Help with Water Pump

[salty joe]

With the pump hooked directly to the PS set at 24V, it took about .4 sec. to get up to speed according to the stopwatch on my phone. I ran this with the pump dry.

 

[alec_t]

I ran this with the pump dry.

Tch, tch! Not good for the bearings.
Hmm. With the pump in air it will accelerate faster than in water. For designing current protection we'd better assume that the start-up time is ~ 1 sec in water if running from, say, 18V (which is what I reckon the long-term average for the OEM controller is). From KISS's analysis, and without knowing the power-handling characteristics of the transistors in the pump, I'd be wary of running the pump 24/7 from 24V. At 18V the pump output should be ~ 10klph (cf 15klph at OEM peak). Will that be enough?

 

[ronv]

@ronv

The LF150 PTC limiter in your post 523 looks a promising candidate. Rotor lock (11A) would trip it in ~ 2 sec according to the spec. Not clear how long it would take to reset itself?

From what I've read they reset in a few seconds if the fault is removed. There is some "leakage current" in the tripped mode (milliamps) that keep it tripped. So it would reset in between each pump cycle. Certainly seems like the simpelist solution.

Need to check the safe operating area of the slow start FET to see how long it would stand up to the high power disapation (14v X 5 amps) is a lot of watts.

 

[KeepItSimpleStupid]

@alec, @Joe

Are you considering using a switching regulator to drop from 24 V nominal? This thing will run 24/7/365, so there may be an incentive to go green. Say 6 V drop at 3A or 18 W conservatively. At $0.10 USD /KWH that's $15.78 USD per year per operating pump that isn't doing anything but wasting money. Agreed, the analysis isn't perfect, assumes 1 pump 24/7/365. From what I remember in the thread that effectively means two pumps at 24/7/365 for one of the controllers. The power supplies may be 85% efficient. e.g. the OEM switching regulator and the supply Joe bought to run these pumps. But then, the amount may not be too much in the grand scheme of things.

The reg reguires 1.2 V differential and Joe specified the max output of 24 VDC which I don't beleve for a 24 VDC supply. Joe, could you measure the min and max output again?

**broken link removed** has inductors in onesie quantities that could possibly be suitable.

Is there anything you absolutely need at my end?
e.g.
1. More thermal analysis?
2. A modification to turn the OEM controller into a variable 3A power supply for testing purposes?

It still bothers me, that the commutation doesn't make any sense.

 

[alec_t]

@ronv

Need to check the safe operating area of the slow start FET to see how long it would stand up to the high power disapation (14v X 5 amps) is a lot of watts.

Yes, we'd need to look into that.

@KISS

Are you considering using a switching regulator to drop from 24 V nominal?

Yes, I was. I've drafted a design for a voltage-controlled reg so that 18V would be the normal output but it could be programmmed to go lower for the low-speed 'feed-time' mode if required.

A modification to turn the OEM controller into a variable 3A power supply for testing purposes?

I wouldn't class that as 'absolutely need', but it would be handy for Joe to have if it's a straightforward mod and you have the time (and Joe isn't concerned about voiding the warranty!).

It still bothers me, that the commutation doesn't make any sense.

If you take the BLDC explanation that I linked to and simplify it down to a 2-pole stator and a 2-pole rotor does that make sense? The Hall sensor only has to provide an on/off output to drive one transistor base/gate.

 

[KeepItSimpleStupid]

So one can stay on indefinately? If so, does that explain the solder blob? That would work.

 

[4pyros]

If you take the BLDC explanation that I linked to and simplify it down to a 2-pole stator and a 2-pole rotor does that make sense? The Hall sensor only has to provide an on/off output to drive one transistor base/gate.

So one can stay on indefinately? If so, does that explain the solder blob? That would work.

I dont think so, one on, one off. Back and forth like a flip flop.

I looked around some last night. I fould out from the RC guys that the less poles the faster the motor will run but with less tork.

Two poles will work maybe better then 4.

Then I ran across the "Dyson Digital Motor".

Dyson has developed gone to 104,000rpm brushless DC technology to combine efficiency and manufacturability in its latest handheld vacuum cleaner.

"Because it is two pole, it is very simple, and when you make it run at high speed, you can make it incredibly small," Dyson's Andy Clothier told Electronics Weekly. "It is 84% efficient, which is high at this small size. At this voltage and power level, to get a brushed motor this efficient is very difficult. Our old motor was 40% efficient."




Overall the motor, dubbed DDM (Dyson digital motor) V2, is 55.8mm in diameter and weighs 139g.

Notoriously tricky to start predictably, the motor uses asymmetric poles. "You have to have enough saliency on the poles to make it start in the right direction," said Clothier.

With brushless motors, there is a choice: sensored or sensorless - incorporate a magnetic sensor have tell the electronics when to switch coil polarity, or use a more powerful processor and sense rotor position from back-EMF.




"The way we designed it was to integrate the electronics into the motor. It is the least expensive way of doing it," said Clothier. "The PCB is in exactly the right place to carry a Hall sensor."

Control comes from a simple 8-bit Microchip microcontroller, not one that has special motor control peripherals, said Clothier: "We used our own motor control technology. To get the absolute best, we make sure the motor produces constant power regardless of speed and the battery voltage."

The power control is largely open loop - determined from detailed knowledge of the motor and impeller dynamics, combined with motor speed derived from the Hall sensor. Up to 3,300 adjustment per second are made.

The battery is either six or four lithium ion cells depending on the vacuum cleaner model: DC31 (pictured) or DC30 respectively.

Up to 10A at 20V, and up to 13A when the battery voltage drops, is switched into the motor by an H-bridge of mosfets.

To get current to change direction fast enough with such a low supply voltage requires low-inductance windings - in this case twin coils wound in parallel.

The whole motor, and its mechanical and air environment, was modelled extensively.

"That is where most of the work went in: we developed our own simulation tools to model the whole motor including its electronics," said Clothier. "We also used some commercial finite element software for spot checks and detailed work, but 90% was designed by our own software."

Modelling, for example, showed the sintered neodymium permanent magnet rotor was small enough not need a carbon fibre sleeve to stop it flying apart at full speed.

"This is the kind of thing that looks simple and needs a lot of work," Mathew Childe told Electronics Weekly. "We modelled the motor dynamics and made sure it was stable against vibration right up through its acceleration range, checked the acoustic noise and checked the resonances."

The team also built prototypes that were tested using accelerometers and laser displacement instruments, then fed-back the results. "All the way through, you learn to improve and adapt the modelling process," said Childe.

High rotational speed put means the impeller can be small, but means it is subjected to high forces. "Most people would use aluminium," said Childe. "Through simulation we designed out as much stress as possible and so we can make the impeller out of carbon fibre-reinforced polymer."

A plastic impeller and steel shaft means welding is out of the question. "Everything in the vacuum cleaner is dependent on bonding," said Childe. "We have had an engineer working for two years on adhesives for the product."

The motor has been dubbed DDM (Dyson digital motor) V2.

What was effectively DDM V1 was actually dubbed X020 and is the switched reluctance designed used in the company's Airblade hand dryer

View attachment 64875

 

[KeepItSimpleStupid]

The Dyson unit is pretty cool.

The only thing I ever worked on that spun fast, >40,000 RPM was a Pfeiffer Turbomolecular pump. An earlier model of one of these: **broken link removed**

I don't know what was worse: replacing the bearings or repairing the 3 phase BLDC drive unit. Actually, the latter was worse.

The Helium compressor Cryogenic pumps were a lot simpler EXCEPT 5 minutes without power was a day lost defrosting and regenerating and powering up the pump. Re-generation was a process that had to occur about every 4-6 weeks. The only maintenance was a charcoal canister and a Helium filter and adding Helium once in a while. A very clean unforgiving pump. Don't get oil in it or your in for 2 days of unexpected work.

The turbo could shatter. Luckily that never happened.

Replacing a 1000 Watt light bulb was fun too. The envelope was pressurized to about 10 atmospheres or about 100 PSI. A fingerprint or installing it upside down could cause an explosion.

My only significant accident was when I dropped an ampule of Red Phosphorous that I was sealing under vacuum with a Hydrogen/Oxygen torch. Red phosphorous will burst into flames when dropped hot from 3 feet. So, I just smothered the fire. It didn't smell good for a while.

 

[salty joe]

@alec, @Joe

Are you considering using a switching regulator to drop from 24 V nominal? This thing will run 24/7/365, so there may be an incentive to go green. Say 6 V drop at 3A or 18 W conservatively. At $0.10 USD /KWH that's $15.78 USD per year per operating pump that isn't doing anything but wasting money. Agreed, the analysis isn't perfect, assumes 1 pump 24/7/365. From what I remember in the thread that effectively means two pumps at 24/7/365 for one of the controllers. The power supplies may be 85% efficient. e.g. the OEM switching regulator and the supply Joe bought to run these pumps. But then, the amount may not be too much in the grand scheme of things.

Thanks, every nickle counts. Efficiency has been a major theme for me with this entire project. If it is an easy fix, I'm all for it. BTW, the idle pair of pumps in the tide simulation "flick" or give a little spin every minute or so to keep fish from sleeping next to pump.

The reg reguires 1.2 V differential and Joe specified the max output of 24 VDC which I don't beleve for a 24 VDC supply. Joe, could you measure the min and max output again?

I have 2 PSs. I was going to run the tide with one and the wave with the other. The min/max for one is 19.1V-24.1V. The other is 20V-30V.

Tch, tch! Not good for the bearings..

I thought I might get busted for that. I figured a second or two won't hurt anything, plus wasn't every rule made to be broken?

Hmm. With the pump in air it will accelerate faster than in water. For designing current protection we'd better assume that the start-up time is ~ 1 sec in water if running from, say, 18V (which is what I reckon the long-term average for the OEM controller is). From KISS's analysis, and without knowing the power-handling characteristics of the transistors in the pump, I'd be wary of running the pump 24/7 from 24V. At 18V the pump output should be ~ 10klph (cf 15klph at OEM peak).

I set the PS at 19.1V and the pump still ramped up at about .4 sec. dry. I put it in water and at 19.1V the ramp up time was the same .4 sec. as far as I could tell with a stopwatch. I expected it to take a little longer in water.

Will that be enough?

I won't know about the flow until I see how well the water gets moved around. I think if we can get the controllers you made to handle these pumps then yes.

 

[KeepItSimpleStupid]

Thanks Joe.

Just a thought. Depends on the pump requirements, but having a wired spare would make some sense. They could be WIRE-ORED with diodes, so with one supply set a little higher than the other (Final regulation would be done later) the supplies could act as an automatic back-up if they are both on at the same time or a manual backup if they were not.

On high reliability systems there is something called an OVP or over voltage protector which "crowbars" a power supply if an overvoltage is detected. A "crowbar" puts an electronic short across the power supply, blowing the fuse.

When designing this, one could keep in mind possible alarms. Possibly some easy ones. Maybe ones to add later. For example there is the pump alarm which might require a silence switch and light/buzzer. Basically like a fire alarm panel with the alarm, the trouble light. Agreed, not important now.

Here is a company that makes some nice outdoor enclosures. https://search.l-com.com/search?keywords=enclosures They are nice. I have one at home.

 

[alec_t]

@4pyros
Interesting link. The DDM could be rather similar internally to the Resun pump.
@all
If I've interpreted Joe's pump pics correctly the pcb wiring could be this :-
View attachment 64881
and its circuit could be :-
View attachment 64882
I think that, with the motor having only 2 stator poles and only 2 rotor poles, to achieve commutation the Hall IC (custom design?) must include 2 sensor elements; one which gives a high output when close to a N pole and another which does likewise for a S pole. A virtual ground might be provided within the IC by diode-ORing the element outputs (which are pulled to true ground by R1 and R2). Although the circuit shows M1 and M2 as MOSFETS they could equally be BJTs.

Edit: The assumed commutation sequence for either stator pole is Off, N, Off, S, Off, N, Off ....... The pole geometry (saliency) would determine the rotation direction.

I have 2 PSs. I was going to run the tide with one and the wave with the other. The min/max for one is 19.1V-24.1V. The other is 20V-30V.

In that case we could probably get away with using both at their minimum setting. A further 0.6V or so would be lost by using a series power diode (for PS protection against back-emf, as KISS suggested).

 

[4pyros]

alec, I still think the driver circuit could be like the one in the link you posted in post #547.

 

[alec_t]

Maybe...... except that I can't make that driver circuit fit the pcb layout in Joe's pics. The resistors would be in the wrong place.

 

[4pyros]

Maybe...... except that I can't make that driver circuit fit the pcb layout in Joe's pics. The resistors would be in the wrong place.

Is that the real layout of the board in post #571?

 

[alec_t]

Is that the real layout of the board in post #571?

As far as I can tell from Joe's pic (post #411). I'm guessing the transistor terminals but I think the tracks are right. KISS might be able to confirm?

 

[()blivion]

I'm think you have it just about right Alec, (and nice graphics BTW). Though considering how simple and "budget driven" the design looks, I would have assumed the Hall sensor to be simpler and more main stream than what you have. Still not a bad theory. It looks exactly as the picture does and is about how me and KISS figured it would work. He was saying something about body diodes being backwards from normal or something too..... IDK, he can explain what he meant by that when he gets on.


(Edit: Changed "would assume" to read "would have assumed"... IE past tense, sorry about that.)

 

[alec_t]

Though considering how simple and "budget driven" the design looks, I would assume the Hall sensor to be simpler and more main stream than what you have.

Perhaps. Dual-element N- and S-sensitive Hall ICs do exist it seems, e.g. the Micronas HAL740. That's a 4-legged beast, and its outputs are open drain, so it doesn't tally with what I've assumed in the circuit.
KISS, can you check if the Hall IC in the pump could have had 4 legs? I see a detached wire on the right of the second of Joe's pics (post #411) which perhaps could have gone to the IC?

 

[KeepItSimpleStupid]

A further 0.6V or so would be lost by using a series power diode (for PS protection against back-emf, as KISS suggested).

Nope, not there. The diode just gets connected reverse biased to the power supply, not in series with it.

but I think the tracks are right. KISS might be able to confirm?

Tracks are right on the money.

KISS figured it would work. He was saying something about body diodes being backwards from normal or something too..... IDK, he can explain what he meant by that when he gets on.

The body diodes are backward from the way you drew them. They are fwd biased from the side lead to the tab. From alec's Source to drain. The power supply pins need to be inverted too. If you invert the power pins, you have to invert the drawn body diodes.

KISS, can you check if the Hall IC in the pump could have had 4 legs? I see a detached wire on the right of the second of Joe's pics (post #411) which perhaps could have gone to the IC?

No question, 3 pads, so 3 leads? Cutting off a lead would make no sense. The hall package doesn't have clean edges.

KISS, can you check if the Hall IC in the pump could have had 4 legs? I see a detached wire on the right of the second of Joe's pics (post #411) which perhaps could have gone to the IC?

Very clever suggested circuit for the hall device. Maybe RESUN has some friends in the IC business?

Could this device A1202 from Allegro help? : https://www.allegromicro.com/en/Pro...media/Files/Datasheets/A1202-3-Datasheet.ashx

 

[alec_t]

Thanks for confirming the pcb track layout KISS. One slight puzzle, though: if the track I've shown as V+ is indeed V+, why is it narrower than the track I've shown as V- ? They would both be carrying the same full motor current.

The diode just gets connected reverse biased to the power supply

We've already got reverse-biased Schottkys from drain to V+ and from drain to ground (V-). Not enough?

No question, 3 pads, so 3 leads?

I was just wondering if the Hall IC package had a stubby tab (as per the HAL740) opposite the 3 leads and perhaps that stray wire had become detached from it during the pump dismantling; though the photo doesn't show any obvious tab-like bit nearest the camera. For the HAL740 the tab is the ground connection.

Maybe RESUN has some friends in the IC business?

I'm told custom chips are pretty cheap from China.

The more I look at the post #411 first pic the more likely it seems that solder blob under the gate (?) leg of the right-hand FET (?) was the cause of our woes. A (partial) drain-gate short would account for all the damage (apart from that inflicted by Joe and his saw/blowtorch ) and may even have allowed this puzzling motor to run, in a hobbled way, initially.

 

[KeepItSimpleStupid]

We've already got reverse-biased Schottkys from drain to V+ and from drain to ground (V-). Not enough?

Then that's probably OK.

If switching or any regulators for motor loads are used, you also need a reversed biased diode across the whole mess. e.g. Forward biased from output to input.