Help with Water Pump

[4pyros]

Why do I know where to look for worms?

It was Sandy LOL

 

[KeepItSimpleStupid]

A work-around seems to be to connect a 4k7 resistor between +12V and C4. That results in the current-limit being temporarily raised to ~5.6A ~200mS after power-up, which should give the necessary nudge to get the pump going.

alec:

Does the above mean that the modification could be made safely on sequenced pumps?

Do you expect power glitches to be bad unless the modification is made?

 

[alec_t]

Let me ask this-could a revised PDM built to specs be easily modified to run a pump nonstop?

The addition of the 4k7 resistor should do it.

Does the above mean that the modification could be made safely on sequenced pumps?
Do you expect power glitches to be bad unless the modification is made?

I'll play around with sims and let you know re both. We're fortunate here in the UK in that power outages are very rare, so I tend to forget such things occur elsewhere on the planet! Must try to remember to include them in future sims .

 

[KeepItSimpleStupid]

Failures are rare for me too. Sandy just caused a glitch. Enough to reset a digital clock. Have a couple of AC operated analog clocks so you can tell about how long they were. The last few were under 5 minutes and less than an hour. There is always fried squirrel or some other rodent. Water in a manhole, or a car or tree hitting a pole. Not all of the utilities are underground. I think a mini-substation is less than a mile away and a now idle auto plant next to a railroad uses the power on the large steel towers.

The water tower is about 1000 feet away or less. Trees were planted over top the water mains. They are mostly gone now.

 

[alec_t]

Sims show that power glitches would be a problem without a slight mod because, depending on the glitch timing and duration, the alarm could trip and the pump might never get the current boost to re-start reliably after the glitch. An added diode (D3 in the attached Mk14) overcomes the nuisance trip (at least for a variety of glitch timings I tried, but I can't guarantee for all) by dicharging C3 when the power is lost.
So here's PDM-Mk14, with R6 and D3 (in dotted rectangle) the only difference from the Mk13.
R6 pulls up C4 at power-up to give the current boost.
View attachment 68585
And here's the results for a simulated glitch of 100ms duration occurring at 1.3s from power-up
View attachment 68586
Coil current at power-up pre-glitch rises to ~5.6A (left half of pic). At power return following the glitch the coil current depends on the state of charge of C4, so can be higher (right half of pic) even if Vt is tied to 12V. These values will depend also on the rise/fall times (modelled as 200ms) of the 24V power supply and 12V supply, and of course the realism of the pump model. For the case where Vt is toggling the results would be similar to the right half of the pic.
I'd appreciate someone else running/modding the sim to pick up any errors/sins of ommission.
View attachment 68587

 

[salty joe]

Good eye KISS-thanks, and thanks Alec for the fix. It seems that at least two-three times a year, the power goes out long enough to make all the clocks blink.
It was no problem to slip that little mod into my two existing PDMs. Tonight I'll put the finishing touches on those two PDMs and bench test a pair of pumps.

 

[alec_t]

...but before you do that Joe, here's (yet another!!) slight mod which should make the power-up pump current profile more consistent, less dependent on FET parameter variations, and more like the profile when toggling, hence better if you want to use the PDM for a continuous-run pump. It involves connecting C4 differently (note the polarity, and connection to 12V instead of ground) and using 10k instead of 4k7 for R6.
View attachment 68594
Sorry about all these changes, but that's the nature of prototypes .

 

[salty joe]

OK, after dinner, I'll run downstairs and the switch those parts around. Thanks for putting so much thought into this project.

 

[salty joe]

I connected a wire on the bottom of the board at the D7/R8 junction for the prolong hook. Should this hook be fed into a terminal block, just taped off, or should it be on the top of the board?

I got the latest mod done, tomorrow I'll run a bench test.

 

[alec_t]

Should this hook be fed into a terminal block

Yes, if you're using terminal blocks (as recommended) for the other off-board connections (pump+, pump-, 12V, ~24V, ground, Vt)

 

[salty joe]

OK, thanks I fed both prolong hooks into a terminal block.

I have two revised PDMs on the same board and they are toggling two pumps quite nicely. edit:tried to insert big grin right here, but something is goofy with my computer. After 20 minutes, nothing is geting warm. I'm going to let them run all night.

Since these PDMs will run a pair of pumps, all 12V from both PDMs go to the same terminal, same with V+, ground and 20V.

The alarm does not always sound when the trip test is pushed but the pumps and LEDs turn off until trip test is released. If the alarm does sound, both pizos sound. Turning the trimmer to the lowest setting does not usually cause the alarm to sound even after a startup.

Since I had already added a 4.7K R6, a diode and ran it to 12V, I replaced the 4.7K with a 10K. Then I switched the polarity on C4 and ran the + leg to 12V. So R6/D3 and C4 are run to 12V seperately. Not exactly like the schematic shows - could that affect the alarm?

Thanks.

 

[KeepItSimpleStupid]

Since I had already added a 4.7K R6, a diode and ran it to 12V, I replaced the 4.7K with a 10K. Then I switched the polarity on C4 and ran the + leg to 12V. So R6/D3 and C4 are run to 12V seperately. Not exactly like the schematic shows - could that affect the alarm?

In this case, the +12 is low current, so any +12 will do.

Since these PDMs will run a pair of pumps, all 12V from both PDMs go to the same terminal, same with V+, ground and 20V.

+12 is done fine.
+20 is done fine

Is V+ is the motor and are there two of them?

Now, I have issues with the grounds especially since there are multiple boards and multiple PDM's and multiple controllers and one power supply.

I need to look at the schematic again, but IDEALLY you should separate the low current +12 ground from the High current +12 ground and have TWO ground points on the module. Totally separate.

When working on ONE MODULE alone, you can connect the two grounds together at the module or the negative of the power supply terminal.

When the SYSTEM is together, the low current ground and the high current ground should connect at the power supply negative terminal only.

We have to get the grounds sorted out now. It's just one of those construction techniques that may bite you later.

 

[KeepItSimpleStupid]

Joe:

I'm not sure you know how a house is wired? Someplace near the service entrance there is a ground rod or in older houses the cold water pipe was allowed. This rod is driven about 8' into the ground.

This is the now the reference for the house. The plumbing fixtures, the gas pipe are all at this reference potential. The cable TV and the telephone also get referenced to the same point more or less.

The idea is that any lightning strike will go just to the ground rod and just raise the reference of everything so you don;t have the plumbing at one potential and ground at another.

The ground system is basically called protective ground and should only handle fault currents, but some very tiny currents do flow in the ground system.

The split phase 240/120 system is referenced to this ground potential as well.

This is very similar to how you want the PDM's to work. The negative terminal is the "GROUND ROD" and the single point "NEUTRAL". The motor return currents go to the "GROUND ROD" and the +12 return currents got directly to the "GROUND ROD".

This is so any motor currents in the negative leads, don't affect the +12 V supply. As you have done your "grounds" as each motor starts, the ground will be lifted by (the resistance of the wire to the power supply) * the current. We don't want that. We want a nice clean ground for the circuit.

With one PDM it isn't an issue. With 6, it very well might be.

For now, the ground of the controller, should be at the PDM, not the power supply the way you have things wired.

Now at the main fuse box ONLY, Ground and Neutral are tied together (AT ONE POINT).

 

[alec_t]

I fed both prolong hooks into a terminal block.

To separate terminals, I hope?

So R6/D3 and C4 are run to 12V seperately. Not exactly like the schematic shows - could that affect the alarm?

No, that's fine.
When testing, what were your toggle on/off timings? If they were very short (2 secs or so) the capacitors associated with the trip timer may not have been reaching their intended voltages and so the alarm might not have tripped.

 

[salty joe]

KISS, I have + & - from the power supply going to the controller. At this point, the PDM is getting everything from the controller. My plan was to run + & - from the power supply directly to each PDM board once this is wrapped up, to prevent so much current flowing through the controller. I have two PDMs on one board. There is a 19g wire along the perimeter as my ground for both PDMs. This 19g ground wire is connected to a single terminal point. Is that cool?

Alec, I soldered both prolong hooks to the same terminal. I can fix that.
I have the toggle time set pretty close to the anticipated run time of about 30 sec run/ 5 sec off.
This morning I checked on the pumps and they were not toggling and I'm groggy thinking what the heck- everything is cool to the touch- and then I got a flick. Ha Ha!
Thanks you guys.

 

[alec_t]

So we've established that
1) 3A is not enough to start all pumps reliably; a bit extra is needed
2) Delaying the 3A current limit to give the pump an initial 'kick' of ~ 6-8A, as per the latest set-up, is one way to provide the bit extra and seems to start all pumps reliably.

The other way, as yet untried, to give the bit extra would be to raise the current limit and omit the delay. Are you willing to give that a try Joe? If so, just disconnect C4 and connect a 4k7 resistor across the base/emitter of Q1; that should give a current limit ~4.4A. (3k3 would give ~5A).

The advantage of the first way over the second is a greater starting torque, which would reduce the chances of a stuck rotor. The disadvantage is that the greater torque involves a higher peak current which may exceed the spec of and pose a long-term threat to the (unknown type) Hall effect commutation IC inside the pump, or even the pump coils. The advantage of the second way over the first is that the current limit would be similar to that provided by the old OEM controller and hence should be within the IC/coil specs.

It's your call, Joe. Do you want to go with the first way or the second?
Regardless of your choice I think it would be worth giving the second way a try, if only in the interests of science .

and then I got a flick. Ha Ha!

That's a relief!

This 19g ground wire is connected to a single terminal point. Is that cool?

May be ok, providing that point is very close to the grounded end of Rsense of both PDMs.

 

[KeepItSimpleStupid]

Ground mods.

The suggested mod is really easy. The ground end of the 0.22 resistor is brought to a separate point/terminal and isolated from the ground that you have in your circuit.

That's it!

Two grounds, now get returned to the power supply negative terminal.

EDIT: Later in this thread, it was decided that moving the "Break" in the picture over a little bit, so the wide black line includes the Emitter of Q1. and adding parallel 10 uf/0.1uf combinations across the +12 and the +20 Volt rail to their respective grounds.


@alec
Would it make any sense to put a small cap like 10 uf and say a 0.1 uf ceramic one each board for the +20 and +12 V supplies, each to their respective grounds?

 

[KeepItSimpleStupid]

Brushless motor controller IC anomoly

Good news (maybe):

I got a hold of someone at the suspected manufacturer of the brushless motor controller IC and an email went out to her directly. She could not promise an answer though for whether the suggested circuit for currents > 500 mA is wrong concerning 24 V operation when the absolute maximum voltage for the IC is 18 V.

 

[alec_t]

Would it make any sense to put a small cap like 10 uf and say a 0.1 uf ceramic one each board for the +20 and +12 V supplies, each to their respective grounds?

Won't C1 be enough for the 12V decoupling? A 10uF/0.1uF combo might be worth adding for the 20V rail, but is the FET likely to be affected much by 20V rail ripple? There's already a TVS to catch spikes.
Your suggested division of the ground rail is good, KISS.

the absolute maximum voltage for the IC is 18 V.

Where did that figure come from? I thought the IC was of unknown type?

 

[4pyros]

The suggested mod is really easy. The ground end of the 0.22 resistor is brought to a separate point/terminal and isolated from the ground that you have in your circuit.

That's it!

Two grounds, now get returned to the power supply negative terminal.


@alec
Would it make any sense to put a small cap like 10 uf and say a 0.1 uf ceramic one each board for the +20 and +12 V supplies, each to their respective grounds?

I would put C1 at the chip and add the 10uf and .1 uf combos at both supply inputs. Just for good measure. It may help with motor noise and other adverse effects that may be generated after the motor ages.