[time-nuts] NIST isolation amplifiers

[Bob Camp]

Hi

For a modern build, the PZT3904’s and PZT2222’s are a pretty good way to go with this amp. 

For normal distribution to instruments, there’s really no need to do anything this complex.

Bob


> On Nov 25, 2014, at 2:34 PM, Charles Steinmetz <csteinmetz at yandex.com> wrote:
> 
> A couple of people were asking about NIST isolation amplifiers recently.  I'm attaching circuit diagrams of the 5-10 MHz amp from 1997 and the 1-200 MHz amp from 1990.  I think Bruce has the papers linked at his ko4bb.com pages.
> 
> I built some of the 5-10 MHz amps with minor variations and they work very well (I used a separate capacitance multiplier for the base divider string, and changed the first 4.3k resistor to 6.65k to achieve symmetrical clipping and a small increase in headroom).  I used 2N3904s for the two lower transistors and a 2N2219A for the top transistor, which dissipates over 300mW.
> 
> I tried some fancy transistors with very low base spreading resistance, which reduced the noise -- but the increased junction capacitance made the AM to PM conversion worse, so the overall residual PM was worse.  On the other hand, GHz transistors had higher noise due to lower gain.  So the 3904/2219A combination appears to be just about optimum.  (Note that the 200 ohm resistor at the input contributes about half of the circuit's noise, and you can't use the Norton trick because it would ruin the isolation.)
> 
> The transistor stack draws 32mA and the base divider stack draws ~1.5mA.  The amplifiers have an input impedance of 250 ohms, so paralleling the inputs of 5 sections creates an overall 50 ohm input impedance.  When a circuit has reverse isolation of well over 150dB, as this one does, you need to pay very careful attention to shielding.
> 
> Best regards,
> 
> Charles
> 
> <NIST_200_iso.png><NIST_5_10_iso_amp.png>_______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.