On Sun, 20 Jan 2013 19:51:37 -0700, David DiGiacomo
<daviddigiacomo@...> wrote:
On Sun, Jan 20, 2013 at 3:53 PM, David <davidwhess@...> wrote:
It is 2N918 and not 2N198. The 2N918 was a very common early VHF
transistor. My 184 has 2N3605 transistors as well.
The 2N918 is:
900 MHz @ 4 mA
30 Vcbo 15 Vceo 50 mA
From the Tektronix parts book the 2N3605 is:
300 MHz @ 10mA
18 Vcbo 14 Vceo 200 mA
The ubiquitous and inexpensive 2N2222, 2N3904, and 2N4401 should all
work as replacements.
I think those general purpose parts are a bit too slow to be good
substitutions. I would try something faster like PN918, PN2369A,
PN4275, 2N5770, MPSH17, etc. Everyone should have at least one of
those types in the parts box.
Tektronix's choice of the 2N918 to replace the 2N3605 was bothering me
so I did some research and then some tests on my 185.
I had to dig out my National discrete products data book to find that
the 2N3605 is a fast saturated switch with a short storage time. The
storage time is what matters in this case. I think what happened is
the supply became unreliable unless you were buying production
quantities (I remember having this problem when I needed fast
saturated switches for prototyping), so Tektronix recommended the
2N918 as a field replacement. This may also explain why the 2N3605s
were scavenged.
I did some tests using 2N4401s to replace the 2N3605s in various
stages and at around 50uS and faster, the middle transistor (Q175,
Q195, or Q215 for example) delayed turn off for too long and left a
glitch in the leading edge of the output pulse when the faster timing
pulses were superimposed which makes sense from the schematic. On
slower stages, the added storage delay was not enough to even show up
because the pulse width is so long.
I verified this by adding a 1N270 germanium diode (a small signal
schottky diode would work just as well or better) as a baker clamp
between the base and collector and that solved the problem nicely.
The other places where the 2N3605 was used for buffing the fast
outputs where storage delay should not matter (Q123, Q133, and Q153)
worked fine with 2N4401s. The change in the output pulse from the
substitution was almost insignificant and certainly would not affect
operation.
That leaves a couple of spots (Q175, Q195, and maybe Q215) that used
the 2N3605 where a low storage delay transistor is needed. A baker
clamp worked just as well in my tests though.
Of the ones you listed, I think the PN4275 would be the best by far.
They are cheap, available, and are characterized for exactly the right
application. The RF transistors are not characterized for switching
so it is unclear what their delay or off time is. They should still
work though.
David Hess