BJT as a switch, power consumption?

Thread Starter

Travm

Joined Aug 16, 2016
266
When using a BJT as a switch, power consumption is the product of Ic and Vce(sat)?

I'm looking at using a TIP127 to power a 26w resistive load @12v.
26w/12v = 2.17A
TIP127 Max Ic = 5A, so it "should" be able to switch this. But when I start figuring if it can handle the power, i'm not sure this is right.
TIP127 Pmax = 2W
Vce(sat) = 2v (max) @ 12ma Ib, Ic=3A
Vce(sat) x Ic = 2v X 2.17 A = 4.34W

Is this just a bad choice of transistor? or am I calculating the power dissipation wrong?

Should I just use a Logic Level Mosfet instead?

The load is a small heater, the transistor is to be driven by a MCU, and if necessary I can use another small transistor to drive the base, but the TIP127 should be driveable straight from MCU pin.
 

pmd34

Joined Feb 22, 2014
368
Hi Travm, Mosfets are usually the best choice for straight switching of devices, they can have very low on resistance, so negligible power dissipation.
 

OBW0549

Joined Mar 2, 2015
2,985
TIP127 Pmax = 2W
That's the maximum rated power dissipation with no heat sink. Adding a heat sink, even a small one, will increase that power limit, depending on the thermal resistance of the TIP127 plus the heat sink's thermal resistance.

As has been pointed out, though, a power MOSFET (such as an IRLZ44) would probably be a better choice.
 

Thread Starter

Travm

Joined Aug 16, 2016
266
I guess I'll look at mosfets, I was aware of that fact, but figured this is relatively low power (isnt it?) and a BJT might be cheaper. I don't want to use a heatsink.
I am calculating the power dissipation correctly?
 

AnalogKid

Joined Aug 1, 2013
8,107
As shown in two separate charts in the datasheet, Vcesat (the collector-emitter saturation voltage) is at or below 1.0 V when Ic = 2 A. That is fine for your application.

The 2 W power rating is for the bare transistor with no heatsink. The max power rating for the part is 65 W. For 5 W you will need a small-to-medium sized heatsink. You may or may not need a small fan, depending on the max ambient air temperature and any air movement.

Power MOSFETs certainly can have lower power dissipation. BUT ... In my experienced opinion, I recommend bipolars for this type of environment. By environment I mean a one-off, hand built project being done by a relative newbie. That is not a dig, just my read of the situation. Bipolars are far more survivable in non-ideal conditions and non-ESD-compliant handling.

As with most things, there are trade-offs. If the heatsink is a major size/weight problem, then an oversized power MOSFET is a well-understood solution. But the TIP product line is so freaking reliable that I'd have to be pushed away from it.

ak
 
Last edited:

Thread Starter

Travm

Joined Aug 16, 2016
266
As shown in two separate charts in the datasheet, Vcesat (the collector-emitter saturation voltage) is at or below 1.0 V when Ic = 2 A. That is fine for your application.

The 2 W power rating is for the bare transistor with no heatsink. The max power rating for the part is 65 W. For 5 W you will need a small-to-medium sized heatsink. You may or may not need a small fan, depending on the max ambient air temperature and any air movement.

Power MOSFETs certainly can have lower power dissipation. BUT ... In my experienced opinion, I recommend bipolars for this type of environment. By environment I mean a one-off, hand built project being done by a relative newbie. That is not a dig, just my read of the situation. Bipolars are far more survivable in non-ideal conditions and non-ESD-compliant handling.

As with most things, there are trade-offs. If the heatsink is a major size/weight problem, then an oversized power MOSFET is a well-understood solution. But the TIP product line is so freaking reliable that I'd have to be pushed away from it.

ak
I saw the charts, but to be honest I didnt understand them. Especially considering Vce(sat) is in a table as 2V @ 3A Ic.
I didnt say it, but you summarized exactly why i'm looking at a TIP127 BJT and not a mosfet.
I dont want to use a heatsink because I'm planning on potting it, or wrapping the whole thing in shrink wrap. If Vce was actually 1.0v or less at 2.5A Ic that might work. Or possibly I could use a larger TIP (I think they go up to 8 or 10A? i'll have to look).

What am I doing wrong when looking at the chart value for Vce(sat) vs the Graphs? Looking again at the chart, it says for 300us pulse duration, <2% duty cycle, so this is specific to very low duty cycle switching?
 

AnalogKid

Joined Aug 1, 2013
8,107
The chart lists the max value, while the charts are more in the typical range. Vcesat is temperature dependent. Testing it with a low duty-cycle pulse prevents device heating. Note that the table is for 25degC; that is difficult to maintain when dissipating a few watts continuously.

An electrically larger part (10 A rating but still in a TO-220 package) does not get you anything. The saturation voltage (and hence the power dissipation) will be slightly lower, but not enough to eliminate the heatsink. What *will* cut dissipation is changing to a non-darlington power transistor (TIP42) with a small driver transistor (2N4401, etc.). The chart shows Vcesat = 0.2 V @ 2 A. Now you're at around 0.7 W dissipation; maybe low enough to eliminate the heatsin, depending on the environment. Definitely low enough to eliminate the fan. Note - this inverts the logic polarity from the uC.

HEY - wait a minute - your uC cannot drive a PNP power transistor directly as a saturated switch because the base sits at +11.3 V, way outside the uC output stage rating (unless it has open collector drivers on board). It could drive an NPN darlington directly, but that changes the power connections to the load.

Potting compound generally is an insulator, not good for your application. Thermally conductive potting compound can actually act as a heat dissipator (good), but be sure it is electrically insulating.

What is this for? What is the final application? What is the operating environment? Temperature range? Outdoors / high shock and vib / MIL?

sk
 
Last edited:

Thread Starter

Travm

Joined Aug 16, 2016
266
The chart lists the max value, while the charts are more in the typical range. Vcesat is temperature dependent. Testing it with a low duty-cycle pulse prevents device heating. Note that the table is for 25degC; that is difficult to maintain when dissipating a few watts continuously.

An electrically larger part (10 A rating but still in a TO-220 package) does not get you anything. The saturation voltage (and hence the power dissipation) will be slightly lower, but not enough to eliminate the heatsink. What *will* cut dissipation is changing to a non-darlington power transistor (TIP42) with a small driver transistor (2N4401, etc.). The chart shows Vcesat = 0.2 V @ 2 A. Now you're at around 0.7 W dissipation, maybe low enough depending on the environment. Definitely low enough to eliminate the fan. Note - this inverts the logic polarity from the uC.

HEY - wait a minute - your uC cannot drive a PNP power transistor as a saturated switch because the base sits at +11.3 V, way outside the uC output stage rating (unless it has open collector drivers on board). It could drive an NPN darlington directly, but that changes the power connections to the load.

Potting compound generally is an insulator, not good for your application. Thermally conductive potting compound can actually act as a heat dissipator (good), but be sure it is electrically insulating.

What is this for? What is the final application? What is the operating environment? Temperature range? Outdoors / high shock and vib / MIL?

sk
HAH, yes your right. I promise I would have noticed that before I hooked it up and released the blue smoke.

Plan is to hook this up to a set of cheap Ebay heated mirror pads, connected to a cheap 8 pin pic to run a timer, so when the switch is pressed, a light comes on to indicate the heaters are running, run the heaters for 5 minutes, turn the light and heaters off. Also if the switch is pressed when the heaters are on, it turns the heaters off. Environment will be cold/hot, the module would probably be inside the cab, but I might want to stuff it in the door. I suppose I could put it in a tiny project box, that would keep it safe and dry. I was aiming for maximum simplicity. I could use an NPN transistor and drive straight from the MCU, or if the darlington is causing Vce issues, I can as you suggest, use a non-darlington and then drive it with another smaller transistor.
I was hesitant to outright explain my entire project as "Automotive" modifications are taboo on here iirc, possibly against the TOS. This really is just turning on and off a 12v heater. -30c to 60c. Doubtful it would ever be actually used at 60c, i guess I might accidentally hit the button in the summer.

My experience with transistors as switches to this point has been limited to turning LED's on and off, or small banks of resistors to simulate some theoretical load.

So temperature swings, transient voltages, water, salt, vibration, general abuse. Failure is of no serious consequence, although I'd prefer to do it once. Also if it were to start on fire that would be bad. very bad.
 
Last edited:

AnalogKid

Joined Aug 1, 2013
8,107
There probably is a programmable timer module on ebay that will do this. OTOH ..

One CD4060 oscillator/divider and one-half of a CD4013 dual D flipflop (plus 5-R's and 4-C's) can do this with no programming.

Power up: OFF state
Button press: start 5 minute ON cycle.
After 5 minutes: Turn OFF, wait for next button press.
Button press while on: Turn OFF, wait for next button press.

ak
 
Last edited:

Thread Starter

Travm

Joined Aug 16, 2016
266
There probably is a programmable timer module on ebay that will do this. OTOH ..

One CD4060 oscillator/divider and one-half of a CD4013 dual D flipflop (plus 5-R's and 4-C's) can do this with no programming.

Power up: OFF state
Button press: start 5 minute ON cycle.
After 5 minutes: Turn OFF, wait for next button press.
Button press while on: Turn OFF, wait for next button press.

ak
The pic and programming for me is the easy part. I've already got a small assortment of TIP transistors in a drawer, a couple pics too.
The other method, sounds like an adventure for another day.
I will see what ebay has however for programable modules. Doubtful it will beat a pic, 7805, and a lighted switch for price. But hey it is China.
 

Thread Starter

Travm

Joined Aug 16, 2016
266
https://www.digikey.ca/product-detail/en/on-semiconductor/KSD1691YSTU/KSD1691YSTU-ND/1051426

Bringing this back up, winter is almost over so I'll probably start do this over the spring/summer.
I originally arbitrarily selected a tip127 because with PNP's on the low side, the base current is still load current. Not really important for my application. Also i have some TIP127's and TIP125's in the drawer.

In the interest of keeping my power dissipation low enough that I can pot the thing, no reason why another BJT with an ultra low Vce(sat) wouldn't do this for me? I looked at TIP41/42's and they would probably work, but I dont need the high voltage ratings, and Vce(sat) is still around 1 to 1.5v, puts me dangerously close to 2W.

This is almost worth a new thread, but is kind of an aside to the original questions.
 

dl324

Joined Mar 30, 2015
8,920
I was hesitant to outright explain my entire project as "Automotive" modifications are taboo on here iirc, possibly against the TOS.
It used to be prohibited, but the restriction was lifted 2-3 years ago. As long as what you're doing is legal and safe, there shouldn't be any issues.
 

Thread Starter

Travm

Joined Aug 16, 2016
266
There is an 'automotive' forum here so it can't be against the rules.
I read the TOS when I joined, and I vaguely recall there being issues with over parity (still a thing), and automotive modifications (apparently no longer a thing). Smart play for the board as a whole IMO, automotives have so much electronic garbage in them now. I spent some time troubleshooting a wonky window and lock switch, turns out theres a computer module in my door that was made with poor capacitors, and when the capacitor fails some of the switches dont work (others still do, something about grounding and power requirements or something something).
 
Top