Over voltage on hardware

[guilio2010]

We have hardware that accepts a voltage range of 20-28.8V, but we are inputing in 32 volts in the system. the reason is our power system outputs a voltage that is temperature based so in cold weather, we see a higher voltage. There has been no issues to this date and I'm guessing that the reason is more based on that yes we are inputting a higher voltage which is probably producing more heat in the hardware but due to it sitting outside in the cold weather and the fact we are only seeing the overvoltage for maybe an hour or two in the day is the reason we haven't had any issues. I'm not saying that this is the right way of dealing with it and we should regulate the voltage, but would this be a probably cause of why we haven't had any issues with the hardware?

 

[IRstuff]

Those are certainly plausible reasons for not gettting a failure, but no real system has a perfect threshold between functionality and failure, and it's certainly equally plausible that you are creating latent damage each time the system is powered on.

On the other hand, given the rather curious coincidence??? of having a high end at 28.8V, perhaps the system was designed for MIL-STD-704 power, in which case 32V is within its acceptable input limits.

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

 

[guilio2010]

Thats the question of what the system is designed for vs. what the manufacture actually puts on their specs. I did call the manufacture and explain the situation but they were no help other than that it has no over voltage protection. I figured that since the manual didn't mention anything.

Based that we been running like this all winter, we had 0 failures and other equipment in the system hasn't failed (although it has a 20-30V range), I can only come to the conclusion above of why we haven't seen a failure. However, I do have to add in your plausible reason that we are creating latent damage each time. Can't argue with that. The vairable there is how long will it last before it quits....

Thanks IR. Just looking for other insight.

 

[itsmoked]

If your unit has a constant draw you could easily shave off a few volts with a simple resistor in series with your load.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[guilio2010]

I was acutally thinking about installing a smaller wire. We don't have a constant draw, but I think the draw isn't that much where we would burn the wire up.

 

[OperaHouse]

Higher voltage when cold and for two hours a day sounds like solar panels. If your load could unexpectadly drop then the voltage could theoretically go up to 44V. The voltage you are seeing is just under the power point. A thoughtful design would use a dump regulator to place an extra load on the panels. This wouldn't have to be that complicated as the current past the power point is limited. This would protect from overvoltage and still provide maximum power in lower sunlight conditions. Could be as simple as a zener and a resistor. Adding a transistor or a FET greatly enhances operation. If you want some more suggestions post the panel power.

 

[VEBill]

If you can examine the schematic for the power input circuitry of the device then you should be able to see if you've just been lucky or if there's some margin in there. It really depends on what's inside the 'hardware'.

"28.8" is an odd and unusually precise number. Typically such limits are rounded down to a nice sensible number. They also typically have a reasonable margin above the spec'd limit, but you're not allowed to depend on that.

 

[MikeHalloran]

As VE1BLL said, when you overexcite or overclock or over-anything something, you're working within someone else's safety margins. ... the extent of which you don't actually know.

How much money is at risk, and who you gonna call when the magic smoke leaks out?




Mike Halloran
Pembroke Pines, FL, USA

 

[itsmoked]

If the load varies then do NOT even bother with the small wire gambit. It would turn out badly. You'd need the zener and its friends.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[iop995]

Depending of application / schematic details, you may try to add 4 - 5 series diode. Maybe it's a charging system and I think it tolerate such small difference.

 

[guilio2010]

I agree that the design was not well thought out, but I have to deal with it. I looked at regulators to run all the equipment since we are exceeding all the ranges for everything, but the power consumption would counteract why the temp compensation is in the first place. It was be cheaper and better to just take out the temp compensation and let it run at the ~27.5V. So looking at that, why spend the money or time to regulate the voltage that would just end up losing power?

The question becomes to as why has it survived this long as the temp compensation has been in place for a few years? Are we damageing the equipment? I'm going to say yes. How badly? Don't know. How long still it goes? Don't know. Could we be running in a different range of voltages that the manufacture is not telling us? Yes. We either remove the temp conpensation and in our area, increase the lost of load, or we continue with what we have, come up with the best reasoning why it hasn't failed and keep an eye on it.

For the smaller gauge wire, I was going to take the max load the device could take and use that as the max current.

Thanks for the comments. I appreciate it. Also, I can't post what we are using.

 

[guilio2010]

I'll have to get back into my circuit, but I'll look into a zenar solution...good idea..a

 

[IRstuff]

" come up with the best reasoning why it hasn't failed "

You keep harping on this, and I and others have already pointed out that overstress at the levels you are describing is merely operating into the design margin, or effecting only latent damage. Without knowing any more about the circuit and the technology used, the device may fail tomorrow, or may fail after you retire. The only way to even know for sure is to take something like 10 devices and stress at two different input conditions until failure and extrapolate the results to your specific installation. This is how some people determine device failure rates for reliability analysis.

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

 

[VEBill]

Depending on the magnitude of the current, there are inexpensive DC-DC converters that will accept a wide range of voltages (3-40Vdc) and emit a adjustable voltage (1.5-35Vdc). Up to two or three amps (perhaps with additional heat sinking) they're cheap and cheerful (example, example). Good luck.

 

[OperaHouse]

So, I'll assume this is a solar application.

28.8 would be the normal voltage for a bulk charge on a 24V system and that voltage range would be to keep the boards in spec. I would presume safe operation of several volta over that from a voltage breakdown perspedtive. If linear regulators are employed more heat would be generated which could be a problem. If switching regulators are used, current would lower as the voltage increases. That would imply lower IR losses and maybe better life expectancy.

As I stated before indications are that the solar panel is operating near the peak power point. Panels can be modeled as a current source. As panel voltage drops from indirect sun a switching regulator will demand higher current dropping voltage even more. A resistive solution would be the worst possible choice as voltage drop would increase in low light conditions resulting in fewer hours of operation. Diodes in series would be a more fixed voltage drop. It would still limit the available hours of operation. A shunt regulator is the onlyacceptable solution if one is needed at all. If there are multiple devices on this same supply line, the loss of one load could result in voltage over 40V. Many linear regulators are only speced for the high thirties.

I submit for your inspection and amusement the following shunt regulator.

http://www.ghurd.info<<<-----Information

on my Controller
First time I saw this I only quickly glanced at the schematic and assumed that the ZM was a house numbered LM431 amplified zener in a circuit with just too many parts. I never thought about it again till reading in a post that RX could increase the on/off span to a volt. That got me looking again and the ZM part was a POR with an open collector normally used for microprocessors. Hysteresis appears to be a function of base drive current. A clever off data sheet gimick I assume works. Likely a design that happens when given a couple thousand free. A LM431 circuit would be far fewer parts. Still an interesting idea worth some thought.

 

[OperaHouse]

That didn't copy well

http://www.ghurd.info

 

[guilio2010]

IRStaff. I am happy with that and that is a strong reason and I can't argue with it. I'm going to use that as part of our reason of why it has not failed. Again, I thank you and everyone as my circuit knowledge has dropped off alot over the years so this has been a good refresher. Really, with our setup, and what we are doing, it does fall in the testing procedures you are mentioning so I'm happy and it has strong evidence. The reason I have to answer is if we need to do something, were talking a good bit work so the more reasons of why it will not fail vs. taking action and regulate the voltage, works with me. With VE1BILL, were talking alot more amps also so a small converter isn't the best option. Thats why I'm now looking at a zenar and power transistor setup as a contingency. Not sure what kind of power loss there is there, but it would be a good step for me to look into and relearn.

Thanks,

 

[mikekilroy]

do u have some aversion to telling what the actual current draw is? why the mystery? If you want solutions, give the rest of the story as Paul Harvey would say!