Help with Water Pump

[alec_t]

I guess I still don't see the problem.

I don't think there is one .
BTW a sim of the ronv charger in post #1513 shows that when the 20V supply fails the battery tries to drive >10A back through the supply via D1. We don't know if the supply can handle that. Looks like we'll need a power diode (D1 re-purposed?) between the LM317 output and the battery.

 

[ronv]

I think that is an artifact of the voltage source in the simulation since it will have a low resistance to zero volts. With the real supply, assuming its a switcher, the only thing on the output of the supply would be the filter caps in the supply. But we can check it.

 

[ronv]

Actually why don't we tie it to +pump? Then the diode that is already there would block it from feeding back and we wouldn't have 2 diode drops.

 

[alec_t]

Actually why don't we tie it to +pump?

We need to diode-OR the 20V and battery voltage. The diode already in the PDM could be part of the OR, but then it wouldn't be in a position to block the reverse current. Those ideal diodes that KISS posted a link to could be an alternative to the discrete power diodes already considered.

 

[alec_t]

Looking again at the charger in post #1513, the 1Ω current sense resistor is in the path of the battery charge current (as required) but is also in the path of its discharge current so will drop several volts when the battery powers the pumps. Here's a mod which has the resistor in the charge path only. It needs a second transistor (Q2) (unless you can see some other way of doing it, Ron?).


Edit: Joe, don't build this yet; it's a draft only.

 

[ronv]

Yea, I noticed that when I was drawing in the diodes. The other thing I noticed was the 7812 would be back biased and burn out during a power fail so I think we can just take it out.
Here is a simple version that I think solves all the problems so far.... Cost reduction

 

[KeepItSimpleStupid]

You can always put the recommended back diode from out to in on the 7812?

 

[ronv]

I was thinking that to, but since we don't know for sure what the power supply looks like we might blow the diode as well. The other possibility is to only power the 12 volts for the 2 pumps from the battery and leave the others on the regulator, but I'm not sure it is that modular. Since the +12 only pulls a few milliamps it might be best just to power all the 12 volt circuitry from the 317 regulator.

 

[KeepItSimpleStupid]

As the voltage drops when do we loose the turn-on threshold of Vgs on the FET?

At what voltage, (Speaking of the 12 V tiny supply only) would the FET, not turn on? i.e. If you had the 20V running and had a variable supply for the 12 V, when would the system not work?

 

[ronv]

That's a good question.
In the sims it looks like we still get 4 amps with a gate voltage of 5 volts, but I would guess it could vary a bit. Probably 4 volts would be the low end. The logic should run quite a bit lower so the failure mode should be the FET not turning on over a few ma. around 3 or 4 volts. Not sure where the pump will stall out either, but I quess as long as the logic runs the circuits will protect it although the alarm may sound funny.

 

[salty joe]

Not sure where the pump will stall out either, but I quess as long as the logic runs the circuits will protect it although the alarm may sound funny.

The pumps stall at just under 3.5V.
Thanks very much everyone!

 

[alec_t]

Re post #1726

Here is a simple version that I think solves all the problems so far.... Cost reduction

It's certainly simpler, but means higher power dissipation by the LM317. If, for example, the battery has discharged right down to 9V the LM317 will be dissipating nearly 9W at a charge current of ~0.85A. So the loss of Q1 and Q2 will probably mean a bigger heatsink will be needed, so not necessarily a cost reduction. Joe's choice. Also, there's no obvious protection from reverse current through the LM317 or the 20V supply. Here's a mod showing the connection to the two filter pump PDMs, and with D1 to block reverse current. According to the datasheet the reverse current from the 10uF cap discharging won't harm the LM317.

The other thing I noticed was the 7812 would be back biased and burn out during a power fail so I think we can just take it out.

...as in the version in post #1725

You can always put the recommended back diode from out to in on the 7812?

The effect of power failure was something I didn't take account of for the 7812(s) used in the Tide and Wave Controllers . Apologies, Joe. Those should be revised by connecting a reverse-biased diode between the input and output terminals of the/each 7812. If you have a spare 3A Schottky that will do nicely (although a bit OTT).

Since the +12 only pulls a few milliamps it might be best just to power all the 12 volt circuitry from the 317 regulator.

That's an option (which would avoid having to add diodes across the 7812s). Or Joe may prefer to keep the control of the other pumps entirely separate.

 

[salty joe]

Re post #1726
It's certainly simpler, but means higher power dissipation by the LM317. If, for example, the battery has discharged right down to 9V the LM317 will be dissipating nearly 9W at a charge current of ~0.85A. So the loss of Q1 and Q2 will probably mean a bigger heatsink will be needed,

A bigger heatsink means a greater percentage of energy from the battery is converted to heat. It seems like that must be true anyway. So, if there is a choice here and keeping a couple transistors makes the backup system more efficient, I’m all for it unless it makes everything way more complicated.

The effect of power failure was something I didn't take account of for the 7812(s) used in the Tide and Wave Controllers . Apologies, Joe. Those should be revised by connecting a reverse-biased diode between the input and output terminals of the/each 7812. If you have a spare 3A Schottky that will do nicely (although a bit OTT).
That's an option (which would avoid having to add diodes across the 7812s). Or Joe may prefer to keep the control of the other pumps entirely separate.

I planned to be able to control the two pumps controlled by the wave, the four pumps controlled by tide and the two pumps controlled by speed/battery backup independently. So, three independent controls handling eight pumps. As to the best way to get there, I think the best thing I can do is defer to you guys.
Many thanks.

 

[ronv]

Re post #1726

It's certainly simpler, but means higher power dissipation by the LM317. If, for example, the battery has discharged right down to 9V the LM317 will be dissipating nearly 9W at a charge current of ~0.85A. So the loss of Q1 and Q2 will probably mean a bigger heatsink will be needed, so not necessarily a cost reduction. Joe's choice.

I thought Joe had some heatsinks for the FETs that are pretty big. The 317 is thermally protected but if the power is to high for his heatsinks we can increase the .22 ohm resistor. I chose it because I figured Joe might have a leftover.

Also, there's no obvious protection from reverse current through the LM317 or the 20V supply.

I think where the input to the 317 was tied before - on the cathode side of the existing diode protects it.

Here's a mod showing the connection to the two filter pump PDMs, and with D1 to block reverse currentAccording to the datasheet the reverse current from the 10uF cap discharging won't harm the LM317..

I don't think D1 is required if the input to the 317 is moved back to the other side of the PDM diode

The effect of power failure was something I didn't take account of for the 7812(s) used in the Tide and Wave Controllers . Apologies, Joe. Those should be revised by connecting a reverse-biased diode between the input and output terminals of the/each 7812. If you have a spare 3A Schottky that will do nicely

Under normal power down it shouldn't be a problem. The problem is when we have the battery the output is held positive with respect to the input. I don't think the diode would help in this case because the battery is on the other side of the regulator.

That's an option (which would avoid having to add diodes across the 7812s). Or Joe may prefer to keep the control of the other pumps entirely separate.

I guess I didn't know there were multiple regulators. If so having just the 2 pumps on the battery would save a little power for the battery. Joe, I know it is kind of modular. Is it easy to isolate the 12 volts for the filter pumps from all the other PDMs?

Sorry, I didn't show the 2 pumps. So yes, we would need 2 diodes from the battery to each one.

**broken link removed**

 

[ronv]

Hi Joe,
I was busy typing so I missed your post.

The heat is created when charging the battery, so it's running up your electic bill, but not draining the battery.
Do you have a link or part number for the heatsinks you used on the FET and big diodes? I was hoping to use the same one for the regulator.

If it is easy to isolate the other pumps +12 from the filter pump +12 we could do that. Otherwise it is probably easiest (at this stage of the game) to just run all the low power logic from the battery when the power fails.

 

[salty joe]

Early on, I put a 7812 reg in the wave controller and another 7812 in the tide controller. There was concern that they would be in conflict with each other, so I disconnectedthe 7812 from the wave controller. The 12V for everthing flows from the single 7812.

Glad the battery is not running a heater! Here's the heatsink. https://stores.ebay.com/Tayda2009/_i.html?_nkw=heatsink&submit=Search&_sid=872591586

I only want to run the two pumps on speed control with battery backup, hopefully on a slow speed to buy more time.

 

[ronv]

Early on, I put a 7812 reg in the wave controller and another 7812 in the tide controller. There was concern that they would be in conflict with each other, so I disconnectedthe 7812 from the wave controller. The 12V for everthing flows from the single 7812.

So unless the filter pumps are on a separate board it would not be easy I guess.

Glad the battery is not running a heater! Here's the heatsink. https://stores.ebay.com/Tayda2009/_i.html?_nkw=heatsink&submit=Search&_sid=872591586

Hmm. Little smaller than I thought. I'm suprised the FETs don't get hot when the pump stalls. Alec, Do you think they are ok? So lets make the 0.22 ohm either 0.44 or 0.5 if you have one. You can use 2, 0.22 in series or buy a .5 ohm 1 watt one I guess.

I only want to run the two pumps on speed control with battery backup, hopefully on a slow speed to buy more time.

Yes, The logic probably draws less that 0.025 amps so it's not a big deal to run all the logic. The 2 pumps running on the battery will only draw about 1.5 amps so your 70AH battery should be good.

 

[KeepItSimpleStupid]

Yes, The logic probably draws less that 0.025 amps so it's not a big deal to run all the logic. The 2 pumps running on the battery will only draw about 1.5 amps so your 70AH battery should be good.

The buzzer will be the greedy device. I think it was around 100 mA @ 12 V.

 

[alec_t]

The only snag I see with running all the 12V logic from the battery is that if the battery died, or for some other reason wasn't available, then there would be no water movement at all; which the coral might not like (but I know zilch about coral, so Joe can advise if static water would be a problem).
So how about let the two filter pumps get their '12'V from the battery; let the other pumps get theirs from a single 7812 (part of either the wave or tide controller)?

 

[ronv]

The only snag I see with running all the 12V logic from the battery is that if the battery died, or for some other reason wasn't available, then there would be no water movement at all; which the coral might not like (but I know zilch about coral, so Joe can advise if static water would be a problem).

With the battery out the voltage from the regulator will be set at 13.6 volts. So it there is 20 volts there should be 12 volts (13.6) as well.

So how about let the two filter pumps get their '12'V from the battery; let the other pumps get theirs from a single 7812 (part of either the wave or tide controller)?

This seems the best unless the 12 volt lines are used all over the place and not isolated by controller type.