[Ronja] Spectral analyzer up to 8 GHz

[Karel Kulhavy]

---- Gregory Maxwell <gmaxwell at gmail.com> schrieb:
> On Tue, Oct 14, 2008 at 4:08 PM, Karel Kulhavy <twibright at hispeed.ch> wrote:
> ---- Gregory Maxwell <gmaxwell at gmail.com> schrieb:
>> On Tue, Oct 14, 2008 at 4:42 AM, Karel Kulhavy <twibright at hispeed.ch> wrote:
>> > Due to low achievable gain I had to upgrade the 2N3904 in the design (Ft=300 MHz)
>> > to BFR90/91/91A (5GHz). 2N3904 oscillated at 500 MHz so I expect these to oscillate at
>> > 8GHz. Does anyone has a spectral analyzer like this or a scanner (communications
>> > receiver)? When I build the limiting amplifier cell I have to check it doesn't
>> > oscillate.
>> >
>> > Or do you know any technique how to check for oscillation without such high frequency instruments?
>>
>> I wonder if you could get a confident idea if its self-oscillating by
>> simply measuring the current draw?
>
> Unfortunately not - the current draw will be enforced to a preprogrammed value by current sources.

> I would expect that your traces would have enough enough induction to
prevent the system from having above unity gain at 8 GHz … but I guess
you need the test equipment to be sure of that then. :(

I have tested the stability in a simulator. Put estimates ranging from 5nH to 60nH
depending on length of trace. Denis L Feucht from Tektronix claims Tektronix used a rule of thumb estimate of 10nH per cm of trace.

The circuit seems to be perfectly stable. Neither the input circuit nor the emitter
capacitor nor the power supply sees any negative resistance at any frequency. I still
have to test for stability in a clipped state with the coils (I tested just without
coils) and also check what effect the coils will have on bandwidth.

Do you know the Supaplex game? 
http://archive.gp2x.de/cfiles/screenshots/supaplex.png
All what is missing here is the pacman eating resistors
and transistors and coils falling on his head:

http://ronja.twibright.com/sim/stab90.png
http://ronja.twibright.com/sim/stab90_s.png

Karel