Help with Water Pump

[KeepItSimpleStupid]

@ronv
Bingo! From P = I^2(R) we get (9)*2.x ~ 18 W. I ran into this with an audio amp. Power is defined by what's limiting it, not what's available

 

[KeepItSimpleStupid]

@alec
Think about what might be required to disassemble a pump non-destructively to remove the PC board and sensor, replace it and re-encapsulate with a fish friendly encapsulant. The PC board can be destroyed in the process which is OK. We need access to what looks like 3 wires. Two attached to the corners near the FET and likely one attached to (+) power. There was some Epoxy that glued these wires to the bottom of the PCB.

That gel in an earlier post looked pretty neat as an encapsulant and would be easier to cut away. You now know how it goes together, so you know where to cut say with a Dremel. We know that MEK is an Epoxy solvent and supposedly vendors are available that can remove an encapsulant and an Epoxy solvent is available.

Increasing the reliability of the pump hedges on the ability to do the above. More on what's required for breadboarding coming soon.
Meanwhile, you have bigger fish to fry.

How easy would it be to attach a temperature sensor to the pump such as this one?: https://www.electro-tech-online.com/custompdfs/2012/06/lm34.pdf

Can a "test pump" be instrumented with two winding temperature sensors and an ambient temperature sensor?

 

[KeepItSimpleStupid]

Pump 100 Hz results

Z = 5.7 ohms 38.5 deg
L = 5.6 uH 79.5 deg
C = 447 uF 1.25 DF Surprise - a Huge change in capacitance Factor of 10
R = 4.47 @ 38.5 deg
ESR = 4.48 ohms; at 1 Khz it's 6.29 Ohms (Not measured before)
DCR = 2.15 ohms (No surprise)

Huge change in Z, approx. 4X lower than 1 kHz
Huge change in capacitance, about 10 X higher than 1 kHz
24 V/ 5.7 ohms = 4.2 Amps

 

[()blivion]

That capacitance looks *WAY* off. There is no way it's that high just from intrinsic parasitic capacitance, there has got to be a coil short or something. You could also just be getting false reading with inductance and capacitance being together, depending on how exactly your meter reads inductance and capacitance. A RC time based measurement could be thrown off by a short, or by the inductance when using AC probe voltage. There could also be some resonance effects taking place using AC.

Either way that capacitance looks just plain wrong.

 

[ronv]

I agree. Almost sounds like there is an electrolytic hidden in there somewhere.

Here is a datasheet picture of a 2 pole driver. Pretty simple. Notice the zeners across the external FETs. I didn't hear you mention any clamp or freewheel diodes. Does it have any?
If no zeners it might be a good idea to provide a path back to the power supply caps for when the slow start FET turns off. A reverse diode would probably do it.
Since you still have one good coil it might be a good idea to get a feel for the temperature rise vs. current.

 

[KeepItSimpleStupid]

This is an entirely different way of making a wave: https://www.reefcentral.com/forums/showthread.php?t=1411393

ronv

Well, since it's not a capacitor, it could just be an artifact of the measurement. Transformers are usually measured at much higher currents.

No. No clamp or freewheel diodes inside the motor. Just a 3 pin hall sensor, 2 FETS? and two 520 ohm resistors. I haven't drawn a schematic yet. That's what I'm complaining about. The hard part is is getting a system of components that will work on 24 V. See how super simple this one is? https://www.electro-tech-online.com/custompdfs/2012/06/AH287-2.pdf, but it won't work on this pump without some extra circuitry.

And it's encapsulated and pretty much destroyed when it was taken apart. There is a pic taken by sj in this thread that shows a FET with a ring around it like it got really hot and their is a solder blob that should not be where it is that connects the tab to the left lead. This is on the opposite side of the discolored coil. I also havn't removed the solder blob and tried to test the FETS. The 520 ohm resistors are OK.

The slow turn-on circuit is alec's. Not sure what he used for protection. It wouldn't be needed if it was in the motor.

Since you still have one good coil it might be a good idea to get a feel for the temperature rise vs. current.

I could do that easily with an IR thermometer and a CC power supply which I have. It might be worth a shot.

 

[()blivion]

The slow turn-on circuit is alec's.

Correction. The slow start method was my idea. He was going to use a PWM system and I was like...

"Why don't you just RC delay the FET's on time, shouldn't be a problem if time is sub 100ms and power is low"

Not the greatest way to do it but simple as dirt and should work well enough.

 

[KeepItSimpleStupid]

@ ob():
Sorry. Just thought that alex did the design that sj followed. I was just trying to call the prototyped design something so, I called it "alec's circuit". The OEM motor and the OEM controller also work for names. Do you have a good name for generation 1? Do we call it pump control G1 (Generation 1) like the various cell systems or do we use program versions such as V 1.01.01.00 where you have major and minor revisions. Suppose this is version 1 with version .01 of the sequencing, version .01 of the motor driver and version 00 of the alarm system (e.g. it doesn't exist). That's a major rule of communication. Agree what to call things. Whatchamacallit won't cut it.

Besides, I don't think this motor will survive PWM.

-----

Anyway, Tunze seems like a neat system: **broken link removed**[predid]=-infoxunter026 but they surely want to lock you into their stuff. Why can't one buy a Tunze controller, connect it to a piece of crap, realize that what it's connected to is crap and then buy a tunze pump. The instructions get lost in the translation. I like how easy the pumps come apart.

Expensive for a reason, the pumps have what's necessary, although I didn't see a physical alarm.

Their feeder mode is to turn the pumps off. Joe's sounds better.

Here is a interesting comparison of some various pumps.

This is definitely a wow!: http://zubbie.com/reef/240g/default.htm

 

[()blivion]

Besides, I don't think this motor will survive PWM.

Neither did I, That's why I was against it from the start. Though I knew from page one that they were brushless also.

Just thought that alex did the design that sj followed.

Oh don't get me wrong. It's 99.999% alec_t's design to be sure. I did nothing more than observe and make comments that he incorporated. Then I made a M$ paint layout based on his design. Then stuff happened to my life and I had to jump ship. A few months passed, and this thread still lives. So I'm back. If my skills are needed soon I'll voice my usefulness.

Do you have a good name for generation 1?

I was thinking "The salty special", as is tradition for electronic projects like this. Or something like "Aquamarine box" if you want to use old hacker code. Or we could just call it "The controller" if you want a some what darker feel to it, lol. As for version numbers (ha ha) I haven't thought of it. If we are talking strictly medicinal naming, then the G1 G2 G3 ect ect is the most clear cut way to show our progress. As a computer technician though, I never really liked version numbers. Naming projects that go stale quick with a date and time is more useful IMHO. But whateves.

┐(￣ー￣)┌

I was just trying to call the prototyped design something so, I called it "alec's circuit".

Sure, that works for me honestly. I was strictly speaking about the slow start section/idea. It really doesn't matter to me what we call the whole unit. He actually did all the work, I did nothing (I prefer sleeping than working myself). So he should get to name it if he wants. Him or salty.

 

[salty joe]

This is an entirely different way of making a wave: https://www.reefcentral.com/forums/showthread.php?t=1411393
QUOTE]

Yes, the wavebox was patented by Tunze. They do provide a little more wave per pump at the cost of an unsightly box. A very cool idea, but the box is not neccessary to create a wave. With or without a wavebox, the pump needs to be adjustably turned on and off rapidly until the tank's individual resonance is found. I'll be using a pair of pumps opposing one another. Only one pump runs at a time. Alec wrote a wave circuit to do that on post 237. I'm hoping that a fix can be found for the tide circuit and then applied to the wave circuit.

I filled a couple big coffee cans with concrete and set a stick each one. Every 6" or so I tied a string to the sticks. Then I can see how the water motion is affected by pump positions. Kinda like a wind tunnel for aquariums.

 

[alec_t]

Thanks for clarifying the BLDC motor operation KISS.

What is your take on using a protected low side switch for your driver.

I'm fine with that, but if we mimic what the OEM controller does, i.e. varies the drive voltage, then a variable v-reg chip having a shut-down input could be used as a high-side switch instead and has inbuilt protection (hopefully!). I've got a design on the go for that. It also limits current and if the limit is exceeded for a preset time (i.e. assumed rotor lock or fault) causes shut-down and beeps. After a while the shut-down signal is released and the process repeats cyclically.

What's your take on the suggested approach for a new commuting system?

If a coil rewind on the dead pump were possible then I think a new commutating system would be an interesting project in its own right. However, I don't think Joe will be happy having to gut his good pumps. Bearing in mind this home-brew controller is only his second ever electronics project it would be a big ask too.

Basically the Fan driver would have all of the recommended parts

But the commutating chip you linked to has only 400mA output capability .

Pullups would be use on the open drains (total of 4)

The existing circuitry in the pump seems to have only 2?? Does it use inversion?

 

[KeepItSimpleStupid]

I'm fine with that, but if we mimic what the OEM controller does, i.e. varies the drive voltage, then a variable v-reg chip having a shut-down input could be used as a high-side switch instead and has inbuilt protection (hopefully!). I've got a design on the go for that. It also limits current and if the limit is exceeded for a preset time (i.e. assumed rotor lock or fault) causes shut-down and beeps. After a while the shut-down signal is released and the process repeats cyclically.

So, if I get bored, I could reverse engineer the OEM controller with respect to the power supply, but I don't think it's much more than the application note? Sounds like a way to go. Earlier OEM controllers apparently lacked a fuse and used a different style inductors.

However, I don't think Joe will be happy having to gut his good pumps. Bearing in mind this home-brew controller is only his second ever electronics project it would be a big ask too.

Agreed. If he ends up with a dead pump while testing or operating, he has a cadaver.

But the commutating chip you linked to has only 400mA output capability .

Yep. No longer super simple, but simple. Greater than 12V operation is difficult with most of the fan controllers and hall sensors. One particular complex chip (PWM and alarm capable) and high parts count is seemingly hard to source.

The existing circuitry in the pump seems to have only 2?? Does it use inversion?

Haven't got there yet. The temperature rise experiment sounds like a good idea.

So far:
1) We know a little about the OEM controller
.....a) Varies voltage to about 22 V Max and not for long
.....b) Average power is less than 100%
.....c) Is 3 A current limited
.....d) Drastically different than Generation 1 non-OEM controller

2) We know a little about the motor
.....a) Failure evidence so far
............1. Solder blob
............2. One winding overheated
............3. Generation 1 non-OEM controller had the ability to trash the motor
............4. two resistors good
............5. FETS - not there yet
............6. Some Motor characteristics (particularly that 3A seems like the maximum current)
............7. Temperature rise is a good experiment
............8. Schematic - 5 parts (Not there yet)

 

[KeepItSimpleStupid]

Pump thermal analysis

Pump out of wate ris like a fish out of water. Won't live too long.

Didn't pay attention to spot size or emissivity. Spot size would be 1.5" at 12". Black areas register high temps, than grey or the bare coil. Thermometer Fluke 561 was laser aimed at the coils at about 10". Not the best choice for a scientific measurement.

Pump was 72 and ambient was 70 F (pump was in a different room)
Looking at an exposed winding with an IR thermometer in air

30s 80F
60s 89F
2:15 150F

So, I stopped the test. It heats up pretty quick in air

The water test basically suggested that the pump would be happy in the water drawing 3 Amps at about 7.1 Volts all day. The amount of water was probably 1qt to 2 qts.

AT t=0, the pump coils were 100 F from the previous test.
(I didn't turn off the current to measure the pump, so those readings are not included.)
At 10 min and then turning off the power and measuring.
The temperature was 90F, so the temperature actually went down. I didn;t measure the temp of the tap water. I would have had to use a different method. 65 F is typical for tap water. 76 i typical for tropical fish. Not sure about salt water fish.

So the pump looks like it would be happy in water with a 3 A current limit indefinately. It does not take into account the thermal resistances of the potting compound or the potting compound to case.

Don;t forget that this was at 100% duty cycle. A real pump would operate at a 50% duty cycle per coil.


Conclusions:

1) Don't run the pump more than a few minutes in air. About 2 min 30 s in air max.
2) The pump coil seemed pretty happy at 3A (7.1V) sitting in tap water for 10 min which probably means indefinitely with a large enough water sink.
3) Thermal monitoring is a way to people proof your products.
4) Future pump failures when using a 3A current limit could be more easily blamed on internal pump electronics.

If need be, I can repeat the water test paying attention to details such as. Have pump sit in water overnight so pump and water temp would be equal. Always turn off current before taking a measurement. Remove effects of taking the measurement from the data. I don't think it's necessary.

 

[ronv]

Would it be possible to adjust the current so that the temperature in air stabilized at 105C? That would give an upper level spec in air for locked rotor.

 

[KeepItSimpleStupid]

I didn't get close to that. I reported F temperatures and 150 F is about 66 C and it was still climbing.

I don't think 105 C is necessary. Let's see what everyone else has to say.

 

[()blivion]

I think I don't have to say this, but 105c is the upper limit for most electronics. 80c is realistic too. 100c is boiling water by definition.

 

[ronv]

105 C is probably the spec on the wire insulation. How are you planning on setting the spec for current limit? With luck perhaps a dry pump can also be protected.

 

[KeepItSimpleStupid]

That insulation on the windings is probably Formvar. **broken link removed** Very tough stuff. Breaks down at 261 C.

This thing is more likely an in-the-water jet. Not likely to run dry.

@Joe
What is the adjustment range of your power supply?

@Anybody
Does it make sense to try to modify the OEM controller to run at 24V or be adjustable?
That way joe can take a pump and run with it. Drawback, protection not as robust as intended.
Just looking at the idea if Joe could take a 24 3a supply or an adjustable V supply at 3A and run a pump 24/7 with the regulator being a modified OEM controller?

 

[alec_t]

Does it make sense to try to modify the OEM controller to run at 24V or be adjustable?

Disabling the voltage ramp ramp and Small/Mid/Big cycling would enable the OEM controller to put out 22.6V limited to 3A by the regulator, presumably? Or running the OEM controller from the 24VDC (non-OEM) supply and taking an output from the top of the OEM switcher output cap should do something near it?

Joe: Unlikely, but does your power supply have the luxury of a settable current limit?

 

[KeepItSimpleStupid]

The top of the switcher cap (diode) is the output. I checked that. It was easier to clip probes there. I didn't check the min voltage for the step-down process in the datasheet, but with no load 24V in gave 24 V out. It could be within a few diode drops anyway.
I checked now and it's 1.2V.