Transistor biasing - fan mod - UK Vintage Radio Repair and Restoration Discussion Forum

WaveyDipole · 7th Apr 2015, 6:30 pm

I am working with a circuit that drives a fan. The simplified version is in attachment Fan1.jpg. The circuit is fed with a positive voltage of 24v. Normally, the thermostat is closed causing the base of the transistor to be at zero volts. I have omitted the extra logic circuitry that sets the signal low but the transistor is turned off and the fan does not run. As the termerature rises and the thermostat is activated, the connection to earth is severed, causing the signal to rise to 12v which activates the transistor producing about 12.5v at the emitter which drives the fan.

In order to maintain the original logic circuit and to run the fan continuously at a lower RPM but also activate it at full speed when the temperature rises sufficiently as per original design, I have modified the circuit as per the diagram in Fan2.jpg. I added a 1N4148 diode to isolate the base of the transistor from earth via the thermostat while still passing a signal to the base from the logic ciruit. I realise that there would be a 0.7v drop across the diode, but figured that this should not affect fan speed noticeably. I also added two biasing resistors to bring the base to a suitable voltage. However I have observed something curious.

I started with RX1 and RX2 at 1k. This produced a base voltage of about 12v as expected. I varied this to RX1 of 1.2k and RX2 at 680ohm. This ran the fan slower but still rather noisy - but when the thermostat was activated the fan ran faster but not at full speed. I surmised that at 680ohm RX2 was dragging the signal to earth. I increased RX1 to 27k and RX2 to 10k. The speed increased with the activation of the thermostat but still did not reach full speed and the voltage at base was still only around 9v. I increased the resistors to 47k each. This now produced 12.5v at the anode end of the diode, but about 1.1v at the base and emitter of the transistor. I experimented with other values and found that RX1 at 39k and RX2 at 47k gave around 5v which ran the fan very nicely and quietly. When the thermostat was activated, the base voltage was close to what it was orignally and about 12.5v at the emitter of the transistor, which drives the fan at full speed.

I appear to have arrived at a circuit that works, but I do not understand quite how it works. A potential divider of two identical resistances should produce a voltage mid-way between earth and the full potential which is 24v. So why was I getting only 1.1v? Reducing the value of RX1 should increase the voltage above mid point but we are still way below mid-point. My main concern is whether it is safe to run like that?

ukcol · 7th Apr 2015, 8:18 pm

Quote:

Originally Posted by WaveyDipole

So why was I getting only 1.1v?

The TIP31C is not getting enough current via Rx1 to get the circuit to behave as you would like it. Replacing the TIP31C with a power darlington may be one way of improving things.

John_BS · 7th Apr 2015, 8:30 pm

Yes: we don't know how much current your fan takes, nor the precise current gain of the TIP31c (around 100), but from your measurements we can deduce that the base current taken by the TIP31c is about 0.3mA for 5V on the fan. It is this base current which is affecting your voltage-divider calculations.

It's also not clear from your circuit how the IC is operating: the thermostat shorts its output to ground, which may or may not be OK, depending on which opamp is used and how it is biased. We assume that the opamp supplies about 12V to set the fan speed at "high"?

John

WaveyDipole · 7th Apr 2015, 9:20 pm

John, yes, I measured the output of the op amp at 12.5v when the thermostat was disconnected. The other part of the logic circuit is another op amp which has its non-inverted input set to zero when the thermal switch makes a connection to earth. This I beleive would turn off its output which is fed into the non-inverted input of the op amp shown. This would, in turn, switch off output to zero I think. I expect that the resisor network around the two op amps sets the output voltage. It seems to be a rather rather over-complicated circuit just to turn a fan on and off... Incidentally, the fan is 12v at 2.6w.

ukcol, that explains it and makes sense perfect sense. The transistor is not getting fully turned on and therefore not behaving as one might expect. The power darlington is an interesting idea and I believe I have one in my bits box somewhere, however in the meantime, I have found a much simpler solution on another thread. It involves just adding one 1k resistor in series with the thermal switch. I have tried this and it seems to work nicely, so I will opt for this solution.

I have attached the complete circuit and the one resistor mod. I might even opt for a 680ohm resistor and a 500ohm preset which should allow some control over the slow fan speed.

Herald1360 · 7th Apr 2015, 11:19 pm

LM339 is an open collector NPN output, so shorting its output to ground will only affect the external circuit.

Otherwise it still seems unnecessarily complicated for what's required.

Voltage rails don't make sense on the 339, it can't operate between ground and 0V!

WaveyDipole · 8th Apr 2015, 8:57 pm

Herald, I just checked the ciruit diagram. Actually the bottom left rail is held at 15v via a zenner. It was my error in and assumed it was 0v, but as you say that doesn't make sense.

7th Apr 2015, 6:30 pm	#1
WaveyDipole Nonode Join Date: Mar 2013 Location: Leicester, UK Posts: 2,535	Transistor biasing - fan mod I am working with a circuit that drives a fan. The simplified version is in attachment Fan1.jpg. The circuit is fed with a positive voltage of 24v. Normally, the thermostat is closed causing the base of the transistor to be at zero volts. I have omitted the extra logic circuitry that sets the signal low but the transistor is turned off and the fan does not run. As the termerature rises and the thermostat is activated, the connection to earth is severed, causing the signal to rise to 12v which activates the transistor producing about 12.5v at the emitter which drives the fan. In order to maintain the original logic circuit and to run the fan continuously at a lower RPM but also activate it at full speed when the temperature rises sufficiently as per original design, I have modified the circuit as per the diagram in Fan2.jpg. I added a 1N4148 diode to isolate the base of the transistor from earth via the thermostat while still passing a signal to the base from the logic ciruit. I realise that there would be a 0.7v drop across the diode, but figured that this should not affect fan speed noticeably. I also added two biasing resistors to bring the base to a suitable voltage. However I have observed something curious. I started with RX1 and RX2 at 1k. This produced a base voltage of about 12v as expected. I varied this to RX1 of 1.2k and RX2 at 680ohm. This ran the fan slower but still rather noisy - but when the thermostat was activated the fan ran faster but not at full speed. I surmised that at 680ohm RX2 was dragging the signal to earth. I increased RX1 to 27k and RX2 to 10k. The speed increased with the activation of the thermostat but still did not reach full speed and the voltage at base was still only around 9v. I increased the resistors to 47k each. This now produced 12.5v at the anode end of the diode, but about 1.1v at the base and emitter of the transistor. I experimented with other values and found that RX1 at 39k and RX2 at 47k gave around 5v which ran the fan very nicely and quietly. When the thermostat was activated, the base voltage was close to what it was orignally and about 12.5v at the emitter of the transistor, which drives the fan at full speed. I appear to have arrived at a circuit that works, but I do not understand quite how it works. A potential divider of two identical resistances should produce a voltage mid-way between earth and the full potential which is 24v. So why was I getting only 1.1v? Reducing the value of RX1 should increase the voltage above mid point but we are still way below mid-point. My main concern is whether it is safe to run like that? Attached Thumbnails Last edited by WaveyDipole; 7th Apr 2015 at 6:49 pm.

7th Apr 2015, 8:30 pm	#3
John_BS Octode Join Date: Dec 2008 Location: Wincanton, Somerset, UK. Posts: 1,782	Re: Transistor biasing - fan mod Yes: we don't know how much current your fan takes, nor the precise current gain of the TIP31c (around 100), but from your measurements we can deduce that the base current taken by the TIP31c is about 0.3mA for 5V on the fan. It is this base current which is affecting your voltage-divider calculations. It's also not clear from your circuit how the IC is operating: the thermostat shorts its output to ground, which may or may not be OK, depending on which opamp is used and how it is biased. We assume that the opamp supplies about 12V to set the fan speed at "high"? John

7th Apr 2015, 9:20 pm	#4
WaveyDipole Nonode Join Date: Mar 2013 Location: Leicester, UK Posts: 2,535	Re: Transistor biasing - fan mod John, yes, I measured the output of the op amp at 12.5v when the thermostat was disconnected. The other part of the logic circuit is another op amp which has its non-inverted input set to zero when the thermal switch makes a connection to earth. This I beleive would turn off its output which is fed into the non-inverted input of the op amp shown. This would, in turn, switch off output to zero I think. I expect that the resisor network around the two op amps sets the output voltage. It seems to be a rather rather over-complicated circuit just to turn a fan on and off... Incidentally, the fan is 12v at 2.6w. ukcol, that explains it and makes sense perfect sense. The transistor is not getting fully turned on and therefore not behaving as one might expect. The power darlington is an interesting idea and I believe I have one in my bits box somewhere, however in the meantime, I have found a much simpler solution on another thread. It involves just adding one 1k resistor in series with the thermal switch. I have tried this and it seems to work nicely, so I will opt for this solution. I have attached the complete circuit and the one resistor mod. I might even opt for a 680ohm resistor and a 500ohm preset which should allow some control over the slow fan speed. Attached Thumbnails Last edited by WaveyDipole; 7th Apr 2015 at 9:34 pm.

7th Apr 2015, 11:19 pm	#5
Herald1360 Dekatron Join Date: Feb 2007 Location: Leominster, Herefordshire, UK. Posts: 16,536	Re: Transistor biasing - fan mod LM339 is an open collector NPN output, so shorting its output to ground will only affect the external circuit. Otherwise it still seems unnecessarily complicated for what's required. Voltage rails don't make sense on the 339, it can't operate between ground and 0V! __________________ ....__________ ....\|____\|\|__\|__\_____ .=.\| _---\__\|__\|_---_\|. .........O..Chris....O

8th Apr 2015, 8:57 pm	#6
WaveyDipole Nonode Join Date: Mar 2013 Location: Leicester, UK Posts: 2,535	Re: Transistor biasing - fan mod Herald, I just checked the ciruit diagram. Actually the bottom left rail is held at 15v via a zenner. It was my error in and assumed it was 0v, but as you say that doesn't make sense.