I go by finger touch. If it's too hot to touch then it's running too hot.

Oh, I forgot... I assume that also goes for the second case -the power supply regulators-. They are also too hot to touch for more than a couple of seconds; but the current is well within the limits. They have the hole to attach heat sinks; so I guess I should have added them from the beginning.

 

Let's keep the two problems separate. I will move your posts about the power supply to its own thread with this title.
"When is a transistor too hot for its own good?"

I will rename the oscilloscope thread to
"Hitachi V-522 oscilloscope troubles"

... in about 8 hours (1300h UTC), after you get a chance to read this.

 

I think a heat sink would always be a good option to play your card safe.

 

You can calculate the wattage that each regulator has to dissipate.
Take the voltage difference across the regulator and multiply it by the load current to give you the power in watts.

For example, suppose your supply voltage is -20V and the load current is 1A. The LM7912 has to drop 20-12 = 8V.
The power, 8V x 1A = 8W has to be dissipated by the regulator and that's a lot of power.

Your solutions are limited.

1) Reduce the supply voltage to within 2-3V of the regulator output voltage.
2) Transfer the power over to a high wattage series resistor to drop the supply voltage.
3) Change to a switching regulator.
4) Put heat sink on the regulator.

In your case, it appears that a heat sink is appropriate.

 

Calculate the power being dissipated in the transitor. Multiply the power dissipated by the thermal resistance from junction to case. Add this to the temperature of the case. If this exceeds the max junction temperature from the datasheet, the transistor is too hot.

Bob

 

When is a transistor too hot for its own good?

Answer: always.

Semiconductors rediffuse over time.
The higher the temperature the faster the rediffusion process, it occurs (very slowly) even at room temperature.

So if you have an emitter follower that is running too hot to touch, say case temp 60C to 80C, it will likely fail after about 10 years of service.
This is why many high power audio amps fail : inadequate heatsinking.

But then pvc is essentially a rigid plastics material and contains 'plasticisers' to make it flexible.
The evaporate out over time, leaving the insulation brittle on the wiring.
In the same way the longer and hotter the wiring the more quickly this occurs.[/quote]

 

Since you recommend calculating the temperature, I assume that the infrared thermometers (which I just bought exclusively for this purpose), are not only unreliable, but not suitable at all for this purpose.

 

Since you recommend calculating the temperature, I assume that the infrared thermometers (which I just bought exclusively for this purpose), are not only unreliable, but not suitable at all for this purpose.

If you are referring to my post, I did not recommend calculating the temperature of the case, I meant to measure that. From it, and a couple of parameters from the datasheet you can calculate the temperature of the junction. That is the important temperature.

Bob

 

Heat sinks are not the only option, you might want to look into pre-regulators....

HTH Steve

 

Let's keep the two problems separate. I will move your posts about the power supply to its own thread with this title.
"When is a transistor too hot for its own good?"

... in about 8 hours (1300h UTC), after you get a chance to read this.

Back in my TV repair days, I was called out by a little old lady with a battery mains portable that had been playing up, and died completely when she thumped it.

To make the chassis compact, the manufacturer had made a frame to assemble the boards like a box round the back of the tube - the top board being upside down. On closer inspection, I discovered a TO5 transistor (BFY50/2N3053 type of thing) was missing from the top board - it had got hot enough to unsolder itself, and fell onto the bottom board when the set was thumped. When I tested it, there was no leakage or other fault, so I put it back in - the messed up picture led me straight to the main 12V reservoir capacitor which I replaced to wrap up the repair.

Another story dating back to when 20 - 40Mb hard drives were becoming not worth bothering with, I used a collection of old Conner drives to even out the temperature in the bottom of a table top oven I used for various experiments.

The first I knew of the oven thermostat welding shut, was when I took out those drives and noticed that most of the components had fallen off the boards - they had some very useful small IRF MOSFETs which all tested OK and were subsequently used in various projects.

 

BFY50/2N3053 never had much gain to start with!

The old TO3 transistors in TVs started out life with a gain of 30 -50 if you were lucky and finally died off at 5 -10.

 

Oh, I forgot... I assume that also goes for the second case -the power supply regulators-. They are also too hot to touch for more than a couple of seconds; but the current is well within the limits. They have the hole to attach heat sinks; so I guess I should have added them from the beginning.

Bob Pease "5 second rule": If you can hold your finger on a hot device for 5 seconds, the heat sink is about right, and the case temperature is about 85 deg C.

 

I think a heat sink would always be a good option to play your card safe.

Not if you want to make any money or stay in business. Now, if you are doing a project that has weight/size/cost budgets that will allow you to take this approach profitably, then the marginal savings in design cost (since you aren't spending time deciding whether to add a heat sink) may be worthwhile. But what size heat sink?

But if you are designing a product that has tight budgets in one or more areas and especially if you are designing something for a high-volume market that has competition, then you are being irresponsible to put heat sinks where they aren't needed -- to the degree that designing products that are not competitive in the market place is "irresponsible".

 

BFY50/2N3053 never had much gain to start with!

The old TO3 transistors in TVs started out life with a gain of 30 -50 if you were lucky and finally died off at 5 -10.

You're thinking of the TO3 2N3055 - there was the 2N3054 in TO66 IIRC, and the 2N 3053 which was TO5.

The TO5 device was driving a larger series pass transistor that didn't fall out because it probably didn't get that hot - either that or because it was bolted on.

 

Oh, I forgot... I assume that also goes for the second case -the power supply regulators-. They are also too hot to touch for more than a couple of seconds; but the current is well within the limits. They have the hole to attach heat sinks; so I guess I should have added them from the beginning.

When I worked for a firm that made faradic muscle exercisers , we discovered a cheaper off the shelf transformer that was much cheaper than the custom parts previously used. It was my job to incorporate it in a new design. On the prototype, the pulses didn't look right so I decided to check the temperature of the TIP120 driver transistor by grabbing it with finger and thumb - it branded my thumb with an imprint of the TO220 heat tab.

It turned out the transformer was saturating - a small reduction in pulse width and the now slightly blued transistor was chugging away like nothing had happened.

 

You're thinking of the TO3 2N3055 - there was the 2N3054 in TO66 IIRC, and the 2N 3053 which was TO5.

No, perhaps I did not make it clear that my second paragraph was not referring to the same transistors as the first, but some additional information.
Sorry if there was a misunderstanding.

The above quote is correct information, however.

 

When I worked for a firm that made faradic muscle exercisers , we discovered a cheaper off the shelf transformer that was much cheaper than the custom parts previously used. It was my job to incorporate it in a new design. On the prototype, the pulses didn't look right so I decided to check the temperature of the TIP120 driver transistor by grabbing it with finger and thumb - it branded my thumb with an imprint of the TO220 heat tab.

It turned out the transformer was saturating - a small reduction in pulse width and the now slightly blued transistor was chugging away like nothing had happened.

Reminds me of a design project I gave my students one semester a key -- and with malice aforethought -- part of which was thermal management. I told them upfront that this was going to be important and as I gave lectures I pointed out repeatedly where the examples I was working were directly applicable to their project. But many, many students didn't pay attention and were of the opinion that a transistor's a transistor (hey, they have the same symbol in the schematic, after all). So they would use 2n3904 and 2n3906 transistors with no collector resistors as their final output totem pole transistors for a linear power amp spec'ed to run from +/- 32 V rails and deliver up to 5W to an 8 Ω load. Needless to say, they let the magic smoke out of a LOT of transistors (which I had bought for them with my own money because the school didn't have the budget for parts). Students would come to me puzzled by why their transistors were smoking and I'd ask what transistor they were using for their output and, usually, they'd say one of the small signal transistors. But one night a student came in complaining that his output transistor was getting really hot and when I asked him what he was using he looked at his thumb and said, "2N3055". He had branded the part number into his thumb. He was using the right transistors, but wasn't using a collector resistor to drop the voltage and dissipate the power and also didn't have a suitable heat sink on the device even if he had used the resistor. I had bought a collection of heat sinks, some which I knew would be too small, and a range of power resistors, some of which were the wrong size or wrong power rating, so that the students would have to run the numbers to choose the right ones. I ended up buying three sets of components in order to keep up with the carnage, but it was money well spent as the entire class really gained an appreciation for reading data sheets and taking power and temperature effects into consideration.

 

Reminds me of a design project I gave my students one semester a key -- and with malice aforethought -- part of which was thermal management. I told them upfront that this was going to be important and as I gave lectures I pointed out repeatedly where the examples I was working were directly applicable to their project. But many, many students didn't pay attention and were of the opinion that a transistor's a transistor (hey, they have the same symbol in the schematic, after all). So they would use 2n3904 and 2n3906 transistors with no collector resistors as their final output totem pole transistors for a linear power amp spec'ed to run from +/- 32 V rails and deliver up to 5W to an 8 Ω load. Needless to say, they let the magic smoke out of a LOT of transistors (which I had bought for them with my own money because the school didn't have the budget for parts). Students would come to me puzzled by why their transistors were smoking and I'd ask what transistor they were using for their output and, usually, they'd say one of the small signal transistors. But one night a student came in complaining that his output transistor was getting really hot and when I asked him what he was using he looked at his thumb and said, "2N3055". He had branded the part number into his thumb. He was using the right transistors, but wasn't using a collector resistor to drop the voltage and dissipate the power and also didn't have a suitable heat sink on the device even if he had used the resistor. I had bought a collection of heat sinks, some which I knew would be too small, and a range of power resistors, some of which were the wrong size or wrong power rating, so that the students would have to run the numbers to choose the right ones. I ended up buying three sets of components in order to keep up with the carnage, but it was money well spent as the entire class really gained an appreciation for reading data sheets and taking power and temperature effects into consideration.

You usually put the resistor in the emitter lead so the increased volt drop tends to offset a small part of the base drive.

 

The resistor in the emitter circuit provides negative feedback.

 

You usually put the resistor in the emitter lead so the increased volt drop tends to offset a small part of the base drive.

There were a few things at play that argued in favor of putting them in the collector paths. First, there was no feedback of any kind, though this wasn't a huge driver since the spec stipulated an 8 Ω load. The spec they were given was that the amp should deliver an average of 5W to an 8 Ω resistive load when the input signal had an amplitude of 1 Vpp. Second, at full output they needed to drop quite a bit of voltage across the resistor to minimize the heat dumped into the transistor. Depending on where they set the supply rails (they were actually adjustable between 24V and 36V, which I had forgotten about until just now), they might need to drop quite a bit of voltage across those resistors. Third, one of the things that students would be asked to do in the next phase of the project would be to explore the crossover distortion and ways of reducing it, such as Vbe multipliers and I wanted to keep that as simple as possible.