Help in selecting transistor and heat sink

jpborunda

Joined Apr 9, 2014
55
Hi all.
I have been working on designing a basic constant current sink circuit, based on the one in the attachements. My question is about how to select the correct heatsink and power transistor. First I'll try to explain the circuit specs and requirements:
R1 is a 33Ω resistor, Vs goes from 0 to 3.3V in order to generate a current from 0 to 100mA(maximum) in R1, which would be aprox I_RLOAD since Ic≈Ie. The +HV source goes anywhere from 0-200V, which allow RLOAD to vary from 0 to 2kΩ and still maintain the constant current (ideally).
The current will be applied for very small periods of time, say pulse width range from 0 to 500µs, and a frequency of 20 to 100Hz (As the sample figure attached).
Therefore I believe that my worst case scenario, would be when the transistor "sees" 100mA(max current) at the full voltage range of 200V, which turns to be a 20 W power dissipation.
Now, I could use some help with selecting the correct power transistor and heat sink. Where do I start the analysis to select the correct components (transistor and heat sink)? Would a mosfet be a better choice? Any help would be greatly aprecciated.
Jay Pe.
Thanks!

Attachments

• 16.5 KB Views: 48
• 39.8 KB Views: 32

#12

Joined Nov 30, 2010
18,224
A mosfet seems the obvious answer because there is no contamination by the drive current.

First, we need to know if there is any inductance in the load that would throw voltage spikes at shut-off. Next, we need to know that the op-amp can provide enough voltage to turn the mosfet completely on. 12 volts would run just about any mosfet. Then, there is exactly which op-amp. Some of them aren't very good at delivering the highest voltage available.

This is simple enough for a pro, but you have to talk some more so we don't get caught by lack of specifics.

ps, the length of the wire running to the load counts. If it's several feet, that matters.

AnalogKid

Joined Aug 1, 2013
10,176
The good news is that your power level opens up a lot of possibilities.

Your calc is correct for steady state, but your max pulse is 500 us and your shortest repetition period is 10,000 us, so the average power in your transistor is more like 1 W.

You'll want a 300 V minimum rating on the transistor for margin.

At 100 mA the sense resistor has 3.3 V across it, and this comes out of the gate drive budget if you go with a FET. Still, you're not looking for hard saturation so you should be ok with Vcc = 12 V.

A power darlington (either a single device or home grown with two transistors) will require less drive headroom than a FET and reduce the already small base current error by 99%.

For most power transistors, a TO-220 package usually is good for over 1 W dissipation at room temperature without a heatsink. It will get warm, but well within its ratings.

ak

#12

Joined Nov 30, 2010
18,224
Right on the nose with your calculations, AK. Simple enough job, right after we're sure JP isn't running something with a coil in it.

jpborunda

Joined Apr 9, 2014
55
Thanks for the response. Im not familiar with the part of the inductance in the load. Can you elaborate on that part please? The load would be composed physically of a resistor 0-2k, and a small capacitor in parallel (0.01µF) with it, so I would think that there will be no inductance it the load.
With the op amp, I was thinking of getting the MC34071 http://www.onsemi.com/pub_link/Collateral/MC34071-D.PDF, which I believe would be capable of driving the mosfet the way you say.
Sorry if Im leaving some info out, I'll try to be more specific next time.
Thanks a lot!

studiot

Joined Nov 9, 2007
4,998
Forgive me for asking, but what happened to the base - emitter voltage of the output transistor?

jpborunda

Joined Apr 9, 2014
55
Your calc is correct for steady state, but your max pulse is 500 us and your shortest repetition period is 10,000 us, so the average power in your transistor is more like 1 W.
Thank you, Thats in fact GREAT news! I just have a few questions now... I see that I was calculating power for the steady state, but how did you calculated the ~1W power? Is it with the VRMS voltage? And also, for this 1W of power to be correct, would I have to switch the 200V source on and off at the same time/frequency that the input signal of the op amp changes from 0 to X? I mean, what would the power dissipation of the transistor be if I leave the 200V source "on all the time", and just switch the input signal?

@#12: I wont be using any inductive devices, so I think there will be no problem with that.

Last edited:

#12

Joined Nov 30, 2010
18,224
That amplifier should work just fine. It misses the highest supply voltage by less than a volt, so 12 Volts as the power to the amplifier should be plenty. Now let's find a transistor.

ps, 500 us/10,000 us = .05
That's 5% "on" time.
.05 times whatever the steady state power would be is the answer.

What country are you in? We need to find where you can buy parts.

jpborunda

Joined Apr 9, 2014
55
That amplifier should work just fine. It misses the highest supply voltage by less than a volt, so 12 Volts as the power to the amplifier should be plenty. Now let's find a transistor.

ps, 500 us/10,000 us = .05
That's 5% "on" time.
.05 times whatever the steady state power would be is the answer.

What country are you in? We need to find where you can buy parts.

I see.. Thats exactly what I wanted to know. Thanks. I live in Mexico, Chihuaha, but I live at a border from El Paso Texas, so I actually have a friend that buys all my ICs for me and I just pick then up. I always struggle finding stores with no minimum quantity requiered by the way.

jpborunda

Joined Apr 9, 2014
55
That amplifier should work just fine. It misses the highest supply voltage by less than a volt, so 12 Volts as the power to the amplifier should be plenty. Now let's find a transistor.

ps, 500 us/10,000 us = .05
That's 5% "on" time.
.05 times whatever the steady state power would be is the answer.

What country are you in? We need to find where you can buy parts.
I see.. Thats exactly what I wanted to know. Thanks. I live in Mexico, Chihuaha, but I live at a border from El Paso Texas, so I actually have a friend that buys all my ICs online for me and I just pick then up. I always struggle finding stores with no minimum quantity requiered by the way.

#12

Joined Nov 30, 2010
18,224
As far as power while it's "off".
You can leave the 200 volts on all the time and nothing will happen.
Only when you turn the transistor on will heat happen.

ps, the shipping charges for Mouser are too high, in my opinion, but you can buy one part if that's all you need.
pss, those transistors are rated for more voltage than you need, but they are the cheapest in the category you need.

jpborunda

Joined Apr 9, 2014
55
As far as power while it's "off".
You can leave the 200 volts on all the time and nothing will happen.
Only when you turn the transistor on will heat happen.
Ok, well yes that actually makes scence if I think about it... I guess my problem was simpler than I tought. All there is left to do is the component selection. Could there be a problem with going with a transistor thats voltage rated much higher than what I'll be using it for?
Thank you very much.

#12

Joined Nov 30, 2010
18,224
The problem with higher voltage ratings won't affect this.
Higher voltage rating for a BJT means lower current gain, but you aren't using a lot of current.
Higher voltage rating for a MOSFET might mean it's slower, but you aren't pushing megahertz speed.

jpborunda

Joined Apr 9, 2014
55
Ok, I see. I guess it will come down to costs and time of arrival for me.
Thank you, you've simplified a lot of things for me today. As you can imagine Im sort of a beginner, so some insight from someone with more experience is really useful!
Thanks!!!!!

#12

Joined Nov 30, 2010
18,224
Like I said in post #2, it's simple enough for the pro's.
You just need to talk more so we can get the bigger picture.

ps, welcome to AAC. We're open 24 hours a day.

C'mon back if you need help with the resistor between the op-amp and the transistor, whichever one you choose.

THE_RB

Joined Feb 11, 2008
5,438
I wouldn't worry about a darlington. Any decent modern TO-220 power transistor suitable for higher voltages will have a gain of at least 100 at such a low collector current (100mA). So only needs 1mA or less to drive it's base.

My vote would be for a good rugged HV transistor with a proven track record, 2SC2335 is a good one.

A bipolar transistor should be more stable than a FET in this application because of the low (and relatively fixed) Vbe.

AnalogKid

Joined Aug 1, 2013
10,176
Your worst case on time is 0.5 ms and your fastest repetition rate is 100 Hz, or 10 ms. That's a 20:1 ratio so the average power is 1/20th of the steady state power. By coincidence your steady state power is 20 W so the two 20's cancel out to 1 W average power; the two 20's are not related and your next project probably will have less convenient math.

The bipolar transistor base-emitter voltage is alive and well. For such a low collector current it probably is around 1.3 V, considerably less than a MOSFETs Vgs for saturation (worst case).

For load driving I prefer bipolar transistors; they are just plain more reliable, more resistant to transients and abuse, etc. (FET-people, get over it). Anything whose internal structure and operation depends on the integrity of a sheet of glass that is 10-20 atoms thick is just asking for trouble.

Yes that ultra-low drive current is appealing in some situations. But with a darlington the already very small base current error is reduced to less than 0.1%. Good enough?

ak

jpborunda

Joined Apr 9, 2014
55
Your worst case on time is 0.5 ms and your fastest repetition rate is 100 Hz, or 10 ms. That's a 20:1 ratio so the average power is 1/20th of the steady state power. By coincidence your steady state power is 20 W so the two 20's cancel out to 1 W average power; the two 20's are not related and your next project probably will have less convenient math.

The bipolar transistor base-emitter voltage is alive and well. For such a low collector current it probably is around 1.3 V, considerably less than a MOSFETs Vgs for saturation (worst case).

For load driving I prefer bipolar transistors; they are just plain more reliable, more resistant to transients and abuse, etc. (FET-people, get over it). Anything whose internal structure and operation depends on the integrity of a sheet of glass that is 10-20 atoms thick is just asking for trouble.

Yes that ultra-low drive current is appealing in some situations. But with a darlington the already very small base current error is reduced to less than 0.1%. Good enough?

ak
Thanks for the explanation about the power, I understand now.
I also considered using a BJT first since it would require less voltage than a FET, but the FET would reduce the current error and I read somewhere they are better for switching circuits. I was kind of weighting both options but I think I'll go with a Darlington pair, since as you said, it will provide a smaller current error.

I would just need to make sure that the Op amp provides enough current/voltage to the base right?

Thank you all