Transistor Case temp vs junction temp question

[Damo666]

Hi guys. I've done a lot of googling as well as searching these forums, but i can't find any information re the subject. If my transistor quotes a maximum junction temperature of 125oC , what is a reasonable temperature that the case should be without using an heatsink? I ask, as one of my transistors in a rf pa stage gets quite hot. I can only put my finger on it for about 5 seconds until it gets uncomfortable to touch. The transistor is a 40965 made by RCA, but i can't find any info about it. It was made in the 70's to early 1980's. It's also operating at approx 16vdc and drawing about 170mA. Can anyone help please?

 

[Sceadwian]

That depends on the junction to case thermal resistance. If you don't have that value there is no way to answer your question. Given the package type you should be able to look up any package of similar type to find the thermal resistance, it doesn't have to be the datasheet for your specific chip.
Just from personal experience, if you can put your finger on it for 5 seconds, it's not too hot.. I've gotten 2nd degree burns from chips operating at safe temperatures in less than 5 seconds.

 

[RCinFLA]

According to data sheet pg 2 it has theta j-c of 50 deg C per watt. This is a TO39 package and theta j-c is surprisingly quite high. Look at collector pin to see if it is insulated from case. If so, it would explain the high theta j-c.

4096 datasheet and Application Note, Data Sheet, Circuit, PDF, Pinout | Datasheet Archive scroll down the page to 40965

Just for reference, a die mounted to the pedestal of a TO220 package (like LM317 regulator) has a theta j-c of 3 degs C per watt.

For a power BJT the collector is usually the die substrate. If it can be hard mounted to case then theta j-c is low. If it requires collector to be insulated from metal case then the insulating spacer drives the theta j-c up.

For RF devices, it is hard to keep stability when the entire rather large TO39 metal case is the output collector.

 

[Damo666]

Thanks for the good input guys, i'm really very grateful. I can't tell if the collector is insulated from the case unfortunately as its soldered flush to the pcb. In your opinion then guys, do you think that being able to touch this transistor for a good few seconds without any burn will be running it within safe limits? I do all my browsing from a phone so can't read pdf files at the moment. Thanks again. Damian.

 

[crutschow]

You can check if the case is insulated with an ohmmeter between the case and the collector pin solder-joint on the opposite side of the board.

 

[Sceadwian]

I don't know about others, but I've burnt my hand on chips running at safe temperatures several times. I would cautiously say that 'a good few seconds' without a burn would be sensible, but if you want surety you need to work the math out. Your transistor could destroy itself from pulse heating even if the average case temperature remains the same, especially with that kind of package thermal resistance.

 

[MRCecil]

The RCA 40965 crosses to a 2N3948 per this database:
transistors_10-10000

If your transistor is in a TO-5 or TO-39 case, it's a good chance it's a match. I looked up the thermal specs for a 2N3948 in an old Motorola Semi data library (1972). Junction to case is 35 deg C/W max and junction to ambient is 175 deg C/W max. So if you were able to hold your finger on the transistor for 5 seconds or longer with just discomfort and no burn, it is likely the case is probably below 140 deg F or less, which is far below the maximum temp. I learned in the Navy decades ago that if it's workin' don't fix it.

Hope this helps.

EDIT: Here are the temp conversions:
35 deg C = 95 deg F
175 deg C = 347 deg F
140 deg F = 60 deg C

 

[Damo666]

Ok guys. I've checked for continuity between the collector pin and the case and there's full continuity with no resistance. On this note, i presume this is good news for me? I do apologise about the numerous newbie questions, but i tend to worry about things vintage like this breaking down. @Cecil - your reply as also made me feel more at ease, so cheers for that. Thanks again. Damian.

 

[Damo666]

According to data sheet pg 2 it has theta j-c of 50 deg C per watt. This is a TO39 package and theta j-c is surprisingly quite high. Look at collector pin to see if it is insulated from case. If so, it would explain the high theta j-c.

4096 datasheet and Application Note, Data Sheet, Circuit, PDF, Pinout | Datasheet Archive scroll down the page to 40965

Just for reference, a die mounted to the pedestal of a TO220 package (like LM317 regulator) has a theta j-c of 3 degs C per watt.

For a power BJT the collector is usually the die substrate. If it can be hard mounted to case then theta j-c is low. If it requires collector to be insulated from metal case then the insulating spacer drives the theta j-c up.

For RF devices, it is hard to keep stability when the entire rather large TO39 metal case is the output collector.

Hi. I did a check, and the collector is directly connected to the case of the transistor. One last question then; Do you think the tranny will be ok running it with a case temp of a max of 75 Degrees C intermittently? I just cant find the thermal resistance of this thing anywhere.


Regards,
Damian.

 

[mneary]

Is this an amplifier that you've recently constructed, or something that's already been working for 25-35 years?

If you're the same Damo66 that asked the same question here, then what you're doing by operating out of band is forcing the stage to dissipate about 2.25 watts instead of the original 2.03 watts. If that was the only problem, I would say the reliability is probably compromised, but not seriously.

Equally important is that you've moved it from 465MHz down to the 70cm (433MHz) HAM band and you're unable to re-tune the class-C final stage. I would be concerned that some of that 'lost' power is being transmitted as spurious signals and harmonics.

Suggestion: A lot of the power is probably being lost to antenna mismatch, and so adding an antenna tuner (might be as simple as a 3-15pF variable cap) could restore some of the efficiency. Use one of the antenna tuner calculator websites to guess what the existing match network looks like, and then solve for the new frequency. This might lead you to some simple components you can insert into the antenna line.