EEVblog 1599 - TOP 5 Jellybean Bipolar Transistors

#25 Reply
Posted by G0HZU on 28 Feb, 2024 23:18
Quote from: Kleinstein on 28 Feb, 2024 08:50
The gold doping should mainly effects the storage time, not so much the ft part of the speed. So the gold doping would help with switching, not so much with linear amplification.
The gold doping can also be at a moderate level. There is not just the question doped or not. Similar to the fast diodes: there are the really fast ones with a trr of some 4 ns and the moderately faster ons with some 100 ns with a lower level of gold (or other dopant to add recombination centers).

For really fast saturated switching there is the classic 2n2369, that is also available as SMD version MMBT2369 (and similar). For this application (though already a bit special) one can consider it a yellybean type. Here the high speed comes with a rather high leakage. So this transistor is quite a bit different from the normal small signal types.
Yes, the 2N2369A is a classic high speed switching device (process 21). Many years ago I used the 2N2369A quite a lot as the active device in classic Colpitts crystal oscillator circuits. It performs well in this application.

#26 Reply
Posted by T3sl4co1l on 29 Feb, 2024 00:12
I have a number of them (Philips TO-92 PH2369A) removed from CRTs, back in the day, along with some BFQ221/225/241. Together, they were the bottom/input (2369), top/cascode (BFQ225), and complementary emitter follower (221/241), in the cathode video driver circuit. Probably was from around the time when ASICs were just taking over; late 90s/early 2000s? -- the N3 chassis Trinitron I was using for a long time, dated from 2001 I think and used such a chip.

Not exactly pushing bandwidth boundaries for ~100MHz video, but shows their use in a linear/amp circuit anyway.

Also have some... I'd have to go look up the part, but they're something like 20V 1A 1.something GHz transistors, in a... wide thick plastic package somewhat like a TO-202 without any tab at all, which were the bottom cascode in some other monitors I took apart way back when. Interesting devices to be sure.

Tim

#27 Reply
Posted by David Hess on 29 Feb, 2024 13:27
On a different subject, I went hunting for current jellybean TO-92 parts:

BC546C/BC556C 65V 100mA High hfe Amplifier
2N3904/2N3906 40V 200mA Amplifier/Switch
2N4401/2N4403 40V 600mA Amplifier/Switch
2N2222A/2N2907A 40V/60V 600mA Amplifier/Switch
BC337-40/BC327-40 50V 800mA Amplifier/Switch

The KSA1381/KSC3503 is now the only more powerful extra linear part that is still available, albeit at 10 times their original cost. The TTC015B/TTA008B and MJE340/MJE350 might be possibilities, but would need to be tested because their datasheets are incomplete. The figure of merit here is something like Ic/Cre. I did not check for surface mount alternatives.

Quote from: Kleinstein on 28 Feb, 2024 08:50
The gold doping should mainly effects the storage time, not so much the ft part of the speed. So the gold doping would help with switching, not so much with linear amplification.
The gold doping can also be at a moderate level. There is not just the question doped or not. Similar to the fast diodes: there are the really fast ones with a trr of some 4 ns and the moderately faster ons with some 100 ns with a lower level of gold (or other dopant to add recombination centers).

Plenty of RF transistors are gold doped. I tested a bunch of old and new small signal RF transistors and none were suitable replacements for fast saturated switches even when Ft was much higher. I concluded that there is some basic difference in construction besides gold doping.

Baker clamping of an RF transistor reduces storage time, but increases saturation voltage.

Quote
For really fast saturated switching there is the classic 2n2369, that is also available as SMD version MMBT2369 (and similar). For this application (though already a bit special) one can consider it a yellybean type. Here the high speed comes with a rather high leakage. So this transistor is quite a bit different from the normal small signal types.

Where they are still available, the remaining ones are difficult to find. One thing to look for is low Vceo of 12 volt for NPNs and 15 volt for PNPs.

My most recent survey from a couple years ago for SOT-23 parts includes:

NPN:

MMBT3646
BSV52
MMBT2369A
MMBT2369

PNP:

MMBT5771
MMBT3640

I also have the list of saturated switches that Tektronix used:

NPN, in order of Ft:

2N2501 350 MHz
2N5769 400 to 500 MHz
2N2369 500 to 650 MHz

PNP, in order of Ft:

MPS3640 500 MHz
PN4258 700 to 900 MHz