Francesco
francesco messineo wrote:
Hello Bruce,
On 12/9/09, Bruce Griffiths<bruce.griffi...@xtra.co.nz> wrote:
Francesco
Yes, just keep the gain of each stage low and cascade 2 or3 if you need
higher reverse isolation.
Since the input impedance is around 1k (predominantly capacitive) at 10MHz
you can drive several in parallel from a 50 ohm source.
If needed you can add a bridged T-coil network at the input to improve the
broad band matching.
I would have one stage from the thunderbolt and then two cascaded
stages for each output all from the single first stage, would that
work? I'm not sure that for ordinary frequency distribution I need a
better level of isolation. I may be wrong of course.
That should be fine giving a reverse isolation of about 120dB or so,
with a channel to channel isolation of around 80dB.
You can always add an external amp to improve isolation when required
(eg at mixer inputs).
Note the output is intended to drive a 50 ohm load.
If you need 100MHz operation just substitute a 4GHz ft transistor for the
2N5109/2N5943.
only 10 MHz!
Just indicating how to extend the operating frequency for those that may
need it.
In this case you may need to reduce the supply volatage dependent on the
transistor ratings.
I've simulated cascades of 3 such amplifiers and found that the phase noise
remains low.
The only thing to watch when cascading for improved isolation is reverse
coupling via the common power supply.
However with well planned supply decoupling this won't be a problem.
ok, I got the hint :-)
The ft of the BC548/558 is only 300MHz compared to ~ 1GHz or so for the
2N5109/2N5943, so reverse isolation at high frequencies will be worse when
you use the BC548/558.
as I said I just need this to work at precisely 10 MHz, and the
BC548/558 are a free (really) item here!
The reverse isolation depends on the current gain of the npn transistor
at the operating frequency.
Thus you need to keep the operating current in the 10mA to 30mA range
for the BC548 and maintain Vce at 10V or so.
The best way to check performance is to breadboard a single amplifier
and measure its reverse isolation (easily done with an oscilloscope) by
connecting the thunderbolt output to the amplifier output and measuring
the resultant signal across a 50 ohm load connected to the input.
You can easily hand wind a suitable transformer for this.
Another point to watch is the power dissipation as the npn dissipates at
least 300mW.
If necessary you can always parallel 2 transistors as shown :
http://www.ko4bb.com/~bruce/OCXOBufferAmplifiers.html
<http://www.ko4bb.com/%7Ebruce/OCXOBufferAmplifiers.html>
The value of the emitter resistors are selected for best output match
(depends on transformer turns ratio).
Otherwise one can lower the transistor power dissipation by using a
tapped output transformer so that the transistor dc collector current
can be reduced for a given RF output.
But maybe there could be a problem if one equipment puts back higher
frequency noise on its 10 MHz input? This could be solved by filtering
at each output?
Maybe one day I will use a similar stage to bring the IF signal out of
some of my amateur radio transceiver, but that's not something I plan
to do soon (I don't see the need for it currently) and in that case I
will use better transistors for sure.
Thanks
Francesco
Bruce
Bruce
francesco messineo wrote:
Hi all,
indeed this is very interesting, can this buffer amplifier be used as
a building block for a distribution amplifier for the 10 MHz signal of
a thunderbolt? I remember having seen on the list a similar version
but with european transistors (like the ubiquitous bc548/bc558?) that
are very common here, but I can't remember when it was to surf the
list archives in the right time-period. Anyone?
best regards
Francesco IZ8DWF
On 12/9/09, Ed Palmer<ed_pal...@sasktel.net> wrote:
I was thinking that I might be reinventing the wheel by designing a
buffer
amp. You guys have almost built the entire car! Thanks again for the
ideas. It'll take me a while to get it built and tested.
Ed
Bruce Griffiths wrote:
John Miles wrote:
If it helps I can send you some LTSpice schematics so that you
can
simulate the circuit for yourself.
The breadboards behave as predicted by the simulations at 10MHz.
John Miles has done some preliminary phase noise measurements on
his
version.
The transformers are wound on binocular ferrite cores.
I used some 14mm (long) cores intended for 40MHz to 220MHz (I
had
some)
operation in my breadboard which works well at 5MHz and 10Mhz.
You can also use an off-the-shelf Mini-Circuits transformer for
low-power
applications. The T13-1 was the one I tried. I'll stick some of
the plots
up on the web later tonight if possible.
See http://www.ke5fx.com/norton.htm for
measurements
and connection details
of the copy of Bruce's amp that I added to my 5061A.
-- john, KE5FX
Note the LED I used in the schematic was merely for simulation
purposes
(ie LTSpice had a model for it).
A standard red or amber LED is just fine.
Another point is the LTSpice LED model isn't particularly accurate for
simulating the effects of temperature variations.
Does anyone knows of more accurate LTSpice compatible LED models?
The LED model voltage drop increases with temperature even at low
current,
whilst the voltage drop across a real LED at low currents decreases with
temperature.
In practice the variation in the LED forward drop tracks the variation
in
the pnp Vbe quite well.
Bruce
_______________________________________________
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
_______________________________________________
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
_______________________________________________
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
_______________________________________________
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
_______________________________________________
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
