Overvoltage protection circuit w/ TVS diode and resettable fuse?

[thinkfat]

Hi,

I'm making a small adapter board for my Efratom LPRO-101 and I want to add some kind of over-voltage protection. The idea I came up with is to use a TVS diode in combination with a resettable fuse, like shown in the below schematic.

Does this make sense?

[thm_w]

Yes, makes sense.
Can change to a unidirectional TVS (what is shown in the schematic is bidirectional), to protect against reverse voltages as well.
Also the diode is only capable of dissipating about 0.3W or so, so it might blow up before the SRF (fuse) does. What is the fuse spec?

[thinkfat]

The fuse is an ECE SD200-8, it's rated at 2A hold and 3.5A trip current. But I realize that for reverse polarity protection the voltage rating isn't high enough. In that case, the fuse will have to withstand whatever voltage is applied in reverse - that's kind of impractical. But now looking at it, the TVS diode is also not suitably rated. I'll have to use something that can take more then 3.5A, I guess I'll use a LDP24A.

For reverse polarity protection I'll probably go with a suitably rated p-channel mosfet.

[thinkfat]

Updated schematic. It doesn't feel quite right, though, combining the LDP with the Mosfet like that. When the LDP trips, the FET would still have to bear the load until the polyfuse trips as well. But if I exchange the ordering, the FET would be essentially useless.

[nuclearcat]

Please take my opinion with truck size grain of salt, as i'm not experienced at all in this question

I think you might like to check this discussion, about protecting circuit with P-channel FET and zener:
https://electronics.stackexchange.com/questions/121670/zener-mosfet-overvoltage-protection

But it looks like polarity protection, and it should be connected different way.
I am not sure FET is anyhow useful in current way of connection, especially with transients, as mentioned in that link(if it is correct) - broken FET most likely will short out and silently become useless. Also internal diode might be be burned on overvoltage, even if you manage to switch it off and keep same way of connection.

I did simulation, and it seems this part of circuit does exactly nothing: https://easyeda.com/editor#mode=sim,id=522581dc59d447bba902f6b4457daf8b

[thm_w]

Some more discussion here: https://www.eevblog.com/forum/renewable-energy/how-to-combine-ovp-with-surge-protection-without-burning-out-tvs/
Your currents are quite high, so the SRF/Fuse in those ratings start to get big.

You can use something to switch the FET based on a current sense resistor, but it starts getting more complex. There are quite a few chips that will be able to do this for you, if the budget is available.

[thinkfat]

Yes, thanks for the heads up. I already noticed, the Zener needs to be connected to the source side, not the drain side.

[thinkfat]

Some more discussion here: https://www.eevblog.com/forum/renewable-energy/how-to-combine-ovp-with-surge-protection-without-burning-out-tvs/
Your currents are quite high, so the SRF/Fuse in those ratings start to get big.

You can use something to switch the FET based on a current sense resistor, but it starts getting more complex. There are quite a few chips that will be able to do this for you, if the budget is available.

The device to be protected is rated up to 32V, so I have a bit of a headroom here. The SRF has a trip current of 3.5A and a hold current of 2A, both rated at 23°C, the device will consume 1.7A max, typically 1.5A. So, I think at room temperature (it'll be used only in my lab) it should work just fine.

I might look into an integrated solution, but it's a one-off project (unless I find a source of broken LPRO rubidium standards that are worth fixing ). So if the simple, brute-force SRF/TVS solution works, I probably won't come back to this problem for a while

[Zero999]

Some more discussion here: https://www.eevblog.com/forum/renewable-energy/how-to-combine-ovp-with-surge-protection-without-burning-out-tvs/
Your currents are quite high, so the SRF/Fuse in those ratings start to get big.

You can use something to switch the FET based on a current sense resistor, but it starts getting more complex. There are quite a few chips that will be able to do this for you, if the budget is available.

That will work but beware it might not be enough to protect against fire, since MOSFETs have a nasty habit of failing short circuit. Additional fire protection should be provided by a traditional one shot fuse.

[thinkfat]

I've done a bit of research on crowbar circuits and it looks like a combination of a zener diode and an SCR is probably the better solution for the overvoltage protection. The advantage IMHO is that the overvoltage condition can be defined a bit more precisely than with a TVS, since the zener diode has a better defined breakdown voltage. Also, the SCR would have to drop much less power since it would effectively become a short circuit when it fires, unlike the TVS diode which is a clamping device. A 24V TVS that fires at around 30V or so would have to dissipate around 100 watts until the SRF trips. That in itself might be a fire hazard.

[thinkfat]

This is what I ended up building.


PCB:


What I learned about PTC fuses: You need to select them very carefully. In particular, you need to be sure that your power supply can provide enough current to trip the fuse reliably and quickly. Otherwise, you'll have a heating match between the mosfet, the SRC and the fuse. It might not be the fuse that looses. If you're unlucky, the SRC will fail open, or something will start burning.

[Zero999]

That looks like it should work. You might want to consider another resistor, about 1k, in series with D1 to slow it down to avoid nuisance tripping and a higher voltage, around 33V, transient suppressor diode in parallel with the whole thing to to clamp, brief high voltage spikes.

Now it may seem like I'm being a little picky, but the schematic is drawn backwards. The convention is power in on the left and out on the right. My first thoughts were it won't work, then I realised the power in is from J3. Mirror the top, so J3 is on the left and it will be much clearer.

[thinkfat]

Quote from: Zero999 on Today at 00:23:38

That looks like it should work. You might want to consider another resistor, about 1k, in series with D1 to slow it down to avoid nuisance tripping and a higher voltage, around 33V, transient suppressor diode in parallel with the whole thing to to clamp, brief high voltage spikes.

Now it may seem like I'm being a little picky, but the schematic is drawn backwards. The convention is power in on the left and out on the right. My first thoughts were it won't work, then I realised the power in is from J3. Mirror the top, so J3 is on the left and it will be much clearer.

Understood your remark regarding the drawing flow of the schematic.

I actually considered putting a resistor in series with D2 for current limiting, D3 might create an interesting region where Vgs drops into the linear region of the mosfet and that'll make it burn up real quick, I think. But a resistor for that would be seriously big. I also should have put more copper around the mosfet to help cooling it, but I'm a bit size constrained here, I cannot make the board any larger as it has to fit onto a 100x150 mm heatsink together with the LPRO. What I might do is put a thermal plane on the bottom side of the PCB under the mosfet and connect it with some thermal vias. And maybe choose a different type with less than 60 milliohms. Rds_on.

[thinkfat]

Here's a photo of the adapter mounted to the LPRO.