Help with Water Pump

[alec_t]

D1 looks to be hooked up correctly.

Good. Might be worth re-soldering its connections, just in case, then re-checking those readings but this time with the pump and 20V supply disconnected.

it steadily increased to 1333

That was due to the 10uF cap charging up.

Could you 'borrow' the big 470uF cap from the output of one of the OEM controllers and attach it temporarily (observing correct polarity) between the Pump+ and Pump- terminals? That would provide a low impedance path for spikes which must presently be going through the varistor (though simulation shows it would also introduce some 'ringing' on the coil current waveform at pump start-up).

 

[ronv]

Joe,

Do you know the actual part number on the varistor. I'm wondering if it is unidirectional and not bidirectional. One has a C suffix and the other an A suffix.

 

[4pyros]

Joe,

Do you know the actual part number on the varistor. I'm wondering if it is unidirectional and not bidirectional. One has a C suffix and the other an A suffix.

Thats a good question.

 

[salty joe]

Here’s the varistor. Manufacturer Part Number 1.5KE33CA-T
l switched DMM probes back and forth on a new varistor using both diode and resistance settings, but the DMM only showed 1, as if it read nothing.

With that 470uf cap across the pump leads, the varistor stayed cool. So did the 470uf cap. Nothing got hot, but the pump LED did not light up. Trip test for alarm worked as planned.

Pump unplugged and power supply disconnected. DMM on diode setting. With common DMM probe on the anode and + probe on cathode of D1, DMM showed 1, as if it read nothing. With probes switched, DMM showed 165.

Thanks very much one and all.

 

[KeepItSimpleStupid]

Bi-directional 33 V nominal 1500 W. This kind of diode would read nothing both ways.

 

[ronv]

OK. Lets try one more thing. Short out D2. Move the 470 Ufd. cap to the +20 volts and the - side to power ground.

 

[salty joe]

OK. Lets try one more thing. Short out D2. Move the 470 Ufd. cap to the +20 volts and the - side to power ground.

Short out D2 would be connect the two legs of D2 together, is that right?

 

[ronv]

Yes, thats right.

 

[alec_t]

D1 readings sound ok.

With that 470uf cap across the pump leads, the varistor stayed cool. So did the 470uf cap. Nothing got hot.

Glad everything's now running cool. Presumably the pump ran ok? Do you mean the new varistor? Seems like the old one was faulty, or else lack of the 470uF was causing it grief. Ronv's suggested test makes the 'top end' of the circuit a bit more like that in the OEM controller (which doesn't have the equivalent of blocking diode D2).

I don't recall you mentioning the varistor getting hot before you built the speed control? With or without speed control I think it advisable (even if it's an inconvenience ) to add that 470uF cap to each PDM, to reduce the chances of varistor heating. The cap needs to be a 25V 'low ESR' (high ripple current capability) type.

but the pump LED did not light up

Not sure why (with the pump at max speed). Is it completely dead (with oscillations it could have been repeatedly reverse biased beyond its breakdown voltage), or does it light on occasions?

 

[ronv]

Oh, Maybe I misunderstood. Is everything working except the LED? If so if it ain't broke don't fix it.

 

[salty joe]

Yes, the pump ran just fine.
I don't think the varistor got hot before adding the speed control. I feel pretty sure I'd have smelled it. I have not changed out the original varistor. I grabbed a brand new one to check with my DMM.

The speedcon wires are still disconnected, so the pump was at max. The LED was completely off. I'll double check for a broken wire. Flipping these components over a bunch of times is kind of hard on them-I am sooo glad you guys had me use terminal points. I was going to solder all the wires from one component to the other. That would have been a nightmare.

Thanks Alec for the cap fix-I'll chase some down. The 470uf from the OEM controller had barely any legs left so new ones will be better.

Thanks for your help Ronv, are you saying not to bother with your last test?

 

[ronv]

Hi Joe,
Lets see how this goes before we change anything more. See how it works with the speed control hooked up.

 

[alec_t]

I have not changed out the original varistor.

So if it now stays cool that proves two things: 1) the varistor (TVS) is still working ok and 2) the 470uF across the pump was the cure. Good .
@ronv
I ran a sim with the -ve leg of the 470u grounded and the current waveform reverted to what it was like with the TVS overheating .
I think I can now explain why the overheating happens without the 470u across the pump. When the Hall IC in the pump switches off the coil current during commutation there is a brief inductive voltage spike. That exceeds the TVS withstand voltage and it starts conducting. Even when the spike has decayed there is the voltage produced by the pump acting as a generator and it is my hypothesis (I'm out on a limb here and someone will probably shoot me down) that this will sustain the current avalanche through the TVS even when the voltage is below the withstand voltage, i.e the TVS behaves somewhat like a tunnel diode with a negative resistance section in its voltage/current characteristic curve. Without the speed control that current will only flow briefly, hence the heating effect is small, but with the speed control it will flow for a longer period while the FET is held off and so the heating effect will be greater. The 470u (if connected across the pump rather than from pump+ to ground) provides a parallel path for current circulation other than through the TVS, reducing the TVS average current and consequent heating.

The LED was completely off. I'll double check for a broken wire

Do a diode check on the LED too.

Flipping these components over a bunch of times is kind of hard on them

Yes. That's the nature of prototypes . For your next project give strip-board (Veroboard) a try. Component legs aren't then sharing the same board hole and it's easier to unsolder things when necessary.

 

[ronv]

Or maybe during commutation the -pump line goes negitive and the FET is off. Then the only path is thru the TVS?
A clamp diode to ground (cathode to -pump) might prevent this one.

 

[alec_t]

Or maybe during commutation the -pump line goes negitive

Can't replicate that in simulation, with or without the 470u in circuit. Here's the asc file for the recent sims :-

 

[ronv]

Forget that one. The FET body diode should clamp that one anyway. So -pump can't go below ground and +pump can't go above +20 so the only thing left - I think- is that the 20 volts is going above 33 volts.

 

[alec_t]

so the only thing left - I think- is that the 20 volts is going above 33 volts.

That's my conclusion, or rather that Pump+ (nominally 20V) is going above 33V (though the sim doesn't show it, presumably because of over-simplification of the pump model). The inductive spike would be above 20V if not clamped fully by the pump's internal IC, and if blocked by D2 then could well be >33V. Without the 470u the only place it can drive a current is via the TVS.

 

[ronv]

Thats where I came out when I suggested shorting out D2 and adding the big cap to the power supply. But I guess what ever works. I guess the speed control circuit will still work with the big cap?

 

[alec_t]

I guess the speed control circuit will still work with the big cap?

The sim says it will; Joe will soon tell us if it doesn't .

 

[salty joe]

Hey! The speed control works like a champ and everything is cool. The safety/alarm works like a champ too. I had the pump at min and the alarm kicked in, so I cranked up the PDM 50K trimmer a little bit and let it run for half hour or so. A really nice plan-Thank you Alec. Also, outstanding troubleshooting by one and all. Thank You.

My El Cheapo DMM can't read the LEDs I'm using. I checked a new LED with the diode setting and the DMM read 1, as if it saw nothing, but the LED lights up dimly from the DMM battery. Tomorrow, I'll disconnect the speed control and hook up the tidal controller and see what the LED does. Meanwhile, the low ESR caps are on the way. I think I'll use all new PDMs, now that the bugs are worked out. This last pair took a beating. You guys are the best!