Help with Water Pump

[ronv]

That's the one!

 

[KeepItSimpleStupid]

The bands should be red red green gold, if it's a 4 band resistor for a 2M2 or 2.2 M resistor. A 5% resistor should only be 4 bands.

The way it works is each color is assigned a number and, you can think of the 3rd band as a multiplier or the number of zeros to add.
The tolerence is none 20%, 10% silver, 5% gold, so for a Red Red Red Gold resistor,( 2) ( 2) (2 zeroes)5% or 2200 Ohms or 2.2 K 5%
I missed some details, but close enough.

As a mnemonic: Bad (Black-0) Boys (Brown-1) Rape *Red) Our (Orange) Young (Yellow-4 Girls (Grey) But (Blue) Violet (Violet) Gives Willingly (Wjite-9). You have to sort out Grey and Green and Black, but you can change the phrase to remember black. I can mostly recognize the colors on sight, now. The phrase is sufficiently embedded, which is good.

Measure one out of circuit? See if one odd resistor got mixed in the batch? Capital and small letters can confuse at times. Like S being Siemens and s being seconds and mS being milliseimens and ms being milliseconds. m being milli and M being Meg.

 

[alec_t]

The link uses upper case M, so they should be ok (i.e. 2.2 Megohms). I don't imagine there's much call for 2.2. milliohms (a short length of hook-up wire would have that sort of resistance ).

 

[KeepItSimpleStupid]

Yep. Drivel. Just general information. But, it is possible that a rogue resistor got mixed up in the bag.

 

[salty joe]

Thanks guys!

 

[salty joe]

Hi Guys. I replaced R9 with 2.2M resistors on the four good drivers and built two new drivers.

(driver schematic) **broken link removed**

I set the 50k trim pot at about 25 and plugged in a pump. It will only run for two toggles. If the power supply is unplugged and plugged back in, the pump will go for two toggles. Before correcting R9, the pump would only run for the first toggle. The leds continue to toggle happily along. I'm tired so all done goofing with it tonight. Any ideas to maybe get this running would be greatly appreciated.

 

[()blivion]

I was worried about dielectric leakage in the RC timing circuit of the first driver and forgot to mention it. Now might be a good time for that.

Edit: Errr.... leakage in general with such low timing current.

 

[ronv]

Do you mean the led that indicates the pump is on continues to toggle?
If so it could mean the circuit is in current limit. -- the led might still have enough voltage to come on but not the pump. But.. If that is what is happening the alarm should have gone off... But.. There might be a hole there where the current limit is tripped but the threshold of the latch hasn't been reached.
Getting closer.

Good to see you back. I was worried we had lost you.

 

[salty joe]

Would dielectric leakage be caused by residual flux on the board? I have not scrubbed them down yet. I am concerned about getting the trim pots and the trip test wet.
Yes, the led that indicates the pump is running continues to toggle. If the trip test is pushed at that point the alarm sounds and will not go off after releasing trip test.

Good to see you back. I was worried we had lost you.

I was worried about losing all of you. It took a little while to get the parts and another little while to build the last two drivers.

Thanks guys.

 

[KeepItSimpleStupid]

Very possible. The one item I would worry about is the buzzer. If the buzzers were new, they would have a removeable piece of tape on them.

You will also find that the earlier you scrub the boards, the easier the flux is to get off. 20/20 Hindsight.

When you scrub the boards, soap and water usually works. Finish the scrub with a spray of alcohol, like rubbing alcohol. The alcohol will displace water. Dry with a hair dryer or set outside in the sun to dry.

Spot cleaning flux, say after a repair, you should be able to use Q-tip with alcohol. A Foam tipped swap with alcohol would be a better choice.

Commercially, flux cleaner and dental picks come in handy.

Remember, the earlier you clean, the easier it is to clean.

 

[alec_t]

At least we know the current sensing works.
The trip circuit is designed to trip the pump off after ~1/2 sec of over-current then re-try a turn-on every ~30 sec. Are you saying it makes two such attempts then keeps the pump off? If so, try putting ~4-10k in parallel with R6 temporarily to reduce its effective value. The R5/R6 values are critical in setting the operating threshold of the oscillator formed by the U1 gates.
If the inter-trip time is significantly less than 30 secs then I would suspect current leakage could be a culprit; otherwise not.

 

[salty joe]

Almost always, the pump runs for the initial toggle only. It ran for two toggles and three toggles, but very seldom. Meanwhile, the leds continue to toggle. I switched the pump out for a 40W bulb and the bulb dimly lit with the led as normal. I put the pump back on and nothing. If I unplug the power supply and plug it back in, then the pump will run, but usually only for one toggle.

Should I go ahead and reduce the effective value of R6?
Thanks.

 

[alec_t]

Meanwhile, the leds continue to toggle.

1) At what time intervals?
2) Which LEDs (the PDM_Mk10 has D8 only)?
3) What is the voltage across Rsense while the LEDs are (a) on and (b) off?

Should I go ahead and reduce the effective value of R6?

Before you do that, try setting the trimpot to max (50k) and see if the pump then runs for a longer period.

 

[salty joe]

Setting the trimpot at max, 53K or so, did not seem to make a difference. The pump will run for one or two toggles, occasioally three toggles.
I have the toggle set at about 9 sec. on, 4 sec. off.
The V across Rsense when the leds turn on is momentarily 0.04, then 0.03. It is 0.00 when the leds are off.

There are two leds because the tide drivers will run two pumps as a pair. I put two drivers on one board. Both drivers share the 12V, ground, 20V, and control.

When the trip test is pushed the alarm sounds for about 7 sec. for both drivers, but on one driver the led continues to toggle while the alarm sounds.

 

[ronv]

Maybe you could take several measurements.
Base of Q1, Collector of Q1
Drain of FET, Gate of FET
Cathode of D7.
I would put aside the driver where the led continues to flash when the test switch is thrown. -- maybe 2 bugs.
Are you sure of the measurement across Vsense?

 

[alec_t]

I have the toggle set at about 9 sec. on, 4 sec. off.

Ah. For testing the PDM you should not be toggling its control input on and off with the timer signal; the control input should be connected instead by a SPDT switch to 12V as we discussed (and you implemented) some posts ago. We are trying to run the pump continuously for at least 30 secs to enable the trimpot to be adjusted.

The V across Rsense when the leds turn on is momentarily 0.04, then 0.03

Hmm. I think your DMM didn't have time to register the voltages correctly. Pity you don't have an o'scope

 

[salty joe]

Ah. For testing the PDM you should not be toggling its control input on and off with the timer signal; the control input should be connected instead by a SPDT switch to 12V as we discussed (and you implemented) some posts ago. We are trying to run the pump continuously for at least 30 secs to enable the trimpot to be adjusted.
Hmm. I think your DMM didn't have time to register the voltages correctly. Pity you don't have an o'scope

OK, I'll get my head back in the game

The rest of today is shot-I have to fix my wife's car and my girl is running cross country. Tonight when things settle down maybe I can slip downstairs.
If I could, I'd put a lot more than a single +1 on your last post, Alec!
Thanks all.

 

[KeepItSimpleStupid]

Joe:

Sometimes it's hard to understand what's happening. Troubleshooting needs to be precise and methodical. Try to make sure your posts are understandable.

"I think" we should be at the point of using the "BIG resistor" or dummy load.

One would hope that with the dummy load:
1) 50K pot at minimum (the alarm LED should stay on after a single toggle)
2) 50K pot at maximum (the alarm LED should always be off and your motor LED will blink)

@Alec
Will Joe run into issues with such a short cycle time of 9s/4s?
Check my logic.

@Joe@Alec
Earlier we kinda went through a possible mechanism of adjusting the 50K pot, but it has to not shut down at one of the 50K extremes. The light bulb, the resistor and the motor are suggested progressions with the 50K pot at both extremes.
1) The light bulb is a low enough load that it's not likely to be able to cause it to go into alarm at either extreme of the 50K pot.
2) The resistor may or may not cause it to go into alarm at one extreme of the 50K pot.
3) The motor may cause the driver to go into alarm all of the time at either extreme of the 50K pot.

We want the circuit to go into alarm when the 50K is toward minimum and out of alarm when the 50K is toward maximum.

The motor draws a large current when it initially turns on. The light bulb does the first time it turns on or when the bulb is cold. The resistor does not.

 

[alec_t]

@Joe
Hope the car fix went ok.

@KISS
Your analysis of the situation has it in a nutshell.
I suspect the circuit isn't coming out of trip state because the lower Schmitt threshold of the U1 circuit isn't being reached. It's unlikely to be if the control input toggles at 9s on/ 4s off. Recovery from trip state is designed to be ~30s (by virtue of R9/C3).
What value resistor did you suggest? Can't remember if Joe has it to hand.

Edit: Just ran a sim with a 9s on/ 4s off control signal toggle. Circuit should still trip and recover as designed, but trimmer adjustment is best done without that toggling.

 

[alec_t]

Joe, it occurs to me that a 110V ~2kW load (electric kettle?) would make a good dummy load for the PDM, if you don't already have the 'big resistor' KISS mentioned. Might even be able to brew a coffee with it while testing .