Help with Water Pump

[alec_t]

Sim shows voltage readings quite different from yours, but I think that's due to all the approximations/estimations involved in my pump modelling.
I'm attaching a revised schematic (SM-Mk4) showing the Pulse Delay Section (PDS) as I think you now have it, and the Speed Set Section with a small mod to cause ~ 1 sec delay at power-up before the PDS kicks in. This should have a similar effect to your serendipitous shorting of pins 1 and 2 of U1, i.e. will keep Speedcon high initially to give the pump a chance to start up at full speed. As before, changed areas are indicated in dotted rectangles. Note in particular the re-purposing of C4 and its modified top-end connection. I've used component values which I think you should already have available.

 

[salty joe]

Thanks Alec. The PDS shown is exactly how it's hooked up.
I should get a chance to modify the speed set section tomorrow.

 

[salty joe]

I added the mod to speed set section. The pump starts reliably. With the SPDT on the "fast" wire, I get a hi speed and a low speed and the middle position is maybe a little faster than low-it's hard to tell. I did not get any speed adjustment using the 25K trimmers. These trimmers are different from the others, they do not have a stop for max and min. Just a screw head that spins. The alarm works as planned. The pump LED is back on, but is flickering.

Sorry it took a couple days. I got up early this morning before work for the second mini session and just wrapped it up.

 

[alec_t]

I added the mod to speed set section. The pump starts reliably.

Good news indeed!

With the SPDT on the "fast" wire, I get a hi speed and a low speed and the middle position is maybe a little faster than low-it's hard to tell

So far so good.

I did not get any speed adjustment using the 25K trimmers.

Hmm. Surprising. What values of the 'spd' voltage do you get as you vary those trimmers (a) with 'Fast' at 12V and (b) with 'Fast' at 0V ?

These trimmers are different from the others, they do not have a stop for max and min. Just a screw head that spins

Sounds like they're the multi-turn type?. If so, you'll probably have to rotate them several turns to get a noticeable voltage/speed difference.

 

[KeepItSimpleStupid]

Multi-turn pots: 10 turns are common, but 5 and 15 are available. They usually start "clicking" when you go past the endpoints.

 

[salty joe]

The speed control does most definitely work. You figured it out Alec-thanks. Brilliant work! The V at spd with “fast” connected to 12v (high) is from 5.71 to 10.95. V at spd with “fast” connected to signal ground (low) is from 3.91, and when the V was adjusted much lower than 3.5V, the pump stopped.

These trimmers do need to be turned quite a few times. I had not noticed the little click before-thanks for that useful tip KISS.

The pump was running for a couple minutes while I ran the test and I smelled pre-cook. D2 on the PDM was pretty warm and the varistor on the PDM was hot. I don’t know what the temp was, but it felt hot enough to get burned if I held onto it.
Could the flickering pump LED and the heat be correlated?

 

[alec_t]

The V at spd with “fast” connected to 12v (high) is from 5.71 to 10.95. V at spd with “fast” connected to signal ground (low) is from 3.91, and when the V was adjusted much lower than 3.5V, the pump stopped.

Sounds about right.

I smelled pre-cook. D2 on the PDM was pretty warm and the varistor on the PDM was hot.

Well observed. Will try to resolve that. Does the varistor get hot even when the pump is running at max speed for a few minutes? Was the FET heating noticeably too?

Could the flickering pump LED and the heat be correlated?

Possibly both are due to some residual oscillation, where the pump is hovering around an on/off state.

Edit: Can you remind us which particular Varistor you have?
What is the Isense voltage with pump at (a) max speed and (b) min speed?

 

[KeepItSimpleStupid]

I smelled pre-cook. D2 on the PDM was pretty warm and the varistor on the PDM was hot.

Well observed. Will try to resolve that. Does the varistor get hot even when the pump is running at max speed for a few minutes? Was the FET heating noticeably too?

Snubber values? @alec: Thinking too high of an RC constant

What is the p/n for the diode across the motor? @alec: Fast recovery?

 

[alec_t]

@KISS
Don't know if Joe ever got the 1uF caps for the snubbers. AFAIK 0.39u is currently in use, so less than ideal.
By 'diode across the motor' I presume you mean D1? I'm wondering if it might be better to have D1 cathode connected to D2 cathode instead of to D2 anode. That way D1 could absorb some surges and so share the duty with the varistor, hopefully reducing its heating. What do you think? Can't simulate this, as according to LTSpice there aren't any spikes/surges to absorb!

@Joe
Can you remind us of the product codes for both the Shottky diodes D1 and D2 that you have in the PDM ?

 

[ronv]

Very strange. Seems like any positive going pulse should go thru the fuse then diode to +24 before it goes thru the TVS. Maybe just put it across the TVS and take the fuse out of the equation at the same time?.
I'm not sure what things look like with the speed control, but it might be the only time besides power on that the FET is turned completly on and off.

 

[alec_t]

I'm not sure what things look like with the speed control, but it might be the only time besides power on that the FET is turned completly on and off.

The speed control turns off the FET completely only when the coil current has already dropped close to zero by virtue of the commutation, so in theory there shouldn't be any voltage spike at that point due to the speed control. However, the commutation must produce a turn-off spike (which is handled by the pump's internal IC). Perhaps D1/D2 need re-configuring to allow D1 to absorb that? Subsequent delayed turn-on should result in a ramp-up of current due to coil inductance; again no voltage spike in theory.

 

[salty joe]

This is the poly cap I have for snubber. I disconnected the snubber when it did not solve the problem with the pump not running. It’s still on the PDM board and could very easily be hooked back up.
https://www.ebay.com/itm/260848704784?ssPageName=STRK:MEWNX:IT&_trksid=p3984.m1439.l2649
This is D2. **broken link removed**
This is D1. https://www.ebay.com/itm/5-x-1N5822...822?pt=LH_DefaultDomain_0&hash=item3a6969ae96
I'm going to have to do some digging to find the varistor.
The V across Isense with pump at max is 0.27, at min is 0.09. The only flicker I could see in the pump LED was at low speed. The slower the pump, the more pronounced the flicker.
The varistor did get hot with pump running at max. I could feel no heat at the FET. D2 started warming up.
Thanks all.

The varistor. https://www.digikey.com/product-detail/en/1.5KE33CA-T/1.5KE33CADICT-ND/151034

 

[alec_t]

I disconnected the snubber

Reconnect it. That may cure the varistor heating (but the 100 Ohm in the snubber may get hot instead).

The V across Isense with pump at max is 0.27, at min is 0.09.

Sounds right.

The only flicker I could see in the pump LED was at low speed.

As expected. If low speed < 600 rpm (say), there are < 20 pulses per sec which would probably be noticeable.

The varistor did get hot with pump running at max.

Hmm. Does it get hot when Speedcon is disconnected (i.e. the speed control is inactive) ?

I could feel no heat at the FET.

Good.

D2 started warming up

To be expected. Current is ~ avge 1.2A, diode drop is ~ .45V, so power dissipated in D2 = ~0.5W. If it gets beyond 'warm' consider attaching a small heatsink (bit of tin can or ali sheet).

 

[salty joe]

With the speedcon wire disconnected from the speed module, the varistor heated right up. With speedcon still disconnecteded, I re-connected the snubber but the varistor still got hot in a hurry. I could not tell if the snubber resistor was heating up-it's in a tight spot and my fingers are big and calloused. If we need to know, I can get one of my girls to check.
I like your homemade heatsink. I'm gonna give that a try.Thanks.

 

[KeepItSimpleStupid]

@alec
What about driving that FET with a real driver. I smell oscillations.

 

[ronv]

The fact that the TVS gets hot sounds like the clamp diode is open. To be honest I've lost track of the latest build but here is what I think the FET stage looks like. I used this to play with the oscillations. It seems better without the 1000pf and the snubber seems to clean it up.

 

[alec_t]

@KISS

I smell oscillations

Probably. But the fact that speed control now works and the voltages across Rsense are reasonable count against that.

@ronv
Thanks for the sim. The 1000pf has been absent from the circuit for ages...because it definitely caused oscillations . Presumably there's a clamping diode inside the pump, as part of the Hall IC, since the OEM controller for the pump has no clamp diode. It does, however, have a 470uF cap directly across the pump terminals. Perhaps that is the primary clamp mechanism?

 

[ronv]

I don't see any way for the TVS to conduct if D1 (D2 in my simulation) is in the circuit unless the 20 volt supply is somehow being "pumped up". I suspose a big cap on the 20 volts would preclude that.
Don't forget the stock controller never really turned on and of only up and down.

 

[alec_t]

I don't see any way for the TVS to conduct if D1 (D2 in my simulation) is in the circuit

Nor can I. The TVS (varistor) has a 28V withstand rating.

Joe, can you check forward and reverse resistance readings of D1 (power off, DMM on diode or resistance range), also verify its connections?

 

[salty joe]

D1 looks to be hooked up correctly. With DMM in diode setting and common probe on anode of D1 and positive probe on cathode, I got 690, then it steadily increased to 1333. With probes switched, I got 163.

Also, the + leg of a 10uf, 50V cap is connected to D1 cathode-D2 anode where the power comes in and the - leg of the cap is connected to power ground.