Hi there,
I think I finally fixed it. My DP832 is now running without reboot for more than 24h.
If you have a Rigol DP832 with a DigitalBoard V02.00 2012-12-07 (front-side board with MCU)
and your Rigol is not doing sporadic reboots you may just be lucky.
TopBoard: v2.10 (large heat-sink for 5V regulator)
DigitalBoard V02.00
My DP832 started to develop the sporadic reboots over time, from once a month to about once a
day to multiple times a day by now. Let's face it - unusable as a bench supply. And yes, my DP832
is running 24x7 - but that's something you'd expect from a bench-supply.
Some days ago I dismantled my Rigol and started poking around on the digital board.
To summarize, aside from the linear 5V regulator/toaster on the top board that we all know about,
I identified 2 issues while checking the voltage rails on the digital front board.
The "Battery" rail:The AC noise from the 5V rail was passed straight through to the output of the regulator (1117 ADJ).
I added input and output caps (although there are some ceramic caps on the output side, but I
don't trust them). Also the deltaV (Vin-Vout) was very close to the dropout voltage. My guess is that
this is the main reason for the ripple being passed straight through (the regulator was not regulating
anymore).
Studying the iMX28 (Application processor / MCU) datasheet a bit, shows, that this beast
is designed to be powered from two power sources. A 5V source as well as a Li-ion battery
source (3.1 ... 4.2V). The MCU automatically switches between those sources.
The "VDD4P2" rail:This is an intermediate power rail (4.2V) from the MCUs integrated switching supply. I did
measure a *sh%@tload* of switching ripple (>2Vpp) on this rail. According to the reference
schematic for the iMX28, there *must* be a handful of decoupling caps installed on this rail.
However, by just quickly probing around I could not find any serious caps (just one tiny 0402 cap)
on the Rigol board (I don't say there are no caps, just say that I did not find any - hard to tell
for sure without schematic). Luckily this rail is accessible from the back side of the board, no
need to disassemble the front.
I added some caps (ceramic and electrolyte) to this rail and the switching noise got *way* better.
I'm still measuring up to 400mVpp switching noise, but stability has improved massively.
Before this modification, with the voltage reduced to 4.9V on the 5V rail, I got multiple reboots
per minute. After adding the caps it just kept running. No reboots observed. :-)
So my guess is: fix #2 is actually the one that cures the spontaneous reboots. Problem #1
probably has no direct impact on that (depends maybe when/if the MCU actually takes
the power feed from the "Battery" rail - if at all).
A lower voltage an the 5V rail causes the reboots to happen faster / more frequent.
My guess is: lower voltage -> higher currents (switching supply in the iMX28) -> more switching ripple on 4P2 rail
-> at some point the MCU resets (likely the built-in reset circuit or brown-out detection triggers)
This could even explain why it happens more frequent on devices with the small heat-sink (old top board).
If the LM317 starts to cut back due to thermal conditions -> lower voltage -> higher probability for reboot
Note: This VDD4P2 rail issue might me systematic! A cross check with a properly working (no reboots)
DP832 of a friend (same DigitalBoard rev) did show massive ripple of close to 2Vpp on this rail as well.
To summarize my modifications:1) I modified the 1117 "Battery" voltage regulator circuit to deliver 3.5V
(instead of the 3.8V) to get some voltage margin (dropout voltage 1.25V!)
and avoid issues there. I also added input (10uF) and output (100uF) caps.
]
To adjust the voltage, I put a 4k7 resistor in parallel to the existing 680E
(going to ground). Yeah, ugly...
2) Add a handful of decoupling caps to the VDD4P2 rail10nF // 100nF // 1uF // 47uF
The VDD4P2 is on the cathode side of the fat shottky diode on the back side of the board.
I scraped away some solder-mask on a ground-plane and botched in a handful of caps.
Fragment of the reference schematics:
I'm quite sure that modification 2 alone will fix this issue, but I didn't bother to undo my other modification.I hope this will help some Rigol DP832 owners to get their device working properly again!
/Thomas
Update: Running for one week now, without issues.