Transistor power rating maximum

[davelectronic]

So something I've not been certain of for a while now. I see schematics of power supply, and single or multiple pass transistors. But whether it's a 3 terminal regulator, or something like the LM723 IC with pass transistors. The usual amount of permissible current drawn is normally around no more than 5 Amps per each series pass transistor.

The thing I don't understand, is some of these NPN or PNP transistors are rated at three or even four times that 5 Amp current in the transistor data sheet. Is there a reason why most linear power supply circuits limit the maximum transistor current to 5 Amps or so. Say you had a transistor of 30 Amps continuous current, could you not draw say 10 to 15 Amps from each of those high power transistors. I would have though half the rated current would be usable from such high power transistors. Is there something there I'm missing, because every PSU schematic I look in to always has the pass transistors at no more around 5 Amps per transistor. And a 20 Amp linear PSU might have 4 X high power transistors, which ratings might be 15 Amp or 25 Amp for each transistor. So the PSU circuit with 4 X transistors for 20 Amps maximum current, could just two high power transistors still manage the same PSU maximum current. I genuinely don't know why this can't be achieved, so long as the device is kept a good margin under its maximum rated current.
Thanks for reading, any help appreciated.

[calmtron]

Check the safe operating area in the datasheet. Also, calculate cooling requirements per transistor at minimum power supply output voltage (max VCE) and maximum current.

[Gyro]

Aside from avoiding exceeding maximum temperature (heatsink dependent) the most important piece of data is the SOA (Safe Operating Area) of the transistor. This is something that doesn't appear in the headline figures at the front of the datasheet, but as one of the graphs in the back.

The SOA graph shows you the acceptable window of operation of the transistor, it has Vce on one axis and Ic on the other. Transistor operational limits are constrained by things like maximum bond wire current and avoidance of Secondary breakdown, the SOA graph shows you the region that you cannot safely leave, for both DC and pulse operation.

SOA graphs vary widely between transistors. Switching devices - particularly Switching MOSFETS and IGBTS have very poor SOAs in the linear region, other MOSFETS are designed to have decent linear operational regions. It is not just types that are different - for instance the good old 2N3055 had a much bigger SOA than the more modern Epitaxial version - the SOA graph is something you should check for every power device.

[magic]

As above, plus check out thermal resistance specifications in your favorite transistor's datasheet. Hopefully °C/W should be self explanatory enough

The maximum current spec is often only applicable to switching and the maximum power dissipation spec only applicable with submersion in freezing water or something close to that.

[TimFox]

Marketing will always choose the highest (best, lowest, whatever is relevant) number to highlight.  For power transistors, that is often the allowable power dissipation for 25 C case temperature , which is higher than the heat sink temperature, which, in turn, is higher than the ambient air temperature.  We used to describe that as “being welded to the hull of a battleship in the North Atlantic”.  In practice, the designer starts with that power and derates it by the stack-up of thermal resistance from case to ambient.  (This relates to steady-state power.)

[davelectronic]

Thanks for your replys, I did forget to take into consideration thermal properties when I ask my opening question. Like staying well with in temperature limits, and power ratings in watts. What I didn't take into account was the safe area operation of any of these transistors. Stupid of me really, I don't know how I missed this in data sheets. My maths isn't fantastic, but I think graphs are ok to realise the devises safe area of operation.

So manufacturer's of these power supply really do stay well in the safe operating limits of transistor devices. I need to take a closer look at the data sheets. Most of the data I do understand, occasionally I do wander how they arrive at a certain figure. It could be hfe stated, tested at different loading etc, it is the maths that I get hung up on from time to time. I did get an education many years ago, but my maths still fell down then. I will try and interpret the information on the data sheet more thougrly.

[rstofer]

The thing to look at is the voltage drop across the series losser times the current through it - Watts!

In a variable supply rated 30V 10A, full voltage/current isn't the hard part to deliver.  Just to throw numbers at it, assume a raw voltage of 32V regulated down to 30V at 10A.  The losser elements are dissipating 2 volts at 10 amps = 20W.

Now set the output voltage to 1V and leave the 32V raw source unchanged.  Now there is 31V at 10A across the losser elements and that's 310 Watts.

[David Hess]

There are a few limits to the maximum current:

1. Power dissipation is the major limit.  Common TO-3 packages can handle between 150 and 300 watts but that is at 25C; once the temperature rise from the junction-to-ambient thermal resistance is taken into account, only a fraction of that power is available.  In practical designs, somewhere between 1/3 and 1/2 of the maximum rated power per package is available.

2. Secondary breakdown in bipolar transistors and thermal instability in MOSFETs further limit power dissipation at high voltages.

3. For bipolar transistors, current gain and bandwidth decrease at high current.

Multiple parts may also make thermal design easier; the power density of a single 300 watt part may be more difficult to cool than two 150 watt parts which can be physically separated.

[davelectronic]

I hadn't realised that a lot less power was available with temperatures above 25 C I just thought if maximum temperature wasn't exceeded, then maximum power was available. Can't see the reason for say 300 watts at 25 C, that scenario in real life wouldn't be common I think. The power loss through burned up heat and dissipation of voltage differential, say needing 12 Volts, but the rectified and filtered voltage value is some what higher. But I assumed keeping the input to output voltage as low as possible to minimise this wasted power value. I did try and interpret the graphs of a couple of power transistors SAO safe area of operation, but it didn't seem as easy as I first thought.

[TimFox]

As I stated above, the "300 W at 25 C" is not a believable operating condition, but the start of the calculation for derating the power depending on what case temperature can be achieved in the total thermal design.  It looks good on the data sheet headline, however.
Together with the maximum allowable junction temperature (before the silicon melts), which is typically 150 to 175 C, the power level at 25 C tells you the thermal resistance from the junction to the case in K/W and the derating co-efficient in W/K.

[perieanuo]

well, if we cool it down with liquid nitrogen we can do even better
I see lately on eevblog every scenario is possible, everyone knows a thing or two about electronics

[TimFox]

If you can maintain the case at a temperature below 25 C, then you apply the derating factor in the opposite direction.  However, with more than 300 W dissipation, you may run up against another limiting specification, such as maximum current.

[Siwastaja]

A good article about thermal design explaining the basics:

http://ludens.cl/Electron/Thermal.html

[magic]

(before the silicon melts), which is typically 150 to 175 C

True in spirit but not exactly accurate. Silicon surely doesn't melt at 175°C

Trivia: I once desoldered an SMD transistor by overheating it accidentally. Still works fine.

[TimFox]

The maximum junction temperature for silicon power devices is usually 150 or 175 C.  Although metallic silicon has a higher melting point (about 1400 C), the device's junction will still fail at some temperature higher than that specification.  "Melting" here can be taken as equivalent to "smoking", even if no smoke appears.

[davelectronic]

Thanks for the thermal design link, definitely go over that in a bit. Liquid nitrogen, err no that's just daft. I will except what the ratings are for a given level of power in the data sheet. Although I would pull out everything I can for efficient cooling. It's looking like multiple pass transistors in purchased power supply are equated to give the rated current, but not stressing just one or two transistors if it's a 10 Amp and upwards PSU. The addition of extra pass transistors gives trouble free longevity. I'm going to do the same, so the linear PSU I've got in mind will deliver a maximum of 20 Amps at 12 Volts. But more typical 15 Amps at 12 Volts, and a a something close to a 50% duty cycle, for driving a HF mosfet linear amplifier. Thanks for the replys.

PS. Forgot to add the transistor type and number for the above current, 4 X TIP36C.