Random reboots on MICRO

Started by Bert, December 09, 2013, 03:39:39 PM

Previous topic - Next topic

Bert

I've built an animatronic fursuit. The iMX233 used as main processing unit and located in a cheek of the head. In a cavity between of inner thermal shield (rubber foam) and external faux fur cover. Of course there's a few coolers to keep an artist happy and to reduce temp/humidity inside.

All works fine but sometimes I'm getting random reboots with no reason. I'm using stepdown DC-DC converter from 11.1V to 5V with 3 amps max as power source. Checked - it's not a power problem.

As far I see - there's a few reasons which can cause rebooting.
* Underpowering (the sympthoms are same)
* Humidity (I suppose it is)
* Overheating (What's maximum allowed temp?)
* Static discharges from faux fur (never tried and unsure about possibility)

The "strangeness" is that sometimes it's working with no problems for hour or more and sometimes reboot happens every 3 or 5 minutes. The ambient temp/humid are same. The battery is fresh and charged.

What most likely can cause these reboots?

Kean

Bert,
What revision of Micro?  Early ones need R17 removed to be stable.
Reboots are most commonly power related (e.g. noise/dropouts from servo motors).  Suggest you run separate DC-DC converters for iMX233 & any other high current devices
Otherwise, heat much above 40C can be an issue for some microSDs, and ESD can cause all kind of weirdness.
Does it misbehave if separated from the fursuit?
Kean

Bert

Quote from: Kean on December 09, 2013, 04:15:37 PM
Bert,
What revision of Micro?  Early ones need R17 removed to be stable.
Reboots are most commonly power related (e.g. noise/dropouts from servo motors).  Suggest you run separate DC-DC converters for iMX233 & any other high current devices
Otherwise, heat much above 40C can be an issue for some microSDs, and ESD can cause all kind of weirdness.
Does it misbehave if separated from the fursuit?
Kean

> What revision of Micro
Micro is a latest one. Yep I know about this R17 issue

> Suggest you run separate DC-DC converters for iMX233 & any other high current devices

Yep I made three-channel DC-DC (2 x 5V + 3.3V) based on LM2576-T with decoupling capacitors


> Does it misbehave if separated from the fursuit?

Sometimes (very rare) it does. It seems it's a power issue. Because I can't blame humidity or high temp in that case (was fully outside and off for a hour minimum).

I haven't oscilloscope so I unable to test spikes in a power line. Should I put also linear regulator at the output of DC-DC?



Kean

Nice work on that fox head Bert.  Looks great!

What capacitors do you have near the output of the LM2576?
I'd recommend trying with 470uF or 1000uF (low ESR electrolytic - but not below ~80mohm) + 100n (SMD ceramic) + 10n (SMD ceramic).  Also make sure you have at least 100uF on the input of each LM2576.  These must be close to the regulators, and are separate from the caps on the Olinuxino PCB.

Try to use thicker wire from PSU PCB to Micro PCB to reduce any voltage drop - especially when WiFi is active.

You could also try adding an LC filter to the 5V supply between PSU PCB and Olinuxino - e.g. 22uH & 100uF.

Finally, you could surround the Olinuxino with some wire mesh to try drain away ESD.  But be careful or you'll end up with a faraday cage around the WiFi adapter!

Kean

Dave

I'd suggest relocating the WiFi adapter to some point further from the iMX.  If you can run without the WiFi installed I'd try that first, though.

Very impressive build!

Bert

Quote from: Kean on December 10, 2013, 05:58:02 AM
What capacitors do you have near the output of the LM2576?

Aluminium electrlythic 100uF and 1000uF. And yes 100uF attached near input of LM2576. In fact all "surrounding" schematics was taken from LM2576 datasheet. 
However I thought about "remaking" of PSU to add ampermeter along with voltmeter and I'll also include SMD capacitors and LC filter you suggested.

Quote
Finally, you could surround the Olinuxino with some wire mesh to try drain away ESD.  But be careful or you'll end up with a faraday cage around the WiFi adapter!

I'm thinking about totally remove WiFi dongle and using nRF24L01 chip as "modem" in some kind since radio control using "glove" were bad idea.


Quote from: Dave on December 10, 2013, 04:17:04 PM
If you can run without the WiFi installed I'd try that first, though.

I noticed that reboots occurs often while WiFi is in adhoc mode so I'll try with WiFi removed