5VDC project, 110VDC overvoltage protection

[Bouteille]

Hello all,

I spent some time programming a PIC microcontroller and I would appreciate some help to design a circuit for it.

My first concern is overvoltage protection. In normal usage, the circuit expects regulated 5V.
However, this project will be used on ebikes and I know that sometimes people (like me) mix cables and put battery voltage (24V to 120V DC depending on the battery) where 5V DC is expected... oops !
I would like the circuit to survive to this kind of event (function not required in this condition).

My electronics skills are very limited so I searched a bit online and found some suggestions incuding zener diodes, polyswitch, crowbar circuit, or dedicated chips.

Easiest solution seems to be a Zener diode in parallel to circuit input, downstream to a polyswitch.
Knowing the following characteristics of the circuit (basically just a microcontroller), I would like to know how relevant this is.
I would also appreciate some help for selecting appropriate components.

Circuit characteristics:

Operating input voltage: 5V (2.5V to 5.5V, absolute max 6.5V)
Typical consumption : 50-200mA (never above 250mA)

Max input voltage protection (survival required, no operation): 26S li-ion battery (110V max, several amps capable)

For the polyswitch, I selected this part: PolySwitch TRF600 Series, 150 mA, 300 mA, 250 VDC
Datasheet: http://www.farnell.com/datasheets/2282572.pdf

Questions:

1. The polyswitch looks big but I guess specifications must be over the maximum protection (and not operation) voltage, correct ?

2. If I understand correctly the Zener diode must handle the short-circuit current until the polyswitch 'trips'.
But how to estimate the correct 'trip-time' from polyswitch datasheet, and then select an appropriate Zener to use in this application ?

Must I use the polyswitch 'Typical Resistance' (datasheet page 238, ie. 8.0Ω), and 'Typical Time-to-Trip Curves at 20°C' (datasheet page 245) ?

3. I assume that the regulated 5V from ebike controller will not be super-stable. Assuming a 5V Zener, is it a problem if the input voltage fluctuates between 4V and 5.5V ?

4. From EEVBLOG Zener diodes video, their use is not recommended for voltage regulation because of varying characteristics with temperature, current, etc...
Would you recommend another protection strategy for this project (crowbar, tps2400) ?


Any other suggestion appreciated, thank you very much for your help !

[Mazo]

I don't think clamping the voltage is needed here(It is harder to clamp than to just switch off and "hide" from the overvoltage)
Here is a fast and dirty circuit.It is dependent on the Vbe of Q3 so not exactly the best accuracy.A TL431 used as a comparator can fix that if it's a problem.

[Zero999]

You could just use a series resistor to limit the current to a level safe for the MCU input and add a zener to the 5V rail to prevent it from exceeding a safe level.

[james_s]

Also select connectors wherever possible that mechanically prevent this sort of oops. A connector carrying high voltage should be quite different than one carrying 5V.

[Yansi]

Here you have, eat that! 

http://www.ti.com/lit/an/snva717/snva717.pdf

[T3sl4co1l]

Also a series diode to prevent backwards operation, and maybe a MOV to clamp anything further (including transients, if they should show up).

If the load current isn't much, it's rather nice to make a crude "LDO" with a depletion mode MOSFET, resistor and zener diode.  This kind of thing:

from: https://electronics.stackexchange.com/questions/237566/typical-use-of-depletion-mosfet

Tim

[T3sl4co1l]

Here you have, eat that! 

http://www.ti.com/lit/an/snva717/snva717.pdf

Hah, this is somewhat better, but also worse, than the other appnote I've seen.

This one: they mention load dump in the opening paragraph, then never again.  It does seem to have an actual voltage threshold (only), and it should be fairly fast acting.  But they didn't test it and show the results.  So... 

The circuit is awfully complicated, lots more parts than should be needed.  Not sure if the intent was towards protection, or just having more features (like the filtered output, and the enable signal).

The other one I've seen was,
http://www.ti.com/lit/an/snva681a/snva681a.pdf
This is actually sensing dV/dt plus V.  Note the output disables before Vin crosses the threshold!  That's a nuisance in waiting, if ever I've seen one.  The best part is, the load they're trying to protect, doesn't even need a protection circuit for the transient in question, so there's no absolutely motivation behind the appnote.

Tim

[Bouteille]

I don't think clamping the voltage is needed here(It is harder to clamp than to just switch off and "hide" from the overvoltage)
Here is a fast and dirty circuit.It is dependent on the Vbe of Q3 so not exactly the best accuracy.A TL431 used as a comparator can fix that if it's a problem.

Thank you for your suggestion. Actually your attached file forced me to install LtSpice which is a very nice discovery !

Your circuit seems to protect the microcontroller by disconnecting it from source when V1 hits a max value.
I have to admit I don't understand how it works in the details... But playing with the simulation suggest that voltage cutoff varies with current load... Can't it be a problem ?

You could just use a series resistor to limit the current to a level safe for the MCU input and add a zener to the 5V rail to prevent it from exceeding a safe level.

Seems reasonably simple but I don't see how to arrange that in real world, could you provide an example (sorry my understanding is very basic) ?

Also select connectors wherever possible that mechanically prevent this sort of oops. A connector carrying high voltage should be quite different than one carrying 5V.

Good point. However there is no real standard so connectors often end-up being replaced to adapt with what is already in place.


Yansi and T3sl4co1l, thank you for your suggestions. I'll need more time to read and digest that before I can comment though.
At first glance it seems very informative, but isn't it overkill for my application ? Looks like it will cost more than the microcontroller I want to protect (I'm maybe wrong).


I have not seen comments on the polyswitch + Zener solution, is it really too primitive or inadapted ?
I thought something basic but efficient could have been implemented with very few discrete components.

[Mazo]

The load current dependence comes from the Vce(sat) of the pass transistor.Can't do much about it short of putting a different transistor.Is a diode drop in series with the supply acceptable?If it is I will think about a better circuit

[nick_d]

I have a similar idea, in my case it is retro 5V electronics that would be irreplaceable if destroyed, so I don't want to trust the power supply. Therefore I bought a large number of PTCs (what you call a polyswitch) and high current Zeners to experiment with. However, I have not built the circuit yet. So I don't know, but I'm pretty sure this would work. Why don't you build it and do some tests and let us know how it goes?
cheers, Nick

[Bouteille]

Why don't you build it and do some tests and let us know how it goes?

Will do ! I just ordered some (Littelfuse TRF600 160ma).

I thought PTCs (thanks for the correct term) have been very common parts used in many systems, and I expected a ton of comments about them. But I am maybe wrong.
Is this something relatively new or confidential (or useless) ?

I preferred to ask advice before placing order but hey... time for experimentation !

[Zero999]

PTC resistors are very slow. Nowhere quick enough to protect a semiconductor device. They're used for protecting cable, motors, transformers etc. which are big things with a large thermal time constant.

[Bouteille]

PTC resistors are very slow. Nowhere quick enough to protect a semiconductor device. They're used for protecting cable, motors, transformers etc. which are big things with a large thermal time constant.

Hence the Zener for clamping. It dissipates power and waits for the PTC to save its life. Almost romantic   
Not realistic though ?

[TrickyNekro]

I know that might come as a bit of shameless self promotion, but have a look here.
https://www.eevblog.com/forum/oshw/openservocan-(-for-hobby-standard-size-servos)/

Look at the input protection side of the project (in the .pdf).
Replace the MOSFETs with high voltage ones and the regulator with a HV variant with enough cooling.

You can not got wrong like this. You got reverse voltage protection and possibly also over voltage protection with some modding.
The circuit is not new by the way, I´ve seen it used in many laptop designs, take it as you want it and mod it accordingly.

Sure building this circuit is more expensive than a polyfuse but I guess you ain´t making a cheap product either.

Cheers!

[Bouteille]

Thank you ! That looks more advanced than what I can handle yet but I'll come back to it as soon as I feel more confortable with basics.

I'm still a beginner (started by programming pics and loved it) and I wanted to believe I could stick to something very simple, but it looks like I'm gonna have to put some more efforts into it.
I tried to read/watch introduction about individual components but when it's time to assemble them and get something done, I feel stupido

I thought I'll be able to get something done after watching Dave's Zener video but apparently... not yet 

Sure building this circuit is more expensive than a polyfuse but I guess you ain´t making a cheap product either.

Actually the device will not be that expensive, basically just a pic microcontroller...

[Zero999]

PTC resistors are very slow. Nowhere quick enough to protect a semiconductor device. They're used for protecting cable, motors, transformers etc. which are big things with a large thermal time constant.

Hence the Zener for clamping. It dissipates power and waits for the PTC to save its life. Almost romantic   
Not realistic though ?

A zener diode is a semiconductor device and would have to be very big to withstand the surge for the amount of time it would take for the PTC to trip.

[Bouteille]

A zener diode is a semiconductor device and would have to be very big to withstand the surge for the amount of time it would take for the PTC to trip.

Yes, that was one of my initial concerns actually:

2. If I understand correctly the Zener diode must handle the short-circuit current until the polyswitch 'trips'.
But how to estimate the correct 'trip-time' from polyswitch datasheet, and then select an appropriate Zener to use in this application ?

Must I use the polyswitch 'Typical Resistance' (datasheet page 238, ie. 8.0Ω), and 'Typical Time-to-Trip Curves at 20°C' (datasheet page 245) ?

[T3sl4co1l]

Yeah, TVS diodes are pretty robust actually, but the sheer amount of energy let through by a PTC is ridiculous in comparison.  Probably the diode would blow (fail shorted, most likely) in a few milliseconds, then the PTC opens in tens of milliseconds.

Not quite as bad as a transistor dying in a hundred microseconds, and quite tractable with the right component choice (large enough TVS, PTC with enough internal resistance -- or add a resistor in series to help out).  But such a solution will be more expensive and less reliable than a regulator or switch.

PTCs handling over 60V are hard to come by, and they will be quite large and therefore slow, making things worse.

The depletion MOS, resistor and zener is damn good for low current loads (under 100mA, say?), and a switch method is better for heavier loads.

Tim