Help with Water Pump

[salty joe]

How does the OP amp trick work?

 

[KeepItSimpleStupid]

Hi Joe:

A quick search found an expensive part (about $6.00), an LT1152 https://www.electro-tech-online.com/custompdfs/2013/04/lt1152.pdf. Note the circuit on PDF page 1, a rail to rail buffer. Alec says the input Z is about 300K, so this will go rail to rail. With any IC, linear or digital bypass caps are installed from the power pin to ground. So you would need a 0.1 uF polyester cap from +5 to ground.

Assuming, the wiper will go from 0-5V with it disconnected, the wiper would go to the input of this OP amp wired as in the diagram. The output would go to to where the wiper was.

The OP amp which stands for (Operational Amplifier) has a very high input impeadace, so it doesn;t load the POT. It only draws about 200 pico amps from it's input. That's like 200 e-12 amps. It also has a low output impedance, so you don't get the divider effect from say 50K and 300K. It's more like <1 ohm and 300K.

If you can't get 0-5V with the wiper disconnected, then that design won't work. If you can't get 0-5V, then we can buffer the input to the POT. So, do the test first.

If you wanted to get smaller, you could use an SMT package and a small SMT breadboard. We could probably find a cheaper part. The one randomly picked out is "chopper stabilized, zero drift" which is more expensive.

 

[salty joe]

The highest V was 4.9, with the wiper disconnected on a 50K pot.

 

[alec_t]

Actually an opamp is a bit OTT. A PNP buffer transistor in emitter-follower mode will do the trick. Here's an option using bits you probably already have in your 'spares' box:


Talking of spares, if you have a spare resettable fuse as used in the PDMs I would suggest using one between the 24V supply and the input of the LM317 in the back-up system, to protect the 24V supply in the event that the LM317 failed short-circuit.

 

[salty joe]

I do have those parts. Only three parts plus a trimmer....wow, that's cool! I'll put it together and let you know.
I also have an extra resettable fuse for the backup. Thanks.

 

[alec_t]

Just thinking it might be useful to have a warning LED on the backup system to show when it's running on battery (though the absence of mains power would be pretty obvious anyway ). You could add it like this:

 

[salty joe]

Alec, I don't know what that arrangement is called, but it is pure good. I could tell the pump was cranking-it had 5.09V, so I turned the trimmer down a little. I ran it down until the pump stalled, then turned it back up a little right away. Every mode is adjustable-it's beatiful!

It's rough looking but if I trim the mini board, I think I can tuck it inside the controller box and mount the trimmer on the side. It works like a champ-and only three more parts. Hard to believe.

I like the red LED for the backup. It would tell me if a GFI tripped. Thanks a million!

 

[alec_t]

Alec, I don't know what that arrangement is called

It's an emitter-follower with a diode to lower its output by ~0.6V.
That was a quick bit of building! Glad it works as predicted. The 5.09V won't harm the pump electronics, but can't boost the speed more than the 5V max . The 5.09 arises because the base-emitter voltage drop of the 2N3906 is slightly greater than the voltage drop across the diode. V_at_pump = V_at_controller + Vbe - Vdiode.
Looks like you can shrink your circuit board quite a bit, so should fit in the controller box.

 

[salty joe]

I trimmed the mini board down to micro board and snuck it in the controller box. Little trimmer is fastened to its side. Works very well-thanks.
The rest of the pumps should arrive late this month. If two pumps on the wave controller (synchronised opposition) is not quite enough, could the controller handle running four pumps in wave mode? Or would that fry the Vreg or something?

I decided I like super glue for electronic work. The heatsinks I have bend the mosfets with the smallest bump, so I glued the heatsinks to the board. I drilled one of the panel mount jack plugs out of round, so I installed the plug and made up the difference with super glue. With all the wires and such, the board could not be laid flat or clamped so... This is a little rough, but a third wire broke on the controller board and I got it too hot while repairing and the whole mess lifted up. I took the wire to the next connection, and was quick this time. Then I super glued the wire where I cooked the board. Then I hurried up and put everything back inside the controller box. Looks just fine now.

This site is really acting up for me tonight.

 

[KeepItSimpleStupid]

Superglue decomposes into nasty things when heated.

Anyway, a specialized version of superglue is used for wire tacking, just like you did.

So, have you used it to glue wounds on the skin together? Again, there is a medical version of it.

 

[salty joe]

Superglue decomposes into nasty things when heated.
.

So do mosfets.

I use superglue for cuts all the time. My fingertips split open from dry skin or just because. My fingertip is glued shut right now!

 

[alec_t]

If two pumps on the wave controller (synchronised opposition) is not quite enough, could the controller handle running four pumps in wave mode? Or would that fry the Vreg or something?

No probs. The Vreg won't even get warm. Here's a revised wave controller schematic. Note that all inverters in the CD40106 are now used. U1c/d provide respective 'Vt' signals for the pair of PDMs driving the OEM controllers of one synchronous pump pair; U1e/f provide the 'Vt' signals for the other pair.

 

[salty joe]

Thanks. We'll see what these pumps can do in a couple weeks. I have a feeling I'll be using the setup you just posted.

 

[alec_t]

Just be careful you don't create a tsunami with all that new pump power .

 

[salty joe]

Hi all,
It's been awhile, I hope I can talk you into seeing me through to the finish. I just got a bunch of these high powered pumps. A fellow reefer down south was kind enough to let me in on a group buy. He purchased directly from the factory in China and passed the savings along. What a guy!

I am working on a couple more PDMs and am waiting on a CD40106 to run the wave in pump pairs. I thought I had a couple CD40106 laying around. Rats.

I'll get the PDMs done and hook things up and touch base as soon as I can.
Thanks very much.

 

[alec_t]

Hi Joe. Hope the build goes well. Keep us posted.

 

[salty joe]

Thanks Alec.
I just finished the PDM pair for the two circulation pumps that will run non-stop. I'll be using a pair of the original pumps driven at 21V. I found if the Vt line is disconnected, the pumps run non-stop. Is this the right way to make the pumps run non-stop?

 

[KeepItSimpleStupid]

Vt should be tied to +12 for speed control to work, otherwise it's an OK method.

 

[alec_t]

otherwise it's an OK method

Not really. Vt should NOT be left floating or the FET is likely to oscillate. For continuous running, with or without speed control, it should be tied to +12V. For timed running it should be driven by the wave/tide timer.

 

[KeepItSimpleStupid]

I can see oscillating as a possibility. The FET must be turning on via the leakage currents of D4 and D5 when Vt is open. Thanks Alec.