High power current sense resistor resistance/voltage drop during inrush current

[aiq25]

Hello. I'm using two 150mOhm current sense resistors in series to measure low side current of an H-Bridge. During normal operation my current is below 1A (voltage on the high side > 150V). During inrush events currents can be as high as 15A. I'm seeing something strange: during an inrush the voltage across both of these resistors is about 1V lower than it should be, according to a current probe placed on the output of the H-Bridge. I'm measuring the voltage across the two ends of the current sense, not individual resistors.

In short, I have a current probe on the output of the H-Bridge which measures the current to be 2-3 times greater than what the current sense resistor is measuring. I measured the voltage drop by directly connecting leads to the resistors. I have verified this is the correct voltage reading by using different methods and oscilloscopes. I have two theory: there is some going on in the PCB that is causing this OR the resistance of the current sense resistors is dropping during these high current peaks.

The strange thing about this is that by reading lower current through the sense resistors, it implies that there is another current path. i have verified the schematic and PCB design and they look correct. I'm confused as to what is causing this. Any feedback would be appreciated.

Due to the lower voltage drop across the sense resistors, the HW protection circuit is kicking in at much higher output currents. The inrush current events are 10-50 us in length and are pulsed currents. The pulsing can last 10-40 ms with about 10-50 us of dead time in between the pulses.

If anyone is curious I'm using 2 of these series parts: Susumu KRL1632E-M

Disclaimer: This is for a work project, so I can't share too many details.

[matbob]

Hello aiq25,

    Do you have any decoupling capacitors connected near the H-Bridge? Where are the current sense resistors placed? Does the current through the decoupling capacitors flow through the sense resistors?
 

[aiq25]

Hello aiq25,

    Do you have any decoupling capacitors connected near the H-Bridge? Where are the current sense resistors placed? Does the current through the decoupling capacitors flow through the sense resistors?

There is one capacitor that is placed across the two lines of the H-Bridge going out to the load. The current through that capacitor should flow through the resistor.

[ajb]

If you have a capacitor directly across the output of the h-bridge, current flowing out of this capacitor will NOT be passing through the current sense resistor.

It would be helpful to see a schematic just so we're all on the same page, and an h-bridge shouldn't be anything proprietary unless you're doing something really unusual.

It's possible that your measurement method is introducing some error.  Can you show is that, if only in diagram form?

[aiq25]

Attached is the block diagram. The capacitor is directly across the AC lines, so the current from this would have to flow through the current resistor, at least that's my understanding. Please let me know if this is incorrect.

[ajb]

The charging current for the capacitor (from the HV bus) flows through the sense resistors. The discharge current (through the load) does NOT flow through the current sense resistors. Nor does flyback current from the load into the capacitor in the case of an inductive load. In any case, a 100nf cap won't account for such a large discrepancy.

Your numbers don't add up, though. At 2x 0.15mOhm, to should only be seeing 4.5mV at 15A.  Did you mean 150mOhm aka 0.15Ohm, or are you talking about the output of the sense amplifier? If the latter, what is the sense amplifier? Schematic/part number? What's it's gain and expected error? What does the layout of the sense resistors and amplifier look like? (This is important, especially if your resistors really are 0.15mOhm!) Are you sure that your current probe is accurate? Is it nuked properly?

[aiq25]

Your numbers don't add up, though. At 2x 0.15mOhm, to should only be seeing 4.5mV at 15A.  Did you mean 150mOhm aka 0.15Ohm, or are you talking about the output of the sense amplifier? If the latter, what is the sense amplifier? Schematic/part number? What's it's gain and expected error? What does the layout of the sense resistors and amplifier look like? (This is important, especially if your resistors really are 0.15mOhm!) Are you sure that your current probe is accurate? Is it nuked properly?

Sorry I meant to say 150mOhm. I was measuring right across those two resistors. I will have to get the part number tomorrow, don't remember it off hand. The gain is around 12. I can give a block diagram of the PCB layout tomorrow as well.

All the tests were done with non-inductive loads. I tried it with two different probes and they both read the same. Also one of them is calibrated, so I'm pretty confident in the probe results. The probes I don't remember the part number but they are Tektronix 50A probes with current amplifier.

The charging current for the capacitor (from the HV bus) flows through the sense resistors. The discharge current (through the load) does NOT flow through the current sense resistors. Nor does flyback current from the load into the capacitor in the case of an inductive load. In any case, a 100nf cap won't account for such a large discrepancy.

Can you explain the path for the discharge current? I understood the charge current but I would expect the discharge current to eventually flow through the current sense resistors, I'm guess I'm wrong.

[Tomorokoshi]

A couple things:
1. The Susumu part number provided is 100 mOhm, not 150 mOhm. This may explain most of that missing 1 V. Closer to 0.75 V perhaps?
https://www.digikey.com/products/en?keywords=%20408-1582-1-ND

2.  That 1206 package is rated to 0.75 W. You would need to limit your current to below 3 A.

[aiq25]

A couple things:
1. The Susumu part number provided is 100 mOhm, not 150 mOhm. This may explain most of that missing 1 V. Closer to 0.75 V perhaps?
https://www.digikey.com/products/en?keywords=%20408-1582-1-ND

I tried 100mOhm before but posted the wrong part number, the part I'm using is intended a 150 mOhm. I also measured both resistors on PCB to be around 0.3 Ohms. I used the lowest resistance setting on the multimeter I was using but it is possible I got a wrong reading since this is such a low resistance.

2.  That 1206 package is rated to 0.75 W. You would need to limit your current to below 3 A.

Yes, the nominal operating current is max 0.75A, 1.5A for 10 seconds or so, so the part should be good for nominal operation. They are not rated for surge power or repetitive pulses so that's what worries me.

[aiq25]

Today I tried another brand and part (0.3 Ohm) and still saw the same results. This resistor was just something we had on hand, still not surge or peak repetitive rated though.

[Tomorokoshi]

Do you have a way of determining the frequency response of your current sense amplifier?

Can you also measure the Vo / Vi transfer function of it?

[matbob]

Are the gate drive currents for the low side MOSFETs flowing through the sense resistors?

[aiq25]

Do you have a way of determining the frequency response of your current sense amplifier?

Can you also measure the Vo / Vi transfer function of it?

I really don't have a way of determining the frequency response because this is part of a much larger circuit. I will try to do it in the future though.

I measured the voltage right across the sense resistor. The amplifier feeds into a micro. Under normal operation the amplifier is measuring correctly and the gain is correct. I have not looked at the output during an inrush because I was just measuring across the sense resistor.

Are the gate drive currents for the low side MOSFETs flowing through the sense resistors?

Yes. The H-Bridge is driven using an IC and the ground of the low side FET is tied to the ground of the sense resistors.

[T3sl4co1l]

What dI/dt is this inrush, and what's the layout look like?

Related question: can this inrush be much more than the circuit is designed for?  Are you not limiting it with a current mode controller?  You've got a whole H-bridge there, with current sense -- it would be a shame to throw away such an excellent opportunity.

Tim

[ajb]

We're all just guessing right now, we need a lot more information to provide any useful advice. 

One thing that jumps out is that if the current sense resistors total 200-300mOhm (not 0.2-0.3mOhm as originally stated), that means that at 15A the source of the low side switches will be at 3-4.5V, which, depending on the gate drive situation, could mean you get "free" current limiting, or with enough d/dt and the right parasitics, something that rings like a bell and blows up. 

Can you show the layout, some scope traces, and ideally a better schematic?  Are these "inrush" events (really, repetitive pulses from what you've described) caused by the load itself, or a result of the supply fluctuating, or are they induced by the way you're driving the load?

[tszaboo]

It is either the transient response of the current sense amplifier. It is not that great, some of them go up to a 100 KHz, but that's it. Some of them have terrible offset or non linear gain.

Or it can be the current clamp also. They are even worse. The ones with banana jack are instant disqualified for any inrush measurement. They work fine in DC, they work fine in AC, they dont work fine with anything in between. I give you an example, a Fluke i30 (or some other Fluke clamp) was showing current in the wrong direction when you applied a step response to it. The ones with BNC connector might work, but they will have some problems.

[aiq25]

What dI/dt is this inrush, and what's the layout look like?

Related question: can this inrush be much more than the circuit is designed for?  Are you not limiting it with a current mode controller?  You've got a whole H-bridge there, with current sense -- it would be a shame to throw away such an excellent opportunity.

Tim

The dI/dt is typically ~10A/5us. Unfortunately I cannot share the exact layout. Maybe I can draw something and share, which parts do you think would be helpful?

This inrush can be more but there is a overcurrent protection circuit that kicks in and pulls the H-Bridge driver low when the inrush is more than the set value. There is no controller though. Do you have any suggestion of a controller?

[aiq25]

It is either the transient response of the current sense amplifier. It is not that great, some of them go up to a 100 KHz, but that's it. Some of them have terrible offset or non linear gain.

Or it can be the current clamp also. They are even worse. The ones with banana jack are instant disqualified for any inrush measurement. They work fine in DC, they work fine in AC, they dont work fine with anything in between. I give you an example, a Fluke i30 (or some other Fluke clamp) was showing current in the wrong direction when you applied a step response to it. The ones with BNC connector might work, but they will have some problems.

I'm measuring this right across the sense resistor, so not taking into account the current sense amplifier.

The current clamp is calibrated and I have tried different ones. I have been using this for other projects and it worked, so most likely I'm thinking it is not the clamp. This is with BNC connector and amplifier, 50A Tektronix probe.

[aiq25]

Today I tried something different: I soldered a wire loop before the current sense resistor (I know not the best practice but I wanted to see) and the current through the loop was different than what is measured on the output AC lines HOWEVER the current matched the voltage reading I got across the sense resistors. I'm thinking it is probably not the sense resistors. Everything is pointing towards there is another current path somewhere in the circuit but I cannot figure out what it could be.

[tszaboo]

It is either the transient response of the current sense amplifier. It is not that great, some of them go up to a 100 KHz, but that's it. Some of them have terrible offset or non linear gain.

Or it can be the current clamp also. They are even worse. The ones with banana jack are instant disqualified for any inrush measurement. They work fine in DC, they work fine in AC, they dont work fine with anything in between. I give you an example, a Fluke i30 (or some other Fluke clamp) was showing current in the wrong direction when you applied a step response to it. The ones with BNC connector might work, but they will have some problems.

I'm measuring this right across the sense resistor, so not taking into account the current sense amplifier.

The current clamp is calibrated and I have tried different ones. I have been using this for other projects and it worked, so most likely I'm thinking it is not the clamp. This is with BNC connector and amplifier, 50A Tektronix probe.

Oh, OK, so the amplifier is not measured. Did you take care about ground loops then?

Also, the current clamp are not "broken", they just are not made to handle this situation well. It is combination of the core characteristics, and the hall effect sensor's characteristics, and the surrounding circuit's with their own problems.

[aiq25]

Oh, OK, so the amplifier is not measured. Did you take care about ground loops then?

Also, the current clamp are not "broken", they just are not made to handle this situation well. It is combination of the core characteristics, and the hall effect sensor's characteristics, and the surrounding circuit's with their own problems.

I'm thinking it is some kind of parasitic component. I have tried to take care of ground loops and reviewed the layout extensively, don't see any way for it to be a ground loop. The reason I'm thinking its a parasitic component is under normal operating (load current < 1-2 Amps) the circuit behaves as expected. It is only during high inrush currents it has this weird behavior.

The current probe is 50MHz, so it should have quite good frequency response/bandwidth.

It is not the current sense resistors because I tried high power, pulse withstanding ones now.