Help with Water Pump

[salty joe]

Thanks guys.
I was told that unless these DC pumps have a soft start, they will fail prematurely. I am under the impression that hard start issues are confined to AC motors. Could I get your thoughts?

 

[4pyros]

Thanks guys.
I was told that unless these DC pumps have a soft start, they will fail prematurely. I am under the impression that hard start issues are confined to AC motors. Could I get your thoughts?

I will have think about that one, but its the first that I have hard of it.

 

[alec_t]

I was told that unless these DC pumps have a soft start, they will fail prematurely.

How reliable is your info source? Is that confirmed by the manufacturer?
If the info is correct then it would be advisable to re-design the circuit. So I'd delay building it for now.

 

[salty joe]

I was lurking on my reef forum and a guy who seems fairly knowledgable mentioned that Tunze and Ecotech, both controlled DC pumps similar to the one I'm using, use a soft start for protection. I'll find out if that's the case.

I do notice that the controller that came with these pumps has a repeating pattern that always starts with 6.5w then to 12.5w then 25w.

 

[salty joe]

I talked to a couple guys from Ecotech about wether or not their pumps have a soft start-they said no, they never heard of such a thing. Still waiting on Tunze's response, I have a feeling I'm good to go.

Would it be a good idea to install a fuse in this controller?

 

[alec_t]

Would it be a good idea to install a fuse in this controller?

I'm guessing you power supply already has one. But it would also be a good idea to have individual fuses for the pumps. Perhaps 10A each (without knowing your pump start-up characteristics it's hard to be more precise)?

 

[salty joe]

Here is the response I got from Tunze concerning soft start;

"The DC pumps when connected to our controllers start at 10% power and rapidly (milliseconds) step up to the programmed running power, this is a controller function however and not built into the pump. It would only be accessible if a controller is connected to the DIN connection and not at the power donnection"

I assume this prevents premature wear on the pump. The motors are similar in they are both DC. The Tunze is 18v, 28w. The manual said the motor is electronic and can be run on any DC source from 10v to 30v. The pump I'm using is 24v, 30w.

Do you think something like is needed to prevent premature wear? If so, can it be an add on to your schematic?

 

[alec_t]

Do you think something like is needed to prevent premature wear? If so, can it be an add on to your schematic?

I really can't judge whether premature wear is likely or not; it depends how the pumps have been engineered. If you're concerned about it, then a soft start featurecan be added to the schematic. The circuit I have in mind involves one additional IC and a handful of extra passive components; so the added cost shouldn't be great . I'lll try to run a simulation and post a schematic mod in the next day or so.

 

[salty joe]

Thanks. I'd rather be safe than sorry.

 

[()blivion]

:GERRRRR: I HATE THE Xbox 360.. so.. much... Anyway sorry I'm late, but I'm back now and I'm working on your layout again. Guna be a while before it's completely finished. I want to make it nice as I can after all.

You say the pots are not through hole units Salty? So is there no way you can get them on the board/into some holes? You can use wires from the pots to the board if you want. And mount the pots on your case with the provided nuts and lock washers. Just know this layout assumes through hole pots. It won't be hard for you to take the layout I give you and change it just a bit to fit your actual parts though, so it's no big deal. You just have to have a good idea of what goes where and how the connections are normally made. It should be a piece of cake honestly.

You said you wanted to salvage the old jacks for the pumps? I was going to have you drill holes and use nut's + bolts and ring eye or fork terminals... but we can do it differently if you want. The nut's and bolts would be exactly like the ones you would use for the transistors and regulator to hold them to a heat sink (those parts may have came with some actually). This method would involve drilling holes in your board. But it would make a more solid connection that can handle more current, though a bit more difficult to attach and detach your pumps and power supply on demand.

If those pumps use the same standard size M coaxial power connector that I think they do (https://en.wikipedia.org/wiki/Coaxial_power_connector), then reusing the female end is going to be a bit hard (relative to your skill level). They don't normally use through hole pins for those sockets. They use more like what the pots have than anything else. You would have to modify and widen some of the holes by hand with a hobby knife or something and make them odd little slots to get the sockets to fit on the board. Not to say they can't be recovered and reused... (QED), just that it's a little harder than what I had planed for you. Unless you don't own a drill or can't get the nut's + bolts + ring eye terminals. Then salvaging the jacks you already have might be easier.


Jacks *CAN* be easily reused. . .
View attachment 61089

Or you could use a setup like this. . .
View attachment 61091

Let me know what you think so I can change things if I need to.

 

[salty joe]

Hey ()blivion, hope you made out on that job. Yes, the pots I have will be mounted through the case & have wires soldered to the right spots on the board. I ordered some 2mm jacks, hope they are the right size. The picture looked right. It was under $4 to get them delivered, so since I thought it was the right part it was worth the risk. I ordered a few that get soldered to the board and a few that get mounted through the case. If they don't fit I'll salvage from the controller that came with the pumps. I'm thinking the ones mounted through the case will be easier. It looks like a tight fit on the proto board as is.

No shortage of hand tools around here. Not shy about using them either.

 

[salty joe]

I'm guessing you power supply already has one. But it would also be a good idea to have individual fuses for the pumps. Perhaps 10A each (without knowing your pump start-up characteristics it's hard to be more precise)?

I don't know if this sheds any light on what size fuse to protect the pumps but the controller that came with these pumps has a 3a 250v fuse as the first component hooked to the red wire coming from the transformer.

 

[alec_t]

Is the red wire on the primary (mains) or secondary side of the tranny? Whatever, individual fuses would enable 3 pumps to carry on running if one blew its fuse.
Attached is the soft-start mod. I've only shown one module; you would need a respective identical module for each pump. Note that the module now includes a ramp generator (R1, C9, D17) and comparator (quarter of an LM324). N.B. the LM324 is a quad opamp IC.
In addition there is one PWM oscillator shared by all four modules. The oscillator comprises one of the hitherto unused gates in your CD40106, runs at ~22kHz and its square-wave output is shaped by R3, C11 into an approximation of a triangle-wave.
From start-up, the pump current (even if the pump motor has a low inductance) ramps up from zero to full current (1.5A) over ~ 100mS. The ramp period is set by R1, C9.
Hope you can find room on your board for the extra components . You will need to accommodate :- 1 x LM324, 6 caps, 6 resistors, 4 x 1N4148 (or similar) diodes.

 

[()blivion]

Hey ()blivion, hope you made out on that job.

I wish I had, some times the RRoD just isn't fixable though. At this rate I'm basically doing it for charity. Oh well, I got my fingers in to other investments.

I was told that unless these DC pumps have a soft start, they will fail prematurely.

This is just my opinion (Completely biased opinion, I DON'T want to add more to this layout... LOL) But I don't think a soft start is going to be realistically necessary. Despite what you were told and what the manufacture said. This stuff's pretty tough.

Judging from the pictures of the pumps you linked us, these units are sealed inrunner brushless motors, with the controller installed in the pumps housing. The parts that would be the weakest and most likely to break in such a system under the stress of hard starting would be the internal FET's. But they are probably heatsinked to the case of the pump, making them effectively water cooled. If not then your pumps were not designed very well. Even so, with the FET's positive temperature coefficient, the hotter the internal controllers get, the less current they conduct, causing negative feedback and inherent stability... to a certain degree.

Now... if you REALLY just have to have some form of soft start... for the pumps sake if nothing else... then you *COULD* hook up big capacitors to the gates of the IRF3205 FET's. Then a (larger) resistor up stream of that. The delay will make the FET's switch slow enough to more or less make your soft start circuit, based on the RC time constant and the FET's threshold. (Go ahead and simulate it if you can Alec_t) This would take less parts and less soldering and assembling for a soft start than has been suggested so far. The only draw back is that our boards FET's will have to dissipate a little bit extra power. This is do to operating in linear mode for a longer time than the original circuit would. But that time will be short, not very often, and with such low power compared to the max power of the FET's that there is no worry of destruction at all. At less than 1/10 a second of linear operation time, with a cool down period of ~30 seconds or more where the FET's are either fully on or fully off, they probably will not even get warm. This is what I would at least try if I even cared about soft starting at all, since it's easy to do and takes few parts. I can add this as an option to the layout later if you want.

Speaking of layout... below is a preliminary final draft of the original pump circuit (post 99, I think). Note that it's not perfect yet, I'm sure there are mistakes in it. So were going to need some more eyes to look at it and find them before you commit to building the real thing. Also note that I used my own personal style on this when routing all the connections. This style may not fit your soldering ability. Don't worry, I do plan on making a more conventional layout eventually when all ideas and suggestions have settled down. Plus you do have two boards and more extra parts should you mess up *THAT* bad. The wire connections to the solder side are all color coded for easier navigation and identification, except the red main 24v line across the top, which is coded blue. Also, the connections/controls to pump 3 and 4 are swapped, a little confusing but it doesn't effect the way it works. *FINALLY*.....!!!TAKE SPECIAL NOTE OF WHICH WAY ALL THE PARTS ARE TURNED!!!. Get a FET, Diode, or the regulator backwards and . . . *POOF*. . . The magic smoke that makes it work gets set free, never to return.

View attachment 61124
IF (WHEN) YOU SEE MISTAKES, PLEASE TELL ME THANKS!

Usually when you build with point to point proto board you don't do a lot of running lines on the solder side of the board. But when I do it I like to make sudo-traces with bare copper wire, solder, and the excess wires that poke through from the parts I am adding. I like to think this makes things more sturdy since you reduce the number of actual wires going every which way, which are loose and can catch things. I also like to think this improves a bunch of other stuff like current, cooling, ground loops, and appearance. The problem with this way of constructing is it involves more soldering and the soldering of longer joints. Both these can cause problems for people that are less experienced at soldering. It's relatively easy IMO though, considering you could be soldering 48-pin TSOP I FLASH chips (My WRT54G). Or surface mount capacitors and resistors slightly larger than grains of salt (DS Light I just fixed). Or you could be reballing BGA Xbox 360 GPU's by hand (Take a guess who's doing this right now...?), All of which is MUCH HARDER than what I'm asking you to try doing, trust me.

 

[salty joe]

The controller that came with the pump has a wire that plugs into a transformer. The plug has a pin about as thick as pencil lead and a thicker blade. The fuse is on the pin side of the plug.

So, more than one way to get a soft start. I'm kinda tired right now to try and figure them out.

I like the idea of making solder runs with bare wire. I think I could do that. Think so anyway. I've always been pretty good at soldering and brazing big things. My high school shop teacher taught us how molten metal follows the heat but this is way way smaller scale. The layout makes it look like there's room to spare-that's encouraging.

Thanks guys I really appreciate the help!

 

[alec_t]

Following ()blivion's suggestion, here's a simpler soft-start circuit to replace the PWM version. Only one pump module is shown. Simulation shows that, assuming the pump is on for 30 secs then off for 30 secs, the FET dissipates an average 188mW. During turn-on, the pump current ramps up over ~100mS and in that time the average FET power dissipation is 19W. During turn-off the FET dissipation averages ~34W over the 6mS turn-off period. So power losses are a good bit more than for the previous version, but it is marginal whether a heatsink is needed.

Edit: Nice graphics ()blivion. Did you create all the component icons yourself, or use an existing library? Should the alphabet at the board edge be mirrored horizontally rather than vertically?

 

[()blivion]

here's a simpler soft-start circuit to replace the PWM version.

OH! looks good. You even improved my basic thought to have different on and off slopes. Nice work. I can add that to the layout easy enough I think. It looks like it does dissipate a little more power than I anticipated, but still looks well with in requirements.

Nice graphics ()blivion. Did you create all the component icons yourself, or use an existing library?

Thanks, yes I made them myself just for this project's layout based off parts I had laying around and an educated guesses as to what other parts look like. I did it using 100% M$ paint in Vista (Vista's paint is far better than Se7en's IMO.)

Should the alphabet at the board edge be mirrored horizontally rather than vertically?

I don't know, I made the top board by eye to match the picture of the E-bay auction as best as I could. Then I just used copy/paste then paint's "Flip vertically" function to make the bottom board. Are you saying it should maybe be different? I only did it this way because it reinforces the fact that the board is *FLIPPED*, Which is a really important fact.

 

[salty joe]

Simulation shows that, assuming the pump is on for 30 secs then off for 30 secs, the FET dissipates an average 188mW. During turn-on, the pump current ramps up over ~100mS and in that time the average FET power dissipation is 19W. During turn-off the FET dissipation averages ~34W over the 6mS turn-off period. So power losses are a good bit more than for the previous version, but it is marginal whether a heatsink is needed.

QUOTE]

The pumps will be on for about 30 sec then off for only 5-10 sec.

Instead of a fuse to protect the pumps, could a reset button be installed?

 

[alec_t]

The pumps will be on for about 30 sec then off for only 5-10 sec.

Ok. Doesn't affect the sim result significantly.

Instead of a fuse to protect the pumps, could a reset button be installed?

Not sure what you want to reset. Do you envisage some sort of electronic re-settable over-current trip? That could be done, but would complicate the circuit quite a bit if you wanted to build it in (as opposed to using a commercial external unit), especially if you had one trip per pump.

 

[4pyros]

Instead of a fuse to protect the pumps, could a reset button be installed?

Low voltage panel mount breakers. Something like these?
**broken link removed**