Help with Water Pump

[KeepItSimpleStupid]

There is always the old adage that says "The transistor failed to protect the fuse".

 

[alec_t]

Haven't heard that one before. LOL. In some circuits (not this one) it might actually be cheaper to use a sacrificial transistor than a fuse.

 

[salty joe]

There is always the old adage that says "The transistor failed to protect the fuse".

That's funny!
I was looking to protect the controller, but I guess protecting the power supply and pumps beats the heck out of protecting nothing.

 

[KeepItSimpleStupid]

When I started a job some many years ago, my department was eating fuses at 25 to $30 a pop. These were the expensive ( I^2)t semiconductor fuses. They were operating triac controllers into a Variac.
1. They were not using the current limit (some controllers were purchased with them, some not)
2. They were using 10 A (I^2)t fuses and they really should have been using 25A.

I fixed the above problems and added a 3AG fused sized to the load. Cost dropped.

Later, I stuck my neck out and said that you can replace the entire home-built panel (I meter, V meter, Fuse, Variac) with A DC power supply.
The controller was changed to a programmable ramp/soak controller. The heaters were custom wound and operated in a vacuum system. Power requirements were about 30 V and 10 A.

 

[salty joe]

[QUOTE=alec_ Try 0.22uF or 0.47uF or anything close, and change R3 to 470k or 180k accordingly (better yet, make R3 a 500k trim-pot, which would allow adjustment of the 'flick' time).

This is in reference to C10. I have a .47u cap, but it is polar. It seems that the positive side should go towards +24V, but I've been wrong before when I felt almost certain. Can I get a confirmation? BTW, R3 is now a 500K pot.

This build is a little tougher than the last one. I'm cranking through it-sure hope every single connection is correct. Double checking and all that good stuff, but still...

 

[alec_t]

It seems that the positive side should go towards +24V

Correct. I've attached the relevant section of the schematic showing R3, C10 and flick timing revisions to date. As C10 has been increased in value I've also added R6 as a precaution to limit current into the inputs of gates U3c, U3d when C10 pulses high.

This build is a little tougher than the last one.

Ah, but you're getting to be an expert by now .

 

[alec_t]

Oops. I've just spotted an error in the schematic. The clock pulse input (pin 10) of the 4020 should connect to the left side of R2 (= output of gate U2a), not the right side as currently shown. Apologies.

 

[salty joe]

Good catch Alec-thanks. I'll just move that connection between the output of U2a and R2. Should only take a minute. I'm sure you saved me a lot of grief- thank you, thank you, thank you.

Also the revision is much appreciated. I was not sure if a polarized cap would work since the original C10 was non polar. The latitude for switching parts is very cool and very interesting. I'm pretty sure there are some 330K resistors bouncing around in my electronics junk drawer in the basement. You make electronics fun! I think I'll run downstairs and fire up the soldering iron.

"Ah, but you're getting to be an expert by now . "
Yeah, I've been doing this stuff since all you guys have been in diapers.

 

[alec_t]

I'm pretty sure there are some 330K resistors bouncing around in my electronics junk drawer

Well if you haven't got 330k anything from 100k to 1meg would do.

 

[salty joe]

Outstanding, I was just going to ask if 390k would be ok. Thanks, back downstairs.

 

[salty joe]

I installed two LEDs-one for each set of pumps. One LED is working as anticipated, can't tell you how thrilled I am. I expected the other LED to just barely blink every minute or so, but nothing. I set my DMM to 20VDC and held the probes to the LED wires and got nothing. I adjusted the flick pot to both extremes and watched the DMM for a couple minutes each time. Any chance I might get some help trouble shooting?

BTW, the LED that is blinking for the running pumps does not completely turn off between run times. It keeps a slight glow. It is on for maybe 30 sec and off 5 sec, rough guess.

 

[alec_t]

I expected the other LED to just barely blink every minute or so, but nothing

If the blink is too brief it may not be visible. I've just knocked up a breadboard version of this part of the circuit. With my 'C10' = 0.47u and my 'trimpot' = 330k my LED blinks on for ~ 0.2 sec.

I set my DMM to 20VDC and held the probes to the LED wires and got nothing

DMMs in general are rubbish at measuring briefly pulsed DC. I tried that on the breadboard and got a fluctuating reading between 0 and ~ 0.4V.

the LED that is blinking for the running pumps

During the run phase the LED should be on constantly for 30 secs (not blinking) then off for 5 secs. Or did you buy flashing LEDs?

does not completely turn off between run times. It keeps a slight glow

Strange. Could be due to FET or Schottky diode leakage current, or spurious FET turn-ons due to noise (circuit presently unscreened).

It is on for maybe 30 sec and off 5 sec

Hooray!

Diagnostics:
I assume each LED has its own current-limiting resistor and the combo is connected + side to +12V (or +24V) and - side to the respective FET drain (pump connection). Why does the LED in your pic have 4 legs?
Measure the voltage (20VDC range) at the left side of C10 (= Q4 out from 4020). It should toggle between 0 and 12V every 21 sec or so.
After taking anti-static precautions, temporarily short the bottom end of R6 to ground. Both LEDs should light (all 4 pumps on).

 

[salty joe]

DMMs in general are rubbish at measuring briefly pulsed DC
Oh, had no idea, can't complain for what it cost.

During the run phase the LED should be on constantly for 30 secs (not blinking) then off for 5 secs.
Yes, that's exactly what it's doing.

Hooray!
I was pretty darn excited.

I assume each LED has its own current-limiting resistor and the combo is connected + side to +12V (or +24V) and - side to the respective FET drain (pump connection).
Two pump leads are connected to a bare copper wire. On that same wire, I connected the "bar" side of D6 and the "butt" side of D5, and the drain from the mosfet. The other two pump leads for that pair are also connected to a bare wire along with the bar side of D5 and a 2.2K resistor for the LED. That wire then connects to 24v+. Those two wires are close together and parralell. Two 0.1 uf caps straddle the wires. I repeated for the other pair of pumps on a different place on the board.

Why does the LED in your pic have 4 legs?
It just looks funny in the pic. It's a regular old 2 legged LED.

Thanks for the diagnostics-tomorrowI'll take the readings.

I left the controller running and after nine hours, the same LED continues to operate. Adjusting the clock pot does not seem to matter.
Thanks.

 

[alec_t]

I left the controller running and after nine hours, the same LED continues to operate. Adjusting the clock pot does not seem to matter.

That indicates the 4020 isn't operating, and hence accounts for why there's no flick. Checking the Q4 output at the left side of C10 will confirm that. Check also that you're getting a clock signal (0/12V toggle every 2.6 secs) at the top end of R2 (= output of gate U2a). Then if that doesn't identify the problem it's a matter of double-checking all the pinouts and soldered connections of the 4020.

BTW the 'bar' side of a diode is the cathode; the 'butt' side is the anode.

 

[salty joe]

temporarily short the bottom end of R6 to ground. Both LEDs should light (all 4 pumps on).

That did light up both LEDs. However, I got nothing from Q4 (pin 7) on the 4020. Also got nothing from between Ua2 and R2.

It sounds like you have it narrowed down to the 4020. I'll very carefully check to make sure it's hooked up correctly as well as solder joints. Thanks, I did not have a clue where to start.

Cathode and anode it is.

 

[salty joe]

Well, it was the 4020 alright. It was not even close to being right. I must have gotten turned around. What a dope I am. ugh. Anyway, I can't see trying to salvage the mess, so I'll start over. Maybe even take enough time to get it right.

I'm pretty sure I have enough parts for another so no big deal. One thing I'll do differently is face all three ICs the same way. I'll also be more careful not to let the soldering iron touch wire insulation. On the plus side, I have learned how to apply quick little dot sized mounds of solder. I still need a little more consistancy and sometimes still use a little too much solder, but not at all scared to solder an IC. BTW, it was a really good call to have me use an IC socket last go round. Please bear with me as I go after it again.

 

[KeepItSimpleStupid]

I'd like to tell you a story that happened about 35 years ago. I had taped an audio amplifier design from Audio Magazine. Taping was the method before widespread computer use. I had the boards made and somehow the board manufacturer made a mirror image of my layout.

It took me a LONG time to figure that out and with a HIGH POWER audio amp, I blew up lots of components.

Finally when i did figure it out, I thought about the design and it appeared that i could salvage it. I was able to swap the NPN and PNP transistors except for the bias regulator. I cut 2 traces and swapped the connections and all was well. It was a very good lesson.

TIP:
When soldering, clean and "tin" the tip and wipe with a damp rag. "Tinning" is adding solder to the tip and wiping it off. For an IC/socket, wedge the pin against the hole and apply solder to the OPPOSITE side. The connection and not the soldering iron tip.

It's also best to clean the flux off the joints as soon as possible. e.g. Wash the board with soap and water and a scrub brush. Acetone (nail polish remover) or Denatured alcohol can also be used as a solvent. Some fluxes are water soluable.

You could also pretend to lay out the board by drawing the parts on a piece of vellum or paper and make the connections not as a schematic, but as a pictorial. They don't have to be to scale at this stage.

You could also lay out the board using a free PCB layout package, but that may be too much effort. it might make sense at a later time though and actually have the boards made.

 

[salty joe]

TIP:
It's also best to clean the flux off the joints as soon as possible. e.g. Wash the board with soap and water and a scrub brush. Acetone (nail polish remover) or Denatured alcohol can also be used as a solvent. Some fluxes are water soluable.

.

Are you saying to take the completed circuit to a sink and scrub it?

 

[KeepItSimpleStupid]

Yep, and I'll throw this at you: https://www.finishing.com/Library/flux.html

If the flux is left on, there is the possibility of corrosion. I'd do your final scrubbing with the IC's not in the sockets and then air dry out side in the sun. There was a recent discussion in a major test equipment group where boards coming in for repair were placed in a dishwasher.
Some components could be damaged by water such as transformers, LCD displays and buzzers. Buzzers may come with a sticker over top of them to keep out the flux solvents. High temperature will accelerate the corrosion. I've seen that too.

There are commercial products available: https://www.mouser.com/Search/Refine.aspx?Keyword=flux+remover&No=25&FS=True

I had to deal with boards where fingerprints were a no no and so was water vapor. The boards had to be dried with Dry Nitrogen if repaired.

 

[alec_t]

Joe, I can't remember if I mentioned it before, but here's a method I use when building a circuit on matrix or strip board:-
Print out the schematic.
Plan the component layout by sketching the components roughly to scale on squared paper.
As you solder up each component on the board, ink over the respective conductor path on the schematic as a record of progress; that way no joint should get missed.