D882 vs BD139 in qrp finals

[iMo]

Hi, operating the ebay's Rockmite 40m CW kit (7023kHz) with a D882 final PA transistor.
In past I did some experiments with replacing it with a BD139-16 (ST), resulting in half of the output power (measured the amplitude of the signal with an oscope on a dummy load).
I got ~6W with 882, about ~3W with BD139.
Today I've smoked the 882 final finally  and I've replaced it with the BD139 again.
While reading the datasheets I can see the D882 has got around half of max voltages and twice the max currents against the BD139.
Also the DS of the 882 shows Ft=100MHz, while BD139's DS does not show any Ft at all.
Still investigating how to get the BD139 working in that circuit with a similar output power.
Any hint on it?
PS: my pcb version has got a protection zener (1n4755 in series with 1n4148) at transformer's output (it seems to me the protection should be at the 882's collector, imho). The newer pcb lacks the protection (from the photo there is none zener).

[iMo]

I've simulated above PA with 4 BD139 models in my LTSpice lib and with 2SD882. The simulation shows there is a BD139 model wch gives some relevant output power. Thus it seems you have to be lucky in getting a good, 20-30y old BD139  ..

[CJay]

Not all BD139 are created equal.

Just buy the 2SD882, it's not expensive, only a few pennies more than a BD139

[iMo]

The issue with 882 is they pretty easily smoke in that circuit (even with the collector protection).
I've got 2 dead 882 already, first one passed away after 1 week of qsoing, the second one after 20 minutes.
The new BD139 (ST) works (tried 4), but it gives you a half of the output power.
I think the 882 has got higher Ft than the modern BD139 (people say the new ST/NXP ones are of different chip construction compared to the old Philips/Fairchild BD139 with FT=190MHz), but the max voltages of the 882 are much lower than the BD139.
The question is whether the BD139 is matched properly in that circuit, it could be the input/output Z differs to the 882. No idea what the inp/out impedance of the transistors is, though.

[CJay]

The issue with 882 is they pretty easily smoke in that circuit (even with the collector protection).
I've got 2 dead 882 already, first one passed away after 1 week of qsoing, the second one after 20 minutes.
The new BD139 (ST) works (tried 4), but it gives you a half of the output power.
I think the 882 has got higher Ft than the modern BD139 (people say the new ST/NXP ones are of different chip construction compared to the old Philips/Fairchild BD139 with FT=190MHz), but the max voltages of the 882 are much lower than the BD139.
The question is whether the BD139 is matched properly in that circuit, it could be the input/output Z differs to the 882. No idea what the inp/out impedance of the transistors is, though.

Ahh, I see, it's not because of scarcity.

The general opinion whenever I've seen the BD139 discussed is that 3-4W is about the most you'll manage to get at 12V, you might be able to optimise the matching and squeeze a little more out but it's not likely to make a lot of difference in range and I suspect it would probably cause reliability problems if you're running the transistor closer to the limits.

I'd maybe try and work out why the 2SD882 is dying, you could also consider different transistors, I've just got a bag of 2SC5706 transistors which I think should be good for a solid 5W and from a quick look at the specs seem more rugged than the 2SD882

People like Harry Lythall SM0PVO and others have useful information on transistor impedance calculation and matching if you want to explore that.

[iMo]

A friendly OM here gave me 3 new 882 (NEC, ebay) from a larger batch he ordered in past (he also messed with the kits, with similar experience). Two of them were DOA and the third worked great, but only for 20minutes of qsoing. Then a smoking bubble appeared on the package.
The PA transistors there in the kit heats up basically as the heatsink is a joke (it is say 1-2w heatsink, imho).
I added a copper wing to help with cooling. So the overheating could be the issue too. I am running the PA at 14-16V (4xLion cells) so the "overvoltage" could be a reason.
This morning I replaced the BD139 with 2sc1969 (a 16W CB final) an it works too, the output power similar to the 882. It does not fit the pcb however (it is mounted outside the board on a larger heatsink).
It could be the 882 works there more reliably when powered from max 12V. The new board (without the zener protection) is advertised as the "8W" one, btw. From my experimenting the 882 works there best, but with reliability issues..