PWM to Analog converter circuit
- Thread starter Tealc
- Start date
I tested it as best I can using my PC and my CPU fan and it still works as expected. I had to crank the potentiometer right down as my CPU fan is a super low current Noctua NF-P12.
I'd normally have 2 or 3 fans and load it up to one of my oscilloscopes rather than faffing about with my PC but my scopes are in storage.
I don't have any heatshrink here so will have to grab some from my stocks, unless you want it naked and fancy wrapping some tape around it or something.
I'd normally have 2 or 3 fans and load it up to one of my oscilloscopes rather than faffing about with my PC but my scopes are in storage.
I don't have any heatshrink here so will have to grab some from my stocks, unless you want it naked and fancy wrapping some tape around it or something.
That Zalman is a different thing. I've never seen, tested or bothered to read much about them but it may be a controller that outputs a 5v variable duty cycle that'll control attached fans. I have made one of these and used it to test my own circuit. I built it out of a 555 timer ic and a few passive parts, simple enough.
Now if you wanted 8 3 pin fans controlled of my converter circuit then it would need to be two individual standard circuits. I was developing a high power circuit but haven't got a workshop at the moment so never got that completed. Phanteks sort of came along with their non adjustable version and orders for mine dried up.
I could still possibly make some circuits though. I quite enjoy throwing circuits together.
Now if you wanted 8 3 pin fans controlled of my converter circuit then it would need to be two individual standard circuits. I was developing a high power circuit but haven't got a workshop at the moment so never got that completed. Phanteks sort of came along with their non adjustable version and orders for mine dried up.
I could still possibly make some circuits though. I quite enjoy throwing circuits together.
This sort of thing.
Schematic shamelessly copied from the internet.
The 555 timer basically uses the 100k potentiometer to adjust the duty cycle from one side to the other. It's what I based my own PWM controller on, although I trimmed it to get near 22kHz frequency.
The thing is I don't actually have all the parts and test gear I need at home. I would need to go and get them all from my Mum's house, which would take a while.
The potentiometer can be wired so it is remote from the PCB itself but I'm not sure if it could quite reach a desk without picking up all kinds of stray signals that could piggyback themselves on the PWM wave and cause problems. This is something I'd need a scope for to be sure.
Are you sure there isn't a commercial option for something like this? The Zalman thingy could be easily modified I would expect just to get the variable PWM from it. Edit: Scrub that as it looks like you can't get them any longer.
If I was building such a device I'd like it to have a low threshold, say 30% duty cycle which would need designing in and it would be ideal if the circuit also threw the fans into 100% duty cycle for a short time on power start, just to make sure the fans get started.
Edit: Looks like you can pick up a decent range of PWM controllers off ebay for not too many £s. Such as this one Not home made but cheap and would easily handle 8 fans. It has a decent frequency. The fans would be driven as if they were 3 pin fans, which makes sense. You'd lose tach feedback of course.
Schematic shamelessly copied from the internet.
The 555 timer basically uses the 100k potentiometer to adjust the duty cycle from one side to the other. It's what I based my own PWM controller on, although I trimmed it to get near 22kHz frequency.
The thing is I don't actually have all the parts and test gear I need at home. I would need to go and get them all from my Mum's house, which would take a while.
The potentiometer can be wired so it is remote from the PCB itself but I'm not sure if it could quite reach a desk without picking up all kinds of stray signals that could piggyback themselves on the PWM wave and cause problems. This is something I'd need a scope for to be sure.
Are you sure there isn't a commercial option for something like this? The Zalman thingy could be easily modified I would expect just to get the variable PWM from it. Edit: Scrub that as it looks like you can't get them any longer.
If I was building such a device I'd like it to have a low threshold, say 30% duty cycle which would need designing in and it would be ideal if the circuit also threw the fans into 100% duty cycle for a short time on power start, just to make sure the fans get started.
Edit: Looks like you can pick up a decent range of PWM controllers off ebay for not too many £s. Such as this one Not home made but cheap and would easily handle 8 fans. It has a decent frequency. The fans would be driven as if they were 3 pin fans, which makes sense. You'd lose tach feedback of course.
Last edited:
I can still make them. I don't do them with 8 fans as I find the voltage drops at full speed and stuff. What I have do e in the past is make two boards fed off a single pwm. This isn't too dissimilar to phanteks as they sort of do the same thing.
I'm pretty sure I've posted enough information in the thread for someone to work out the schematic themselves. It isn't particularly complicated.
I haven't made one of these for about 2 years. The last thing I remember doing with this was developing a high current version that drove quite a few fans but I don't remember exactly what I did but the results were decent. The lack of workshop and time didn't help.
I do see a lot of complaints about the Phanteks implementation that I don't tend to come across with mine, however I would imagine they sold more than the 30 odd I made.
I haven't made one of these for about 2 years. The last thing I remember doing with this was developing a high current version that drove quite a few fans but I don't remember exactly what I did but the results were decent. The lack of workshop and time didn't help.
I do see a lot of complaints about the Phanteks implementation that I don't tend to come across with mine, however I would imagine they sold more than the 30 odd I made.
Excellent. I've made a little start and also went down my Mum's to get all the bits I need, plus one of my oscilloscopes which while I don't actually need it I love playing with so thought I'd grab it while I was there. My wife is over the moon as it really goes with our dining room decor, not.
I haven't made one for quite some time so am looking forward to this and I think this'll be the first build to go over to Spain.
I haven't made one for quite some time so am looking forward to this and I think this'll be the first build to go over to Spain.
Thanks Resident.
Considering that far more complicated transistor circuits still work after more than 40 years in service I suspect that these things will last a good amount of time. The heatsink I use these days really prolongs the life of these things and in my tests only got warm to the touch
Naked transistors are not good when heat needs to be dissipated.
Later this week my aim is to build a couple of circuits up ready for when someone shows some interest, that way there'll be no appreciable delay.
Madmalkav sent me a name of a website that could spin me a few PCBs for a fairly decent cost so may look into that and as alternative to the veroboard I have been using. There's nothing wrong with veroboard of course and I quite like the long, easy to insulate footprint but it may make things easier to build. I could even do kits. If I redesign the footprint it would likely need a project box and the price would go up as they are pricey beasts. I can't really spin a custom shell like Phanteks do unless I had or could justify the expense/running costs of a 3D printer.
Considering that far more complicated transistor circuits still work after more than 40 years in service I suspect that these things will last a good amount of time. The heatsink I use these days really prolongs the life of these things and in my tests only got warm to the touch
Naked transistors are not good when heat needs to be dissipated.
Later this week my aim is to build a couple of circuits up ready for when someone shows some interest, that way there'll be no appreciable delay.
Madmalkav sent me a name of a website that could spin me a few PCBs for a fairly decent cost so may look into that and as alternative to the veroboard I have been using. There's nothing wrong with veroboard of course and I quite like the long, easy to insulate footprint but it may make things easier to build. I could even do kits. If I redesign the footprint it would likely need a project box and the price would go up as they are pricey beasts. I can't really spin a custom shell like Phanteks do unless I had or could justify the expense/running costs of a 3D printer.
Last edited:
Congrats Mike. That's excellent news.
madmalkav said:
That is true. Am looking into this. I've looked before, perhaps at UK options and they've been expensive. China, though can be had for less than £1 a board.
PWM signals are usually of limited current, 5mA absolute max, so when we draw current from multiple fans eventually the PWM signal will deteriorate until it can no longer trigger the attached fans.
Corsair may use BJT transistors rather than a Mosfet, or have some sort of pull down resistor that draws the current away. A BJT draws an amount proportional to its output current, somif it uses 100mA to drive the fan it may need to draw 1mA from the PWM source. I don't own any Corsair PWM fans to check what they use, or what the structure of the fan is.
To boost the PWM you'd use another transistor, typically a logic level mosfet to repeat the signal. A mosfet in an ideal world draws zero current from the source, although in reality it always draws something. You would take a 5mA PWM signal and use the other side of the mosfet that could allow you to draw several Amperes.
With my PWM converter circuit I draw some 40mA for the power transistor base, way above the 5mA allowed by Intel specification. The BS170 I use draws less than 1µA from the source, according to its data sheet. I've never measured the current draw from PWM and even if I wanted to I doubt I've got anything with the required resolution.
Corsair may use BJT transistors rather than a Mosfet, or have some sort of pull down resistor that draws the current away. A BJT draws an amount proportional to its output current, somif it uses 100mA to drive the fan it may need to draw 1mA from the PWM source. I don't own any Corsair PWM fans to check what they use, or what the structure of the fan is.
To boost the PWM you'd use another transistor, typically a logic level mosfet to repeat the signal. A mosfet in an ideal world draws zero current from the source, although in reality it always draws something. You would take a 5mA PWM signal and use the other side of the mosfet that could allow you to draw several Amperes.
With my PWM converter circuit I draw some 40mA for the power transistor base, way above the 5mA allowed by Intel specification. The BS170 I use draws less than 1µA from the source, according to its data sheet. I've never measured the current draw from PWM and even if I wanted to I doubt I've got anything with the required resolution.
Took a look at my Apache PWM fan and it has a PNP BJT transistor ( W1P 01) on the PWM line. It is however in series with a 10k resistor, meaning that at maximum 0.5mA will flow through the transistor. I must say I was surprised to see a BJT there, not a little FET. Maybe it is a cost thing. FETs are more expensive than BJTs and as a system would generally have just one fan attached maybe the designers felt it wasn't really an issue.
Oh in that case you would just need a basic PWM converter for your PWM fans with the extra feedback connector.
You'd plug your PWM fans into your splitter. One of the spare spots would take the PWM converter circuit for your 3 pinners.
Or I could make a hub for the PWM fans if you prefer the neatness.
Something like this super word art
You'd plug your PWM fans into your splitter. One of the spare spots would take the PWM converter circuit for your 3 pinners.
Or I could make a hub for the PWM fans if you prefer the neatness.
Something like this super word art