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

[Flyback]

-

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.

...Right, but would the current to the motor just go and decrease when the pump suddenly runs dry?...please remember we are regulating the current to the motor with our buckboost...so our feedback loop is actually responsible for the level of current sent to the motor...if our feedback loop fails to detect motor overspeed when the pump suddenly runs dry, then our buckboost will not reduce the current that it sends to the motor.
For reasons hitherto explained, when the pump suddenly runs dry, the motor may not in fact actually speed up, it may just spin irregularly at some random speed, however fast or slow that speed may be.
Our buckboost feedback loop will only decrease current to the motor if it senses the motor speeding up above the nominal value (8000 rpm)

--yes we are trying to make it so that we can use it in our tulip field irrigation system...we don't have it built yet

But surely that's exactly what the 4425 does?

yes it is capable of PWM'ing the IGBTs of the inverter, but we have bypassed that by grounding the ISENSE pin of the ML4425....for us, the ML4425 simply commutates from one coil to the next....we have VREF tied to SPEEDSET, so ML4425 just switches the IGBTs at 400Hz.

 

[tcmtech]

So at peak input you are using ~34 volts DC at 20 amps ~(680 watts) just to pump water to irrigate some flowers.

So why are you not just using an off the shelf AC powered pump or lower voltage 12 or 24 volt DC or variable voltage (direct solar power applications) pump and a simple float switch to know when the tank is empty to begin with?

And don't say it's because of efficiency, cost, size, flowrate or pressure or that you are not using a float switch because there is not enough room or its too expensive/difficult to add to the system.

To me this all sounds like you inherited someone else's mess and are sticking with it for unknown reasons.

 

[alec_t]

yes it is capable of PWM'ing the IGBTs of the inverter, but we have bypassed that by grounding the ISENSE pin of the ML4425

I don't think you've bypassed it (but I could be wrong ). Grounding Isense will (if I've understood the 4425 datasheet correctly) result in a permanent logic high at the lower input of the AND gate preceding the gating logic, so won't prevent the PWM driving the gating logic. I reckon the IGBTs will be switching at ~20kHz, even if the coil commutation is at 400Hz (or whatever).

 

[Flyback]

..don't want AC powered pump because we have a DC supply to work with.
..BLDCs are getting cheaper and cheaper as they get more and more popular...plus they are small size.
..dont want 12 to 24v as its too much current.
..don't want a float switch because surely we don't have to have one?....are we really saying that a BLDC pump cannot be made without a float switch and be able to handle the situation of "Pump suddenly running dry"?

-no the igbt's aren't going to switch at 20khz....in our initial lab set up the ml4425 is switching them at 400hz

 

[4pyros]

The Engineer who originated this design is, I believe, a top consultant motor & drives expert.

The consultant has gone now.

--yes we are trying to make it so that we can use it in our tulip field irrigation system...we don't have it built yet

And now you are trying to fix something you can't even confirm is happening.
Your speed will not change because you have the PWM fixed at 400Hz.
Your current will not change.
So your system bypasses all the built in controls of the controller chip.
Why did not your consultant take into account for the pump running out of water?
Or did he? and you dont even know it.
What happens now the way it is when it runs out of water?
Does it change speed?
Does it change current?
Sorry but this whole thread is very confusing.

 

[Flyback]

4pyros I am asking exactly the same questions as you are.
So do you agree that tying the SPEEDSET pin to VREF pin of the ML4425 means that we only get one , and one only, IGBT commutation frequency?

 

[4pyros]

are we really saying that a BLDC pump cannot be made without a float switch and be able to handle the situation of "Pump suddenly running dry"?

No you can if you use the control chip as designed.
What was your "top consultant motor & drives expert" thinking?
How much did he get paid to make a pump controller that can't run dry?
Was he BSing you?

 

[4pyros]

4pyros I am asking exactly the same questions as you are.
So do you agree that tying the SPEEDSET pin to VREF pin of the ML4425 means that we only get one , and one only, IGBT commutation frequency?

Yes you only get one speed, thats probly why you have to kick start it.

 

[Flyback]

No you can if you use the control chip as designed.

..so you say if we use ML4425 as it is intended, then we can handle "pump suddenly running dry"....but I question this, can we?
..I mean, even if operated from a constant voltage DC link, and with high frequency PWM of the ML4425, then the motor needs to speed up before the controller starts reducing the duty cycle of the high frequency PWM....and we've already discussed why the pump speed wont necessarily increase if the pump suddenly runs dry...because if such a sudden load change occurs, then the pump speed is more likely to become irregular and either slower or maybe faster, -who knows.?

 

[tcmtech]

..dont want 12 to 24v as its too much current.

Then use a 24 - 36 0r a 36 - 48 volt unit. Not really hard finding those either.

I have very strong suspicions as to why the consultant left with the project unfinished and barely functional.

 

[alec_t]

no the igbt's aren't going to switch at 20khz....in our initial lab set up the ml4425 is switching them at 400hz

The only way I can see to do that is to disable the PWM oscillator by grounding pin 6 via a resistor. Is that how you're doing it?

 

[Flyback]

no pin 6 of ML4425 is CT pin and we have a 2n2 to ground on that pin.

 

[JimB]

I have very strong suspicions as to why the consultant left with the project unfinished and barely functional.

I have this awfull vision of the initial project specification, full of words like:
Energy efficient
Modern technology
Carbon neutral
Climate change
Minimal maintenance

And the project badly manged by a bunch of bean counters.

Au revoir Pump Consultant!

JimB

 

[tcmtech]

I would like to know the actual volume of the tank, the volume in gallons per minute or liters per minute required, the working pressure this system requires and how all of that relates to using a overly complicated pump design?

I would not be he least bit surprised if this turns out to be way a over engineered and way over budget way to water a basic greenhouse of flowers from a cistern tank that catches rainwater or something similar using solar panels whose energy could and should be getting used in a far more productive fashion than this.

 

[alec_t]

no pin 6 of ML4425 is CT pin and we have a 2n2 to ground on that pin.

Okay, so Ct is in place (but seems redundant if you're not using the chip's PWM oscillator).
Then tying SpeedSet (pin 8) to Vref (pin 7) presumably sets the PWM duty cycle to 100%, i.e. a fixed logic high feeds the AND gate (the datsheet is somewhat sketchy here).
In summary:
You are not using the IC's current-limiting function or speed-control-via-PWM-control function.
You are using only the IC's back-emf monitoring, commutation logic and tacho functions.
1) Is that right?
2) You don't want to use a current-sense resistor, but your converter is stated to be a constant-current source. How does it regulate current if it doesn't have a means of sensing current?

Edit:
Going back to your earlier question "So do you agree that tying the SPEEDSET pin to VREF pin of the ML4425 means that we only get one , and one only, IGBT commutation frequency?" I'd have to disagree. With the IC's speed control function over-ridden, in the absence of external speed control/limitation the motor coils will be driven hard on and the motor will accelerate until the VCO frequency (commutation frequency) maxes out at ~2kHz (=20,000rpm).

 

[Flyback]

yes youre right about the first bit.
Yes the buckboost converter senses the current with a current sense resistor...its on the same pcb as the buckboost, and we don't have to wire to it, like we would with an ML4425 sense resistor.
I thought tying SPEEDSET to VREF on ML4425 sets the commutation frequency?

 

[alec_t]

I thought tying SPEEDSET to VREF on ML4425 sets the commutation frequency?

Only if you use the internal PWM for speed control, as far as I can see from the datasheet. BTW I've edited post #75.

Edit:
As it's now apparent you have a current-sense resistor you should be able to use that, or the tacho output, to detect a pump-running-dry situation.

 

[4pyros]

As it's now apparent you have a current-sense resistor you should be able to use that,

Not when he is using a constant current source, maybe the voltage would change.
But he still has not said if he has done any testing to see if anything changes when the pump is run dry.

 

[alec_t]

Not when he is using a constant current source

The current control loop will have a finite response time, so a sudden current change would be detectable before the loop responds to correct the error.

 

[Flyback]

The current control loop will have a finite response time, so a sudden current change would be detectable before the loop responds to correct the error.

..The loop determines what the buckboost output current is, -if the motor does not speed up (above 8000rpm) when the pump suddenly runs dry, then there will be no decrease in current from the buckboost converter....this is what worries me.

So you seem to be saying that the ML4425 (when used in 'normal' mode, not like i am doing) has two speed control loops...
loop 1...to adjust the PWM duty cycle as a means of adjusting the speed
loop 2...to adjust the commutation frequency as a means of adjusting the speed.

...it seems strange to me.....I mean, if we want the pump to spin at 8000rpm (only ever that speed), then surely we would want the frequency of the commutation of the igbt's to be 400Hz (which gives 8000rpm)....and only ever this frequency.
Surely the ML4425 should have a facility to be able to fix at a certain commutation frequency?
I would have thought that's how it worked always?
Admittedly at startup, it should ramp up to this frequency instead of just going straight up to it.

Surely in a BLDC controller like ML4425, one controls the speed by putting a voltage on the SPEEDSET pin to set ('in stone') the commutation frequency, and then allow the ML4425 to adjust the duty cycle until that speed is achieved?

Having the ML4425 adjusting both commutation frequency AND duty cycle to regulate the speed sounds like a bit of a who-haa?
...I mean, like juggling too many balls, unnecessarily?