Help with Water Pump

[alec_t]

I'm still here KISS, but have been kept away from the PC by family commitments. Very impressive analysis of the pump and OEM controller. Well done.

Reverse engineering of the control circuit should not be that hard

If so, then a schematic of the controller would be very helpful to keep us focussed on where we're starting from and where we could progress.

SMALL ramps down the voltage from 22.6 to 17.1 VDC over probably 10-30 s (I didn't measure it carefully). MIDDLE ramps the voltage from 17.1 to 14.9 over probably 10-30 sec

I'm surprised the voltage change is as small as that. Presumably the pump throughput drops off quite dramatically with supply voltage.
Those figures suggest the long-term average voltage is ~18V .... a far cry from 24V continuous and implying only ~50% of the power dissipation the pump experienced in the trial run. So not surprising it got hot?

yet they are advertised to work on 120.

Where did you spot that? Couldn't find that feature on the ad I linked to.

 

[KeepItSimpleStupid]

This ad the mfr's website mentions 120 VAC. **broken link removed**

The controller probably isn't much more than changing the output of the switching regulator. The ramping has to do with making waves. True, it doesn't seem to stay at 22 V for long. 12 s max.

Maybe your right, the pumps don't like to run continuously? Some did? The water is probably close to an infinite heat sink.

 

[KeepItSimpleStupid]

alec:

I was thinking of a 2 phase motor driver built from: https://www.electro-tech-online.com/custompdfs/2012/06/AH287-1.pdf fan controller, This dual inverter https://www.onsemi.com/PowerSolutions/product.do?id=NL27WZ06DTT1G, and either **broken link removed** or **broken link removed** as the low side smart switch.

And this 5V regulator LT1129-5 https://www.electro-tech-online.com/custompdfs/2012/06/112935ff.pdf

Real estate wise, the Fan controller has an integrated hall switch, so it's 3 small parts on the PCB, and the sensor off the PCB and the associated "glue". For SJ's sake, "glue logic" is a term used for various gates etc that allow a design to work. In this case the "glue" is pull-up resistors and the recommended other design parts. I know alec and myself could create a design based on these parts.

I think the design would be straightforward and simple and, of course, somewhat self protecting.

The breadboard would consist of the FAN controller on wires and LEDs and resistors for the coils. The motor shaft would be placed in some sort of holder. The shaft would be rotated by hand and the LED's would light.

Then it could be tried in a "good shell", a PCB made and then for final testing.

The hard part is to figure out how to orient the hall sensor and coils drives so it doesn't run backwards. The hall sensor is "off axis"

 

[salty joe]

()blivion, that is a very generous offer. But if we can make the tide controller and the wave controller run these pumps, that would be great. Plus, starting over would be another learning curve for me to deal with, I enjoy learning new things a lot but really want to get the water moving yesterday. Just kidding. All the delays have been my fault, yes it is a very good thing I'm not trying to earn a living at this. The patience and goodwill you guys exhibit not to mention the awesome talent and knowledge-it's incredible! Thank you.

KISS, that controller does not do what I need. I knew there were issues with that controller, but did not care becuase I thought the pumps were solid. I figured you wanted it to see what Resun did to run the pumps. You're talking about surgery on the pumps, right? I'm in.

The guy I bought the jacks from contacted me. He does not know the specs, so he asked the manufacturer. He said it might take a couple days to hear from them.

Thanks all.

 

[ronv]

Are you sure the failure was not just a 1 of a kind? Were the others still running?

 

[()blivion]

@KISS

ob()

lol, I like it... Have I become a function now? Or was that just a coincidental typo?

@Salty

()blivion, that is a very generous offer. But if we can make the tide controller and the wave controller run these pumps, that would be great.

That's about what I expected. And is most likely the best course of action. I too think this should have been done ages ago. Not that I'm complaining. Just know that I am here if you decide that this project needs to go computerized to give you all the features you want.

@ronv

Are you sure the failure was not just a 1 of a kind? Were the others still running?

I'm betting it was just a bad unit too. I don't see something getting hot enough to melt solder while under water *AND* at the same time that solder cutting a connection through the epoxy to the other lead. All in less than a few hours. We would have had to be pumping massive watts into it. It's more likely that they did a poor soldering job, filled it in with epoxy, then tested it, said it was good, then shipped it to salty

....That or gremlins.

 

[KeepItSimpleStupid]

@Ob() I guess so. I just associated () with a function and that's what you get. As long as your not defined as:

void ob()

you should be OK.

 

[KeepItSimpleStupid]

Pump failure analysis

The pump controller PCB is (1.18")30.36 mm x (1.58") 40.2 mm
Sensor is mounted close to the edge along the long edge and centered
With the FET leads down (stator at the bottom) the right side shows discoloration.

The right coil DCR is 0.8 ohms, so it's shorted
The Left hand coil DCR is 2.152 ohms

Values were taken at 1 KHz

Z = 22.39 ohms at 73 deg (L in series with R)
3.41 mh @ Q = 3.4
7.42 uf D=0.293 (surprising)
R = 6.29 ohms (R in series with L)

It's wound with 66.5 feet of 25 AWG wire
25 AWG wire is limited to 2.7 Amps and is 32.37 ohms/1000 ft

At t=0+, the DCR is dominant and therefore it can draw 24/2.152 or 11 Amps
That will kill the 2 Amp connector

It also says that when the system is behaving as an inductor, the system can draw 24/6.29 or 3.81 Amps.

The wire is only good for 2.7 Amps or about 30% overload

Conclusion:
The system was capable of supplying 8 amps and therefore at t=0+, the system tried to draw that amount. Whatever was drawn by the other motors takes away from what is available from the power supply. This is actually a peak current and not an RMS current.

This high current took out the likely 2 Amp rated Jack.

A 24 V 1.5 Amp power supply will further limit the peak currents. The voltage would drop when the system drew more than 30 W RMS.

The homemade controller could supply >3 amps and that exceeded the wire wating of 2.7 Amps by 30%. At steady state, the coils could have been drawing 3.8 Amps.

I was surprised by the 7.5 uf of capacitance. That may dampen the spikes created in the motor.

Recommendation

You MUST limit the peak current to the windings to 3 Amps or Limit the voltage available to the motor to 18 VDC.

The diodes and MOV's/ZNR's should be added at the motor and a diode at the power supply, but not from this analysis.

The connector must have a rating of at least 3A. 5A ratings are available for DC coaxial barrel connectors.

Temperature monitoring could be useful. Not sure it's possible.

You now have something tangible to run with. Having the OEM controller was useful.

Oh, and the ~0.8 ohm short will take out the FET especially without a heatsink.


I still don't like the design of the motor controller. That's next.

 

[alec_t]

Very thorough again, KISS.

At t=0+, the DCR is dominant and therefore it can draw 24/2.152 or 11 Amps

when the system is behaving as an inductor, the system can draw 24/6.29 or 3.81 Amps

Tricky to account for the generator effect of the pump motor. Clearly it reduces the 'average' current to < 1.3A or the 30W tranny wouldn't cope with the OEM controller.
Am working on a current-limit and rotor-lock shutdown design. Given that Joe has invested in a good 24V power supply do you think limiting the current, rather than reducing that supply to 18V, will do the trick?

 

[salty joe]

Very thorough again, KISS.

+1

Given that Joe has invested in a good 24V power supply do you think limiting the current, rather than reducing that supply to 18V, will do the trick?

My power supply has adjustable V output.

 

[alec_t]

@Joe
That's good to know. To be on the safe side perhaps the pumps would be better run from a lower voltage than 24V. However, that would reduce the flow rate of course. Going by Resun's figures for pump powers of 6.5W, 12.5W and 25W (it's not stated whether those are peak or average values), for the Small, Middle and Big waves, the flow rates are respectively 2klph, 8klph and 15klph.
By my calculations the greatest average (long-term) power consumption using the OEM controller and based on Resun's figures is ~15W and occurs with Big at 10 sec, Small at 10 sec and Middle at 5 sec. So, assuming the pump can be run reliably and continuously at 15W perhaps we should be aiming at that value with our home-brew controller? 15W power continuously corresponds to ~10klph flow-rate (by interpolation of Resun's data). Would that be enough for your reef-tank?

@KISS
Re your analysis.
(a) You give the winding resistance as 2.15Ω and also 6.29Ω
(b) Is that 7.4μ the intrinsic winding capacitance or an external component?
(c) I probably should know, but can you remind me what 'D' signifies?
(d) Should the voltage ranges for Small and Middle be swapped?

 

[KeepItSimpleStupid]

(a) 2.15 is the DC resistance of the wire. A special mode on my LCR meter. It would be what you would get with an ohmmeter. 6.29 has a circuit configuration. An R in series with an L circuit element, I believe or an effective AC resistance. e.g. A resistor R in series with an ideal inductor having a resistance of 0. The meter is an Agilent U1733C Circuit mode was auto, so it chose resistor and element (L or C) in series.
(b) Intrinsic and that was surprising. I didn't measure ESR. I can, though.
(c) Dissipation Factor, usually called DF: https://en.wikipedia.org/wiki/Dissipation_factor
(d) No. The very first time, you turn on the OEM controller it does something different for SMALL. A glitch. I believe it stays at the lower voltage that it ramps down to.

Note that Z = 22.39, so A = 24/22.39 or a little over an amp.

@Joe
He needs to check the range of the adjustment. It may be +-5% or +-10%

 

[ronv]

My 2 cents (probably worth a Penny)
The R is consitant with the wire length. I don't know the RPM of the motor but lets say 1800 rpm or 60Hz on the coils would put the total at 3.43 ohms. If you buy the 15 watt number that would put the back emf at 17 volts. Kinda sounds reasonable. 2 amps.
But catch a guppy in the pump and it's toast. I think I would just fuse it as the start up current will be really high and may be hard to filter out.

 

[KeepItSimpleStupid]

@ronv
This thing is a 2 coil Brushless DC motor. Who knows what it's RPM is.

 

[ronv]

But we know at 0 RPM the current is very high and that there will be a significant back EMF when it is running. The inductance will also change when the armature is present. Does the controller have any current limit or just the fuse? What size fuse?

 

[alec_t]

Thanks for the explanations and link, KISS.

Note that Z = 22.39, so A = 24/22.39 or a little over an amp.

But isn't that Z the impedance at 1kHz? I can see that enabled the inductance to be calculated, hence useful to assess inrush current rise, but not sure how it helps to determine steady-state current (which obviously is important for prolonged running). How does the back-emf from the motor when spinning get calculated/estimated?
The more I look into these 2-phase BLDC motors/controllers the more puzzled I get. For example, I can see that rotor lock is likely if the rotor magnet poles happen to be aligned with the field winding poles at start-up, but I don't see how the lock can be overcome without human intervention (i.e. give the motor a shove). And with a symmetrical winding/magnet arrangement how is the correct direction of rotation ensured at start-up? Is a trial and error approach used? With 3-phase presumably rotor lock isn't a problem.

 

[KeepItSimpleStupid]

@alec
I think I see your point. Labeling of rise(middle), break(big) and tail(small) would make more sense.

 

[ronv]

Ahh. 3 amp switcher so it probably protects itself to under 3 amps which would keep the locked roter to under 18 watts (6 volts). Whereas the one now is only limited by the supply and its fuse.

 

[ronv]

The more I look into these 2-phase BLDC motors/controllers the more puzzled I get. For example, I can see that rotor lock is likely if the rotor magnet poles happen to be aligned with the field winding poles at start-up, but I don't see how the lock can be overcome without human intervention (i.e. give the motor a shove). And with a symmetrical winding/magnet arrangement how is the correct direction of rotation ensured at start-up? Is a trial and error approach used? With 3-phase presumably rotor lock isn't a problem.

I think there are some non functional poles added to the rotor that are offset from zero.

 

[KeepItSimpleStupid]

But isn't that Z the impedance at 1kHz?

It's the frequency I picked. We don't know the commutation frequency. I could figure that out with a good motor and a scope. Could probably find the RPM, too just looking at the ripple current and knowing the geometry of the armature. Both I don't have. Using a frequency close to the commutation frequency would be better. I didn't look at the frequency dependence of the measurements. I only have a few frequencies available. Maybe 1 Khz was too high.

An inductor datasheet (later) says measurements were made at 1 Hz.

I can see that enabled the inductance to be calculated, hence useful to assess inrush current rise, but not sure how it helps to determine steady-state current (which obviously is important for prolonged running).

There is some guessing going on here. Z really determines the steady state current, but we are not energizing this thing with a sine wave, so I do think there are more parameters to worry about.

A shorted driver and a power source capable of delivering high currents will take out the winding.

How does the back-emf from the motor when spinning get calculated/estimated?

I know there are ways of measuring it and that for a DC motor V = Vm - IR. The IR drop there is the drop from the DC resistance of the windings. We don't want the Back EMF getting to the power supply. It takes out power supplies. Don't ask me how I know. So diode have to prevent it from entering the supply and make the back EMF jump over the regulator. This is missing in the OEM controller.

When the FET lets go of the coil there will be a spike. Can this take out the commutating FET? Probably. I'm not sure we can solve the who came first, the chicken or the egg UNLESS we know one side is bulletproof.

The best we can do is limit the current to 3A, prevent Back EMF from getting to your driver and placing transient protection on the motor output and reevaluating operating conditions is the only thing we can do from the driver end.

Say the commutation FET failed from a spike: with a fuse there would be no stator damage. Temperature right now could affect the integrity of the commutation FET or the winding temperature. If that FET could shut down from overtemperature, you could isolate the FET failing from a spike vs overtemperature and the winding failing from overheating.

The more I look into these 2-phase BLDC motors/controllers the more puzzled I get. For example, I can see that rotor lock is likely if the rotor magnet poles happen to be aligned with the field winding poles at start-up, but I don't see how the lock can be overcome without human intervention (i.e. give the motor a shove).

No matter how hard you try, nothing is identical. There will be some difference and therefore some preferential way to rotate. Now as friction increases from wear or something gets caught into the impeller we might end up with a "rotor lock" condition. Orientation (additude) will be in our favor too. Rotor lock comes from a failure of the commutating FETS too.

Attempting to "restart" energize, wait, energize might free it.

And with a symmetrical winding/magnet arrangement how is the correct direction of rotation ensured at start-up? Is a trial and error approach used?

Addresssed above. Symmetrical, but not identical. Angle the pump slightly and I think all bets are off. e.g. it's less likely that rotor lock would occur on start-up assuming a good motor.

With 3-phase presumably rotor lock isn't a problem.

Bearing failure and driver failure. But there are probably no issues starting in the right direction.


I did read that with one of the hall effect switches, there is an undetermined state that occurs 10 uS after power is applied. This driver could possibly use that kind of switch. I'm thinking ahead here. I don't yet have data, nor have I presented any to back this up. Could a 10 uS glitch cause the motor to run backwards on startup? I don't know? I think it's too short.

I'm still not convinced as to what the issue or failure mechanism is because of possible issues with the commutation of the motor, stressing it (running it at 100% (24/7/365) and not running with a 3A current limit. I do believe that limiting the current to 3A would likely prevent winding damage, but not necessarily failure of the motor. I lost a post that I was going to post earlier. It's now divergent at this time divergent from the current focus.

Here is an inductor datasheet: https://www.electro-tech-online.com/custompdfs/2012/06/AIRD03.pdf Look at the 3.3 uH data. Note the DC resistance is really small and that inductance changes with temperature. so, it might be possible to model the OEM motor with the 3.3 uH inductor, a 2 ohm resistor (correct wattage) and a two 15 uf caps in series https://www.digikey.com/product-detail/en/EKZE630ELL150ME11D/565-1718-ND/756234 15||15 = 7.5 uf across the whole mess. Unfortunately, the wire gauge isn't right. I guess I'm not sure what the value is, but I'm just throwing it out as data.

We have to have to have a 3A current limit to mimic the OEM controller and I think the measurements indicate why it's needed. A lower voltage reduces temperature and spike generation amplitude and increases reliability.

A nice graph of commutating FET temp, winding temp, ambient temp (water) as a function of supply voltage with a 3A current limit would be really useful.

Again, I'm suspicious of the commutation circuitry. I don't have the case, armature or working motor which can add some more data.

1) If the rotor locks and the power supply is not restricted to 2.7-3A, there will be winding damage. 100% confidence.
2) Why did the rotor lock? We really can't say, but it could be thermal (commutation driver or winding) or electrical or mechanical. 100% confidence it's one of these.
3) Measurements confirm operating current and steady state current allowed in the windings.
95% confidence.
4) Measurements suggest high turn-on winding current. Current vs. Time on a good motor might reveal something. 100% confidence
5) Armature geometry combined with scope info could reveal RPM. 100% confidence
6) No temperature data and it might be difficult to obtain.
7) OEM controller reliability issues positively identified and easily fixed. It only really affects the MTBF and cost.

I do think it makes sense to separate the driver breadboard from the "controller". It may also make sense to separate the jacks from the driver. Another plug. For now, those 2A (likely) rated jacks are fine. When there is a fault, they won't be.

What is your take on using a protected low side switch for your driver. They don't appear to be that expensive, but the major issue is non SMT availability. I might have a 5A non SMT driver around. I'm not sure of the p/n.

What's your take on the suggested approach for a new commuting system? Do you undersatnd how the parts would go together? Basically the Fan driver would have all of the recommended parts except the transient protection on the internal FETS would be moved. A 5 V regulator would be added to supply the inverter. Pullups would be use on the open drains (total of 4). Now you have a logic signal to drive the protected switch. Parts count isn't excessive. Hooked up right, reverse polarity protection could be afforded too.

It's missing a few bells and whistles, but it's simple and much more reliable.