Help with Water Pump

[salty joe]

Car repair went well, yesterday was a beautiful day for the kids to run, today is rainy and cold but I get to be inside goofing with drivers and hanging out with you guys online-thanks all you guys for hanging in there with me.

I've got two drivers on one board that share the ground, 12V, 22V, and control for two pumps to run as a pair. 12V is connected to both R1 and 12V input. There are two 40W incandesant bulbs as dummies. Bulbs and leds are lit. 50K trimmers are set at about 25K.
Since these are the drivers I just built, I ran these tests.
If R1 is grounded, all lights go out.
11.82V at both D4/R7 junctions.
5.4V across both R6s.
11.6 across both trip tests.
Here are the lunker resistors I have. (thanks KISS) They are about halfway down the page.
**broken link removed**
Hopefully this is getting started on the right foot.

 

[alec_t]

Hopefully this is getting started on the right foot.

Yes, good start.

5.4V across both R6s.

That exactly matches the simulated value, but may be a tad high for the particular U1 examples you have in circuit.
Keep the bulb dummy load for now. Set R1 at 12V, hold the trip test button for ~ 1 sec and release. Trip should operate but reset after ~30 sec. If that sequence doesn't happen try a 4k7 or thereabouts across R6 to reduce its effective value then repeat the trip test.

 

[salty joe]

Both trip tests made their pizo sound for 30 sec. right on the money. I never knew 30 sec. was such a long time-very annoying noise-just like they're supposed to be.

But only one of the trip tests shut down the led and the light bulb. The other trip test would only shut down the led and light bulb if I held the trip button down. I double checked that Q1 was 2N3904 and that it was hooked up correctly. Also double checked the trip test connections. Bad trip test button maybe?

 

[ronv]

Check that D4 is in correctly and to the proper points.

 

[KeepItSimpleStupid]

Food for thought

@Alec: Something to think about

Will the range of the Hfe (60-300) of the 2n3904 give any grief?
Worth changing to an OP amp?
Would it be worthwhile to make and compare a few measurements (ie. voltages across Rsense, R3, R4) with a fixed load (4 ohms is it?)?



Schematic for ref: **broken link removed**

 

[salty joe]

D4 was good but I accidentally jumped R7 with a lead. I never would have found it if you had not put me in that vicinity. I snipped out a small piece of wire and now the led and light bulb go out with the trip test. Thanks Ron.

 

[ronv]

That would do it. I bet the little diode and IC are glad to see that go. That point was tied to +12 to keep the pumps on wasn't it?
Anyway it sound like they survived so alls well.
Was that the last bug?

 

[salty joe]

I hope so. I think we're getting ready to run this through it's paces, so we'll see. Thanks all.

 

[alec_t]

Glad to hear the bug was found and the test was then successful. Next step is to repeat the test using a 'big resistor' dummy load. Check that the trip works and, if you can, measure the Rsense voltage in both the trip and non-trip states.

@KISS
I'll look into the effect of Q1 gain but my hunch is it won't be a problem.

Edit: sim shows trip current is ~2.7A with gain = 60, 2.4A with gain = 300. Exact trip point isn't critical but can be adjusted if nececessary with a pull-down resistor on Q1 base.

 

[salty joe]

With the big resistors hooked up, there was 0.5V across both Rsense when leds were on and 0V when leds were off and the pizos sounding.
I did not get a chance to push the trip test-it went off on its own after maybe 20 sec. I barely had enough time to take a reading from the two Rsense with the leds lit before the pizos started. Sure enough, after 30 sec. the pizo stopped and the led came back on but this time the led stayed on for only 5-10 sec. before the alarm sounded again.

The big resistors got hot in a hurry.
The alarm part of the drivers work independantly, which is very cool.

 

[KeepItSimpleStupid]

So the resistor puts it in alarm? Adust the trim pot for max and see if it stays out of alarm, but LED on.

 

[alec_t]

The fact that ~0.5V on Q1 base trips the alarm indicates Q1 gain is probably >300 and the trip current is ~2.3A. That's fine. We can work with that for now.

the led stayed on for only 5-10 sec. before the alarm sounded again.

Hmm. That's far too long and accounts for the big resistor getting hot. Should be on for < 1 sec. Check the value of R8. Have you perhaps got 470k (yellow, mauve, yellow) instead of the specified 4k7 (yellow, mauve, red) ? (Sim gives ~7 sec on if R8 = 470k).

 

[salty joe]

I double checked both R8s and confirmed they are 4K7, 4.7K.

I wanted to put a stopwatch on the trip times, but this time the trip did not occur. I let it go for about a minute and a half then pushed the trip test and it turned off the led and sounded the alarm for 30 sec. just like it’s supposed to. I waited for about another minute with the led on then unplugged the power supply because the giant resistors were getting so hot. I licked my finger and touched the resistor and it sizzled. The mosfets and ICs did not get warm at all.
This seems odd because I did not touch the trimmers or the power supply adjustment or anything and got a different result.

 

[alec_t]

I double checked both R8s and confirmed they are 4K7, 4.7K.

Ok. Now check D7 using the diode-check function which your DMM hopefully has. It should read ~600mV if conducting properly. If you don't have that function then use the ohms function. Should read ~1k in the conducting direction.

the giant resistors were getting so hot

They would, passing ~2.3A continuously. Now you know why we're using a dummy load and not the pump .

The mosfets and ICs did not get warm at all.

Excellent. And good that you thought to check.

This seems odd because I did not touch the trimmers or the power supply adjustment or anything and got a different result.

Not too surprising. I think perhaps the one big resistor load is just on the limit of operating the trip, so even a small voltage change somewhere may tip the balance.
Next step I'd suggest, if D7 is ok, is to try using 2 of those resistors in parallel as the load for one driver circuit (ignoring the other driver for now). That will raise the load current well above the target 3A threshold and the trip should kick in <1 sec.

 

[salty joe]

D7 read 645mv with my DMM on the diode setting. I checked D4-D7 on both drivers and they all read within a point or two of 645.

With the big resistors in parallel, it tripped immediately. After 30 sec. the led came back on for an instant before it tripped again.

 

[alec_t]

Diodes ok then, which makes me more convinced the single big resistor load was marginal for setting off the trip.

With the big resistors in parallel, it tripped immediately. After 30 sec. the led came back on for an instant before it tripped again.

Now we're getting somewhere. That load (we'll call it 2BR) is a better dummy for the actual pump.
With the trimmer set at max, and using the 2BR load, can you estimate the 'on/instant' time before the trip occurs? Hopefully it should be ~1 sec.

 

[salty joe]

With the trimmer set at max, and using the 2BR load, can you estimate the 'on/instant' time before the trip occurs? Hopefully it should be ~1 sec.

I got 1.39 sec on my stopwatch but I was a hair slow on the stopwatch button-probably more like 1.2 sec.

 

[alec_t]

probably more like 1.2 sec.

Good. Simulation says 1.1 sec. So far so good.
Now for the big test.
Set the trimmer to min. Use the pump as the load.
If the pump runs ok check that nothing gets hot (Rsense should get warm, but not excessively so).
If the circuit trips it will probably be in ~0.2 sec, but it might not.

 

[salty joe]

With the trimmer at min., the pump ran just fine. After about 6 minutes i could not feel any warmth in Rsense or any other component. This is getting exciting

 

[alec_t]

Hurrah! Now we know the pump can get up to speed, and so reduce its concurrent consumption below the trip point, within ~0.2 secs. I did wonder if the current might have been higher for longer.
As a final test for that driver, get the pump running then press the trip test to make sure the pump turns off. If it does then we'll declare the driver as functioning and you can pour yourself a glass of your favourite tipple .