Anti-Overshoot Power Supply

[nctnico]

I have to admit that motor control is not my strongest suit, so here's what I can tell you. It's a 3-phase Brushless DC motor, with hall sensors, we're using sine-drive control, DRV8301 controller with some decent sized FETs for switching. The 4V battery is boosted by a 2-phase synchronous boost converter to 8V which is what the FETs pull from to run the motor at 8V. There's 1500uF of caps on the 8V side. I'd be really interested to know how I can find out whether it is coming from the motor or not.

Unless something is very wrong or its  running as a bidirectional converter the boost converter shouldn't let any power flow back from the motor to the 4V power input. Probe the 8V supply under the fault condition just to be sure but the obvious way to find out if its the motor would be a current probe somewhere along that 4V -> 8V path.

Assuming its just a boost converter you're probably left with it being a characteristic of the test power supply system, including the cabling etc. Remote sense and some capacitance right at the load could be all thats required, fancy 2 quadrant power supplies are when you need to push power back into the source (battery for example).

It can still be induced somehow. At high currents even a small inductance can cause a nasty spike. I one of my designs I had trouble with the self inductance of planar (not wirewound!) resistors.

I think this circuit is a good occasion to get a differential probe and probe around a bit to see where the spike is actually coming from.

[katzohki]

I have to admit that motor control is not my strongest suit, so here's what I can tell you. It's a 3-phase Brushless DC motor, with hall sensors, we're using sine-drive control, DRV8301 controller with some decent sized FETs for switching. The 4V battery is boosted by a 2-phase synchronous boost converter to 8V which is what the FETs pull from to run the motor at 8V. There's 1500uF of caps on the 8V side. I'd be really interested to know how I can find out whether it is coming from the motor or not.

Unless something is very wrong or its  running as a bidirectional converter the boost converter shouldn't let any power flow back from the motor to the 4V power input. Probe the 8V supply under the fault condition just to be sure but the obvious way to find out if its the motor would be a current probe somewhere along that 4V -> 8V path.

Assuming its just a boost converter you're probably left with it being a characteristic of the test power supply system, including the cabling etc. Remote sense and some capacitance right at the load could be all thats required, fancy 2 quadrant power supplies are when you need to push power back into the source (battery for example).

Okay thanks, this was my thinking as well. Can't hurt to look at the 8V side of things certainly so I'll add it to the checklist as I get on with this.

[T3sl4co1l]

I'm not sure what you mean by ride through, but at 4 Volts or adjustable output I'm not finding anything that can provide at least 15A that would still be called an LDO. At any rate, if I can just whack on a linear regulator and call it a day I'm in. Although this shunt regulator idea sounds fun.

You said this --

...we have a component in there that gets taken out at 6V. That this isn't protected is a "whoopsie" in the design, but it's supposed to be battery powered so it's only happening during test.

Design oversight or not, perhaps it is time to add some protection to that component?  A regulator can be added in front of it, to drop essentially no voltage under normal circumstances, but provide the same output voltage even when the input swells too high.  In other words, a regulator, to ride through it.

Tim

[David Hess]

Some regulated power supplies have better load transient response than others.  If this is important then it should be part of the specifications during design.  I have seen commercial supplies which gave it no thought and had obvious design mistakes which made overshoot worse like leaving out base-emitter shunt resistors.

T3sl4co1l mentioned using a shunt regulator and I have done that before.  I essentially made a precision power zener diode using a big power transistor and feedback circuit.

Another common solution is to add enough bulk output capacitance to absorb the transient.

[ogden]

BTW: the 6573A is a switching PSU too. I have the 20V/100A version. And yes these series have a down programmer too although it seems to be unspecified.

Ouch, my bad. Thank you for clearing this out. For example 6654A I used long ago is linear, thou it was "only" 10A or so. Seems like 6x7x supplies are SMPS and 6x4x, 6x5x - linear, right?

[Keysight DanielBogdanoff]

I talked to the power team about this, they mentioned that the motor loads are tricky (for a resistive loads the E362xxA supplies have <1V overshoot, but motors can vary).

You should be able to get a demo of one if you want to try it out. You could parallel an E36233A's outputs to get to those power levels.

[nctnico]

BTW: the 6573A is a switching PSU too. I have the 20V/100A version. And yes these series have a down programmer too although it seems to be unspecified.

Ouch, my bad. Thank you for clearing this out. For example 6654A I used long ago is linear, thou it was "only" 10A or so. Seems like 6x7x supplies are SMPS and 6x4x, 6x5x - linear, right?

Based on the service manual this appears to be the case. The 6573A / 6673A still weighs 28kg though!