Custom PWM reference controller?

[Veraxis]

If you smoothe the PWM with a simple RC filter you get a voltage. Won't the POT type controller accept that as equivalent to the pot wiper voltage?

I think the pot in that controller in the OPs first image is acting as the timing resistors for the 555 timer rather than setting a voltage.

 

[Cole]

So many wonderful ideas and possibilities! I can remove the Controller board from my pc and provide measurements/values to determine if using this board is an option?

 

[Veraxis]

So many wonderful ideas and possibilities! I can remove the Controller board from my pc and provide measurements/values to determine if using this board is an option?

Do you have a schematic and/or board layout of this controller by any chance? It looks like the kind of thing that might come as a kit, so it might include a schematic. If not, some good quality photos of the top and bottom should be sufficient to reverse engineer the schematic. It looks to be a pretty straightforward 555 astable configuration into a MOSFET driver of some kind-- probably a half bridge.

As dougy83 suggested, it may be possible to connect the PWM signal to the reset pin of the 555 timer. I believe the reset pin of a 555 can be run off of a logic level signal, so that shouldn't be an issue. It may also be necessary to disconnect or short one of the timing resistors or reconnect the threshold and trigger pins in order to drive the output to the correct level as soon as it comes out of reset. TBH, I'm not an expert in how the output of a 555 will behave when being constantly reset like that. What is the startup condition on the output of a 555?

I mean, if all else fails, you could probably just remove everything else on the board and just rewire the n-channel MOSFET from the half bridge as a low side driver and run it off the logic level PWM signal. If you reconnect the gate on the P-channel FET to the +12V rail, then the body diode of the FET would work as your back-EMF diode.

 

[Cole]

Here is the link of the exact item I purchased on Ebay. Ill upload the pictures to the forum as well. I do not believe they have a schematic listed, but I will message the seller to ask them for one.

**broken link removed**

 

[Veraxis]

Okay, so I think I've reverse engineered that board (I'll still need to check my work, because it's kinda tricky with all the boards in the images at Dutch angles hehe). I can't read the part numbers off the TO-220 package parts to be sure, but I'm pretty sure it is using a 555 astable to drive two N-channel MOSFETS in parallel as a low side driver. It has what looks to be a TO-220 package double-diode (probably Schottky) in parallel as a back-EMF diode. I'll draw up the schematic when I get home later today, but basically I think it could be as simple as unsoldering the 555 chip and connecting the PWM signal where pin 3 used to be...

 

[Veraxis]

Okay, so here is the schematic I came up with from those photos. It's hard to tell in a few places, and I'm making some educated guesses on what the TO-220 package items are, so you'll probably want to double-check my work on the real board.

Basically, it is a 555 astable to a low side driver, which works perfectly for modifying it to run from the PWM signal from the CPU fan. If those MOSFETs are able to be driven to a reasonable current from a logic level signal (you'll need to check the part number on the various TO-220 parts and look up the datasheets, or just post the numbers here), then it should be no issue to just unsolder the 555 timer IC and solder the PWM signal onto the pad for pin 3. If not, then we may have to change out the MOSFETs or rewire the 555 timer in some way to act as a buffer for the PWM signal. Others may be able to advise on that.

The board seems to use some kind of weird voltage regulator configuration for dropping the voltage to the 555 timer. At first I thought it was going to be a fixed regulator, but it seems like it has some resistor support components which suggest it might be a variable regulator like an LM317 or similar. I didn't feel like tracing it out as it isn't really critical to what we need to know. I also left out the indicator LED, which I believe is also run from the voltage regulator.

 

[Veraxis]

Here is an updated schematic with the regulator filled in. I'm not 100% sure if I'm reading the values for R8 and R9 correctly. Anyone can feel free to double check and/or correct me on anything here if it looks out of place.

Edit: Cleaned up the layout of a couple junctions.

 

[dougy83]

JLNY, thanks for the hard work on the schematic. To convert the board to a buffer, simply remove the POT (R3) and short out C1/C2, and feed the PWM into pin 4 after separating it from wherever it's currently connected.

 

[spec]

Here is an updated schematic with the regulator filled in. I'm not 100% if I'm reading the values for R8 and R9 correctly. Anyone can feel free to double check and/or correct me on anything here if it looks out of place.

View attachment 104152

Edit: Cleaned up the layout of a couple junctions.

Nice work JL

The International Rectifier (now Infinion) IRFP4710PbF is a likely candidate for the output NMOSFETs.

spec

 

[Veraxis]

Nice work JL

The International Rectifier (now Infinion) IRFP4710PbF is a likely candidate for the output NMOSFETs.

spec

Thanks!

What makes you think it is that MOSFET in particular? That part you attached seems to only come in a TO-247 package, whereas the power MOSFETs on the OP's board look to be TO-220 package.

 

[spec]

Thanks!

What makes you think it is that MOSFET in particular? That part you attached seems to only come in a TO-247 package, whereas the power MOSFETs on the OP's board look to be TO-220 package.

Ah, so it does- I missed that.

I only suggested that The IRF4710 may be the NMOSFET because it has been around a long time and has been used in few other inverter designs that I have looked at.

But, at the end of the day, the NMOSFET type is not important because the NMOSFETs are already fitted to the motor controller board.

spec

 

[Cole]

Well i cant really make out any text or numbers due to the tiny and faded text but i was able to get a few pictures of the mosfet.

 

[Veraxis]

Well i cant really make out any text or numbers due to the tiny and faded text but i was able to get a few pictures of the mosfet.

If I'm reading that correctly, I think the top line reads 80NF70. Judging by the ST Microelectronics logo at the bottom left, I would say there is a good case that these are STP80NF70 MOSFETs.

Typical Vth is about 3V (4V max) and the transfer chart shows about 30A Id at +5Vgs and +12Vds (remember, that would be 60A total for two in parallel), so I would think that it is reasonably likely that these will run off of a logic level signal. What do others think?

Were you able to get any pictures of the other components like the regulator or the diode package?

 

[Cole]

They ebay seller got back to me. They are going to ask their engineers for it. As for the other components;

30CT0100

LM317T

 

[Veraxis]

LM317T is a variable voltage regulator. However, for the values of 5.6k and 10k that I have listed on my schematic that would only be an output of 3.5V, so I don't think that's right... I've probably misread one of those resistors... if R8 were actually 30k instead of 10k, that would be abou 7.9V out nominal. You could always try to measure the regulator voltage if you wanted to be sure.

I'm not getting anything on the part number 30CT0100, but I'm guessing that it's the double-diode.

 

[KeepItSimpleStupid]

R7 should be between 120 and 240 ohms. I would need to check the datasheet, Some people get it wrong. The lower number is usually better. What's the regulator output?

12 V rail, R9 should be closer to 1K. A high efficiency LED could put it at 10K.

 

[ChrisP58]

JLNY, thanks for the hard work on the schematic. To convert the board to a buffer, simply remove the POT (R3) and short out C1/C2, and feed the PWM into pin 4 after separating it from wherever it's currently connected.

Be aware that the voltage threshold of the reset pin of the 555 is very low. So unless the PWM output as seen at pin 4 of the 555 goes right down to zero volts, and there is almost no voltage differential between the two grounds, when fan current is flowing, then this may not work reliably.

Section 7.4
https://www.ti.com/lit/ds/symlink/ne555.pdf

 

[dougy83]

Be aware that the voltage threshold of the reset pin of the 555 is very low. So unless the PWM output as seen at pin 4 of the 555 goes right down to zero volts, and there is almost no voltage differential between the two grounds, when fan current is flowing, then this may not work reliably.

That is a good point. The control signal is spec'ed at similar low level of 0.8V max for low level output (http://formfactors.org/developer\specs\4_Wire_PWM_Spec.pdf page 9), which is more than the minimum 0.3V (typical 0.7V) of the reset pin . I would expect it to be very close to 0V though and I expect it will work; if not, then there are some simple things to fix it (namely a series diode and resistor pulldown).

 

[Cole]

Another forum had the same exact question, this was the "best answer".

 

[Veraxis]

Another forum had the same exact question, this was the "best answer".

Seems like a reasonable enough circuit. A very basic low side driver similar to what was proposed here previously, although a bit more general-purpose in nature. The discussion here was perhaps a bit more specific to working within the particular PWM controller you were already using.

IRL530N is a good choice for this application as it is specifically meant to be used with logic level signals. It has a 2V max gate threshold voltage, which means that a 5V signal will be sure to turn it fully on. The resistor and zener diode are being used as a pull up, which may be necessary if the PWM signal is based on an open-collector or open-drain driver without a pull-up on board. The lack of a back-EMF diode is a bit concerning IMO.

Passing the tach through would probably not be a good idea in your particular case, as you are hooking multiple fans together in parallel and using the PWM signal from the CPU fan to drive all of them. Trying to read multiple tachs in parallel probably wouldn't work, and might mess up your motherboard's ability to detect the fan speed.

Were you able to try anything out on your computer? I'm curious to see if you were able to get something working for your system.