So when I run measure axis noise on my new huvud, I get:
Axes noise for xy-axis accelerometer: 1590.706536 (x), 2330.756013 (y), 2173.935340 (z)
It’s not great, well over the 1000 limit defined as “problematic” in the docs. However, I still get some pretty clean measurements, at least I think. It looks like everything I’ve gotten before.
Should I feel confident using these measurements? If so, what’s the point of measuring axes noise?
Second, doesn’t apply to my at this time but the docs say you should be between 1 and 100, and over 1000 is likely something wrong. So what if your noise lands in between 100 and 1000?
Anyway, I connected another adxl to my printer and it’s noise is much lower. What should I look into to get cleaner measurements?
If you have another ADXL with a much lower noise level, I would use that. Having noise that high will probably make your measurements imprecise, and it may cause the input shaper to not be as effective as it can be.
That’s frustrating since the one that is noisy is the one that is built into my huvud.
Perhaps try tightening all the parts first, or it might just be the fan that’s causing this noise. See if you can do a measurement with all the fans off and see if the noise is more acceptable at that level.
If not, a simple double sided tape should be approximately secure enough to use an ADXL to measure the resonnance - not ideal, but good enough.
All of my fans were off. The problem is there is nowhere physically to tape a separate adxl345 to and nowhere on the board to connect it. 
I see… welp, it doesn’t hurt to try using those measurements. It shouldn’t cause too much of an issue, but do some test prints with it with you there the whole time. If it starts getting violet, send ‘M112’ immediate to terminate the print. Once you start feeling confident about it, it should be okay.
According to my tests, noise seems mainly related to the quality of the supply voltage.
With adding an additional capacitor and a ferrite bead into the VCC line, I was able to significantly improve noise.
Regarding your graph:
- Never seen such a zigzag graph
- The first spike at 0 Hz looks suspicious. Could be the noise
You will need to compare against a “clean” measurement, to judge if you can trust it. I would not.
The results are surprisingly clean but I am still frustrated with that noise and you are right I did notice the bumping on the graphs. What kind of wiring issues should I look for? The board is by design pretty simply wired so there’s only two power lines that go directly to the 24 V power supply.
Well, maybe due to the noise the recommended shaper (EI) is pretty aggressive, limiting your max acceleration to around 13k. Depending on your printer this might not be an issue though.
The ADXL takes 3.3V max, so there must be a DC-DC converter (also for the other devices onboard). As it is an highly integrated board, you most likely will not have much luck in adding a cap directly between VCC and GND before the ADXL.
Maybe a 10 uF low ESR cap in the 24V line may already help. Just spitballing, I’m not enough proficient here.
In any case this might be good feedback to the huvud developer. The ADXL datasheet recommends a cap and a ferrite bead directly before the chips supply pins.
I’ve reached out to pontus for information on the wiring on board. I know nothing about kicad. I’m not surprised about the aggressive acceleration limit, this is a voron 2 with pretty tight belts. And that’s just the x axis. the y axis limit was more around 6k.
