Battery Powered System, Overvoltage Possible!? 1

[cmamich]

Hey everyone, first of all --- great website, long time lurker, first time poster. Thanks to all who contribute.

I have a National Instruments data acquisition device and a 900Mhz serial modem being powered by two 12V gel batteries in series. The DAQ and RF modem are in a NEMA rated enclosure and the power is brought in with a marine grade 3 prong plug/receptacle. There is a switch between the receptacle and the devices themselves. The issue I have been having is the RF modem has been intermittently failing. Inspection of the devices shows no evidence of physical damage or burned smell. The failures do not occur at any specific time (ie during battery change) or seem to follow any pattern. The OEM inspection of the failed device claims overvoltage destroyed the device. The device accepts a range of 7-28VDC, two 12V batteries at full charge should check in at around 25.5V. There has never been a failure immediately following a battery change which would stand out to me as the only way to potentially ever have an overvoltage condition but still unlikely. The DAQ accepts 11-30V and has never failed. The application has substantial intermittent impact and vibration. Both devices receive power in similar way with the ground at the receptacle and the positive coming off the fuse block. The RF device is a typical size N radio plug for power while the DAQ is a screw type terminal. What could possibly be happening to cause an overvoltage condition for that device? Is it purely happenstance that the DAQ has never failed in that same manner or perhaps the circuitry inside the device has more protection for such conditions? Is it a bad assumption to think that the voltage of a gel battery is stable even under harsh conditions? My plan at this point is to power the system via 12V to give the device a bit more of a cushion but I'm not answering the true question nor do I think that is really the cause. Perhaps you electrical gurus can help a mechanical guy stumble through this?

 

[zappedagain]

What is charging the batteries? In an automobile you can get extreme voltage spikes if someone ever disconnects a battery terminal while the motor (alternator) is running. This is called 'load dump' if I remember correctly.

Voltage induce failures can occur long after the fault condition has left. Most voltage failures are from a high voltage breaking down a trace, kind of like chiseling on a column that is holding up the ceiling; the chiseler may be long gone before the house collapses.

John D

 

[cmamich]

Thanks for the reply. The batteries are deep cycle, so they are charged seperately, connected and discharged over a period of about 2.5 weeks then replaced with freshly charged batteries. So this is purely battery power which as I understand it should be quite stable under all conditions.

So a circuit is supplied more voltage than its designed for causing higher than expected current in a trace --- heat builds up, physically deforms the trace which causes higher than designed resistance resulting in further deformation overtime under original design conditions that ultimately leads to failure. Is my understanding of that condition reasonably accurate? I feel like I would be able to see a "smoking gun" if this were the case. Granted the device is on a PCB and there probably is multiple layers which may make it impossible to see physical damage.

 

[OperaHouse]

"so they are charged seperately, connected"

Do you actually remove the batteries from the circuit to charge? For gel cells, each battery would go up to 14.1 volts on a charger that properly limited the volts. Less sophisticated bulk chargers could be much higher. I repair a lot of stuff and seldom is there any physical sign of failure.

Maybe the manufacturer is pushing his spec a little. The modem shouldn't draw that much. A 9-12V zener in series with the modem power would get it to a lower end of the voltage range. An easy thing you could try.

 

[cmamich]

Yes, the batteries are removed from the system to charge. The charger I use is a Samlex SEC-1230A. It has DIP switches to change the charge profile for various types of batteries. In the case of of the Gel Cell its a 3 stage charge with 13.5V Float and 14.0V Boost. So I guess it would be possible the modem is seeing 28V in a worst case scenario which is at the top of the range. They have never failed immediately following a battery change, but as "zappedagain" again pointed out perhaps that anecdote is not really indicative of it not being the failure mode.

 

[cmamich]

So the Zener would provide some protection, gauranteeing only a safe voltage is reaching the device --- sounds like a good idea.

I'd really like to uncover the true underlying cause to the overvoltage in the first place.

 

[MikeHalloran]

Both of the loads have internal power supplies, right?
Permanently connected in parallel.
I wonder if it's possible for them to interact badly on their primary sides when the battery switch is opened.

If that's true/possible/plausible, a procedural change might be sufficient. I.e., never open 'the switch', and have two battery connections, so the charged battery is connected before the discharged one is removed. Okay, you might want to include diodes in the battery circuit so the charged battery doesn't try to charge the discharged one.







Mike Halloran
Pembroke Pines, FL, USA

 

[VEBill]

"...accepts a range of 7-28VDC..."

Use one "12 volt" (nominal) battery.

Two gets you too close to 28 volts when they're fully charged. And obviously there's not much margin on the 28 volts.

 

[cmamich]

Mike:
Each device is connected in parallel to a pair of 12V batteries in series, thus 24VDC nominal. The procedure is to switch off the power, now the devices see nothing, change the batteries and reswitch the power on. At no time is there ever just one battery connected to the system and each device is always powered on/off simultaneously as they are connected to the same switch.

VE1BLL:
Agreed, in hindsight it was perhaps not prudent to push the specifications to the limit. My intention at this point is to try running the system at 12VDC and see if failures persist.

All:
Could someone familiar with PCB's explain to me the physical failure mode of a device that lets assume was designed to accept 28VDC max but got say 28.2VDC. This condition would only be possible right after battery change so it would only be for a short time as the voltage would reduce over time/discharge. As I've mentioned before there has never been any correlation between battery change and failure --- it is always some time after the change.

 

[MikeHalloran]

>>>... switch off the power, now the devices see nothing, ...<<<

No, they see each other.

Put a scope on that connection, and open the battery switch.



Mike Halloran
Pembroke Pines, FL, USA

 

[cmamich]

Ahhh yes, I see what you are saying --- yes the are still interconnected. So what could possibly be happening then? What is your theory?

 

[MacGyverS2000]

Could someone familiar with PCB's explain to me the physical failure mode of a device that lets assume was designed to accept 28VDC max but got say 28.2VDC.

Depends upon the device. Semiconductors, regardless of the spec'd max, have a finite life (though if you stay well within spec, that lifetime can be considered, for all intents and purposes, infinite). They are spec'd for a maximum that will provide a reasonable lifetime. If you go over that spec, you are shortening it's lifetime... if you seriously overshoot the spec, the lifetime will seriously approach zero.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[cmamich]

I guess my mechanicalness in me just won't let this go --- so when someone designs a PCB...they specify the trace for a given portion of a circuit, presumably the thickness and width of the trace, to be able to carry the design voltage and current. Someone like me comes along oversupplies voltage and the trace will now be carrying too much and accordingly heat would build up and physically damage/alter the traces physical characteristics leading to failure overtime? I guess my fundamental question is what physically happens in a case where failure has occured by no signs of physical damage.

 

[zappedagain]

It gets a little bit weirder inside the ICs where things are really, really tiny. At that level you can have electrons bumping into the molecules that make the trace and dislodgimg them or even knocking them loose. Look up electromigration if you want to know more. Eventually you have a thin spot that can't handle the current flow so it gets hotter. The heat accelerates this process and the death spiral begins...

John D

 

[OperaHouse]

Are there any relays or solenoids in this box that may be causing spikes? What about signals coming in? The search for a battery related problem may be a red herring.

 

[MacGyverS2000]

0.2V will have zero effect on a PCB... I didn't realize the question was aimed at PCBs as that's kind of a non-starter.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[VEBill]

The "7-28VDC" voltage input specification might be a clue that the troublesome RF modem is using a simple linear (7805-type or similar) voltage regulator. If so (maybe?), then perhaps the regulator is simply getting (way) too hot when pushed to the limit. If this is happening, then there should be some evidence of overheating in the power supply area of the circuit. And sometimes the heat kills other nearby components while the regulator itself (barely) survives.

This is just one possible explanation of many.

 

[zappedagain]

I read your original post again. NI builds their DAQ cards fairly bulletproof, so even at the spec'd 30V the DAQ will most likely still have lots of margin before damage occurs.

Did the modem OEM give any more detail about the failure; what failed? what pin saw overvoltage? Did a device fail or a trace on a PCB (if the latter then they have a very poorly designed PCB)?

John D

 

[LionelHutz]

The claimed over voltage failure could be a red herring as well. It's possible this modem will simply not operate at the high end of it's supply rating.

It's impossible to really make good suggestions without knowing details on the failure. You shouldn't see anything other that maybe some small transient over-voltages and if those devices were well built the power supply would handle these without failure. So, I would just proceed and put the 2 batteries in parallel instead of series.

 

[cmamich]

Okay, I followed up with the OEM to get more detail on their failure analysis. Apparently there is a pair of zero Ohm resistors that are part of the power circuitry that failed. OEM says the only way that could have happened is reverse polarity connection (possible but unlikely, but I would imagine it would fail immediately if that were to be the case --- I'm going to test this) or back current in the ground wire. This would be a good juncture to explain the setup a little better.

Imagine a large heavy steel box that is part of a larger frame. Inside this steel box are two batteries and a stainless steel enclosure. The batteries, as mentioned before, are connected in series --- three wires: one jumping from BAT1 positive to BAT2 negative, one from BAT2 positive to plug connector, one from BAT1 negative to plug connector. The recepticle, mounted in the stainless steel box, is a plastic body with no possible electrically conductive paths to the stainless box from there I have the positive switched which then runs to a fuse block and then onto the respective devices. The grounds pass directly from the plug to the devices. Each device is mounted on what amounts to plastic standoffs so they do not contact the enclosure either. I realize I have a floating ground but how could back current possibly be in the ground wires? The prototype of the system was in a plastic enclosure and ran for a year straight without issue --- so somehow I feel that I'm getting screwed electrically here but I can't see how. I'm probably not understanding some basic rule of EE design or something.