Current sense with INA213A

[drkidd22]

Related to:
https://www.electro-tech-online.com/threads/solenoid-control-with-mosfet.152038/#post-1311712

I have the circuit below and it's working ok so far. But I've ran into an issue where the current sense IC U5 has blown out all of the sudden. I am operating below the devices rated voltage and there;s simply a LPF on the output to smooth out the current so it can be read over ADC. Does anyone have any experience with the INA213A?

At the connector J2 I have a solenoid which draws about 1A so my output from the current sense is about 2.5V and I do see that, when it's working properly.

 

[dknguyen]

I'm pretty sure it's inductive flyback. This is a motor being powered from a 24V supply. The inputs on the INA213A can only withstand 26V, so why is there an SMAZ33 which kicks in at 33V? For that matter, why is a zener even there to begin with? You don't need one. All it's doing is preventing D2 from being able to supress the spike until it exceeds the supply by 33V which is 57V relative to ground. Without it, D2 would suppress the spike when it exceeds the supply by only 0.5V. Even then, 24.5V when the device can only withstand 26V is a little close for comfort.

After removing D3, consider placing a TVS diode or zener across the current sense resistors. Technically you want to size it so it's a little higher than your maximum expected current sense voltage but you won't find a diode with a breakdown voltage that low. But even if you use a 3V, 5V, 10V diode which is many times the current sensing voltage, it would still clamp the voltage across U5-4/5 far below the 24.5V that D2 (and no D3) would.

Why is D3 there?

EDIT: I realize that technically Q2 should be taking the brunt of the the inductive spike and preventing it from reaching U5-4/5, but if U5 took just one of those spikes directly then I would have expected it to fail immediately . However, you imply that it was working okay for the most part so it could be that the spike is making it's way through the parasitics in Q2 and reaching U5 at lower levels slowly wearing it out until it blew. It's hard to say for sure if that is what is damaging U5, but the effect of D3 is significant enough that you can't just write it off and say it's impossible that it's getting through Q2 to U5. So replace D3 with a short first and see if the issue goes away.

 

[ronsimpson]

why is a zener even there to begin with?

If you only have D2 the voltage across the coil is less than 1 volt during "flyback" time. (Just turned off) With a very small voltage across the coil it will reset/turn off very slow. If you allow 33+ volts during flyback the coil will reset the current much faster.

U2 and U5 have a 24V supply. I did not look up the data but that seems high. I think a 100uF cap on the 24V will help. If there is a spike I think it is getting into U5 by VCC. I just can't see it coming from Q2 into U5. (R5, R17 should kill that and the MOSFET should not allow it any way)

 

[dknguyen]

If you only have D2 the voltage across the coil is less than 1 volt during "flyback" time. (Just turned off) With a very small voltage across the coil it will reset/turn off very slow. If you allow 33+ volts during flyback the coil will reset the current much faster.

Hmm, I see. In that case, I agree with Ron.

 

[ronsimpson]

I went back to the beginning of this. It appears the idea is to apply full voltage to the coil until it is on, then drop back to 1/2 voltage to hole the coil on. (often the hold current is 1/2 the turn on current)
Then for turn off the idea is to reset the coil fast. (fast using 33 volts and fast because the coil current is 1/2 so it will take less time)

 

[drkidd22]

Thanks for the tips guys, I will look at the spike and see if its getting in through VCC on U5.
Maybe I can also tie U5 Vcc to a 5V rail instead of 24V.

 

[ronsimpson]

I want a large cap on the supply.

You can add a RC on the power supply to that IC or both ICs.

 

[drkidd22]

I ran into another scenario, the solenoid has a metal casing and this happened when the solenoid was very close to the PCB and the case not earth grounded. So might also be static build up arching over to the PCB.

 

[drkidd22]

So I had some time to look into this and captured the below waveforms. Ch1 is 24VDC line, Ch2, is voltage across the solenoid and Ch3 is voltage across R5||R17.
There is a lot of noise on the line when the solenoid is triggered, how can I attenuate this?

On the below image I tried adding a 100uF to the 24V line and it helped a little, but I want to get it cleaner.

Ch1 below is the output of U5 measured at R12 and Ch2 is the actual current through the wire.