BLDC pump suddenly runs dry...what happens next?

[BobW]

You've totally misunderstood that article. There is nothing there to indicate that a motor should be driven from a constant current source.

Yes, they do talk about switching current between motor windings, but that's a generic term that means the same thing as switching power. You need to go back and rethink your entire motor drive strategy.

In that article, the word "constant" appears twice. The word "current" appears 7 times. Those two words never appear next to each other. You apply a voltage and the current takes care of itself.

...they do say regulate the voltage, but that surely is the same as regulating the current?

No. It's not the same thing at all!!!

 

[Flyback]

OK, but let me show you a block diagram of our setup...(attached)

The motor coil commutation period is actually 2.5ms (not the 1.05ms that is written on it)
...surely its just a simple case of increasing the current when the motor speed goes below 8000rpm, and reducing the current when motor speed goes above 8000rpm?.......but when the pump suddenly runs dry, that's an extreme load removal, and I just don't see how the motor goes faster....the speed feedback loop couldn't possibly be fast enough to account for that...and so the driver would end up pulsing the coils on when they shouldn't be on, and the motor would just spin unsmoothly, not necessarily faster at all?
so what I am saying is, how does the controller know to reduce current to the motor when the pump suddenly runs dry?

 

[alec_t]

As I understand it the controller speed feedback loop is designed to keep the short-and-long-term-average speed constant regardless of pump load, if the pump supply is a regulated voltage (not constant current). At constant speed the short-term-average current will drop if the load decreases (pump runs dry), so monitoring current should enable you to check that and switch off the pump.

 

[4pyros]

surely its just a simple case of increasing the current when the motor speed goes below 8000rpm, and reducing the current when motor speed goes above 8000rpm?

No you change the period like in PWM control

 

[Flyback]

youre right, its set up to regulate the speed to 8000rpm no matter what the load......but if the switching frequency of the inverter gives 8000rpm......then whatever you regulate your current to, the motor will spin at 8000rpm?...is this the problem?......you appear to be saying that if we regulate the vout (instead of Iout) of the upstream SMPS then that solves the problem?

As far as changing the period is concerned...that's the way to do it in low power motors, but we are not pwm'ing the bridge transistors.......the commutation period for each coil is 2.5ms, and so we literally switch the coil in for that 2.5ms with the transistors...there's no pwm'ing of the transistor within that 2.5ms.

 

[alec_t]

You seem to be over-riding the control IC, rather than letting it do its own thing?

 

[4pyros]

youre right, its set up to regulate the speed to 8000rpm no matter what the load......but if the switching frequency of the inverter gives 8000rpm......then whatever you regulate your current to, the motor will spin at 8000rpm?...is this the problem?......you appear to be saying that if we regulate the vout (instead of Iout) of the upstream SMPS then that solves the problem?

As far as changing the period is concerned...that's the way to do it in low power motors, but we are not pwm'ing the bridge transistors.......the commutation period for each coil is 2.5ms, and so we literally switch the coil in for that 2.5ms with the transistors...there's no pwm'ing of the transistor within that 2.5ms.

I dont get it are you using a controller or trying to make one????

 

[Flyback]

We are mainly using the ML4425 to switch the bridge transistors.....we also use the tacho output of the ml4425 to tell us what the motor speed is....we feed the tacho pin output pulse train to a frequency to voltage converter IC , and this then outputs a dc voltage which is proportional to the motor speed......we use this dc representation of motor speed in the feedback loop of the buckboost converter, which varies its current accordingly....ie regulates the motor speed.

 

[alec_t]

we use this dc representation of motor speed in the feedback loop of the buckboost converter, which varies its current accordingly....ie regulates the motor speed

Doesn't that conflict with the IC's attempt to control speed by varying the coil commutation times?

 

[tcmtech]

To me this sounds like you are driving your motor at near (winding and core magnetic) saturation power levels 100% of the time which makes the feedback loops nearly pointless.

By doing so when the motor is under load it converts more of that input power into useful mechanical work but when unloaded almost all the input power just becomes raw heat thus that's why there is little to no input voltage or current changes that can be seen.

The thing is I have lots of small to mid sized BLDC motors and a few large industrial servo units around and when their driving circuitry is designed right and set up properly they all run off of common constant voltage power supplies and exhibit variable current draws as their mechanical loads change just as any common brushed DC or AC induction motor does.

 

[Flyback]

Doesn't that conflict with the IC's attempt to control speed by varying the coil commutation times?

The ISENSE pin of the ML4425 is grounded in our setup.....the ML4425 does not attempt to do any PWM'ing or current limitation.

As far as the ML4425 trys to control the coil commutation times....Well, we have the SPEEDSET pin of the ML4425 tied to the VREF pin of the ML4425....so we don't get much variation in commutation frequency.

 

[tcmtech]

...no because the ISENSE pin of the ML4425 is grounded in our setup.....the ML4425 does not attempt to do any PWM'ing or current limitation.

So what you are saying is that you override the IC's control systems in order to add a overly complex and for the most part completely unnecessary external control system that doesn't do a damn thing gainful to the whole process and application?

What is the pump motors designed input voltage and current rating?
What is the primary power source and why?
What is the driver circuits input voltage and why?
What are you running everything at and why?

Why didn't I just keep out of this thread being I knew damn well who the OP is and what he is known for?

 

[Flyback]

The Engineer who originated this design is, I believe, a top consultant motor & drives expert.
The motor, when at 8000rpm has a voltage of 34V...and an average current to it of 20 Amps.

The primary power source is an 18 to 32V DC supply.....we are in a big agricultural place, and depending on the loading, the power source varies in this way, so our input source is 18v to 32VDC. The buckboost converter is fed from this 18 to 32vdc supply.

"Driver circuit"...?....do you mean the rail that powers the ML4425?....its a 12V rail derived from the 18 to 32v input source with a buck converter.

"What are you running everything at and why"?.....I think, with this email so far, I have now explained all the details of your request?

Are you asking why we are regulating the motor speed by regulating the current sourced to it by the buckboost converter?...well, because of the following...
1....We don't want a current sense resistor downstream of the motor coils, because we would have to site it on a pcb, and then wire to it from the motor, -this would mean that wiring inductance being in the current loop which would be noisy and we don't want that.
2....If we simply switch the motor coils in and out in this simple way, then we get virtually no switching losses in the IGBTs of the three phase bridge...also, we get very little high frequency hysteresis losses in the motor iron.
3.....We need the upstream buckboost converter anyway, so why not then use it as a regulated current source for regulating the speed of the motor?

The problem I suspect, is what happens when the pump suddenly runs dry?.....I can't see how the bridge driver ic (ML4425) commands a resultant increase in speed, which is what would be needed for the buckboost to reduce the current that it is sourcing to the motor?

 

[4pyros]

The Engineer who originated this design is, I believe, a top consultant motor & drives expert.
The motor, when at 8000rpm has a voltage of 34V...and an average current to it of 20 Amps.

As far as I understand it you want us to help you try and figure out some guys unconventional design without all the details.
I still don't know exactly what you are asking.
How it works the way it does? or how to modify it to do something else?
I think you need to ask the "top consultant motor & drives expert" in ether case.

 

[alec_t]

We don't want a current sense resistor downstream of the motor coils, because we would have to site it on a pcb, and then wire to it from the motor, -this would mean that wiring inductance being in the current loop which would be noisy

Can't you use a Hall sensor instead of a sense resistor?

what happens when the pump suddenly runs dry?.....I can't see how the bridge driver ic (ML4425) commands a resultant increase in speed

An increase in speed is the last thing you want! You have to stop the pump.

 

[Flyback]

The consultant has gone now.
I am just wondering about this situation of "Pump suddenly running dry".
I mean,are you in agreement that because we have the SPEEDSET pinof the ML4425 tied directly to the VREF pin of the ML4425, that the bridge IGBTs will be commutated at just one fixed frequency, (in our case 400Hz, which gives the 8000rpm in the 3 ph BLDC motor)

Can't you use a Hall sensor instead of a sense resistor?

...sorry no, a hall sensor has a good FAST bandwidth, but is still not fast enough to resolve the same as using a current sense resistor directly into the PWM comparator of the ML4425. And in any case, even if we used a hall sensor, it wouldn't take away the wiring inductance.
We could however, use a hall sensor as the current sensor for our buckboost converter...maybe that's what you meant?

As regards the question about whether we should stop the pump when it has run dry.....yes that would be nice, but we will settle for simply reducing the current to the motor so that it keeps spinning at 8000rpm...which is harmless, as long as the current does get reduced to that (presumably small) level which allows the motor to spin at 8000rpm.
....yes , we'd like for the pump to stop when it runs dry, but that means having all sorts of external extra bells and whistles like flow sensors etc etc...and we don't want that...after all, we are only watering tulip fields via our irrigation system.

 

[4pyros]

The consultant has gone now.
I am just wondering about this situation of "Pump suddenly running dry".
I mean,are you in agreement that because we have the SPEEDSET pinof the ML4425 tied directly to the VREF pin of the ML4425, that the bridge IGBTs will be commutated at just one fixed frequency, (in our case 400Hz, which gives the 8000rpm in the 3 ph BLDC motor)

OK so now it sounds like you never even tested it to see if there really is even a problem.
Can you meter the motor wet and dry and see what happens?

 

[Flyback]

So far, we just have the motor running open loop with a dynamometer attached, and we just feed it from a 34V current limited lab power supply to get it to spin....(note, in this lab setup we have to physically nudge the motor to get it to align before it spins up)
However, we are yet to attach the motor to our proposed buckboost current source as described.

Incidentally, in our open loop lab setup.....the motor coil current looks highly resonant, and we don't have flat-topped pulses of motor current...but the peaks go up just above 30A even though the average current is 20 Amps....in this initial lab setup, we are again not PWMing the bridge IGBTs with high frequency PWM...but just rather switching the bridge IGBTs at 400Hz...to give the coil-to-coil commutation of the motor coils.

 

[alec_t]

a hall sensor has a good FAST bandwidth, but is still not fast enough to resolve the same as using a current sense resistor directly into the PWM comparator of the ML4425.

The Hall sensor would not be used for messing with the PWM (which is for applying a current limit); it would be solely for detecting the pump load change, independent of any 4425 or converter function.

we will settle for simply reducing the current to the motor so that it keeps spinning at 8000rpm...which is harmless

I'd have to disagree on 'harmless'. Methinks the pump seals/bearings will suffer catastrophic failure.

we are again not PWMing the bridge IGBTs with high frequency PWM

But surely that's exactly what the 4425 does?

 

[4pyros]

lab setup? I thought you said it was in the field watering flowers or something.
Is this something you are trying to make?