[time-nuts] Advice on 10 MHz isolation/distribution

Bruce Griffiths bruce.griffiths at xtra.co.nz
Thu Feb 18 02:01:15 UTC 2010

life speed wrote:
> Date: Fri, 12 Feb 2010 15:32:55 -0500
> From: Bob Camp<lists at cq.nu>
> Subject: Re: [time-nuts] Advice on 10 MHz isolation/distribution
> Hi
> There are a few differences between what you are simulating and the schematics Bruce posted earlier. The collectors of the input stages (q1, q4 and q7) seem have to come unglued from the bases of the output stages. The 95 ohm / 100 nf roll off networks seem to have vanished from the emitters of Q1 and Q7. There are  a few other minor things.
> I suspect that once you get it back to a node-node feedback circuit, the impedance at the splitting point will drop and the isolation will go up quite a bit.
> Bob
> Hi Bob,
> I see I fumbled the mouse, thanks.  I had to spend a couple days working on other parts of the design, but I am back at the 10 MHz circuit again.  I don't see the RC networks in the emitters of Q1, Q7 in the original schematic (design 3), although I tried them and also with a 1.6K ohm resistor and did not see much difference.
> I notice that the gain of 2 (6 dB voltage gain) happens in the first stage, and the second stages seem to have an equal amount of loss, which is a reasonable outcome for buffer/isolator circuit.
> The isolation, while improved, seems to be only 50 dB from Vout_1 to Vout_2.  To make this circuit really effective, I think I would have to parallel two Q4/Q1 two-stage amps.  Either that, or there are still some mistakes.
> Clay

Your design is subtly different from the one I posted.

The output amplifiers should have about 1K6 from the ZTX5179 emitter to 
ground not the 95 ohms  in series with 1k6 shunted by 100nF.
Among other effects your version will exhibit significant distortion 
unless the current in the output transistors is increased.

You've also reduce R9 R10 and R12 to 10 ohms??
This significantly reduces the RF loop gain.
In particular the open loop RF gain of the ZTX5179 input device drops to 
little more than unity.
They should be about 1K or so.
If you do the analysis correctly you will find that they have little 
effect on the RF or low frequency noise.

My simulations indicate Vout-1 to Vout_2 is around 100dB at 10MHz with 
the correct values for R9, R10 and R12.
This may be slightly optimistic.


More information about the time-nuts mailing list