Voltage Suppressor/Voltage Clamping Advice

I have a circuit with a 24V rail that has a device on it that is rated for 26V max. I’m a little worried about transients or something causing the 24V rail to exceed 26V momentarily given the 24V rail will be driving a motor. Are there any voltage clamping devices, TVS, or any other type of component/circuit that remain open/off at ~24-25V for normal rail operation but will still activate/breakdown to ensure the rail doesn’t exceed 26V? Basically something with a very well defined/narrow breakdown or activation window just below 26V?

I haven’t been able to find anything on my own… looking for any suggestions or ideas on how to handle this.

You might look into electronic circuit breaker chips. TI would be a good place to start. Some of them have overvoltage protection which is settable with a resistor and quite precise. The motor would go upstream and the sensitive load downstream. Do look at how they respond to overvoltage. Some latch off which probably isn’t what you want.

Another series option, if the sensitive load is light, would be an LDO regulator set to 24 or 25 V and operating in dropout.

I think a TVS’s tolerance will be too loose for your application - that is a very narrow window! I’m also not a fan of shunting motor energy if it is avoidable. Block small transients, clamp big ones.

Do you need to continue operation during a transient, or do you need to shutdown?

For continued operation you effectively make a low drop out linear regulator with a target voltage of e.g. 25V. You’ll end up with an ohmic loss during normal operation and it’ll clamp during transients. You can do this with an actual linear regulator, an electronic fuse (e.g. TPS2521), or drive a fet with an opamp.

If this is a device that might have supply rail issues due to back-feed, an active clamp circuit may be in order - a comparator to monitor the supply rail against a precision reference can drive a FET that dumps the excess power into a load resistor.

If the issue is that your device just can’t tolerate going over limit, but can operate a few volts below the nominal 24V, you might consider taking a few diode drops from the PSU if you can afford the power dissipation.

Firstly, if you’re expecting the voltage to sit around 25-26V for long periods of time (not transients) or consistently (more often than random instances) then you shouldn’t have picked a 26V chip placed right on that bus.

If the energy (power AND time) level of such a overvoltage is high (i.e. an intentional overvoltage, from a power source that someone has plugged in) then what @ToyBuilder has proposed is what you need (an active crowbar solution).

For a transient, though, that will be too slow.

If you are just worried about spikes and typical transients (EFT or Surge), things with not a severe amount of energy or time exposed to the circuit, then the key is to have a TVS but also some sort of buffering between the TVS and downstream circuitry. The voltage will essentially be clamped, filtered, and buffered, and as such the DC 24V is all that should be seen by the chip, despite the transient being inflicted.

Most circuits that are rated to 26V, would only have a 24V TVS as the transient protection (not a complex crowbar circuit) plus some filtering, spacing, and correct placement. So, it’s just about implementing it correctly (not just placing a TVS, close to the chip you’re protecting, with nothing else in between helping - that’s not going to do anything).

1 Like

Depends on the current. You could use a TL431 with a FET, 1% accurate

Is the specs of 26V abs max or just recommended max?

Often there is a considerable difference

Its absolute max of 26V

Thank you all for your feedback, its much appreciated!

I’m realizing I did a poor job of explaining the problem, let me try again. Please see the attached figure for a better understanding.

I need to provide a motor and motor driver IC (H-bridge) with a constant 24V. So I have a boost converter providing that 24V to the motor to make sure its a steady 24V. Input to the boost stage is a battery (12V nominal). I also have the INA219B current monitor/ADC from TI that will monitor current to the motor for some data analysis type purposes. The INA219B has a max common mode voltage rating of 26V on the two lines that measure the voltage drop across the sense resistor as shown in the diagram.

My concern is that the INA219B will see 24V nominally, which is within spec by a couple volts. However if there were any transients from the motor turning on/off or the Boost stage momentarily exceeding 26V, I could blow up the INA219B. So I would like to find a way of protecting the sense inputs of the INA219B from seeing anything above 26V.

I could find another way of doing the current monitoring and ditch the INA219B but we have some data that shows the performance of the INA219B is good enough for this application. I don’t have enough data to know exactly what ADC specs are needed to get sufficiently good data for this application (i.e. resolution, sample rate, DNL, INL, etc). So I’m worried about deviating from the INA219B as its the only data point I have so far for a current monitor/ADC that I have confidence will work.

So I feel like I’m stuck between the risk of finding an alternate way of doing current monitoring to ditch the INA219B or risk the INA219B seeing voltages above 26V and damaging the chip. Hence why I’m hoping to find a cleaver way of protecting the INA219B while maintaining the 24V nominal voltage to the motor.

Any thoughts are welcome and much appreciated! I’ve lost too much sleep over this one!

My suggestion – investigate Allego’s ACS7xx line.

Have you considered the INA238?

There are lots of high-side monitors with higher C-M voltage ratings, including in TI’s INA line. You could also look at low-side monitoring. Much better than trying to protect the INA219B.

1 Like

It’s 100% going to spike over 26V with that setup. However I think the problem isn’t so terrible if you’re committed against changing. A few things I’d consider:

  • Test the INA219B - drop 40V into it and see how quickly it fries?
  • Add series resistors & caps - the input impedance of that is pretty high, no need to wire the shunt directly into the device. If you don’t need super-fast edge sensing you can add caps here too to reduce spikes from ringing.
  • Add zener or TVS on the over-voltage - a zener might be enough w/ the lower current after the series resistor.

You could also add buffer amplifiers etc. But at that point that you are adding active circuitry, probably better to find an alternate part unless you have super-specific reasons for the INA219B.

Motor power can be pretty dirty, so unless this is an ultra-cheap design I’d personally (a) change to something with higher ratings and (b) add the series resistors+caps, even as a board stuff option.

Thank you everyone, I can’t thank you all enough!

Based on the feedback, I think I just need to stop being such a chicken shit and move away from the INA219B despite it being the only device I have good results with so far. I was too worried about selecting another device with better common mode specs and then finding out that the ADC performance doesn’t provide good enough results but now I think that may just be what I have to do

Many thanks!.

1 Like

Place the sense resistor in the low side. You are using the STSPIN 840, and that is designed for having external sense resistors. At max current the voltage on the sense resistor should be less than 500mV so not to interfere with the internal gatedrivers of the STSPIN. If you need more gain, add an external opamp

For motor drives you need a big DC link capacitor that can absorb the energy when you are breaking the motor. During breaking, the motor act as a generator. For larger systems one often have a break FET to clamp the energy