Help with Water Pump

[salty joe]

Thanks for posting those ron; QUOTE]

+1. I'll get it printed and compare it to my working PDM.

BTW, Since Ron gave me PDM14, I think I have every recent schematic and drawing in my documents. A few days ago, I was unable to paste into the reply box. If anyone is interested, I can email if I still can't paste.

 

[ronv]

Alec,

Can the hooks be put into the MK14 to give the added features? How would the current pulse of the MK14 work with the speed control? Why did we go away from that? Was there a "buglet" or ?

Joe, I think we just need to zero in on the version to go forward with.

 

[alec_t]

Can the hooks be put into the MK14 to give the added features?

Well that's what the Mk15 does (using U1d), plus reinstating R11/R12/D9 (to equalise trip times for first and subsequent trips) and removing R5/R6 (redundant with the removal of C4).

How would the current pulse of the MK14 work with the speed control?

Should be the same for Mk14 and Mk15.

The current limit feature of both Mk14 (as modified and found to operate all pumps) and Mk15 is in theory the same, so it's a puzzle as to why, with D4 disconnected so that the trip circuit isn't affecting the FET, only some pumps work with Mk15.
Before we give up on Mk15 perhaps a few more tests?
Joe, can you confirm:-
a) your soldering iron bit is grounded (connected to its plug earth pin),
b) D4 (not D5) is disconnected.
I'd suggest doing the following tests, taking due anti-static precautions, with the trimmer at max and a toggle 'on' time of at least 10 sec, in each test measuring the volts at the FET gate, drain and source both with and without the pump running/humming and then restoring the circuit to its previous configuration before doing the next test:
Test 1) re-flow the solder joints at all 3 legs of the FET and ensure R2 is connected with a very short conducting path to the FET gate. Also check R2 resistance value in circuit.
Test 2) temporarily short the base of Q2 to +12V (i.e. short out newly-numbered R6 in the Mk15).
Test 3) temporarily insert a 2k2 resistor in series with D5.

 

[salty joe]

Joe, I think we just need to zero in on the version to go forward with.

Working on it. I wonder if all those links are gone forever. What a bummer.

Test 3) temporarily insert a 2k2 resistor in series with D5.

Does it matter which side of D5 gets the 2K2?
Thanks for those tests-I'll get it done today.

 

[alec_t]

Does it matter which side of D5 gets the 2K2?

No. Whichever is more convenient (or should I say whichever is the least inconvenient ).

 

[alec_t]

I wonder if all those links are gone forever

I did retrieve the missing text. I'll post the links if I can find them.

Edit: Nope, can't find any from the 8-14 Dec gap

 

[ronv]

Well that's what the Mk15 does (using U1d), plus reinstating R11/R12/D9 (to equalise trip times for first and subsequent trips) and removing R5/R6 (redundant with the removal of C4).
Should be the same for Mk14 and Mk15.

I think it was C4 in the MK14 that kept it from oscillating. The bandwidth is so low the motor inductance is not an issue.

The current limit feature of both Mk14 (as modified and found to operate all pumps) and Mk15 is in theory the same, so it's a puzzle as to why, with D4 disconnected so that the trip circuit isn't affecting the FET, only some pumps work with Mk15.

I think the oscillation path is D5 not D4 as the alarm does not always go off.

You might try the snubber mentioned back about 119 or 120

 

[alec_t]

I think it was C4 in the MK14 that kept it from oscillating.

Could be. But the modified Mk14 doesn't have C4 and there is no problem; all pumps run ok.

I think the oscillation path is D5 not D4 as the alarm does not always go off.

I thought Joe said the alarm function is now (Mk15) ok with U1d inputs properly connected?

You might try the snubber mentioned back about 119 or 120

That's one option. Another might be to slug the D5 action (albeit not as much as in the Mk13 version) with the series resistor suggested in Test 3 above, plus a cap to +12V.

 

[salty joe]

First the good news. Fussy pump A ran reliably for every toggle with a 2.2K resistor at D5, even after setting the on/off times about 3 sec. There is no bad news unless the numbers look ugly.
Power supply set at 20.6V. The soldering tip did not show conductivity with ground until the tip collar was removed and the tip twisted and rubbed against the soldering gun shaft. I reconnected pin 13 from U1d to Vt. Pin 12 (fault) was grounded. I confirmed that D4 was and is disconnected. Toggle time about 20 sec. run, 10 sec. off. I re-flowed solder at all three legs of FET. R2 is installed at the very next hole to the gate. When I re-flowed the solder, I got a total solder bridge at the gate/R2 junction. With the DMM set at 20K, resistance at R2 is 0.10.
In V, before anything was done;
run gate 11.5, run drain 0.3, run source0.3
hum gate 10.8, hum drain 0.3, hum source0.1
off gate 0.1, off drain 20.1, off source 0.0
After re-flow at the FET, I got the same numbers except run gate was 11.4 and off source was 0.1.
When I jumped R6, all numbers were the same except hum gate was 8.7.
With a 2.2k resistor at D5, all the numbers were the same except no hum numbers. Party time!
The alarm did not sound when the pump was in hum mode. The alarm did work a few tests ago, at least when the trip test was pushed.
Thanks a ton.

 

[KeepItSimpleStupid]

@alec

Confirm what conditions the alarm should activate for. I do remember that it's not comprehensive. I THINK it's only supposed to work when the pump draws too much current and you abandoned the "Window detect" for current. We know now, that a "humming pump" won't generate an alarm and I don't think the electronics will detect that.

Not sure we're ready for a real locked rotor test. I wonder if the best way to accomplish a locked rotor test is to remove the armature from the pump. That way nothing is resisting an big force, but it should be EQULIVELENT to a locked rotor condition.

Nice job Joe!

So where are we?
I think the alarm electronics won't detect humming or a broken wire to the pump. Do we go there?

Does this PDM make all the pumps work?

Are we ready to try a stall test and see if it alarms (i.e. remove the motor shaft). This assumes the pot is adjusted properly.

Speed control?

My $0.02

You might have a milestone before Christmas.

 

[salty joe]

Feels like a milestone to me! Thanks KISS, it's been you guys for sure. When I think about the high powered brains and effort that's gone into this, it's almost overwhelming.

 

[ronv]

Merry Christmas - I hope.

Alec, I think I understand the oscillation better. Take a look at this sim with a FET with higher gate capacitance.

 

[KeepItSimpleStupid]

Mmm:

Time to add more parts?: See: https://www.electro-tech-online.com/custompdfs/2012/12/dn80.pdf

 

[alec_t]

Fussy pump A ran reliably for every toggle with a 2.2K resistor at D5

Good news indeed, Joe. I think that confirms that a certain combination of FET and pump characteristics causes the FET to oscillate. Seems the 2k2 plus the gate capacitance slugs gate voltage changes sufficiently to damp or kill the oscillation. So regard inclusion of the 2k2 as part of 'Mk15-modded'.
Next step is to re-connect D4 and re-test with the toggle 'on' time at least 10 sec. Then the alarm should trip automatically if the pump current exceeds the 4.4A limit for more than the trip time set by the trimmer. Start with trimmer at minimum.

@ronv
Thanks for the oscillation insight.

@KISS
The alarm presently is for over-current only and operates as stated above, but won't detect the hum state. Hopefully the proposed Fault module would do so, provided there is no coil commutation in the hum state.
I've been unable so far to reproduce the hum state in simulation, but will work on it. Modelling the pump internals, particularly the IC functions, is the tricky bit.

 

[salty joe]

Thanks guys. I'll reconnect D4 tomorrow and test. I am so excited to use the Mk15. I thought it was a goner. Terrific save Alec!

Getting ready to take off on a minor road trip for an all day wrestling meet. They are kids only one time so we go.

 

[salty joe]

With D4 reconnected, I got the same V numbers at the FET:
run gate 11.4, run drain 0.3 -0.4, run source 0.3-0.4
off gate 0.1, off drain 20.1, off source 0.0

Pump toggling right along.

With the trimmer set at min., the alarm did not sound. The alarm worked as planned when the trip test was pushed. While the controller kept toggling, the pump and the LEDs did not come back on after the trip test. I think there was a cracked wire at the Vt terminal point. Everything seems good now. I'll check it out some more in the morning.

 

[salty joe]

I don’t think there was cracked wire after all. I found that the trimmer acts up if it is pegged all the way to min. I backed it away just a blonde hair and that was the end of the bug. I am going to assume that pegged all the way to max is likely to cause a bug too.
With D4 connected and the trimmer at min., fussy pump A toggles right along, the trip test makes the alarm sound and shuts down the pump for IDK 20-30 sec. then the pump fires back up.
As far as I can tell, it’s all systems go!
The 2k2 at D5 really did the trick. Thank you Alec. Are there any other parts to finish fine tuning the Mk-15? Which reminds me-I still need to install caps where 20.6V and 12V enter the Mk-15.
Is it time to build the speed control?
Thanks all.

 

[alec_t]

I found that the trimmer acts up if it is pegged all the way to min. I backed it away just a blonde hair and that was the end of the bug. I am going to assume that pegged all the way to max is likely to cause a bug too.

Well observed. I've found on some trimmers the wiper can move past the ends of the resistive track and then, of course, it's open circuit.

Are there any other parts to finish fine tuning the Mk-15? Which reminds me-I still need to install caps where 20.6V and 12V enter the Mk-15.

Those caps should hopefully be all that's to do.

As far as I can tell, it’s all systems go!

Before going ahead with the speed control we need to bite the bullet and test that a locked rotor trips the alarm. Set the trimmer to a hair past minimum. Jam the rotor. Fingers crossed. Power up. If the trip doesn't operate in ~ 2 sec then power down!

Provided ALL pumps are happy with the Mk15-modded then yes, try the speed control. Good luck with the build.

A very Merry Christmas to you and yours

 

[KeepItSimpleStupid]

Trimmers used as two terminal devices, should have one end and the wiper tied together. It's common practice. I believe it will prevent that situation. So, if the terminals are 1, 2 (wiper) and 3 (cw), then tie 3&2 together for one terminal and also use terminal 1 for increasing resistance.

I think we had that discussion before.

Before going ahead with the speed control we need to bite the bullet and test that a locked rotor trips the alarm. Set the trimmer to a hair past minimum. Jam the rotor. Fingers crossed. Power up. If the trip doesn't operate in ~ 2 sec then power down!

Earlier, I suggested to just remove the armature. It's the equivalent of a stall and you won't get hurt or break anything when trying to force something that wants to rotate stopped.

This thread still amazes me. Length, content, attitude of everybody, patience of everybody, set backs, design, troubleshooting, knowledge and people from many parts of the world. Try doing this in the 80's. Assume everyone had a FAX machine and lead for their pencils.

 

[4pyros]

Earlier, I suggested to just remove the armature. It's the equivalent of a stall and you won't get hurt or break anything when trying to force something that wants to rotate stopped.

Would the current increase enough without the rotor to trip the alarm?


Happy Holidays