Ya... that'll do it. The best thing to do when you mount it on a heatsink it to use a mica insulator between the transistor and the heatsink. There are better materials than mica, but it doesn't really matter for this application. And you can also get these plastic bushing things that will isolate the screw from the mounting hole on the transistor so that the transistor has no electrical connection to the heatsink.
Rebuilt the circuit over the weekend. I forgot to put in the LM741 and powered it up - fan spun at full speed - ie. 12V coming through. Looked at it and saw I hadn't put the chip in. Put it in and no change - still 12V coming through fan output. Will be doing some checking of all connections today. Are there any points I can measure to narrow down the cause of the problem? btw. I tripled checked the value coming on the CPU FAN header - 5V. At the moment I am using a spare power supply and just using the 5V line for the CPU FAN header. Using normal molexs for 12V.
I'd check your wiring in the output stage. Without the op-amp in place there should be nothing to "pull-up" on the base of the 2N3904 to start any current flowing. just incase to missed it, the pinouts of the two transistors are different. For the MJE2955T: 1 - Base 2 - Collector 3 - Emiter For the 2N3904: 1 - Emiter 2 - Base 3 - Collector
Yes you were right. All fixed and tested. My results are shown below - I think I may need to add a couple of output transistors in parralel. How many do you think I need judging from the results below and what resistors are connected to the collector? In your diagram you have 01 1W Test A: CPU Fan Header : 5V (Via PSU) Fan Output: ~5V No. of Fans connected: 0 Test B: CPU Fan Header : 12V (Via PSU) Fan Output: ~12V No. of Fans connected: 0 Test C: CPU Fan Header : 12V (Via mobo fan header) Fan Output: ~11.9V No. of Fans connected: 0 Test D: CPU Fan Header : 12V (Via mobo fan header) Fan Output: ~11.7V No. of Fans connected: 1 Test E: CPU Fan Header : 12V (Via mobo fan header) Fan Output: ~9.7V No. of Fans connected: 2 Test F: CPU Fan Header : 12V (Via mobo fan header) Fan Output: ~7.4V No. of Fans connected: 3
The circuit seems to be working the way it should but I'm surprised by the voltage drops, when I built and tested it in my lab I only got about a volt or so drop with a 3A load. You shouldn't need to parallel output transistors for the 2A output you initially said you wanted. Can you give me voltage measurements at the 2 inputs to the op-amp and the output of the op-amp (pins 2, 3, and 6 IIRC). Also voltages at the collector and emitter of the 2N3904. Make these measurement with lots of load (3 fans), so the voltage drops show up easily. I'll have a look at the design again to see if there's an easy way to give the output stage a little more oomph without having to parallel the MJE2955Ts.
With 3 Fans (fans rated at 0.55A each) - CPU Fan Header - 11.95V LM741 - - Pin 2 - 4.85V - Pin 3 - 4.85V - Pin 6 - 7.3V (varies +/- 0.2V) 2N3904 - - C - 11.45V - E - 6.6 V Fan output voltage - - 7.4V btw. what is the value of the resistor in the extra power circuit? I don't mind an extra MJE2995T Hope this helps biff!
thanks. Any luck with those voltage figures? Did you manage to reference them back to the circuit design? cheers, alvin
Hi alvins, sorry for all the delays but we got into a busy stretch at work and I'm also getting ready to move at the end of the month. I've been going over the numbers you sent and things just dont add up. The two 10K resistors that go from the red lead from the cpu fan header to groung form a voltage divider. The voltage at the junction of the two resistors should be exactly half of the input voltage, or essentially 6V if the voltage at the cpu fan header is 11.95V, but you're getting 4.85V instead. The same also applies to the resistors between the black lead to the cpu fan header and the output. If the black lead is indeed 0V then the same division should take place in that the junction of the two resistors should be half of the output. So you should be seeing 3.7V there and not 4.85V. Now to get a little more technical... Another thing I find strange is that both the inputs are exactly the same voltage. Now, under normal circumstances this is precisely whats supposed to happen but not when the output stage is being driven too hard as what may be appearing to happen. Let me see if I can explain this. Ignore the odd voltages you are geting at the inputs for a moment. Assume you have 10Vdc on the red input lead and 0V on the black input, through the voltage divider there should be 5V on pin 3 of the op-amp. Say at this precise moment the voltage at the output stage (the collector of the MJE2955T) is 8V. Since there is another voltage divider between the output and the black input lead the voltage at pin 2 of the op-amp should be exactly inbetween the two voltages or 4V in this case. If the voltage on pin 3 is higher than the voltage on pin 2 then the output of the op-amp (pin 6) will rise, driving the output stage harder because the op-amp "wants" to make the two inputs the exact same voltage. In order fo this to happen the collector of the MJE2955T would have to be 10V so that when it is divided there will be 5V on pin 2. If there is too much load on the circuit then the 10V output may not be achievable in which case the voltage at pin 2 will be something below 5V. Since the voltage at pin 3 is now higher than pin 2, pin 6 will continue to rise trying to reach the equilibrium but since that wont happen now pin 6 will sit at as high of a value as it can which is typically about a volt or two less than the supply voltage. Now lets relate this back to whats happening in your circuit. If we assume that the reason that there is only 7.4V on the output of the MJE2955T is because its too heavily loaded then not only should we see pin 3 of the op-amp have a voltage that is higher than pin 2 but we should also see pin 6 at 10.5V or so trying its darnedest to get that ouput voltage higher. But we are infact seeing pin 2 and 3 at the same voltage and pin 6 is hovering at a lax 7.3V, so it would seem that the op-amp is perfectly happy sitting where it is. So we're going to have to trouble shoot this circuit to see where the problem lies. Firstly recheck all the resistor values and that all the caps are the proper orientation. Also lets take the mobo out of the picture, remove any load from the output of the circuit and connect the red input directly to the 12v rail of the psu and the black input to the ground of the psu (the black wires) and let me know what the voltages are that you took earlier. For all voltage measurements keep the black lead of your multimeter on the ground of the psu (the black wires) so all measurements have the same reference. I promise I'll be more prompt with my replies this time around . P.S. sorry for the dodgy explanation of how the circuit works... so let me know if you need anything cleared up. Also I did refine the output a bit using a FET instead of a transistor so that loading isn't an issue but i suppose this wont be an issue now that you have two MJE2955Ts paralleled.
With 0 Fans - CPU Fan Header - 11.65V LM741 - - Pin 2 - 5.75V - Pin 3 - 5.75V - Pin 6 - 2.59V 2N3904 - - C - 11.05V - E - 1.94 V Fan output voltage - - 11.41 V
These voltages seem bang on since the voltage at pin3 is about half the input voltage on the red input and the voltage at pin2 is about half of the voltage of the output. If you give this circuit some load do the voltages start to shift to what you posted earlier? (remember to make all voltage measurments with the black lead of your multimeter on ground) The voltage at pin3 should always be at half the voltage of the red input lead regardless of load since it has no connection with the output. Also if the voltages do start to shift, take a measurement from ground to the black input lead as this voltage level does have a bearing on the output voltage.
Thanks for all the help biff. Was in a bit of a rush last time - sorry for being a little pushy with constant messages . Yes as soon as I put load the voltages shift towards my previous values. The pin3 value is connected to cap and 10k resistor before being connected to ground as per your diagram. Reading from black input lead to ground reads 0V since black input lead is also ground. This is with testing via PSU. If I had this connected to CPU header I don't think this would change.
I didn't get that impression at all. I wouldn't have started this whole thing if I didn't want to help in the first place. If the voltage at pin3 is dropping with increasing load then I'm thinking there must be a component either out of spec or just plain not working right. The only connection pin3 has with the output is either the two diodes across the inputs or the op-amp itself. Try removing the two diodes and see what happens. These diodes have no bearing on the functionality of the circuit, its just considered good practice but if everything works fine after removing them then I wouldn't be too worried about replacing them. If that doesn't make a difference then we'll change out that op-amp. Don't go running out to get a new one just yet though. The LM741 has a low input impedence compared to other op-amps out there, and although it still should be high enough (if in spec) as not to affect this circuit maybe another choice would be more appropriate.
Some good news! Removed the two diodes and the fan controller is working alot better. It seems the max voltage is maxing out at 9.8V. When the CPU header is ranging from 5V to 9.8V the controller is working perfectly. When the CPU header is above that - the fan controller still only puts out 9.8V. Just did some measuring at pin3 - this is always 5.85V as you said it should be.
Good stuff! Thats the one problem out of the way now we gotta get that max voltage higher. Now that things are working more normally, can you send me all of the voltages again when when fully loaded? Hopefully we'll just have to tweak a resistor value.
With 4 Fans - CPU Fan Header - 11.95V LM741 - - Pin 2 - 4.25V - Pin 3 - 5.95V - Pin 6 - 11.12V 2N3904 - - C - 11.3V - E - 10.25V Fan output voltage - - 8.61V (the previous value of 9.8V was with 3 fans) Wonderful world of internet - someone designing and diagnosing a circuit for someone else who is probably on the other side of the world.
Well I'm all moved... now to unpack . Anyway, I'm not sure what to tell you as to why you're not getting a high enough output voltage. When I built this circuit I lost less than a volt through the output transistor and that was with more load than what you're using. The problem is that the 2N3904 and 100ohm resistor isnt sinking enough current through the base of the MJE2955T. There are two ways to go about fixing this. 1 is to beef up this part of the circuit so that there is no doubt that it will be able to sink more than enough current, but this means higher wattage stuff, ie. a bigger transistor inplace of the 2N3904 and more than likely higher wattage resistors. The other way to go about it is to change out the MJE2955T for a FET. A FET is like a transistor but its output is controlled by a voltage and not by a current. So this way we dont need to beef up the 2N3904 and there will be less wasted power and heat. If it were me I would go the way of the FET, but a solution can be reached either way so it's your call. Let me know which way you want to go and I will do up some drawings. Sorry for all the headaches this is causing you but transistors parameters are very unpridictable. For instance the gain of the 2N3904 can be as high as 300 or as low as 15.... thats a 2000% difference! and when you add more transistors into the circuit this number could get alot bigger. So it makes it hard to come up with exact values. Basically the two methods you can take is design way overkill so that in a worst case scenario the circuit will still perform the way it should but this way could get into a lot of wasted heat and resources, or hope you get "typical" components and all will be well with not so much waste. [edit] this is a long shot but if you have a 220ohm or 270ohm resistor handy try puting it in between the output of the op-amp (pin6) and the base of the 2N3904. There may be an issue that the opamp has such a low output impedence that it is "overwhelming" (not a great choice of wording i guess) the 2N3904. If you got one handy its worth a try. [edit]
