> Tom says:
>
> >Now I could have changed the PIN diodes and did all other kinds of
> >nonsense that doesn't help, but I dislike that approach.
>
> Now if the PIN diodes were correctly chosen in the first place (long
> recovery time) and are biased correctly, their contribution to the rx IMD
> performance will be negligible - in spite of various rumours.
>
> >I shorted the active PIN diodes out, and measured the IMD
> >change....zero.
>
> Exactly what should happen!
I can't understand the fascination with changing PIN diodes Peter.
While I suppose it's possible for a manufacturer to have selected
flawed diodes, I suspect the most common performance increases
observed by ear are psychosomatic.
> >It doesn't matter what anyone guesses or estimates, the vast
> >majority of the IMD was from the gate of a dual gate MOSFET left
> >hanging on the IF chain ahead of the narrow filter IF mixer and
> just
> >after the roofing filter IF. Removing forward bias from that
> transistor
> >when the noise blanker was off improved IMD about 40 dB.
>
> At a guess, the gate protection diode wasn't reverse biased when the noise
> blanker is off. Pretty good way to screw IMD performance.
The problem is the second gate is forward biased to full on when
the noise blanker is off, the drain remains at full operating voltage,
and the next stage is disabled and connected. The first FET in the
noise blanker has about a 12 KHz wide window of signals and it
drives a "dead load" at full gain. The drain saturates and reradiates
IMD back into the IF through stray capacitance.
Turning the noise blanker AGC off and on through a diode clamp
that follows the noise blanker control voltage solves the problem.
> >The second largest problem was the RF amp, so much that I use a
> >pair of push-pull CATV bipolar transistors in a "patched-in"
> preamp.
>
> >IMO, four twenty cent FET's in push-pull parallel in a broad-band
> >amplifier (before any filtering takes place) is not the way to go,
> >although everyone is free to think it is.
>
> I wouldn't put any amplifier in without some filtering in front - well,
> maybe an 807 running 60 or 100 mA as a cathode follower, but that sort of
> power handling capability. But when you talk of 20 cent FETs, how much are
> 2N5109s, which are really good linear CATV transistors? Probably don't
> cost any more. Way to go, Tom.
I'm not sure what in the preamp is "bad", but it generates
measurable IMD even at low drive levels. Bias looks good, but
since I had a pair of LT1001 (gold substrate 5109's) in a push-pull
amp already on a board I just patched it in.
I also cranked the IF gain up a few dB so I could reduce front end
gain.
The FT-1000 is competitive with a heavily modified R4C now,
except for a strange effect. When strong CW signals are parked
just outside the filter passband, I hear clicks that aren't as bad on
the R4C. Anyone have any idea what could cause those clicks to
be stronger? Something in the narrow filter?
73, Tom W8JI
w8ji@contesting.com
--
FAQ on WWW: http://www.contesting.com/ampfaq.html
Submissions: amps@contesting.com
Administrative requests: amps-REQUEST@contesting.com
Problems: owner-amps@contesting.com
Search: http://www.contesting.com/km9p/search.htm
|