PWM to Analog converter circuit

[Tealc]

I do somewhere.

I had looked at opamps but think you'd need a rail to rail power opamp and they tend to be a bit expensive, or you'd need to take a volt or two hit on the top end. Might take a look under simulation though and see how well it would work with say a slightly better opamp than a 741. The other tricky thing is that PWM on a PC is not always exactly 5v. I daren't probe my PC with my scope but putting a DMM across my PWM signal showed 3.3v at 100% duty.


In other news feast your eyes on this...

We had some 5mm black braid sleeve in work that needed a new home

I also came across this while I was mooching. It looks more red in real life.

 

[Tealc]

Gave etching another go today.

Did the toner transfer method with ironing to get it to stick. I then patched it up with a 0.8mm paint pen.

Need to get some nail varnish remover on it.

Not much point for this circuit really as it takes ages to print, iron, remove the paper, then touch up and then prepare and etch the copper and it doesn't actually reduce the footprint by much as the fan headers resrict the width. Nevertheless if I need to make a more complicated circuit it could be a way to avoid things like this.

 

[Tealc]

Here's a bit of a side project but in a similar vein.

I've often mentioned diodes as a way of reducing fan speed and this is an example of how to do it in the flesh.

The design is modular so that the diodes amount can be varied between 2 and 5 and give a resultant voltage of 7v to near 10v. It's designed for a watercooling pump.

Anyway here is a video.

 

[Tealc]

I definitely think there's something to be said for ghetto mods. The 12" desk fans tend to be a bit noisy at night and I find blow a little too much even on lowest setting, so if this weather continues after the wet spell they've predicted I might put some of my 120mm fans to use with some PWM goodness to keep the noise down. My desk power supply buzzes a bit so would have to use an old router power supply or something.

I can only imagine the breeze those 11 fans would have created if I'd arranged them into a fan wall. The funny thing is I have another 6 or 7 fans that I could have used but ran out of space and connections. I've no idea why I have so many fans and why I've kept them all. Might be intersting to arrange them into a rough rectangle with cable ties for giggles. Maybe I could create one of those freefall simulators with them.

 

[Tealc]

Hello Lettuce.

Looks like it's picking up the PWM signal as stronger than the RPM tach feedback. This occurs when the fan is driven to low speed and voltage. I'm a little surprised to see it occuring at around 900RPM as normally it's seen when the voltage of the fan gets low enough that it can't make the signal strong enough and 900 I'd have thought would be fine.

My first advice is to increase the speed of the fan by adjusting the potentiometer a touch to see if it settles down. It may be a peculairity of the Gelid fans that causes this.

I also see the fan speed varying a lot between 900 and 2000. That is peculiar behaviour. All circuits I've built have been incredibly stable and vary only by just a few RPM here and there given a stable PWM.

I did test the circuits on my oscilloscope and PWM box before sending but I don't check the validity of the tach signal as that is totally fan dependant, so loading up one of my fans will not give a true reflection of the tach signal. Each fan will have their own hall sensor and circuitry to operate.

I wonder if this is somehow caused by the two fans on the Gelid. Can you check that only one tach wire comes from one of the two fans mounted on the cooler?

All being said I've made several (5 or 6) circuits for Gelids and not had any feedback thus far of any issues so maybe it is a wiring error on my part or something. If you have no joy then please contact me by email and we'll get this sorted.

 

[Tealc]

Seems more stable at the moment.

I should point out that for Gelid circuits such as yours all I do is solder it to another wire, insulate and run it off to the other side of the PWM converter, at no point does it form part of the converter circuit itself. The extra length could be picking up signals from elsewhere I suppose but it really shouldn't be that sensitive.

In addition, just in case you were not aware, the GPU doesn't care what the tach on fan is. All it does is vary the PWM duty cycle in the hope that there is something on the other end to listen to it. You can disconnect the tach and the graphics card will just go on modifying its duty cycle regardless.

 

[Tealc]

Here's a small update.

Thought I'd shove a length of 6.4 heatshrink on the leads to see if it looks better than cable ties. I'm not sure. Maybe it will look better when properly shrunk and not done with a flame. Maybe I will try some smaller bands along the length.

And I ordered some SMD resistors a few weeks back and thought I'd throw a circuit together. I'm a bit of a noob with SMD and as I made a few mistakes and made a bit of a mess. Makes the rear look simpler and aids assembly of the components on the other side. It also saves about 3mm off the height of the board. I really could do with a steadier hand or larger resistors, these are 1206 I think. I think the whole thing looks much better with the bulk of the resistors on the underside, even if it is more difficult to assemble.

I was thinking of videoing myself slapping one of theser together. Not sure how interesting it would be.


@bandyleg - Your converters are on their way to you mate.

 

[Tealc]

Bit of an unneccesary update now.

Have been fiddling around with my Arduino development board and what it can do for me.

The following video shows one such experiment...

...and also has me wobble on about a possible project.

 

[Tealc]

I've been moving house so haven't had internet to join the discussion. Yes moving again...a year, almost to the day, since I moved last time. Really hard work but hopefully the kids will be happier in this house and I can hopefully get a half decent electronics workshop on the go in the garage once I've had time to unpack some stuff. I hope to get started on my fan controller project then since I've actually got the components I need from China (took 2 months for delivery) for it.

Anyway regarding the sharing of information. Sorry but I'm not sharing my many hours of experimentation results with someone who doesn't even ask nicely. There are no problems with Speedfan. Max voltage is supply and minimum is less than a fan will spin at. Tach feedback works down to similar speeds as any convential fan and only messes up once voltage of the tach signal drops below the motherboards ability to detect and determine the signal as valid.

 

[Tealc]

Hi Mate.

The circuit itself isn't ideally suited to driving that amount of load but it should be possible to create a variation of the circuit that changes a component or two and shifts the level of the incoming PWM to 12v but would likely lose tach feedback.

I was experimenting on making a several Ampere fan controller that used a similar circuit with the addition of a couple of components that might work but haven't tested or done anything with it in the real world, mainly because of my home move but also because I'd need to buy parts.

It's not actually that difficult to make a board with two channels, one for fans and one for the pump, once I would have made and proven the design anyway. The incoming PWM signal can be split several times to feed several different devices, especially with my original circuit as it draws so little current from the PWM signal.

You've set up trust so you just need to log in to it, using same name and password as forum. You can then send a message. The forum emails me and then when I reply to you we can chat via our own email.

 

[Tealc]

It's been a while since I made some converters but in the last few days I was asked if I could make 4 PWM converters for single fan outputs.

Here are the boards.

Now using SMD resistors on the underside to cut down on height. Soldering not wonderful but solid and reliable enough. I really need to improve the lighting in the garage over my workbench.

This is how the fans will attach, off board, unlike normal boards where the fans attach on board. I don't have many of these 3/4 pin headers so only use them on single fan designs.

Edit: Oopps wrong image. Corrected.

 

[Tealc]

Right.

Spent a little time yesterday knocking up a test circuit that can handle high loads of approaching 2 Amperes and still give tach feedback.

This is what I came up with. Well actually the first iteration had a 10uH instead of the big inductor choke but that got super hot when loading the circuit down.

I must admit inductors are not something I've ever used or learnt about. I don't have many of them to hand as most on old PCBs are not marked so I tend not to harvest them and even if I did suck them off they can be tricky to measure.

Load, roughly 1.6A.

And the trace on the scope. The lower trace is the PWM and the wavy line is the voltage at the load. It looks very different to the trace on my original circuit, this is all down to the inductor resisting change in the voltage.

I'm not overly impressed at the moment. It sort of works and maintains a fairly decent trace. I can get a range of maybe 4v-9v but it is struggling to do more. The 10uH did better but it still topped out at around 10v. It works better in simulation.

Back to the drawing board I think for the moment.

I'm also thinking about using a different high power PNP (rather than medium power PNP) on the original circuit in darlington configuration, perhaps with another of the same type to boost my current but still get manageable heat. I could always go with a bigger heatsink I suppose.


Update:

Simulation is encouraging with the darlington configuration I mentioned. Darlington configuration, as you all know, is where you have 2 transistors in place of just one. The output of one feeds into the input of the other and you can get very high current gain using the solution.

So I've swapped the B772 for a 2N3906 and TIP42 in Darlington and drive seems to allow 2A with only the loss of 1v in simulation.

I expect the TIP42 to get quite warm. It is a much higher rated device though. Max collector current goes from 3A to 6A and watts dissipated max goes from 12.5W to 65W. The downside of using a high power device is very limited current gain, dropping from maybe 80x down to just 20x or 25x so that's why I need the 2N3906 in there. Without it I'd be lucky to see 0.5A. With it have more than enough.

I would need to modify the stripboard design to incorporate the extra device (it's only a small one) plus have room for a larger heatsink.

 

[Tealc]

An update for Zombiron. I haven't quite finished your converters yet mate. I need to get some hot glue onto the fan connector end so the board doesn't short out with anything. Apart from that they are good to go and have been tested on my oscilloscope as working.

I'm aiming to get them to the Post Office on Friday afternoon for you.

 

[Tealc]

Could you do a unit that is capable of running dcp4 pump rated @ 1.8A

The simplest method would be similar to a circuit I built for Panyan a while back to control the intensity of a strip of LEDs.

This should allow PWM to pass straight to the pump itself and allow it to be regulated by the duty cycle of the CPU. It is not adjustable of course and gives no tach output but if 20% or 30% pump speed is possible without labouring the motor it should be ok. This circuit can drive very high loads.

I don't think there's much need to use an opto-isolator in a circuit such as this as the mosfet works very well at isolating the two voltages from each other and as we use a common ground in PCs there is no point in isolating that off.


I did try the darlington configuration briefly last night but didn't have much success. I may have wired it incorrectly (probably did, or just blew one of the transistors) in my haste to build it on breadboard. As predicted the TIP42 got very hot but I wasn't actually able to adjust the power like in the original circuit.

This high load version will be of some use so I will continue developing and see where it goes. I may even be able to sell one or two to the watercooling aficionados.

 

[Tealc]

Interesting stuff!

Wouldn't that do the fans aswell? Or would a IRFZ44NPbF be better to do all?

This is were it starts to lose me, as I know there's a LOT of variables to specing the correct mosfet in a particular circuit

It would power anything you wish to connect to it but you'd have to lose all tach feedback on the fans. The hall sensor on the fan body needs to have a voltage at all times to operate accurately. PWM fans have this sorted in electronics but 3 pinners and Molex powered pumps do not. The pump will work fine on PWM but the 3 pin fans are much better off on DC volts, hence the converter.

The IRFZ44NPbF should work fine as it is also a logic level and can source enough current. The circuit is not complex enough to worry about speccing a fet that has certain properties. All it needs is to be able to source the current, switch fast enough and be logic level. I only chose IRL510 because I used one before and it did the job. I've no idea what a lot of the specs of fets actually mean and how they matter in circuits.

Managed to do a sketch as to output a pwm signal same as a motherboard does(well hacked others really ),
then added a pot so be able to adjust the duty cycle. Then output the RPM and Duty % to the serial monitor

Was very proud of myself.

No idea why, have a cpu fan on it on my desk just spinning . was tinkering. I can post on here if you want it to test your circuits if its usefull to you! Although going on what you seem to have/ access to, kinda like bringing a knife to a gun battle

My equipment has taken me two or three years to accumulate and is still way short of what it could be, or I'd like it to be. I'm only a very basic hobbyist and tend to learn what I need for a certain project. I do have quite a selection of components though but typically never the right ones for projects I decide to do.

I've only recently begun playing around with Arduino and find it almost like cheating, after working with simple ICs and passives all this time. It's so easy to just throw in a bit of code and do things that would take dozens, if not hundreds of ICs to accomplish. I must admit being quite happy with myself when I managed getting a scavenged LCD panel to display anything I wanted.

You got me thinking about something now. I'm thinking of building a fan controller and wonder about having a PWM duty cycle modifier through Arduino software, where the incoming PWM cycle is checked and multiplied by a factor determined by an external device eg Potentiometer, thermistor or even photo diode. This might be useful if say the duty cycle at idle drops your pump/fan/LEDs too low.

I use a 555 timer built 25kHz PWM box to verify my own PWM circuits. I could have used one of my Arduino boards to do it but I built the circuit before I started Arduino.

 

[Tealc]

Not sure what's going on in that video. Seems to be showing near max voltage throughout. I did hear a squeak somewhere along the line but the fan in shot didn't seem to deviate.

The video quality is even worse than my videos

 

[Tealc]

LOL roxikmode on!!

In this video PWM with 100% duty cicle OUT 12,24 V MAX with IN 11,80V of mother board (need more amper and my circuit go to 20V)
in the video my circuit pilot with 11,80 of mother board connector 4pin fan
4 fan Corsair AF140 quiet edition in overvolt with duty cicle 100%.
squeak if you hear is the low amper of mother board 4pin connector!

So you've created a DC-DC buck boost converter, or utilised an off the shelf part?

I'm not really sure why you would need to overvolt a fan. Running them too much above their designed voltage would wear them out much faster, I'd have thought.

Perhaps you could do a video where you describe in more detail what you are doing. Use speech bubbles if your english isn't that great.

I did look at a DC-DC converter as a method to do what I wanted and I think I posted some info early on in the thread about my experiments but I found that the components are closely matched to the current you are drawing and the voltage you are providing. I felt like I would lose the ability to control fans ranging from 0.05A up to 1A neatly. Sure if I were designing it for one or two of my own fans and I knew the current draw curve then it would have been ok I suppose.

But still a buck converter would be more ideal in this situation, especially one where someone wanted to control a load of fans and maybe a pump or two.

 

[Tealc]

I have 3x 3pin XSPC 120mm 1650RPM fans connected to my radiator.

Do you sell a cable that allows me to run all 3 fans to a fan header on my motherboard but allowing PWM control?

Kind Regards

I usually fit 3 fan spots to my converters anyway so this would be the nearest thing to a standard build for me. It would work just fine off any PWM enabled header.

Drop me a message in Trust if you'd like me to build one for you.

 

[Tealc]

Hi, just found this thread.. Are you still making these I would love to use the on-board VGA to control my Icy Vision on a 290x...

Yes I am. I've made a few specifically for the Gelid. Drop me a message in trust and we'll have a chat about it.

 

[Tealc]

I've put the two boards together for smsmasters and pob9. Just need the black heatshrink put on and these are good to go.

With 4 standard fans attached you can expect fan stall speed-95% fan speed.

I use stripboard because it's cheap and at these sort of frequencies it doesn't make any difference to the function of the circuit. My Oscilloscope agrees with me.

I've not really had professionally made boards priced up but expect that at maybe 20 boards per year they'd be quite expensive and certainly above that which I or anyone else would like to pay for. You need to have thousands made to make it cost effective.