Use PWM to control your Jebao DCT pump using brushless motor controller from ebay

[d0ughb0y]

the board looks exactly the same, except yours have 2 1000uf capacitors on the bottom side while the 50watt controller only has 1. I really think the controller is the same, only a higher amp power supply is provided. Considering the biggest jebao pump use less than 5 amps, those TO-252 mosfets should be able to handle it.

You can hack this now by tapping on the signal output (output 1 2 or 3, output 4 is not connected) of the LM339. Every 6 pulse is 1 revolution. Just filter out the blip noise in the signal (see my screenshot of the signal above).

or if you don't want to deal with the BEMF signal noise, you can tap on to the high side signal, that's pin 7,9,or 11. You can use any one.

 

[d0ughb0y]

I just noticed on your picture of the fets, you can see diodes next to each fet.

 

[karimwassef]

ok. what's next?

 

[d0ughb0y]

I think I got my rpm calculations above wrong.
6 commutation is only equal to 2 high side pulse, not 6. So multiply 745 by 3 = 2235rpm at 70% (lowest setting). if there are internally 6 coils instead of 3, then divide that by 2. I don't really know the typical rpm of a water pump. It may not matter, you can calibrate to determine the number of pulse to get 1 gallon. the head pressure would not matter, as 1 gallon is 1 gallon and will take the same amount of revolution to deliver, the difference being only time.

as to what's next. I'm going to buy some parts to test (diodes, fets, etc).
I can't really do much now since I have a house guest using the room where I have all my stuff for this project setup.

 

[karimwassef]

Ok. Other than the 1K Ohm, is there anything else I should try?

 

[d0ughb0y]

wire the gate diodes like this. use a 1A fast recovery diode. (UF4007)

if you feel like taking a risk, you can just wire a jumper wire between the two points. The white controller is bare bones and is wired directly.

I don't have a diode symbol. The arrow points to direction of current flow, just like diode symbol.

 

[karimwassef]

I don't know if I have any fast recovery diodes. Let me check my inventory.

 

[d0ughb0y]

I snuck into my quest room to do some soldering.

This really put my soldering skill to the test

This is the commutation signal. Running at 100%, the RPM is 60000/25.2367 = 2365 rpm, which is about the same speed as the lowest setting on the stock controller. So definitely the green board runs slower.
Not sure if those blips/noise on the commutation signals are degrading the power or not. probably does, it messes up the magnetic field.

 

[karimwassef]

so can we make it go faster?

 

[d0ughb0y]

I will compare the circuits of the stock, white and green boards and try to make changes and see if anything works.

 

[stage3-s4]

whats the max speed you can got out the pump with the ebay controller?

 

[d0ughb0y]

the max I got on the green ebay controller was the min on the stock. The whilte ebay controller I am assuming the same as stock since the power usage is exactly the same.

I think in order to fix this, the hall input signal coming from LM339 must be fixed to reduce or remove those isolated pulses. The stock controller has it as well, but they are quite narrower and I think the controller chip knows to ignore those, but the one on the green board is twice as wide. It may be the capacitors in the resistor divider is causing the signal to switch slower, so I may try to remove the capacitors and see what happens.
Also, adding the diode or just shorting the gate resistor will also speed up the switching of the FET.

this is the stock controller. On the last two signals, the noise is a small high pulse, and a small low pulse within the larger high pulse.

on the green controller signal, it looks like there are two high pulses within the larger low pulse, and two smaller low pulses within the larger high pulse.

I only have an 8 channel logic analyzer, so HU signal is not captured. I could use a 16 channel logic analyzer for this application.

 

[d0ughb0y]

I compared the signals of the white and green board using an oscilloscope, and the wave form itself looks similar, but the narrow pulses are a bit wider on the green board. That difference might be causing the control chip to slow down the motor.

The bemf circuit is almost identical, even resistor values. The white board has fewer capacitors, and I'm thinking the added capacitor might be causing the pulse to get a little wider. I'm going to try to remove the 3 capacitors on the green board and see if it makes any difference.

The white board also connects the driver signals directly to the transistor/fets without resistor. I can also short the resistors out to see if it makes any difference.
If that is still not enough, the next step is to replace the N channel FETs with ones that have lower gate capacitance.

I think these are all the possible mods that can be done on the green board to see if it improves flow.

 

[d0ughb0y]

BTW, on the stock controller, if you wipe off the heatsink glue on the fets, you can see the part numbers. on the 50watt controller, it's AOD413A and AOD454 and these parts are available from digikey. The N channel is rated 40v 20A and P channel is rated 40v 12A so I think this is the same fet used on the 100w stock controller as the max current is less than 5A. Karim, you can confirm on 100w controller in case you still have it disassembled.

If I have to build a controller from scratch, I will use the same FETs.

 

[Dactylopterus]

I Hope this helps. It is the 100W contoller from a Jebao DCS7000

Sent from my SM-G900W8 using Tapatalk

 

[d0ughb0y]

Thanks. That helps.

I see it is pretty much identical, except for high power mosfets

AOD4186 N channel mosfet 40v 35a
AOD4185 P channel mosfet 40v 40a

and the second 1000uf capacitor.

Not sure why they don't just use the 100w controller for all pumps and just include different wattage power supply.

 

[karimwassef]

The higher power mosfets may cost more or be less available.

 

[karimwassef]

2" flow sensor is in..

<a href="http://s1062.photobucket.com/user/karimwassef/media/IMG_6404_zpssbb5yiq5.jpg.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/IMG_6404_zpssbb5yiq5.jpg" border="0" alt=" photo IMG_6404_zpssbb5yiq5.jpg"/></a>

<a href="http://s1062.photobucket.com/user/karimwassef/media/IMG_6403_zps828xsgvc.jpg.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/IMG_6403_zps828xsgvc.jpg" border="0" alt=" photo IMG_6403_zps828xsgvc.jpg"/></a>

it has a SS clip, but I think I'd use it and find out how it works out.

The fan has "no resistance". I barely breathe into it and it turns like a turbine!

<a href="http://s1062.photobucket.com/user/karimwassef/media/IMG_6400_zpsxa6ibgqz.jpg.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/IMG_6400_zpsxa6ibgqz.jpg" border="0" alt=" photo IMG_6400_zpsxa6ibgqz.jpg"/></a>

<a href="http://s1062.photobucket.com/user/karimwassef/media/IMG_6402_zpsbl6n3l6l.jpg.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/IMG_6402_zpsbl6n3l6l.jpg" border="0" alt=" photo IMG_6402_zpsbl6n3l6l.jpg"/></a>

 

[karimwassef]

doughboy - since you're the master Arduino dude.. would you please write the code to turn the sensor measurements into flow (GPH)?



Sorry - working too hard and need some help.

 

[d0ughb0y]

if you google search, there are plenty of arduino code that works on this type of flow meter. Even adafruit sells this type of flow meter.