exploding batteries
#1
I had my clusterboard run for about two months and today i've turned it of to add new modules. but when i turned it back on, within a minute, one of the bateries exploded. Well started to bubble and sprayed some juice out.

Is this a common thing for normal batteries explode in cluster board, or was it just a rare case? Just a general AA alkaline batteries.

so 20 minutes later, another set of batteries - another explosion Sad

and the think i've did today was: adding 3 SOEdge boards next to 3 SOPINE boards

what is this?
  Reply
#2
Explosion seems a bit hyperbolic, no?

But likely it is related to this: https://forum.pine64.org/showthread.php?tid=5849

Should not happen with fresh batteries though, but IMHO probably better to not use batteries at all.
  Reply
#3
(06-11-2021, 03:42 PM)poVoq Wrote: Explosion seems a bit hyperbolic, no?

But likely it is related to this: https://forum.pine64.org/showthread.php?tid=5849

Should not happen with fresh batteries though, but IMHO probably better to not use batteries at all.

if you can a popping sound that is heard to the second room and close surface sprayed over by battery liquid hyperbolic then yes - however i have other opinion

after removing SOEdge modules and getting back just to SOPINE modules everything fine, no more battery explosions.... however i didn't bought SOEdge modules not to use them
  Reply
#4
It must be (over)charging the batteries. It shouldn't happen, but you could add a diode to prevent it.
  Reply
#5
(06-14-2021, 12:52 PM)alembiq Wrote:
(06-11-2021, 03:42 PM)poVoq Wrote: Explosion seems a bit hyperbolic, no?

But likely it is related to this: https://forum.pine64.org/showthread.php?tid=5849

Should not happen with fresh batteries though, but IMHO probably better to not use batteries at all.

if you can a popping sound that is heard to the second room and close surface sprayed over by battery liquid hyperbolic then yes - however i have other opinion

after removing SOEdge modules and getting back just to SOPINE modules everything fine, no more battery explosions.... however i didn't bought SOEdge modules not to use them

So with 7 Sopine it works? But the SoEdge cause problems? That would be interesting.

So far I assumed it was just a question of the number of modules, because if you read the thead I linked above carefully it seems like each Sopine module feeds back a bit of electricity into the circuit, and when there are a lot of Sopines and weak batteries, it could cause the problems you describe.
  Reply
#6
(06-14-2021, 07:03 PM)poVoq Wrote:
(06-14-2021, 12:52 PM)alembiq Wrote:
(06-11-2021, 03:42 PM)poVoq Wrote: Explosion seems a bit hyperbolic, no?

But likely it is related to this: https://forum.pine64.org/showthread.php?tid=5849

Should not happen with fresh batteries though, but IMHO probably better to not use batteries at all.

if you can a popping sound that is heard to the second room and close surface sprayed over by battery liquid hyperbolic then yes - however i have other opinion

after removing SOEdge modules and getting back just to SOPINE modules everything fine, no more battery explosions.... however i didn't bought SOEdge modules not to use them

So with 7 Sopine it works? But the SoEdge cause problems? That would be interesting.

So far I assumed it was just a question of the number of modules, because if you read the thead I linked above carefully it seems like each Sopine module feeds back a bit of electricity into the circuit, and when there are a lot of Sopines and weak batteries, it could cause the problems you describe.



It works with 3 SOPINE, but it complains about 3 SOPINE and 3 SOEdge
  Reply
#7
I see, so probably just a question of the number of modules. Edit: See message below. Do not put SoEdge modules into the clusterboard.

Anyways the thread linked above suggests several possible fixes, but I just removed the batteries made a small hardware fix to still allow rebooting and run the clusterboard off a ATX power-supply connected to an UPS.
  Reply
#8
My apologies for coming late to the party.

I'm sorry to hear that the batteries are exploding in a Clusterboard, but that actually confirms the earlier analysis peformed in the already mentioned other forum thread.  As @poVoq already explained, the more SOPine modules are installed in a Clusterboard, the higher the chance their combined outputs will overpower the batteries and start charging them, causing issues.  This applies particularly when the installed batteries become older and their voltage drops.

By the way, please keep in mind that neither rechargeable AA batteries are supposed to be installed in a Clusterboard, nor the power provided by the installed SOPine modules would be suitable to charge such batteries.  Mind you, up to this point it was all about installing and using only the SOPine modules.

After looking into the SOEdge schematic, I can confirm that installing SOEdge modules in a Clusterboard makes things much, much worse.  According to the SOEdge schematic, 5 V provided by the Clusterboard (mind you, it's up to 15 A at 5 V) is connected on the SOEdge directly to the VCC_RTC input of the RK809 PMIC.  The VCC_RTC input is also connected directly to the pin #197 of the SOEdge's edge connector, which, using the SOPine schematic and the Clusterboard schematic as references, leads directly to the positive terminal of the Clusterboard's battery pack.  This is obviously really bad. Sad

In a few words, having even a single SOEdge installed in a Clusterboard will surely make the installed batteries pop very quickly, just as reported by @alembiq.  As already described, installing even a single SOEdge module causes the entire power input of the Clusterboard (up to 15 A at 5 V) to be fed into the installed batteries.  As a result, you must not install batteries in a Clusterboard if any SOEdge modules are to be installed.

As already described, the other forum thread describes a few ways to fix this issue, but those apply only if no SOEdge modules are installed in a Clusterboard.  Furthermore, as already described, installing a SOEdge module causes the Clusterboard's 5 V supply to be fed back through the VRTC pin on the SO-DIMM sockets, possibly causing damage to the A64 SoCs on any installed SOPine modules.  According to the A64 datasheet, the absolute maximum voltage for the VCC-RTC power input of the A64 SoC is 3.6 V, which is way below 5 V that becomes present after installing a SOEdge module.

In a few words, unless there's something wrong or missing in the SOEdge schematic, it's clear that unmodified SOPine and SOEdge modules must not be mixed in an unmodified Clusterboard.  It should be possible to mix these two types of modules in a single Clusterboard, but only if each SOEdge module is modified not to feed 5 V back into the pin #197 of its edge connector, or if the Clusterboard is modified to split the VCC-RTC supply between the SO-DIMM slots into two separate, fixed-size groups.  Also, if you want to use batteries as the RTC backup for the installed SOEdge modules, different hardware fixes need to be applied to the Clusterboard.

Of course, it's also possible that I've missed something obvious and important, so please free to provide corrections.
  Reply
#9
this is a nice analysis Smile
do not take any offense, i just hope you're wrong and the problem isn't general, but just mine... modifying hw to this level reminds me of the communism times, when you bought a bicycle and then you've spend few evening modifying it to be able to really use and not to have just a as prop to show of to your visitors
  Reply
#10
So more than a year later it is still advertised for the clusterboard:

https://pine64.com/product/soedge-ai-neural-module/

Did anyone figure out a fix by any chance?
  Reply


Forum Jump:


Users browsing this thread: 1 Guest(s)