Help with Water Pump

[KeepItSimpleStupid]

@ronv
The major reason to change the caps is the Working Voltage is only 25 Volts.

The reverse biased diode on the motor is to protect the power supply. The other diode is again to protect the supply.

Take a look here: https://www.electro-tech-online.com/custompdfs/2012/06/lm317.pdf on PDF page 7 and the note for D1. Just not sure if the same thing is applicable to switching supplies.

I'm still munching on the avalanche breakdown.

 

[ronv]

@ronv

The major reason to change the caps is the Working Voltage is only 25 Volts.

The reverse biased diode on the motor is to protect the power supply. The other diode is again to protect the supply.

Take a look here: https://www.electro-tech-online.com/custompdfs/2012/06/lm317.pdf on PDF page 7 and the note for D1. Just not sure if the same thing is applicable to switching supplies.

I'm still munching on the avalanche breakdown.

I just hate to waste money and time. You and Alec should decide.

 

[KeepItSimpleStupid]

OK, the SHK capacitors https://www.electro-tech-online.com/custompdfs/2012/06/SHK.pdf are rated for high ripple low ESR which is a good thing. 105 C is a good thing too.

So, if a 25 Working Voltage DC is acceptable for 22.5 V supply, we are supposedly good, but I never like operating near the edge. So, what I would do is upgrade them to a higher working voltage if they fail.

 

[salty joe]

I'll second that, but I'm still betting we can get the number of posts over 1000.

But the last 276 posts will be all about what an incredible job you guys did, right?

@Joe
I sent you a surprise envelope today. There was one part that I really wanted to enclose, but I could not find one. I also sent you a weird male and female header. They can be cut. They might be too high, but they illustrate what can be done. Picture mounting putting one or two female headers where an IC is and then the male header facing down from the bottom of the daughter card. There are some lengths of wire wrap wire (insulated and uninsulated) and some odds and ends.

Wow, thanks I can hardly wait.

 

[KeepItSimpleStupid]

I sent you one of each of these headers:

https://www.sparkfun.com/products/115
https://www.sparkfun.com/products/116

 

[()blivion]

I do dislike the 25 volt max caps for a 24 volt supply, very close. But I also see ronv's side, the manufactures were the ones that installed the caps after all.

So then, it's settled....... () Wait?


Now, back to watching Japanese cartoons.

 

[alec_t]

Only one thing bothers me now. That is if the pump will reliably start at 10 or 11 volts.

In normal use it would be running at ~20V. The 10V would be selected only during 'feed mode', with the pumps already running.

On the main board, there needs to be mods for the two diodes and the caps need upgrading. The snubber looks missing too.

Agreed we want some protection. The current-sense resistor also needs to be on the main board.

The main board needs the provisions of either variable voltage outputs or two fixed values.

The provision is already there, using T2 or T3 (whichever has its collector connected to the 2k2 resistor).

It would be nice if a NOT FEED signal was on the daughter card

Since all pumps would (probably?) be operating at the same speed I was considering the TAM module proposed earlier would provide a common 'not feed' signal to all pump modules via a feed-through on the daughter boards, and would also receive alarm input signals via those boards.

This daughter card didn't have a PTC thermister, a pump LED, or alarm LED.

The LEDs would be in the TAM. I think the PTC resettable fuse (if used) would be better placed on the OEM board near the jack.

The daughtercard doesn't seem all that bad.

Thank you (I think!)

It may be possible to use a SIP header on one of the existing IC's to mount the board with or without cutting all of the traces. You could also route the signals through one of the DIP locations too.

....providing the OEM signals didn't clash with the daughterboard signals.

Personally, I'm still worried about the pump.

Personally, I think modding the pump itself is a step too far: we should be aiming just to provide a reliable driver with current-limiting and voltage spike supression. That can be done by modding the OEM controller as above, but my own preference would be to use the post #716 circuit (with protection diodes as necessary) to control an IRF3205 (which Joe already has) instead. The down-side would be losing the thermal-shutdown feature of the 2576 (but a resettable fuse would remedy that to some extent). On the plus side would be simplicity, not having to hack the OEM controllers, and a less bulky overall system.

 

[4pyros]

Personally, I'm still worried about the pump.
Personally, I think modding the pump itself is a step too far: we should be aiming just to provide a reliable driver with current-limiting and voltage spike supression. That can be done by modding the OEM controller as above, but my own preference would be to use the post #716 circuit (with protection diodes as necessary) to control an IRF3205 (which Joe already has) instead. The down-side would be losing the thermal-shutdown feature of the 2576 (but a resettable fuse would remedy that to some extent). On the plus side would be simplicity, not having to hack the OEM controllers, and a less bulky overall system.

The pump should be protected by the Hall IC.

 

[KeepItSimpleStupid]

It may be possible to use a SIP header on one of the existing IC's to mount the board with or without cutting all of the traces. You could also route the signals through one of the DIP locations too.

....providing the OEM signals didn't clash with the daughterboard signals.

I was basically referring to using the headers listed in post #725 or similar.

Way #1 (A standoff without the standoff - probably not practical)
1) IC removed and header installed.
2) No signals routed through header

Lower profile headers or DIP socket could also be used.

Way #2 (Use the IC pads)
1) Remove an IC
2) Cut some or all signals going to pads. Some pads in the OEM controller aren't used.
3) Route the unused or cut pads to the header and to the daughtercard.
4) Use Wire Wrap wire to attach the pads to appropriate points on the main board.
(e.g. Shutdown signal, power, sig ground, HIGH SPEED, Sense resistor ends)
Note HIGH SPEED is the signal required on the main board. What turns on T1 or T2 not NOT FEED.
(I'll have to figure out what the base is connected to)

Way #3 (standoffs)
1) Board mounted on a standoff
2) A SIP polarized connector mounted on daughtercard
3) Sever connections to pads on one side of an IC chip.
4) Place a SIP polarized header on one side of the socket (if 8 or less connections)
5) Connect daughter card using female to female jumper

Way 4 (one side soldered in place)
1) Use a ribbon with pins on one side and pins on the other. (I can detail that technique)
2) Stipulation: Daughtercard connector must be inline with main board connector.

The viability of some of these ideas depend on the use/non-use of the OEM case. Note the board isn't square.
Then there is the issue of getting the control signals to the daughter card.

Yep, construction issues. I do recommend gold flash connectors in any area with significant humidity. Been there, done that. But it doesn't mean it's a must.

Thinking ahead, again, the OEM boards with daughter cards could be placed into a card cage of sorts with a pigtail connector going to the controller.

For the benefit of Joe, this is a card cage: https://www.electro-tech-online.com/custompdfs/2012/06/New20Page2044.pdf

Usually the boards slide into guides and a EDGE connector is used to connect to a backplane. At this point, I'm not saying yes or no to this method.

I did construct one piece of custom instrumentation this way and it turned out really nice. There were a couple of commercial lock-in amplifier cards and one or two custom cards and a fold down front panel. I used vector hardware to do this.

Again, I'm not suggesting it, but the NIM or Nuclear Instrumentaion Module is a High level version of a card cage. See: https://www.electro-tech-online.com/custompdfs/2012/06/Catalog_Wiener_2010.pdf The bins are part standardized and part not. Certain signals are common to all modules, but then there is an area where you an use your own connectors. Yea, a nice expensive system.

Just trying to suggest some ideas such as getting some inexpensive card guides (the plastic pieces) and constructing a cage of sorts with plenty of room for convection cooling of the OEM+daughter cards, but instead of having edge connectors, a pigtail with a connector would be provided.

Remember, these are just ideas.

@alec, @joe
Would the standard voltages of the OEM controller of 22.5 and 10.x be OK, is justifiability necessary or can two other fixed voltages be selected by changing two resistors?

Will Feed really be when the pumps are operating? Is feeding time within a pump cycle? I was assuming that the pumps would have to start at the lower voltage as they do with the OEM controller.

 

[ronv]

In normal use it would be running at ~20V. The 10V would be selected only during 'feed mode', with the pumps already running.

I was thinking about a normal start up where the regulator is in current limit towards the low end of the soec. That would make it a problem that the factory controller also has. Unlikely I suspose, but when you look at the numbers.......

but my own preference would be to use the post #716 circuit (with protection diodes as necessary) to control an IRF3205 (which Joe already has) instead. The down-side would be losing the thermal-shutdown feature of the 2576

The regulator has cycle by cycle current limit as well as thermal so the response is fast. The problem with just shutting down after a few seconds is that with the big power supply the pump fets are toast before the shut down.

(but a resettable fuse would remedy that to some extent). On the plus side would be simplicity, not having to hack the OEM controllers, and a less bulky overall system.

Yup, what it wants to be at this point is a little mico. ()Blivian???

 

[ronv]

The pump should be protected by the Hall IC.

I think that is only when the pump is being driven by the hall. Once the FETs are added I think that feature is lost.

 

[alec_t]

@KISS
Using a card cage would certainly be a (pricey) step up from the present biscuit tin . Nice, though.

@ronv

The regulator has cycle by cycle current limit as well as thermal so the response is fast

The way I would drive the FET would also have instantaneous current limit. See attached PDM-Mk7.
View attachment 65202
Sensed current over ~3A turns on Q3 which pulls down the FET gate via D4, thus limlting the current.

 

[ronv]

That might work. With the current limit at 3 amps the motor voltage at start up will only be 6.5 volts. I'm pretty sure the pump won't start at that voltage. So it could be raised to 6 amps or so - in the order of 12 to 13 volts it should be ok.
The power in the fet is around 50 watts at this point so a fair sized heat sink would be needed even if it's only on for a few seconds.
I would vote for this one..

Joe, Any chance you have a 12 volt power supply laying around that could be used to see if the pump will start at 12 volts?

 

[KeepItSimpleStupid]

@Joe
You could always take it out to the car battery. Put a voltmeter across the battery first. It will likely be little higher than 12.

 

[ronv]

Yup, Even a 6 and 12 volt one would put some bounds on it. It should be under load (under water) just to make sure. Several tries because it can depend on where the rotor stops.
We could target slightly over the minumum voltage on the slow setting. Chances are they knew where it stalled.

 

[KeepItSimpleStupid]

It has to start at 10.25 V?, except the controller always puts out 22.5 V eventually. IF Joe lifts a leg of the two resistors right next to TR1 and TR2 and connects his 24 VDC suppply to the 24 VAC in, he would have a steady 10.25 V to the pump.

The leg to lift is the one connected to the transistor side.

You can wait until the solder braid arrives (the envelope) for desoldering.

 

[ronv]

Cool. Good idea.

And easily test all the pumps.

 

[()blivion]

I agree that some tests need to be done on the pumps to see how low they can limbo. I'm betting they will run fairly well on 12v. Most likely will start all the way down to 10v. 6v would be pushing it, but they should run on it if they were already started. I can get most 12v PC box fans to run down to ~2.5v if they were already spinning. And these pumps are effectively the same circuit design, just bigger scale. Note that it's the current that has been scaled up more so than the voltage. PC box fans nominally are 12v@~130mA but our pumps here are 24v@1.3A if I remember correctly. So double the volts but the amps went up by x10. Much bigger current change. So... lower resistance, meaning more likely to draw more current and start + run on lower voltages. BUT!!! they also have a larger physical load. Which translates directly to harder to start and run. So it's debatable what exactly they are going to do. I'm guessing how low they will run will be quite a surprise though.

Yup, what it wants to be at this point is a little mico. ()Blivian???

We had talked about going embedded system and Joe respectfully declined. But we were talking a little crazy about going and making a complete "supper reef system". I had forgot to emphasize that we don't HAVE to do anything that extreme. I'll explain that now.

Edit: I guess a salt water aquarium is not the same thing as a 10" Burmese python. You may want to feed your fish more than once a week. :-/

Salty, MCU talk may scare you. But when we say "computer control" we are not talking about what you would normally think of as a computer. These are single small chips, very much like what you have already been using (some even smaller). It's just that INSIDE them, they have a computer like system. This allows us to do really neat things. For example, we could do a clock and calender that could feed your fish automatically on what ever day you wanted. The effect of this would be that say once a week, one of the pins on the chip would go from 0 volts to 5 volts for a second or two then back. This would probably activate a relay or surplus RC car servo or something that would drop in some food. The best part is that it's just a program running inside a chip. It doesn't have parts to speak of. And if we want to change it or turn it of, we can just change the program. In a nut shell, it can be as complex or as simple as you want it to be, and can be changed at the drop of a hat. Most times without even re-soldering anything. Just rewrite the program in the chip and *POOF*... magically does something entirely different, or only slightly different.

Anyway... I *AM* still here and we *COULD* still do a simple MCU system. What exactly did you have in mind though ronv? PIC is my (current) brand, but I don't actually care. I have dev tools and PIC's now though. And I can do pretty much anything we need as long as we have a game plan set in place first. I would also be willing to donate a MCU or two to the cause. For spares I have some 20 pin PIC18F13K50 if we want fast 48Mhz CPU, USB, hardware multiplyer, 10bit ADC, and other fancy uselessness. I also have one or two 28 pin PIC16LF72's if we want a few more pins at the minor cost of some speed and peripherals. And lastly, I have a few 8 pin PIC12F509's if we want supper small space savers (and like NO pins, lol.) Also.... MCU's are dirt cheep if salty wanted to just go out and get us a specific unit for this project. So practically any part is on the table. A 10 pin mid range part would be nice. With lot's of code space, though I don't really need it.

Here is a bit of a bummer though. If we went MCU with this, we would be throwing away good parts of Alec_t's fine engineering and schematic work, and I wouldn't think he would like that much. Alec_t, would you be able to work with me schematic and simulation wise if we went full MCU core with this project? Would it bother you at all? Most of the timing and control work you have done so far would be all for not, as we would be much better off to do it all in the micro.

The decision to go MCU is entirely up to you (Alec-t) and salty I would think. I will help as best as I can if we do go this way though.



On another note. I think some thing that would also be neat is if we could hunt down some one (maybe on the forums?) that would etch a PCB's for us for nothing or cheep. I myself have about 80% of the stuff needed to do it, just don't have a decent printer to make the art work. But I'm aiming for photo printing and supper fine traces for my own small pitch digital boards. so I need a color LASER printer to make that happen, which needless to say is not cheep. What I REALLY want is to do though multilayer boards for small BGA chips like what you might find in cell phones and stuff..........................but then I woke up.

 

[KeepItSimpleStupid]

Toner transfer seems to be the way people are mostly doing it. Supposedly an iron works, but a modified laminator (higher temps, thicker substrate) is the way to go. The trick is the paper. Anything from magazine paper, photo paper, pulsar paper to Disolvo paper (dissolves in water) paper. I do have a laser printer.

It's been a long time, but I have done the photoresist method: a few times with real tools and once or twice using the sun. I need to find a source for translucent polyester paper for that worked the best for me.

This http://iteadstudio.com/store/index.php?main_page=index&cPath=19_20 and http://dorkbotpdx.org/wiki/pcb_order are probably the cheapest PCB services around. I friend just used iteadstudeo and was pleased. For a large 4-layer board he used a vendor on ebay.

The other issue is that Joe effectively has the sequencer working. All of this MESS has to to with the motor driver.

I'd like him to hack a single OEM controller and just start running a pump 24/7. For giggles, one could use a $10 120 V timer and a 120 VAC relay to get a dry contact closure and make it do 1/2 hour on, 1/2 hour off or some such nonsense.

Now, learn one of the free packages such as Eagle, DIPTRACE, DesignSpark or KiCAD and go make 8 or 20 boards with a few to spare. No point in etching them "at home". The controller is a separate entity. But, you still have to decide on a form factor, where they will be protected from fingers and the environment etc.

It's a lot easier when you have an enclosure as part of the puzzle. DIN rail and Snaptrac isn't that expensive, but it forces a PC board form factor. It forces the boards to be 69.9 mm wide. See: http://www.te.com/catalog/bin/TE.Connect?C=1&PN=2TK2-6&M=BYPN&LG=1&I=13

So, far this would be my vote for a construction method.
1) Proto any way a single motor driver
2) Run a pump 24/7 with a stupid household timer
3) Test
4) Commit to a snaptrack design
5) have x motor driver boards commercially made.

Micro/vs discrete controller is another challenge. The discrete controller works.

 

[ronv]

()B, Kiss is right, it is almost done (I think).
I'm not a micro guy but it just seemed to me a little PIC would replace all the counter and latch logic.

Kiss, I'm not sure I understand how the transistor are hooked up in the controller, but usually a lower voltage on the FB pin lowers the voltage on the output, so Joe would need to add a resistor from the FB pin to ground?