using potentiometers that fluctuate a lot, or MAX7219 getting messed up by other devices, or also messing up other devices because the data signals are fast and loose wiring or just other traces in close proximity pick up the interference.

Many sim builders are learning this stuff as they build, and having some good understanding of what the pitfalls is probably very helpful.

I found this youtube video that seemed to explain a bit of the fundamentals, but I wanted to also start a thread where we can share our best practices and experience.

https://www.youtube.com/watch?v=ImkvsQEY6OY

YouTube
LearnEMC
Circuit Board Layout for EMC: Example 1

This example illustrates the steps involved in assessing and redesigning a simple printed circuit board in order to meet EMC (electromagnetic compatibility) requirements. Issues of interest on this board include high speed circuitry between external connectors, poor decoupling, and high-speed traces that are longer than necessary. http://www.Lea...

▶ Play video
#

There are also example 2 and 3 on the series.

ripe owl
#

I haven't got around to it yet, but I've wanted to experiment with series resistors for MAX7219 and shift register data lines. This will include reading through the MF- firmware to get some indication of signal speed for those devices. If there's enough time (i.e. the signal pulse are long enough) I might also experiment with small capacitors in the data lines in conjunction with resistors to create a filter network.

I'll be probing with my rather inexpensive USB-oscilloscope to see if I can measure differences, potentially also playing with 2-layer vs 4-layer boards just out of curiosity.

My idea is a board with an RP2040 on it as well as 2 MAX7219 and 2 74HC595 ICs on it. Each MAX will drive 6 digits, each 595 will drive a number of LEDs (not sure how many, yet). My plan is to arrange the board to resemble a radio panel. Once testing is finished I might part ways with the boards to anyone interested in using them for a radio panel.

Should my experiments show any usable results I'll upload diagrams and results here as well as making recommendations for circuit design and layout for these specific circuits.

I won't get around to it for a few weeks as I'm leaving for vacation this weekend, but I've been mulling over this for a while now.

quasi ether
#

I used a dirt cheap logic analyzer to similar ends recently to check how fast a shift register was being polled on a commercial product, to see how much attention I would have tp put into interfacing reliably with it

hot kettle
#

I would test the filter stuff with 50cm crappy Chinese Dupont cables to see if there is a difference. I did a radio panel with 2 MAX chips and input and output shifters and it worked fine

#

and it was my super crappy autoeouted horror in Easy eda. maybe I was just lucky

#

but maybe the most problems come from the jumper wires

#

having those filters in the digit modules would be a cool thing to have in MF shop for example, because a lot of folks do build with wires since making a PCB is a lot of learning

ripe owl
#

I could do that, testing with crappy dupont connectors and wires I mean.

I haven't had noticeable issues on my PCBs. There is a chance, however, that my circuits and layouts just barely work. I'm interested in seeing how close I am to things not working, how bad things need to be for them to fail and what strategies might be implemented to resolve our at least reduce issues within a reasonable cost and complexity margin.

#

I agree that adding a resistor and/or cap to the signal lines on the MF boards might be cool. If there's a noticeable difference to signal integrity this could be a relatively cheap and simple way to resolve an issue several people seem to have struggled with recently.

I'm not proficient enough in code to fix this kind of stuff in the firmware, but I'm hoping I can contribute some help on the hardware side of things🙂

Thanks for the ideas, @hot kettle
Any other ideas are certainly welcome!

hot kettle
#

yea, I meant that a rats nest of dupont wires is a good test scenario

#

the root issue with max7219 issues is also the really bad quality max modules which is another thing

ripe owl
hot kettle
#

but even "good" ones struggle with the wiring interference I would think, the filter stuff might be interesting to try indeed

quasi ether
#

Wondering if there's a use case for ferrites somewhere

#

Seen them used for signal integrity before, but most of the signals I build filters for need much lower pass frequencies than an smd ferrite can provide

ripe owl
hot kettle
#

these are usually the issue

iu_.png
ripe owl
hot kettle
#

but I guess just having the datalines for shifters and maxes close to each other in long bundles of wire is the nightmare scenario

#

just to see if things make a differenc. those modules also have issue with power consumtion sometimes

#

I had some that work fine, and one that needed the capacitor on the power pins

#

to work at all

#

I think there are just the worst batch of out of spec chips they used for them to get them done cheap

ripe owl
hot kettle
#

yea they usually just have the empty footprint

#

also a great way to create a short circuit 🙂

#

but this made the module work in testing

#

anyway, I also want to learn more about the ground return path stuff and in general best practices about pcb design

#

the MAX7219 is probably the point that pushes the requirements of pcb layout a bit further and might be the point where we find the problems

#

the other devices like switches and leds are way less strict on the requiirements

ripe owl
#

The problem with 7 segment displays is the lack of direct return paths... the current doesn't go to ground but rather back to the IC in what can end up being a fairly large loop...

I suspect a solid ground plane under both the anode and cathode traces might couple them and reduce noise. This is specifically why I want to use 2 MAX chips on a 4-layer board with the inner layers blanked out under one of them and it's corresponding display, essentially turning it into a 2-layer board in that area

sly cove
#

This chat rather exploded since I last looked at it, and I haven't had time to thoroughly read through everything here (I skimmed it), but having just gone through several revisions for my radio panel design a couple things I learned...
Long wires for anything having to do with the MAX chips is a bad plan, period. My initial design connected the max chips to an Arduino mega that was not on the same board as my radio panel components and I (with pretty much everyone here's help) spent tons of time troubleshooting erratic 7-segment behavior. The wires I used for that were 18 AWG stranded and were roughly 16-20" (400-500mm) in length. These were the wires for CLK, DIN, and LOAD(CS) connections.
Eventually I tested replacing those wire connections with 6" (152mm) long shielded wires and that ended up solving the interference issue.
Of course at that point I was moving more towards the RP2040 design where everything was on the same board. In researching USB I did read/see several examples of adding a ferrite bead to the USB data lines, but that was geared more toward USB-C/3.0+. I did not incorporate ferrite in my USB micro/2.0 design and have not had an issue.
The data lines from the MAX chips should be run as far away from other high speed signal lines (PWM, USB, etc.) as possible and ideally have a nice solid ground pour on the opposite side of the board. The more complex designs become, the more challenging it is do do this. I struggled quite a while with just the radio panel 2-layer board design.

hot kettle
ripe owl
#

Two nearly identical pcbs with MAX7219 chips on them for another project. Series resistors used in data lines:

Probe 1: 330 Ohm
Probe 2: 10 Ohm

I have yet to get the damned scope to trigger the way I want it to for better resolution. This was measured at 1MHz sample rate at roughly 20 microsecond signal pulses (50kHz clock rate)

I hope to get some better resolution results when I actually do some slightly more scientific testing😅
This was just me probing about trying to troubleshoot code related issues🙈

20260221_182845.jpg 20260221_182752.jpg
quasi ether
#

at these sample rates I'd argue it is pretty much impossible to get a good impression of any interesting transients

ripe owl
# quasi ether at these sample rates I'd argue it is pretty much impossible to get a good impre...

I agree, I'm going to have to figure out how to get my scope to trigger better and then get some 16MHz or more readings.

I was quite surprised tbh by how little was going on in the data lines. The board has an 8V to 5V SMPS and still there was little noise even when sampling at higher rates. I just wasn't able to capture the transients the way I wanted.

Looking forward to spending more time with this and probing around some cheapy modules

#

Who knows, I might look into some nicer scopes as well🙈😅

ripe owl
#

Does anybody have any experience with the low-mid range oscilloscopes they'd be willing to share... I'm having a hard time figuring out which of the stats on all these devices are relevant and which ones just look great on the brochure🙈🙈🙈

#

I'm not looking for anything above 400€, preferably below 300€

hot kettle
#

I saw this, that actually tempts me

#

its just the multimeter form factor but there are comments that say it seems precise

#

no idea

#

but no other data, mine is also about watching youtube videos and trying to judge people whether they do it for the money or actually talk of their experience

ripe owl
# hot kettle I saw this, that actually tempts me

Interesting device... if the product pictures are to be believed this product also incorporates a thermal camera😅
I hadn't considered sifting through AliExpress listings to be honest. Might do that a bit as well before deciding on what to get