Electronics > Beginners
Best Overvoltage & Overcurrent Protection forr failproof operation design
<< < (2/2)
David Hess:
What exactly are you protecting against?

Crowbar protection is more about safety than fault protection.  It is usually intended to prevent catastrophic failure which is why blowing the fuse is not a big deal; once the crowbar is needed, something is already seriously wrong and a blown fuse is the least of your worries.

Sometimes crowbar protection is combined with robust current limiting so it can be reset after a fault is cleared; some bench power supplies have this.

Shunt protection like with a transient voltage suppression diode or power shunt regulator works for temporary voltage spikes but they should not be happening at the output of a regulator anyway.

If you put the fuse before the feedback divider on the output of the switching regulator, then its voltage drop will be compensated for.  Another option is to put the fuse and the crowbar at the input to the regulator with the crowbar's trigger circuit at the output but some regulators become really unhappy if their input voltage is lower than their output voltage; special precautions may be necessary.

JS:
  If the buck converter is protected properly (against output short circuit) you might not need the fuse and rely on the buck protection, that way you don't have the voltage across the fuse.

  Using the protection at the input would only protect from the buck output to feed more than your desired 5V but can do nothing if you are forcing the input to go higher by an external source. Even if your load can't generate higher voltages that could happen if your load has an external world connection which gets shorted with a higher voltage than 5V, let's say a µC with a transistor driving a 24V circuit. If the transistor fails and gets 24V to the base the energy will be coming from the load to the PS. The µC is probably death at this point but you might have some other circuit to protect connected at the 5V rail. In this case you need to limit the output voltage, not shutting down the converter input. Other than that, is quite likely than a buck converter doesn't have much trouble managing input short circuit, but not guaranteed by this post.

  If you are using the fuse as David suggested you need to be sure the voltage drop in the fuse won't cause trouble with the feedback loop. In conventional regulators it probably does, if you do that with an LM317 or it's family the load regulation will be considerably affected, I don't know with your particular regulator. Also, when the sense terminal gets shorted to ground and the output floating so fast it could also bring problems. Check and double check the regulators datasheet. (fine print read might be needed).

You could replace the fuse with a low on resistance mosfet which gets turned off when the crowbar acts. It won't be fast enough so the supply will saw it for a short period but buck converter output will have an inductor which will damp the blow until the mosfet takes place. With just the 5V to work with so you don't need to have external supply and extra circuit you could use a logic level mosfet, as N channel mosfets are much easier to get and cheaper, speccialy for logic levels, you could put it in the low side of the circuit (0V rail) instead on the 5V rail and the gate connected to the 5V rail. Source to input, drain to output. With this, and choosing the mosfet wisely, you could lower the voltage drop by an order of magnitude.

JS
David Hess:

--- Quote from: JS on May 18, 2018, 12:26:05 am ---If the buck converter is protected properly (against output short circuit) you might not need the fuse and rely on the buck protection, that way you don't have the voltage across the fuse.
--- End quote ---

Crowbars are specifically useful against pass element failures which result in the input voltage being applied directly to the output.  If the pass element shorts, then current limiting becomes useless.

Sometimes switching regulator topologies which provide DC isolation between the input and output like Flyback and SEPIC are used to avoid this very problem but a crowbar could still be used to protect from feedback divider failure.

JS:

--- Quote from: David Hess on May 18, 2018, 12:48:07 am ---
--- Quote from: JS on May 18, 2018, 12:26:05 am ---If the buck converter is protected properly (against output short circuit) you might not need the fuse and rely on the buck protection, that way you don't have the voltage across the fuse.
--- End quote ---

Crowbars are specifically useful against pass element failures which result in the input voltage being applied directly to the output.  If the pass element shorts, then current limiting becomes useless.

Sometimes switching regulator topologies which provide DC isolation between the input and output like Flyback and SEPIC are used to avoid this very problem but a crowbar could still be used to protect from feedback divider failure.

--- End quote ---
In the case of the pass element being shorted the converter is already wrecked and the input fuse will blow. Having a fuse in the input of the converter is a good idea anyway I guess and no problems there with voltage drop.

JS
Zero999:
I'd recommend both an SCR/TRIAC crowbar and zener.

Here's an example showing both. The first TL431 triggers an SCR to short circuit the supply. C1 slows it down so it's not triggered by brief transients. The second TL431 and the BJTs form a zener diode, with a higher voltage, than the crowbar and provide protection against brief over-voltage spikes. Another discrete over-voltage protection diode, with a higher voltage than the second TL431 circuit could be used too, because it will be faster still.

https://www.eevblog.com/forum/beginners/2n7000-strange-behaviour/msg1185894/#msg1185894

Navigation
Message Index
Previous page
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod