Help with Water Pump

[alec_t]

@KISS
Here's a two-phase BLDC controller chip
http://www.rutronik.lt/news+M5476286d75f.html
Wonder if there's something like that in the pump or factory controller?

 

[KeepItSimpleStupid]

Your right, that is a problem. Thanks for catching it. My brain was fried yesterday due to the weather. I get weather induced migraines and yesterday was bad.

I forgot that we are using 24 V, 30 W pumps. So that won't work. We either need a 24 V 30 W load of some sort or two 12 V, 15 W loads in series. I don't think at this point we need more than one of them.

Here is a 24 V 35 W light bulb and socket: https://www.bulbtown.com/JC35_24V_35_WATT_24_VOLT_JC_HALOGEN_GY6_35_BASE_p/jc5520.htm

What's nice about light bulbs, is that there is a large inrush current and you can tell if it's working.

When I look at the bad pump, we'll know what to use to simulate a motor load. The simulated load won't be able to do back EMF where the motor acts like a generator.

 

[KeepItSimpleStupid]

@alec
I don't think there is anything like that in the factory controller. From what I read, it appears to be a switching regulator with 3 different voltages.

I worked up a internal motor control design (selected parts) in my head, but I have NO IDEA if it's possible to replace the internals of the pump.

 

[KeepItSimpleStupid]

@alec
I was thinking of a design based on the AN287. **broken link removed** It only has a 400 mA output capability We need a little over an amp RMS (30W/24V). It's pretty simple and well protected. A little tough finding, I think because of ROHS transitioning. The output capability can be solved with a dual inverter and some pull-ups (part selected) and a fully protected dual high-side switch (two possibilities here). I think there is plenty of real estate to implement it, but I have no idea if it can be done.

I bet I know what's taking out the regulator on the controllers. There needs to be a bypass diode around the regulator and a reverse biased diode at the input or output of the power supply. I bet they are missing.

We know a lot of things could have taken out the pump. The one thing you didn't have in your original design was "current limiting" and/or the pump was not fused at the pump rating. A few additional things that can be incorporated at your end is: 1) A reverse biased diode across the motor leads 2) A ~48V MOV/VS across the motor leads 3) fuse - 2 Amp SB for each motor; 4) Current limiting. With 4, 3 may be unnecessary.

@Joe/alec
You should get 4, Shotkey diodes rated at 3 Amps, one for each pump. They will be across the leads of the motor (reversed biased). You keep the other one at the power supply as well. We should pick out 4 MOV/TVS's for alec's circuit too. Again, across the motor leads. This could wait until I get the inductance of the motor.

So what's the plan?

At this point, I think joe should consider building your alternate motor driver on another board. He should also consider headers to either connect the drivers of the board he has to the new driver design and the old design, so he can easily use both.

Full testing can be done using a resistance and an inductor once I find the components to "simulate" the motor.

Then move to pump testing again.

Meanwhile...Can we replace the pump innards?

Comments?

 

[salty joe]

bet[/U] I know what's taking out the regulator on the controllers. There needs to be a bypass diode around the regulator and a reverse biased diode at the input or output of the power supply. I bet they are missing.

Full testing can be done using a resistance and an inductor once I find the components to "simulate" the motor.

Then move to pump testing again.

Meanwhile...Can we replace the pump innards?

Comments?

It was one of the MOSFETS that bit the dust, not the regulator.
IDK if this is significant, but the negative lead between the Schotkey/FET drain and the jack was scorched. The + lead was unscathed.

That sounds great-thank you.

Wow, replace the pump innards? I now know where they live and it is only under about 1/8" of epoxy. I think I could ease the epoxy off and pour new epoxy. The trick would be getting a watertight seal between the new and old epoxy. That's assuming I can swap out parts. With you guys showing me the way, seems doable. That's a fascinating and exciting idea!

 

[KeepItSimpleStupid]

What I was talking about is what takes out the original supplied controllers by the manufacturer. Those failures apparently were the regulator. I would need to verify my suspicion that certain parts are missing.

What is the p/n or datasheet of the jack you are using? We need to check the current rating unless the FET was really close to the jack and the jack melted because of the heat. In which case, the FET needs a heat sink and current limiting so it can withstand a short forever or blow a fuse. The primary of the power supply has to be fused. Conductors not enclosed in a protective case have to be current limited or fused for the conductor size.

This might fall into NEC Class II wiring rules which kinda says that 24 V systems need to be limited to 100 Watts. https://answers.yahoo.com/question/index?qid=20090101093924AASMrXF So, it does make sense to limit the power exiting your controller enclosure. Class II does apply to wiring run within the walls of a structure which probably would not apply here.

When I installed a Hydrogen/Hydride and safety system, I had to be concerned about this. The system was interfaced to the building fire alarm and the fire alarm was interfaced to my system. All wiring in the ceiling space was PTFE/Teflon insulated fire rated cable and power limited. The power supply I purchased was about 8 amps and it had taps for each power limited sections. Two versions were offered: fuses and PTC thermisters. I took the latter.

I had one glitch that I could not have anticipated. Some of the strobes from the same manufacturer would not work. Turned out that they drew more on start-up than the power limited circuit would handle. The manufacturer made changes.

One of my mottos that I like to live by is that when I fix something, I don;t want to see the same failure again unless it's a consumable. Meaning, I fix it right even if it means adding parts. Fans, for instance, are consumables but most of the time the problem is dirt. A floppy drive is a consumable because they are subject to mechanical wear and for the most part cannot be fixed. Same is true with a hard drive. But, if a lamp keeps taking out a thermocouple scanner, then that has to be fixed.

IDK if this is significant, but the negative lead between the Schotkey/FET drain and the jack was scorched. The + lead was unscathed.

This could be. That negative lead should have been returned to where power entered the board. A wire should have gone from the FET to this point of entry, Each wire would then carry 1 AMP or so. When you start using the same length of wire, parts of that wire could be carrying 4 Amps, 3 Amps, 2 Amps and 1 Amp. Furthermore, when you have low level circuits also using that same "ground", the reference point changes for different parts of the circuit. This isn't good.

This is a construction technique that we failed to talk about at all. So, I guess now is the time. I'm sorry I didn't mention it earlier.

The general idea is to have all grounds to terminate at one point and that point would be the third prong of the power cord. The problem is, that isn't practical.
Ideal, but not practical.

What is done instead is to create 3 grounds:
1) Protective - not necessary if in a double insulated case. Your PS would have a connection to protective ground.
2) Analog ground - This circuit doesn't have one. This is the cleanest ground of them all.
3) Digital Ground - This is the return of all of the digital logic gates and the grounds of those chips.
4) High Current Ground - This is the return for all of the FETS. The conductors have to be large enough to handle all of the FETS on together or in your case, two of them.
With your construction technique, this ground should go direct to where the (-) lead enters the board.

The idea is that you keep all of these grounds totally separate, then at one place (where the -power connects to the board), The power supply (-) pin even better.
But, we can also use the sense leads of the power supply too, but you probably don't need it.

With the separate approach, there are ways to insulate the grounds from one another. Insulate here means to reduce the effects of one ground on the other. The first step is to keep them separate.

I have a few final remarks that just came to me: Suppose that the MOSFET/circuit started to oscillate. That would take out both the pump and the controller. A shorted FET means nothing because it's only a switch.

* The pump has design issues. That should be obvious. It's the harder one to fix.

* The OEM controller has issues which I believe are minor.

* I have read more stuff than I have posted.

* While we are on the subject of possibly replacing the pump innards how important is speed control in the future? How might speed control be used in the future, if at all?

 

[salty joe]

Sorry....
Thanks for all your ideas and thoughts, and thanks for the grounding technique. Instead of wrapping the perimeter with bare wire, I'll leave the insulation on and have all the grounds meet at the negative PS. I have been using 22 gauge single strand for everything. Thanks a million! But the scorched, insulated mosfet/Schotkey wire ran straight into the jack that got a plug melted into it. I will look into what those jacks are rated for. Thanks again.

A speed control would be a nice feature if did not hurt reliability or add too much complexity. It can be useful to slow down the water at feeding time.

 

[KeepItSimpleStupid]

Have just each of the FET grounds meet there and one ground from the logic part of the circuit.

If speed control for feeding is the only issue, then let's handle it in a simple way.

Have alec look at his circuit and make sure that pump power and logic power are separate and build it that way. This is what I would call a "hook" in a design. It's a future requirement and it's not needed now, but the "hook" which is keeping logic and pump power separate is easy to implement now.

later, the implementation could be a toggle switch ON/OFF/(ON). This could be labeled "SLOW PUMP" and have values of ALWAYS, NEVER and TIMED. The (ON) means momentary ON.

To feed the fish, you would bump the switch (momentary) to timed. The system would slow down for whatever feeding time was set and then resume to full power. We could use something like this: https://www.linear.com/product/LT1083 or something else. This would post regulate the power supply you purchased for the feeding time. A better choice would be a switching regulator. A relay would be used to insert that into the pump power lead for the duration of feeding.

Now you know what a hook is. Three possibilities present itself immediately: 1) substitute a post regulator (switching or linear) for the feeding time and 2) PWM (Pulse width modulation) at the pump.

If this was a 12V pump, I might try to go with PWM. With it being 24V, there are more complexities.

Again, there are some bell and whistle choices: Some options are: 1) Stall detection 2) A frequency signal proportional to speed, 3) PWM control of speed and 4) Voltage control of speed.

Slowing the speed for feeding sounds like a good option and it's relatively easy to implement.

Alec has some alarm ideas which is not as good as stall signal. Checking to see if the current is in range works.

@alec: You may want to see if it's easy to add an alarm disable just for the duration of feeding. It's a way to approach the problem. It's not foolproof, but it's better than nothing. Alec then has to provide an alarm disable input that can be used later. He may have that, or parts of that anyway.

 

[alec_t]

Plenty there to chew on, KISS.

What is the p/n or datasheet of the jack you are using? We need to check the current rating unless the FET was really close to the jack and the jack melted because of the heat.

It occurred to me yesterday that the power jack is one suspect we need to eliminate from our enquiries. Looking at catalogues it seems the majority of those jacks are rated 1A at 12V. 2A jacks exist, but I haven't seen one rated above 18V. If the jack was under current/voltage stress it could get hot, which might lead to poor contact. That in turn could mean the pump being intermittently powered, hence in a trying-to-turn-on state for a prolonged period. That would certainly raise the current to a point which could damage something, as well as heating the jack further and escalating things. Do you know the jack spec, Joe?

Edit: Just an aside. Why do jacks have such a low voltage rating? I would have expected even the thinnest bit of Chinese plastic to withstand > 100V.

 

[alec_t]

I think PWM will be the way forward to provide most/all the features on the wish-list. It should be relatively straightforward (famous last words!) to use it for slow start/stop, speed, and current-limit control. Thinking cap on.

 

[KeepItSimpleStupid]

Packages arrived. I guessed and opened the motor one. I just poked around and looked a little bit.

Initial impression:
The 3 lead devices have a tab with the middle lead cut off - Assume a FET for now
I suppose that the motor shaft is at the lowest point in the tank or (FET leads down).
The solder blob (Left FET - leads down) looks suspicious and if the leads were down, solder could possibly accumulate there if it got hot.
The LEFT - FET left lead is shorted by a solder blob to the tab.
Could be a manufacturing defect.

The resistors measure about 520 Ohms. The sensor is probably good.

I'm having a hard time locating the motor leads. I did find two that measure 4.2 ohms. That could be 2x the winding resistance or the winding resistance or something else. The windings ends are also embedded into epoxy under the circuit board. I doubt it's possible to destroy the windings.

I don't think there is a way to make this motor live again.

 

[KeepItSimpleStupid]

Yep, if it's a coaxial power jack/plug 5A is rare, but they do exist https://www.newark.com/switchcraft/760k/connector-dc-power-plug-5a/dp/96F9610 I didn't open the other package yet.

1A, 2A and 5 A ratings are common. I bought a couple of locking coaxial power connectors for a project.

5A ratings exist for Laptop computers. When you use an inferior plug, the jack usually melts. Been there. Done that.

It could be a definite weak point.

 

[salty joe]

This is the jack I used. https://www.ebay.com/itm/10-x-DC-Po...995?pt=LH_DefaultDomain_0&hash=item41615f0373

I sent an inquiry to seller as there are no specs listed. Also asked for specs on his other jacks. I would never have guessed the jacks could be a problem. Hope to have an answer in a day or two.

Yeah, that pump is plenty dead. Sure hope you can make it yield some info. Thanks guys.

 

[KeepItSimpleStupid]

OK, I did an inspection tear down of the RESUN 15000 Wavemaker OEM controller.

My findings:

The external power pack is nothing but a transformer: 220-240 VAC 50/60 Hz 1500 mA; No taps; With a 120 V cord.

Major internals of the OEM controller are:

(D11) KBU 808 Bridge rectifier 8 A at 800 V ; Yea 800 V, a little ridiculous, someone had to have had a sale.
(IC5) LM2576T-ADJ (Adustible switching regulator, 3A step down (7-40 V input) Datasheet: https://www.electro-tech-online.com/custompdfs/2012/06/lm2576-1.pdf
Fuse - Yeah
(IC1) 7805 - 5V regulator
NE555 - 555 timer, probably because the timing is out of the range of the 4528 (SMALL)
HEF4528 (MIDDLE)
HEF4528 (BIG)

Small, middle and big are labels for the potentiometers on the controller case.

The coaxial output connector is labeled (DC-OUT)

Off the bat, I see three things that should be changed.

* 2 diodes need to be added to protect the regulator from Back EMF of the motor.
* The capacitor on the switching regulator filter is an SHK 100uf @ 25 VDC. It's a 105 C rated cap, but I have my doubts that it;s high ripple rated and it needs to be. A 25 VDC rating is too small.

So with a 24 VAC input, the DC would be about 33.6 VDC which is too close to the 40 Volt limit. This means that there is no transient protection so either a MOV is needed and/or use a LM2576HV

This basically means the controller would have a habit of blowing up. Precisely the LM2576 will fail and that's what has been reported. It's nicely layed out and put together.

I doubt there is any PWMMing going on in the controller. No electrical measurements yet. Joe forgot to enclose the instructions. It could be powered up using DC.

I do think the motor has more issues than the controller. The OEM controller can easily be upgraded to be more reliable, but so what.

PWMing the OEM motor is probably more detrimental than operating it at constant voltages.
The Zetex ZXBM series of parts is not a bad controller chip. For a motor with PWM input, a Locked rotor output and a 24 VDC supply, there are a lot of parts.
Unfortunately having a Locked rotor flag seems to be important for a reef tank.

In any event, I'm convinced that:

1) The OEM controller is unreliable. The Switching regulator is prone to failures
2) The motor is unreliable because NONE of the practices suggested by FAN/motor controllers are being followed.
3) The 3 AMP current limiting of the OEM controller is beneficial.
4) No current limiting may be the biggest reason for the premature failure of the built controller.
5) Voltage Spikes are probably the biggest killer of the motor.
6) The motor windings are not likely to fail. Large conductors and plenty of cooling.
7) I don't like the choice of the coaxial power connectors. A 5 Amp locking coaxial connector would be better in my opinion.

This isn't very promising for a REEF tank that needs to operate 24/7/365

Joe didn't provide instructions or the other pieces of the puzzle such as the motor armature etc. so I can't make any comments about those parts.

What can be tried for the built controller assuming no motor changes possible:

1) Reversed biased Diode across 8A supply
2) Reversed biased Diode across the wires heading toward the motor.
3) MOV across wires leading to motor.
4) Slow start - that was a good thing
5) Some sort of an alarm (Motor currents out of range)
6) Find a better power connector (5 A locking coaxial minimum) Check temperature at connector wires.
7) Use a fully protected low side or high side driver
8) Current limiting for each motor is probably more effective than a fuse
9) Don't try to PWM the OEM motor.

Most reliable long term, possibly impossible, solution:

1) Redesign the internal motor controller
............Two basic designs possible
.............. a) Use a simple chip, reduced parts count, no error flag, change voltage to motor to
...................change speed.
................b) Use a more complex part (ZXBM series)
......................1) Use a PWM signal (not power) to change speed (suggested)
......................2) Use a control voltage to change speed (possible, but not both)
......................3) Higher parts count, not enough real estate
......................4) Bring out a locked rotor signal (alarm)
......................5) Bring out the FG (Frequency Generator) signal (future)
..............................a) it allows a way to access the health of the motor.
......................6) Stator Temperature monitoring
..............................b) Benefits questionable

Providing one could get a pump motor apart without damaging it or buy one unassembled. Both may be impossible. Putting together a simple better controller seems easily doable.
It might be too hard to use the ZXBM parts because of not enough real estate.

Enough for today.

 

[salty joe]

Sorry, did not think the armeture would be helpful. I held onto it for a spare. Do you want it?
The only instructions was the single folded page. Pretty sure I threw it in the box.
Well, that's a bummer about the pump being unreliable.
I'm up for risking a pump if you are willing to tell me what parts to install.
Quite the analysis. Thank you.

 

[KeepItSimpleStupid]

Sorry, did not think the armeture would be helpful. I held onto it for a spare. Do you want it?

Not sure. I just can't comment on the overall construction.

The only instructions was the single folded page. Pretty sure I threw it in the box.

There was a folded sheet with your note enclosed with the "guts". I don't think the instructions are useful either, but if there are a few sentences about how the OEM controller is supposed to work, I'm all ears mainly for curiosity.

Well, that's a bummer about the pump being unreliable.

Yep. Just an educated feeling. I still have some stuff to look at.

One side of the motor looks burnt and black, but I have a feeling that you did that because of how it looks.

I'm up for risking a pump if you are willing to tell me what parts to install.

I sent an email to RESUN. Have to wait to see what happens, if anything. You have the suggestions of what to do without modifying the pump. What to install and where to install on the OEM pump still needs some more analysis. We don't know what parts were actually used so I don't have 100% certainty, but I would bet that I am right.

We can do our best with external changes and test and wait for a pump to die.

I'd almost would like for you to think about how to remove the circuit board such that the wires to the motor are accessible and a new PCB could be encapsulated in it's place. Waiting for a motor to die might be preferable. The original circuit board would not be used at all.

I looked on the net briefly and found these links:
http://www.esslinger.com/attack.aspx
and this website in general
**broken link removed**

We can't, however, call it plug and play.

Until we can get a pump that is hackable out of the water, we can't try anything. I don't think it's worth adding a couple of parts and hoping, but it is worth replacing the PCB with something that potentially will work reliably. I don;t think it's possible to kill the coils easily.

What I would initially envision is the AN287 IC attached via wires and couple of solid wires might be used to mount it. They would go to the leads, but wouldn't really go anywhere. The leads would just be used for support and would be attached when the orientation is known.

Babbling: The entire replacement circuit would be breadboarded, but it won't likely fit. Connectors might be used to connect to the windings. One of the tests would be to be able to turn the armature by hand with LEDS and resistors attached where the motor coils would go.

So, the parallel task would be to breadboard a circuit with the Hall sensor on leads, with LEDS for the coil, but have the ability to connect a coil. The armature could be turned in a makeshift holder to make sure the LED's turn on.

Once a host becomes available, then the circuit could be tested with a real motor.

Then commit it to a PCB, install and test out of the water and then encapsulate and test in the water.

Quite the analysis. Thank you.

I'm not done yet.

Goals:
1) Somehow we have to obtain an unencapsulated pump. That would be ideal.
2) Somehow the pump has to be able to disassembled non-destructively, modified and reassembled. This can be a dead pump.

If (1) or (2) is attainable then we could work on breadboarding a suitable controller.

 

[KeepItSimpleStupid]

OEM controller

It's perfectly happy operating on 24 VDC, polarity insensitive. 40 VDC is the absolute maximum input. It comes with a 24 VAC powerpac.

What it does:
It has 3 potentiometers with times of (SMALL) 10-47 s, 10-15 (MIDDLE) sec and 4-12 (BIG) seconds.

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 didn't measure it carefully)
BIG goes directly to 22.6 volts

The controller just cycles the DC voltages which can ramp through SMALL, MEDIUM and BIG and repeats.

The fuse seems to be in series with the input power. I did not physically verify this.

So I get the idea on how it makes a wave.

IMHO it needs 1 capacitor changed, 2 diodes and a MOV/TVS/ZNR added. I have not figured out where the other diode and MOV should go on the PCB or selected the diodes, MOV and Capacitor. All of these parts affect the reliability of the switching regulator in one way or another.

 

[KeepItSimpleStupid]

Interesting link: https://forums.adafruit.com/viewtopic.php?f=42&t=14828

Both potting and removal. Magic gel looks cool.

 

[()blivion]

A lot of comments since I have been out... A lot of reading.

I would forget any thoughts of getting the epoxy off of those pumps easily. You would need a very aggressive solvent just to be able to soften it enough to remove with reasonably low force. And any solvent that strong would quickly start attacking the rest of the pump plastic. Probably more aggressively than it would the epoxy honestly. Plus there are a lot of toxicological questions that would need to be raised when mixing mystery polymers with aggressive solvents. Also true when replacing the epoxy. Your fish would likely never survive the chemical warfare.

On another note, I see a bit of talk about maybe taking out our simple controller and instead making a centralized "ultimate reef system"? If your thinking about doing full on computer control type stuff I should take this time to let you know that I have microcontrollers and development systems and experience at my disposal. I could be willing to help with that stuff if it's something you really want and need salty.

If alec_t and KISS will do the PCB, power, analog, and safety design stuff. I *COULD* be convinced to design you a custom programed computer control solution (embedded system) that can do anything you can imagine and more. Like I always say, the only practical difference from a microwave oven's computer timer and a ICBM's guidance chip is the program written to it. So the sky's the limit if you go with this option.

It won't cost you $300 either. I'd do it for nothing minus parts. But no promises it will get done. I'm pretty lazy.




If your debating on whether or not you should go down this route, it boils down to this....

1: Do you want "something that just works" ?
If so, stick with your current design with some tweaks. And maybe even cut back on some bells and whistles if you can. This project is neat and all, and you people are great to hang out with. But let's face it... it's taken several months to develop a water pump controller. Not to compare cocks, but I could have had this done in a week by myself. I'm not complaining, just pointing out the practicality of the situation. If this were a job dead line we would all be fired.

2: Do you want a system that works, *AND* has all the head room in the world to be anything you could ever want?
If this looks better than option 1 to you, then it's time to toss out alec_t's (great) circuit design and play with the big boys in the computer controlled world. You can do anything with a good solid system that was designed from the start to be robust and flexible. Even with a sub 50Mhz MCU (less than 7$) you can do amazing things.

You can start listing features you want and we can start making a block diagram and do other things if you think this is the way you want to go?

Now then... back to sending long strings of A's to stability questionable software.
-Ob

 

[KeepItSimpleStupid]

Alec:

Your awful quiet. What do you make of all of this so far? I'm still not done with looking at the pump guts and I'm having lots of trouble locating the winding ends. A winding might actually be trashed. I will try to draw a schematic of the motor controller. One thing concerns me at this time with the motor: 1) No protection except for possibly a body diode. PWM is bad because of this. I also want to remove the solder blob and check the devices with a transistor checker (Sencore Super Cricket). I think the 3A "current limit" of the OEM controller is saving the pump. A ZNR/TRANSORB/WHATEVER could help the motor. The extra diodes, more-so to protect the power supply.

ob()
My sentiments exactly. Easily done with a micro and you think like me. Decide what you might want and go from there. Onesies are always a problem. So, far the main issue is the reliability of the pumps and the drivers. Until that's solved. computer control, SmartRelay control is out of the picture.

sj
What, if anything, do we do with the knowledge gained by me looking at the controller? Is it useful to anybody? Reverse engineering of the control circuit should not be that hard. IF done, could I offer repair and an upgrade service? This is entirely back burner. We learned a little on how it operates, that it provides DC to the motor, it has verified issues with the switching regulator and for some reason (240/24 VAC xformer) the ones you bought won't work on 120V yet they are advertised to work on 120.