Help with Water Pump

[KeepItSimpleStupid]

alec:

The LT1074 is good for 5 Amps. Do you really think it could handle 2 pumps STARTING at the same time?

The other issue is your "FEED switch" which really doesn't look like a switch in the diagram. The DOT is hard to see.

Might we want both voltages to be adjustable somewhat?

60 uH is close to 68 uH, so you expecting that to be salvaged from the OEM controller. Lower L's are better to assure oscillation from a design point of view. But we aren't sure what the current rating of the inductors are in the OEM controller.

The method chosen for changing the feed voltage might not be compatible with using transistors as as the switching elements. e.g. With feed, it makes more sense, I think, to drive that with a timer. i.e. push button, feed fish, walk away, You have a timed feed mode and it automatically recovers. You might need to think ahead a little bit and think what might happen when an automatic feeder is introduced, not saying that it will be.

(alarm) The alarm needs a tiny bit of help. You need to include a switch labeled Alarm and with labels ON and Silenced. A LED should always indicate the alarm state independent of a silenced buzzer.

(What pump?) You did say that the controller would know what pump has trouble. This is missing. The ULN2004 would work nicely here.
ALARM, PUMP PAIR 1 (at U3c), PUMP PAIR 2 (at U3d), ALARM pump #1, ALARM pump #2, ALARM pump #3, ALARM pump #4.
Yuk, unfortunately an extra pin isn't available for a LOGIC probe. There is an 8 output version of the chip, the ULN2804.

I'd put a 3 pin header at pump 1, pump 2 , pump 3 and pump 4 which will allow disconnection of the drivers. e.g.
OFF ON using 3 pins * * *. The left most is on, the right most is off. There is no connection to the left pin. The right most pin just goes in series with the output. You may not need this if there is an easy way of disconnecting the pump. Suppose, you determine that the pumps need cleaning, but you don't have time to fix it. So, you just disable the pump by moving the jumper.

Comments on the motor driver circuit.
The terminals used for the current sense resistor (0.22 ohm resistor ?) should be brought out to a terminal block where it is easy to probe and so should the voltage to the motor be easily probe-able. These are key troubleshooting points. Assuming the voltage is constant and you measured the voltage across the 0.22 ohm resistor which is proportional to motor current, you have a base line. Bearing friction would show up as more current.

This is also an argument for the jumper. You could re-route signals to turn on a motor with a wire jumper which is another key troubleshooting point. You could even use a 6x2 header with positions of (ON OFF Automatic) for the U3 jumper.

A power/logic ground connection is not shown on the schematic. It doesn't make sense to ground the low voltage side of the system in a marine environment. It does make sense to protective ground any metal cases though.


Salty:
The max output of the regulator will be ~two volts less than the input voltage. So, 20 V hopefully will result in about 18 V.

Alec:
You did real good, but were're not out of the woods yet. I may have a source for the inductors if the ones from the OEM supply are not suitable. 5A ones may be harder to find than 3Amp ones. Who knows.

Your entitled to take the ideas as a grain of salt, but later I could be saying "I told you so".

Remember I think in terms of modular, maintainable and serviceable. So, again thinking serviceable again, one can think about using links to BYPASS/INSERT the secondary regulator or it could be a separate board altogether and you just pull the board and add a jumper.
Don't mind me, I always have stuff to say. Alec, you turned this around really quick.

 

[salty joe]

Alec pulled a fast one on you.

.

Yeah...you gotta watch him like a hawk.

Thanks for clearing up the 2meg2. I was kinda worried it was some kind of exotic creature.
Thanks guys. I hope to find a little time tonight to order parts. Can't wait to bring your creation to life, Alec!

 

[alec_t]

The LT1074 is good for 5 Amps. Do you really think it could handle 2 pumps STARTING at the same time?

Hmm, not sure. Since the PDM limits current to 3A the start-up load would be 6A total. I was hoping the 1074 would handle that ok for ~.4 sec (haven't checked what its internal current limit is) so that one power chip could be shared. After start-up the current drops to ~2A total (well within its spec). Any suggestions?

The other issue is your "FEED switch" which really doesn't look like a switch in the diagram. The DOT is hard to see.

It looks like a switch in the .gif. I guess you're looking at the .asc schematic and don't have 'orthogonal snap wires' deselected in the LTSpice drafting options?

Might we want both voltages to be adjustable somewhat?

Joe, do you want both pumps of a pair to run at the same speed (be it fast or slow), or to have individual fast/slow speed selection?

But we aren't sure what the current rating of the inductors are in the OEM controller.

Well we know it's ok for 1 pump; so that may be an argument for just using one modded OEM controller per pump as power module rather than going for the LT1074 shared power IC.

You might need to think ahead a little bit and think what might happen when an automatic feeder is introduced, not saying that it will be.

And there's me trying to keep things simple ...hence the manual switch. Joe, do you want an auto-timed feed option?

A LED should always indicate the alarm state independent of a silenced buzzer.

There is one alarm LED per pump, which will be on more or less continuously if a respective pump rotor sticks and the circuit trips cyclically.

You did say that the controller would know what pump has trouble. This is missing.

Don't think so; see above.

I'd put a 3 pin header at pump 1, pump 2 , pump 3 and pump 4 which will allow disconnection of the drivers.

Good idea....even more modular!

The terminals used for the current sense resistor (0.22 ohm resistor ?) should be brought out to a terminal block where it is easy to probe

Another good idea.

5A ones [inductors] may be harder to find than 3Amp ones.

Probably. So go with 2 x modded OEM controllers to provide the 18/12(?)V supply for a pump pair?

A power/logic ground connection is not shown on the schematic.

Not sure where the best point for that would be, given that we've still got a choice to make re the power supply.

 

[KeepItSimpleStupid]

I need better glasses. Didn't see D3 to D6 as LED's, so I missed that.

The GIF just doesn't have the expected dots on the arm of the switch and where it would supposedly land. The arrowhead is also missing on the switch arm. Just odd, that I almost missed it as a switch.

 

[alec_t]

Understood re the switch dots. For speed/simplicity when using LTSpice I just draw a diagonal line, rather than bother with a switch model. Agreed, it doesn't look as pretty.

 

[()blivion]

Joe. May want to snag this for your alarm...

**broken link removed**

Ends in two days. Cheep and LOUD!

 

[()blivion]

How about THIS for the alarm circuit instead?

Edit: Some browsers HATE the Falstad simulator, mostly IE. Use Fire Fox or Google Chrome.
And make sure you have updated Java and enabled it in your browser.

 

[KeepItSimpleStupid]

ob():
The third switch doesn't work an no silence switch.

 

[salty joe]

Thanks all, for the input.
Alec, both pumps of a pair running at the same speed would be just right. A manual switch for feeding mode (slow pumps down) would be great. Could it be set up to push a button and the pumps slow down for 10-15 minutes, then pumps automaticly return to normal speed?


Thanks ()blivion for the link. That is a heavy duty piezo. While not nearly as gnarly, I feel certain I would hear these. **broken link removed** That's if they will work in the setup.

 

[()blivion]

Thanks ()blivion for the link. That is a heavy duty piezo. While not nearly as gnarly, I feel certain I would hear these. **broken link removed** That's if they will work in the setup.

Oh, I had seen those but I didn't realize they had the internal circuit necessary to make them buzz (more of a beep/screech really). Most of the units that look like that don't have such things. The other one does for sure, but if those do too then they probably are a better fit for you.

In any case... Good times.
-()blivion

 

[ronv]

Here is a 6 amp switcher, but it would probably want to be built on a pcb.

Here is also a link to a $12 9 volt supply. I have some big P fets I would donate for a switch if you want to go that way.

https://www.allelectronics.com/make-a-store/category/480/Power-Supplies/1.html

Auto feeder. Hmm. How about a peltier to keep the brine shrimp frozen?

 

[ronv]

Consider turning the pumps off during feeding?

 

[KeepItSimpleStupid]

And stress the coral? They have to eat too.

 

[ronv]

Ideally, the environment in which aquarium corals are kept should be as close to those on the reef as possible. A number of coral aquarists have found that feeding their corals at night with the assistance of an automatic feeder and running their mechanical filters and skimmers through timers which turn them off during coral feeding hours has provided good results.

 

[alec_t]

@ronv
The LM3150 looks a better bet than the LT chip. I haven't tried Webench; do the sims run at a decent speed?
Love the idea of the Peltier. LOL.

 

[ronv]

@ronv

The LM3150 looks a better bet than the LT chip. I haven't tried Webench; do the sims run at a decent speed?
Love the idea of the Peltier. LOL.

Webbench is slick. You just plug in the voltage and current and it kicks out a design and a bill of material. You can pick the cheapes, smallest, or most effecient design. The sims are pretty fast. The only downside is that the bill of material is always surface mount.

Joe,
What kind of lighting are you using?

 

[salty joe]

by Ronv, "Here is also a link to a $12 9 volt supply. I have some big P fets I would donate for a switch if you want to go that way.

**broken link removed**

Auto feeder. Hmm. How about a peltier to keep the brine shrimp frozen?"

Thank you Ron for the generous offer of switch parts. Peltier is my new word for today. Amazing stuff.

by ()blivion,"Oh, I had seen those but I didn't realize they had the internal circuit necessary to make them buzz (more of a beep/screech really). Most of the units that look like that don't have such things. The other one does for sure, but if those do too then they probably are a better fit for you."

**broken link removed**
It says an internal circuit and supply voltage is all that's needed to activate these little piezos. Are they a good fit?

Ideally, the environment in which aquarium corals are kept should be as close to those on the reef as possible. A number of coral aquarists have found that feeding their corals at night with the assistance of an automatic feeder and running their mechanical filters and skimmers through timers which turn them off during coral feeding hours has provided good results.

I had not heard about that very cool idea. It makes sense. I'll be relying on an algae filter-no skimmer, no mechanical filtration-so there will always be bits and pieces floating around. So for my tank, I don't think it justifies adding another layer of complexity.

@ronv



Joe,
What kind of lighting are you using?

I have two commercial 48"x48" bubble skylights mounted on the south side of the house directly over the tank. I'm considering using several 10W LEDs with lenses to spotlight any places that need an extra punch.

If the feeder mode is a hassle at all, the tank will be fine without it. It would be a nice feature, but it's not critical.

Thank you, one and all.

 

[()blivion]

**broken link removed**
It says an internal circuit and supply voltage is all that's needed to activate these little piezos. Are they a good fit?

Hummm..... after close inspection, It says...

"internal circuit,(stop) supply voltage is all that is needed"

Which I would think means...

"it has and internal circuit,(stop) so all you need is to supply voltage"

Having an internal circuit at that size is slightly unusual, but not impractical. If it's not marketing, then it should work quite well in reality... I would go for it. And if there really is no internal circuit, we can make a simple circuit ourselves that drives it. A two transistor astable multivibrator for example is a piece of cake to make and should work well. As for the ebay auction it's self... It REALLY should have been worded better, but what do you expect from the little people of Topeka, Kansas?

 

[alec_t]

If the feeder mode is a hassle at all, the tank will be fine without it. It would be a nice feature, but it's not critical.

It's certainly doable, but it does involve some way of changing the voltage supplied to the pumps....and that's what's now the major sticking-point in the project. Without the feed mode I think you could just power the pumps from your existing supply (turned down to its minimum setting, although 18V would have been preferred). Introducing the feed mode means building at least one further supply (e.g. by modding two of the OEM controllers). If you want to complete this project before Christmas (201x?) I'd personally do without feed mode for now, but ensure the modular approach (even without KISS's nice fancy DIN rails etc) is used so that it will be straightforward to add the voltage-switchable supply feature at a later date. Your call, Joe.
Just to tantalize y'all, I've been exploring the possibility of varying pump speed by effectively delaying the turn-on of each current pulse which the Hall device/FET in the pump supplies to the motor coils. Not the normal PWM and doesn't conflict with the motor's normal internal switching (as far as I can tell from simulations). Watch this space.

 

[KeepItSimpleStupid]

FEED MODE:
Idea 1: A "logic 1" turns on FEED mode

Idea #2: Uses two control lines. A "0" is FEED MODE and a "1" is Normal Speed. This allows the use of a transistor to parallel another resistor to the lower resistor of the voltage feedback divider, thus increasing the voltage. This is how the OEM controller does it, but it uses 2 bits for 3 speeds.

Feed mode could consist of buck (lowers voltage) DC-DC converters for each pump (3A) or each pair of pumps (6A). Surge currents of the OEM supply allow up to 5 Amps. Without a measurement or trying, we don't know what's required.
A scope (preferably storage) or a Peak-Hold meter is required. A hall effect current transducer may be required too.

To change Alec's last design requires an additional 1 transistor and 1 resistor and changing the values of possibly 2 resistors, so it's no big deal to be able to accommodate.

Having independently adjustable, rather than fixed, speeds for FEED and normal makes things even messier.
Way #1: Uses a small SPDT relay to select one of 2 dividers. The NC position would be the normal mode.
Probably would use idea #1.

Way #2: Uses two transistors to select the dividers. Requires some additional logic to ensure only one transistor is on at a time. Might use either idea #1 or idea #2 depending on the design.

So, we kinda have to figure out what normal and the feed mode voltages are. One would believe that having rocks, corals, plants etc in a tank vs. an empty one would necessitate independent (2 pump) speed controls.

A DPDT relay could be used for each pair of pumps as well, cutting cost, but reducing modularity. There is a potential disadvantage to the relay method. During the time the relay switches state (10-20 ms) we don't know what the DC-DC converter will do. Will there be a spike or a reduction in voltage?

Natural lighting: Cool!