Help with Water Pump

[KeepItSimpleStupid]

The hall sensor on the two ends measure 1.6V ((-) to right as it hangs from the bottom of the PCB) and 0.898 volts with the diode test. It might be possible to do the diode test for all pins as well as the orientation. For now the hall sensor is detached from the PCB and I know the orientation which is bevel facing away from the edge of the PCB. The middle pin is broken off at the body, but I might be able to probe it. The pins aren't the best in the world. Stay tuned. Not sure what I have that I can epoxy/superglue it too. I wish I had a microscope slide. Maybe even a perfbord. I probably have a slide in the attic, but I don't think I would find that needle in a haystack.

I doubt I can attach a lead to the broken off "bit" (the British word), but maybe I can with some silver Epoxy and some wire wrap wire. I might be able to probe later. I really need to attach it to something solid before I do any more damage. I don't want to silver epoxy the face.

 

[()blivion]

The body diodes are backward from the way you drew them.

Ah... IC.... That's the way the simulator wanted them, so I went with it. (The "Drawing" I made is a screen shot of the simulator)

The more I look at the post #411 first pic the more likely it seems that solder blob under the gate (?) leg of the right-hand FET (?) was the cause of our woes. A (partial) drain-gate short would account for all the damage (apart from that inflicted by Joe and his saw/blowtorch ) and may even have allowed this puzzling motor to run, in a hobbled way, initially.

That's about what I figured too. I just don't see it getting hot enough to melt solder while under water. Especially with out causing some noticeable bubbling in the epoxy around the FET. Though anything is possible I suppose.

Here is my current theory...
I figure that the blob was close but not actually touching, so it passed QC. Then when Joe got it and started really using it and warming it up a bit, that caused the metals to expand closing the gap and causing the failure. The manufacturer most likely figure that if there are any gaps that are close enough to possibly cause a fault, the epoxy should fill them in preventing MOST such pumps from dieing before warranty time is up. But sadly not one of Salty's Too bad using a customized driver consists of user modifications and most likely voids the warranty. (As if the torch and saw didn't)

Still.... I think that is unanimously agreed that premature failure or not, we want decent current limiting.

 

[KeepItSimpleStupid]

ob():
I believe Joe said that there was goopy stuff and epoxy. I think I remember it being clear like RTV or fish-safe RTV Epoxy would wrap around the leads, resistors etc and they would be yanked hard when removed. This failed quickly too if I remember right. The goopy stuff probably doesn't conduct heat very well and neither does Fiberglass. This Novel is getting long.

Marine RTV datasheet: https://www.electro-tech-online.com/custompdfs/2012/06/Marine20Grade.pdf

Thermal conductivity, cal/(cm) (ºC) (sec) ……............. 0.45 x 10-3
BTU per (ft) (ºF) (hr) ...................……........................... 0.11

Fishy safe? Who knows?

 

[alec_t]

considering how simple and "budget driven" the design looks, I would have assumed the Hall sensor to be simpler and more main stream than what you have.

How's this for simpler?
View attachment 64902

 

[salty joe]

ob():
I believe Joe said that there was goopy stuff and epoxy. I think I remember it being clear like RTV or fish-safe RTV Epoxy would wrap around the leads, resistors etc and they would be yanked hard when removed. This failed quickly too if I remember right. The goopy stuff probably doesn't conduct heat very well and neither does Fiberglass. This Novel is getting long.

Marine RTV datasheet: https://www.electro-tech-online.com/custompdfs/2012/06/Marine20Grade.pdf

Thermal conductivity, cal/(cm) (ºC) (sec) ……............. 0.45 x 10-3
BTU per (ft) (ºF) (hr) ...................……........................... 0.11

Fishy safe? Who knows?

There was only epoxy.
This novel is getting long. Thanks for all the help you guys.

 

[KeepItSimpleStupid]

But still no real part to match?

Anyway....Once you get a your design finalized, do you think joe should try to build ONE Driver? One of the things he could do is to force it on and just hook his power supply (Input side) to a simple 120 Volt 24 hr household timer and just exercise a single pump?

Earlier, I thought you had two pumps in parallel, so I suppose now your going to make each pump have it's own separate driver/protection with all of the other revelations. Simple turns complex as opposed to complex turns simple. I like when the later occurs.

 

[()blivion]

@ Alec_t: How's this for simpler?

I like it, good theory. My theory was more simple... but yours more likely has the better benefit of actually being correct. (which tends to help from time to time. )

@KISS: I believe Joe said that there was goopy stuff and epoxy. [snip] The goopy stuff probably doesn't conduct heat very well and neither does Fiberglass.

If not for Joe's latest comment, I would have said it was possible. It's true enough that most polymers are terrible conductors of heat. However solids, and even "goopy stuff" conduct heat FAR better by volume/distance than trapped gasses do in reality. So with the huge gap of solder's melting temp (~185°C) and waters boiling point (100°C), *AND* less than an inch (guessing) of epoxy between the hot part and water... I still think the "paper cup effect" would start to take place before any solder would melt. And even if it did get some what molten... How would it then short out? Being liquid and hot doesn't necessarily stop it from being at least slowed down by the epoxy/goopy stuff. Unless there was already a void there of course. (Which I suppose is totally possible given the location of the short.)

To farther put nails in that coffin though, we can examine the flip side of the coin. If we assume the epoxy did in fact act as a realistic insulator. And that it got hot enough to boil out of the way and allow a short to form. You would also expect it to have caused a crack and/or explode out that section (As I see in blown transistors a lot). Or it should have at least shown some signs of getting supper hot IMHO.

(All of the above presumes that the short that we know exists was the cause of the failure. However the existence of the short does not necessarily exclude the possibility of transistor internal meltdown. Though it does tend to lean away from it quite a lot.)

Salty, You don't need to answer in all honestly but just for the sake of argument; did you happen to notice anything out of the norm before you took a torch to the epoxy? Any cracking or bubbles? Or do you distinctly remember that there was nothing at all wrong with the epoxy's surface, IE it *WAS* glossy and smooth. It really doesn't matter though because....

This Novel is getting long.

I agree, I would have been fine with not doing the autopsy on the pump and just stick to the driver changes honestly. It's only created needless speculation. Truthfully, I never figured it was going to tell us anything we can use that we (I ?) didn't already expect either. It's a brushless motor, so we should current limit the driver as best we can and protect it from transient voltage spikes... I feel like I said this all before... but I'm not sure and I'm far to lazy to reread the whole novel.

:sigh: Wish there was something I could do, but my skill set is not required here. It's starting to make me feel like I actually am "void Oblivion();"

I think I may retire from this thread for a few more moons. But then again, who knows. The company is good at least.



Edit:

Anyway....Once you get a your design finalized, do you think joe should try to build ONE Driver? One of the things he could do is to force it on and just hook his power supply (Input side) to a simple 120 Volt 24 hr household timer and just exercise a single pump?

Good idea.

 

[ronv]

Here is a link to the current limit if you try to go linear.

https://www.electro-tech-online.com/threads/thermal-response.128051/

It would seem better to let the fet suffer with a couple of seconds at 10 amps full on (1 watt) than for 2 seconds at 70 watts.

Kiss, If you are at a loss for somethig to do you could measure the temperature rise of the coil for 2 seconds. I think you could you could apply say 2.15 volts and then measuring the current cold and hot. We could then use the resistance change of copper vs. temperature to approximate the temperature rise of the coil.
Maybe it would help to start a new thread with a summary of where things are at. The schematic, The FET and the motor circuit. It takes a lot of time to go thru it all to find the info.

 

[alec_t]

Thanks for that, Ronv.

It would seem better to let the fet suffer with a couple of seconds at 10 amps full on (1 watt) than for 2 seconds at 70 watts.

I guess it depends which FETs we're trying to protect; those in the controller or those inside the pump.

Maybe it would help to start a new thread with a summary of where things are at.

I agree. But this is Joe's thread. Joe, d'you want to do that?

 

[KeepItSimpleStupid]

Coil wet or dry?

If I do the wet thing, I'll have to let the thing sit in water for a while. I will probably have to dig out an ammeter with more digits unless you think that 2 digits after the decimal place +- 0.01 Amps is OK.

I'd love to see a new thread with the major posts indexed.

e.g (in no particular order)
The intro page
ebay power supplies
ebay pump
OEM Manufacturer's website - pump
Pictures of Joe's teardown
Controller's desired (tide and wave)
Tide controller specs
Wave controller specs
Theory of operation for commutation (this one is mixed up)
OEM pump teardown

For now let's populate the list of stuff, then arrange it with and without links and then transfer to a new thread with a reference to this one.
A lot of VB sites don't like treads to get very long.

Sound like a plan?

You can click on a post number, e.g. #588 and do a "copy link location" to index that post.

 

[ronv]

I guess it depends which FETs we're trying to protect; those in the controller or those inside the pump.

Good point. I guess we don't know much about them. The good news is they have a small board heat sink. Maybe 50 C per watt. Maybe Kiss could measure the one good one to see what the drop is at say 1 amp? Were both the controller and pump fet blown?

I agree. But this is Joe's thread. Joe, d'you want to do that?[/QUOTE]

 

[KeepItSimpleStupid]

Well, you need to protect both FETS and about the only thing we can do about the inside is watch over voltages and over currents.

The external FETS, there is over-voltage, over current and temperature. If possible a protected switch possibly could be used.

The one good FET isn't good anymore. I should have been more careful removing the solder. I did not expect the corner of the package to crumble when I heated and used the soldering iron to more the lead. Usually do it all the time, but this time it didn't work. There is the large tab that offers some amount of heat sinking.

 

[4pyros]

More BDLC info showing a two pole motor. I have yet to find a hall chip with two outputs and just three leads.
**broken link removed**

 

[ronv]

Well, How about 4 of these little guys. The current limit would probably work for the slow start. Might be worth $8 to try it.

**broken link removed**

 

[KeepItSimpleStupid]

I think 4pyros solved the mystery of the controller. The simulation that you linked too seems to make it all click. Note that when they talk about the 2-pole motor, they say that OPPOSING coils are energized at the same time.

So, with two coils, they are energized at the same time. With 4 coils, they are alternately engaged.

So simple, that it had all of us fooled.

Thanks for the link 4pyros.

 

[KeepItSimpleStupid]

I can make out the parts LM2577 and LM2596: **broken link removed**

But it looks like a buck/boost circuit and it has ON/OFF control with some surgery.

 

[salty joe]

If you guys want a synopsis in this thread, I'll do it. If you guys prefer a new thread, fine by me. I'd kinda rather do a synopsis and keep this thread, but whatever you guys want is good by me.

BTW ()blivion, the epoxy showed absolutely no sign of getting cooked. There was only about 1/8" epoxy over the PCB. And if you leave and never come back, I'll miss you.

Could something be cobbled together from the schematics in posts 441, 521,544,549,571, and 584 to add a little armor to the tide schematic in post 367? Not looking for absolutes or anything like that, just looking for a good shot.

The first tide controller with only two soft starts and two MOSFETs for four pumps that got a few cooked parts ran 40W incandescent light bulbs flawlessly for almost a week after being repaired. So I'm thinking it's a solid design but an incompatablity issue showed itself. The very first random current controller I built from post 36 ran the same four pumps all day with no issues for the controller or the pumps. Could having a dedicated MOSFET for each pump be the key?

 

[KeepItSimpleStupid]

Your 40 W, 120 V light bulb is about 360 ohms at 120V, but light bulbs are non-linear, so lets say it's 36 ohms cold. This means, maybe 16 Watts. Maybe about 1/2 the wattage of your pump. A tungsten light bulb is a different animal. It's resistance is 10-15 times lower when it's cold vs. operating.

Resistive and inductive loads are different.

 

[KeepItSimpleStupid]

Joe:

You can do things such as See Post #591

The code to do that is You can do things such as [url=https://www.electro-tech-online.com/threads/help-with-water-pump.124250/#post1065347]See Post #591[/url]

 

[ronv]

Coil wet or dry?

If I do the wet thing, I'll have to let the thing sit in water for a while. I will probably have to dig out an ammeter with more digits unless you think that 2 digits after the decimal place +- 0.01 Amps is OK.

If I remember right the coil is 2.15 ohms. The resistance will go up by .393% / degree C. So if we started with 2.15 volts (1.00 amps) and the temperatur went up 10C the resistance would be 2.234 ohms so .96 amps. Not to bad. If it went up 50C or so it would be easy to see. I'm with you for the water keeping it cool but something made it short out.