Swapping out transistors in fan controller.

[chozo4]

I don't hide the fact that I don't understand a lot about transistors in general as transistor logic is still something I'm learning as I go along. Therefor there will be a lot of questions in here.

However, I've been pondering over time to swap out the transistors in my Sunbeam PL-RS-3 fan controller. So far I've swapped the 470uF 'HUAHONG' capacitors out with Rubycon YK 16v 1000uF for piece of mind. The three channels are rated 30watts and each driven by a heatsinked TIP31C transistor. The reason I want to look into changing them out are for a couple reasons.

1) Lower voltage drop (currently about 1.2v)
2) Lower heat output

Currently the TIP31C is rated for the following:
Polarity: NPN
VCBO 40v
VCEO Max: 40v
VEBO: 5v
Saturation: 1.2v
Max Current: 3A
GBP: 3mhz
Package: TO-220

From what I understand, reducing the saturation voltage would produce less heat generation and voltage drop. However, what I don't understand is if a higher maximum current capability would result in less heat generation as well due to running lower than its' peak current. I'm blindly assuming that being a semiconductor it will be similar to a diode whereas higher current capability will result in lower dropout voltage at lower current throughput.

What I'm aiming at is simply changing them out for something that won't produce as much waste heat and a lower dropout voltage. My fans already run at barely 11volts because of it. Considering they are being used for 12v linear voltage control of the fans based on a PWM input signal the majority of values such as maximum voltage and Gain Bandwidth is non-applicable.

I'm currently looking at the KSD526Y series Transistors by Fairchild with the following:
Polarity: NPN
VCBO: 80v
VCEO Max: 80v
VEBO: 5v
Saturation: 0.45v
Max Current: 4A
GBP: 8mhz
Package: TO-220

This is the closest match for a higher output/lower saturation voltage transistor in the TO-220/TO-220-3 package with higher VCBO/VCEO breakdown voltages and matching VEBO. Am I wrong to assume that the VEBO is also simply a maximum breakdown voltage only when reverse biased and that raising it wouldn't be an issue? Raising the VEBO would increase the possibility for a lower saturation voltage as well.

.. if I'm off my rocker and being stupid please tell me. I'm still trying to understand transistors so probably sound completely bonkers with my assessment and thoughts (also exhausted right now so probably don't make much sense... *yawn*). Some helpful guidance would be appreciated for when I decide to do this in the near future.

[stj]

Re: Swapping out transistors in fan controller.

i wouldnt bother.

if the transistors are being switched full-on using pwm,
then you could use logic-level fets to make a big difference though.

[chozo4]

Re: Swapping out transistors in fan controller.

i wouldnt bother.

if the transistors are being switched full-on using pwm,
then you could use logic-level fets to make a big difference though.

I had thought that as well but I find they aren't in fact. There is a small bi-transistor circuit at the PWM input to convert the PWM to a current-based control for the base current. So it isn't being switched full on like would be expected out of a FET as originally thought. This makes sense considering the use of transistors otherwise they would have not added the PWM->Current control for the transistors base current and instead had the PWM control the fets directly.

Additionally, each of the transistors are alternatively controlled using manual controlled rheostats in manual mode where fets wouldn't be useable due to the hard on/off states. Again, controlling current flow to the base.

[eccerr0r]

Re: Swapping out transistors in fan controller.

If the metal heatsinks aren't getting hot, you aren't losing that much, and no point in swapping them out.

That leaves the question... what fans are you using? If you're just using the typical 2W PC fans you're not really losing much on the TIP31's...

And if it's not PWMing, no matter what transistor you have their, it will dissipate the same power until you're on full-on - then sat voltage becomes an issue - and you might well just get rid of the transistor and hook it straight up to the power source.

[chozo4]

Re: Swapping out transistors in fan controller.

If the metal heatsinks aren't getting hot, you aren't losing that much, and no point in swapping them out.

That leaves the question... what fans are you using? If you're just using the typical 2W PC fans you're not really losing much on the TIP31's...

And if it's not PWMing, no matter what transistor you have their, it will dissipate the same power until you're on full-on - then sat voltage becomes an issue - and you might well just get rid of the transistor and hook it straight up to the power source.

The fans the controller will be handling soon once they arrive are actually a pair of...

Delta AFC0912DE's
Size: 92mmx38mm
Volt: 7v~12v
RPM: 6000 RPM
CFM: 160.22 CFM
Noise: 63.0 dba
Watt: 30W / 37.20W Startup
Amps: 2.50A

(already using a pair of 150cfm deltas on CPU heatsink and exhaust but those are only 17.4watt. They are not on the fan controller but handled by the 2 voltage controlled fan headers respectively. The fan controller only handles a pair of 4-5watt intake fans at the moment which isn't much)

I was also looking to lower the saturation voltage to reduce the voltage drop as well which was another of the reasons to swap out with different transistors. I won't be running them full on at 100% after windows loads so as to control them using speedfan. If I control them using pure PWM directly then I find that these delta's can be a bit noisy at lower speeds without a different PWM frequency than normally supported on many motherboards. That is where controlling them using voltage comes into play whereas they are much quieter at 4.5-5v (only needs 7v to start, can be lower after) than reducing them to the same speeds using PWM control.

If anything the more subtle step up from the TIP31c would be the TIP41c overall to cover the startup wattage without pushing the tip31's too hard. I don't have much far as fan control until I get into windows so they otherwise run full bore or close to it until then. This makes tests with memtest rather... loud unless manually controlled at that point.

[eccerr0r]

Re: Swapping out transistors in fan controller.

Again you'll only be losing power on saturation voltage when the transistor is 'on' 100%. If you're not running 100% all of the time, then the loss during saturation is negligible because you're not saturating the transistor. Because when PWM'ing it's either on or off, then saturation voltage is a very important aspect.

Also remember, when dealing with transistor dissipation, it is transistor dissipation, not dissipation of the device it controls. Now I don't know what the fan is like when you're running at a lower voltage, but likely the current draw will be lower when you reduce the voltage. Though at saturation, using your 2.5A fans will result in a loss of 1.2V*2.5A = 3W. This is a bit on the high side and may need a heat sink if you're running it for more than a minute or so. But when running at a lower speed, perhaps you want 6V to go to the fan. Then at this point saturation voltage makes absolutely no difference as you're not saturating. Assuming a 12V supply here: You're dropping 6V across the transistor. Now the fan will likely draw less current with the reduced voltage (assuming it's more like a linear device than a SMPS), perhaps 1.25A. Now the transistor needs to dissipate 6V * 1.25A = 7.5W - this now clearly exceeds that what a TO220 can dissipate without a heatsink especially if you're generally going to run it like this most of the time. Typically this curve looks like a bell curve with a linear load - the transistor dissipates the least when it's fully on and fully off (obviously) and in the middle is where it hurts the transistor the most. (Also not accounted for: base current draw. These transistors have low betas and actually the base power draw is somewhat significant, but neglected as it's still probably less than 10% of total power loss. The GBP is not a very useful number here as if you're not running PWM you have to deal with the DC analysis beta.)

The result?
1. You need to PWM if you're worried about power consumption and want to reduce speed.
2. If you want to keep simplicity of no PWM, you need to put the TIP31s (or any other transistor) on a fairly beefy heatsink but the transistors will work just fine.

[chozo4]

Re: Swapping out transistors in fan controller.

Understandable and you've made your point. Still learning some of the ropes with transistors so this information proved very useful. I guess one could consider my isue... ECC Corrected

[But really, thanks - it's greatly appreciated! ]

Image attached of the unit as it comes with heatsinks pre-applied.

[stj]

Re: Swapping out transistors in fan controller.

if you have a pwm output fed into a circuit that converts it to current for the transistors.
maybe you can just take the pwm and feed some logic-level fets with it.
then take the fet outputs through a fast diode and into a smoothing cap to convert the pulses to a stable voltage.
maybe.......

[chozo4]

Re: Swapping out transistors in fan controller.

if you have a pwm output fed into a circuit that converts it to current for the transistors.
maybe you can just take the pwm and feed some logic-level fets with it.
then take the fet outputs through a fast diode and into a smoothing cap to convert the pulses to a stable voltage.
maybe.......

That's an interesting idea really and was thinking more on that. If anything I could breadboard up a mockup of one of the controllers like that and see how it works out.

[eccerr0r]

Re: Swapping out transistors in fan controller.

After staring at the actual device more, it does look like it may be running in PWM mode if you have a PWM input. I'd have to look at the reverse side of the board or do a full reverse engineer to see if it's correct but there's no reason why that PWM signal couldn't be used to generate a PWM base signal directly instead of converting it to a DC base signal and linearly controlling the transistors. Are you sure it's linearly controlling the transistors?

Ideally you should do a circuit extract and see how it's connected. You may very well be able to cheaply modify this circuit to use MOSFETs if the fan controller used the TIP31s as common emitter drivers. However the likely case is that it's using them in emitter follower mode and fairly extensive changes would be needed to convert to MOSFET including a need to charge pump (unless you want to really modify and use P-channel MOSFETs which would simplify the re-design a bit).

[budm]

Re: Swapping out transistors in fan controller.

BTW, can we see the bottom side of the board?

[chozo4]

Re: Swapping out transistors in fan controller.

BTW, can we see the bottom side of the board?

Aye, will see if I can expose it later on today or tomorrow. The manufacturer pasted on a foam pad with strong adhesive on the bottom of the board. Was a pain in the rear just to get a 2x2 inch patch open and clear the paste out so I could change the capacitors a while back.