Hi,
had a good look around and have been unable to find anything which is similar.

I'm working on a project to take the variable 12v produced by a pc motherboard header to control fan speed and convert this to a PWM signal to control a device (watercooling pump) which has a greater current draw than the motherboard can handle. (max 1A provided by the header and up to 2A required by the device)
The device has separate power and PWM/tach signal wires.
The motherboard then takes the resulting RPM from the tach wire and plots against input voltage, this can be used to control the device based on temperature

I have a circuit which is working well in simulation based on a 556 timer design largely inspired by this video (and the second part):

Was unable to copy the schematic as the quality was rubbish :/

The circuit runs off the 5v supply provided by the PSU and must be able to provide an output with frequency between 21kHz and 28kHz, max current of 5mA, logic low of <0.8V and max voltage of 5.25V in line with the intel standard.
In simulation it satisfies these criteria.

My problem is with converting the variable 12v signal produced by the motherboard and reduce this to 5v whilst maintaining the variability.
I have looked at using DC-DC converters and voltage regulators however these all seem to fix the output at a certain level rather than allowing for any variation in voltage, this would obviously remove the control element.

The best that I have been able to come up with is a simple resistor based voltage divider taking into account the load of the internal PWM control circuit of the device to be controlled.

I was wondering if anyone has suggestions regarding a better way of scaling this 0-12v to 0-5 whilst maintining the control element and does not rely on a potentiometer (the circuit will be inside a case and therefore inaccessible)

Also, any suggestions for the current source IC able to produce current of >12.2mA, 12.1mA is the minimum for this capacitor on the monostable element of the circuit to function correctly.
So far I have found the PSSI2021SAY which seems to fit the criteria or to produce one from component parts...

Thanks in advance!

Schematic of the circuit is attached.

 

Welcome to AAC!
How much change in the PWM duty cycle do you need? According to Philips Semiconductors Application Note AN170 for the 555 (or 556),
"In the monostable mode, the control voltage may be varied from 45% to 90% of V CC . The 45-90% figure is not firm, but only an indication to a safe usage. Control voltage levels below and above those stated have been used successfully in some applications."
Can you post a link to the pump's datasheet?

 

I understand what it is you want to do, but it is not clear what you have to work with.

1. Is the motherboard signal a variable analog voltage or PWM? If a variable voltage, what is its output range from low fan speed to high fan speed?

2. Is the motherboard output really capable of supplying 1 A, like for driving a fan directly?

3. What is the typical or max operating voltage for the pump?

4. Does the water pump take PWM directly as its power input, or does it take DC for power and use the PWM input as a control signal?

Where I'm headed with this: If the pump, like a PC cooling fan, responds well to a changing power voltage for changing its speed, there doesn't seem to be any need to go through conversion to PWM and back to DC. If the only problem with the motherboard output is that it can't make 2 A, that is a pretty easy problem to solve, and no 555's will be harmed in the process. IF the total project is to reduce the motherboard output proportionally from 0-to-12V to 0-to-5V, and double its current capability, that can be done with two resistors and one transistor on a small heatsink. No switching circuits, no switching noise. It's, like, analog.

ak

 

to alec-t

Ideally i'd like as broad a duty cycle control range as possible as I would like to extend the functionality beyond this particular pump for future upgrades etc.
The pump being controlled is based upon the venerable Laing D5, I have been unable to find a data sheet for the PWM version, the closest I can find is the standard version which is either fixed speed or controlled by a pot, data sheet is here: http://docs-europe.electrocomponents.com/webdocs/0f08/0900766b80f08f97.pdf

As I understand it, the internal circuitry of the pump to interpret the PWM signal uses the intel standard:
http://www.formfactors.org/developer\specs\rev1_2_public.pdf
with the exception of missing the required pullup resistor(s) to make the device run at full speed without a PWM signal instead it runs at approx 60% output when not connected, to rectify this there is a small mod which takes the 12v from the power pins through a 5.1v zener diode with resistors to make it run at full speed without PWM
http://www.overclock.net/t/1474470/ocn-aquaero-owners-club/50#post_21956203
I did this at first however have since found out that the only true PWM header is the CPU fan one and although all headers have four pins, the PWM header is not connected (dont know why they do this).

To analogkid

1) The signal is a variable analogue voltage and seems to range from a few hundred mV to 12V max.
2) The motherboard specification states that the headers are capable of providing 1A max
3) Typical voltage for this pump is 12V however it can run between 8-24V (data sheet above)
4) the pump uses a separate molex connector for 12v constant DC power and the plug that goes to the motherboard provides tach signal and PWM input for control.

so basically I need a way to convert the variable 0-12V analogue voltage from the motherboard to PWM, using the 555/556 was just what I thought would be a good way of going about it.

Someone has suggested that I could use a basic resistor voltage divider to reduce the 12v whilst maintaining proportionality and feed this into an op amp in negative feedback (unity gain?) to separate the input signal from the rest of the circuit

another slight change - rather than use a constant current source which just overcomplicates things, I can just use a standard RC combination for the discharge and threshold pins of the monostable 555 and further reduce the control voltage to 3.33V max to make it all work.

overall im hoping to be able to extend this to other devices too rather than just be limited to this particular device as its a common problem with motherboards.

Hope its all made sense...

 

If the pump runs at full speed with no pwm input, then you could control the pump speed with a linear regulator on its +12 V power input. Lot less work.

ak

 

If the pump runs at full speed with no pwm input, then you could control the pump speed with a linear regulator on its +12 V power input. Lot less work.

ak

not quite sure what would happen with the pump without pwm input, if the control wire is not plugged in it should run at 60%, it only runs at 100% with the mod when getting a constant 12v on the pwm wire through the zener diode (see page regarding the mod) if this isnt connected then the pwm signal would be low and the internal electronics would take over again, setting the pump at 60% and limiting voltage control to between whatever voltage the pump stalls at and 60%

works perfectly if connected to the cpu fan header where it gets a proper PWM input and gets good range of control, from 800rpm to 4.8k rpm, trouble is, that is used for an array of fans which combined, also draw more current than the motherboard can provide.
i'd like to gain proper control of the pump if I can, hence needing to do something about the control voltage and pwm circuit.

could just buy an aftermarket controller such as the aquaero for an easy fix but im interested in the application of the concept just as much as getting it working.

 

Here's my offering of a cheap, simple circuit which will give ~0%-100% duty-cycle variation in a 5V ~22kHz PWM output, using a 0-12V control signal. The push-pull output stage can be omitted if your motor control input is high impedance (>~2k).

 

Cheers, looks good, will check it out when I can

 

Built the schmitt trigger circuit using a single CD40106 and it works great with an added bonus of keeping the fan/pump running below the point where it can generate a pwm signal and once it goes above the upper limit it creates a steady 5v causing the device to run at 100%.
Just need to build a permanent version now.

 

Glad you got it working.