Update: PWM Water Pump Voltage Controller

D

dinofx35

Weaksauce
Joined
Jan 16, 2006
Messages
91
The PBCs arrived today! Still waiting on Futurlec to send me my MOSFETs. Stay tuned for progress. I'm not going to take any orders until I have the first sample built and running in my system.

hardforum.jpg
 
Oooooh.. Aaaaahh... very pretty boards. How's the v-grooving?

Only thing I can see that's iffy, is that the mounting hole looks like it's pretty close to that cap and the 2 pin header. Any mounting screw in there looks like it would bump into those components.
 
Where would you mount it? Wouldn't two holes be sufficient, or you could use rubber fan posts or something else. In my prototype, I put a lexan backing on the board with some spacers, and used UV reactive blue zip ties to hold on the lexan. The I just stuck it somewhere with double-sided tape. Here, there's a blank spot on the top third of the back side for hanging the board somewhere with plenty of stick.

The V-scoring is perfect. I was worried it would cutting into the layout, but it's very thin, and double-sided.
 
Well, the two holes that are available are furthest away from where you are plugging and unplugging the connectors. I'm sure there are plenty of alternative ways to maount it, it just seems a shame that with all the extra board space those mounting holes are crowded.

In all, that was just a minor observation. The boards like real nice. I hope my fan controller comes out just as good.
 
There is no extra space. A heatsink will cover the remainder of the board. I wanted to keep it 1"x1.5" so that I could get 8 per 4"x3" board

Good luck with yours.
 
Futurlec has been a major disappointment. The first time I ordered from them, I ordered 2 quantities of Molex KK crimp terminals. They sent me 2 pieces instead of 20, so I had to wait another 2 weeks for 18 pieces of bent metal to get "rushed" from China.

I ordered the molex parts from them again for this project (connector housings are ~20 cents each stateside), and it seems they still haven't shipped my order after 3 weeks. Insane. Stay away from them unless you absolutely must save a few bucks.
 
So, I had asked Futurlec for a quote on the PCBs. To do this, they ask for a previous order, so I gave them a pending one. They ask for this as a way to pull your credit card info from their system, should you accept the PCB quote. But, they decided it would be cool to delay my parts until the PCBs were ready, even though I would be paying for separate shipping for both orders. Note that they asked for previous order, not a pending order.

Well, because of file format issues, I decided to manufacture with pad2pad, and I never ended up ordering PCBs from futurlec. So they had taken my money and were just holding my order in hopes that I would order boards from them, and that they could combine shipping and pocket the savings (I'm guessing). And if course I had no idea this was happening until checking my order status last week.

Long story short, parts should arrive May 8th.
 
UPDATE: OK, all of the MOSFETs and heatsinks have arrived. The circuit has been tested (with 120mm fan) outside of my computer and is working perfectly. I have all of the parts to finish 9 more boards.

Here's the finished circuit. I still need some nail polish remover to clean up the flux from the solder.
pwmcircuit.jpg
 
The MOSFET is dissapating almost 4W for a water pump running at its lowest voltage, which might be about a 50°C temperature rise for that heatsink. The MOSFETs operating range is up to 150°C, so as long as you can't boil water you're safe. But, when running your water pump at full speed, the circuit produces practically no heat. It has a low 0.2 Ohm on resistance. Were you joking, or are you looking for TO-220 heatsinks?
 
I've realized a potential issue with interpreting the PWM signal. In my Shuttle XPC, the PWM signal is roughly a constant 4.25 volts when the duty is 100%. (And, of course 0 volts at 0% duty). I measured this a while back during the breadboard stage, I used a 100K resistor and a .1u capacitor as an RC filter, and just plugged it into a spare FAN header and set it to 100% with speedfan.

Is there any reason to think that this value is the same for all motherboards? I'm going to guess that there is a very good chance that a single pull-up transistor is being used here to pull up to +5V. The 0.8V difference seems suspiciously like the voltage drop across a transistor.

Is there anyone out there with a voltmeter and a PWM Fan header on their MB willing to measure their 100% duty voltage? Or, does anyone have information suggesting that PWM signals are between 0 and <5 volts?
 
You shouldn't have to care what the PWM voltage is off a mobo as long as it can be interpreted as high or low.

dinofx35 said:
I've realized a potential issue with interpreting the PWM signal. In my Shuttle XPC, the PWM signal is roughly a constant 4.25 volts when the duty is 100%. (And, of course 0 volts at 0% duty). I measured this a while back during the breadboard stage, I used a 100K resistor and a .1u capacitor as an RC filter, and just plugged it into a spare FAN header and set it to 100% with speedfan.

Is there any reason to think that this value is the same for all motherboards? I'm going to guess that there is a very good chance that a single pull-up transistor is being used here to pull up to +5V. The 0.8V difference seems suspiciously like the voltage drop across a transistor.

Is there anyone out there with a voltmeter and a PWM Fan header on their MB willing to measure their 100% duty voltage? Or, does anyone have information suggesting that PWM signals are between 0 and <5 volts?
Click to expand...
 
True, if I were interpreting it as a "digital" signal. But, I'm treating it as analog using an RC filter. I think this is not going to be a problem since fan manufacturers require the signal to be at least 2 volts, and I doubt it would be more than 5.
 
dinofx35 said:
True, if I were interpreting it as a "digital" signal. But, I'm treating it as analog using an RC filter. I think this is not going to be a problem since fan manufacturers require the signal to be at least 2 volts, and I doubt it would be more than 5.
Click to expand...
Sounds like board revision time :D

I'd replace the 741 with a 358 dual op-amp chip, and use the second op-amp to insert a second 'comparator' stage which converts the mobo PWM output to a constant amplitude 'bouncing off the rails of the op-amp' signal. Then feed this into your existing circuit.
 
Thanks for the suggestion. At one point I was going to use a dual opamp for output to boost the tachometer signal back to a 12-volt signal. I didn't do that because it would probably require a rail-to-rail opamp, and because my motherboard had no problem detecting the "7-volt" tach.

The dual-op-amp solution would also need 2 more resistors, right? What about just using a small [low power] MOSFET? Anyway, first step is to figure out if it is really an issue. The symptoms would be that the controller maxes out at a lower duty cycle, worst case 35%. You would still have 35 different output voltages to work with. If this is a problem, I also have a workaround that requires two resistors somewhere in the wire coming from the motherboard. But it helps to know what the 100% voltage is.
 
PWM Specification

According to the specification for 4-pin fans, the maximum voltage allowed is 5.25 volts. This circuit should work as intended on all properly implemented motherboards.

You may not be able to test the pull-up voltage of your header without actually pulling up the voltage. In other words, if you just compare the PWM signal to ground, it will be 0 volts
 
You must log in or register to reply here.
Back
Top