[time-nuts] Oleg' s PN test Re: A new member & PN test set

Bruce Griffiths bruce.griffiths at xtra.co.nz
Wed Mar 30 15:20:19 EDT 2016


The first inductor self resonance is much lower than it need be. EPCOS have a 
range of inductors which have a much higher first self resonance frequency.

The other problem is that inductors have several resonances alternating 
between parallel and series resonance. In short, your inductor model is 
inadequate and gives false predictions. You actually need to measure the filter 
response.

Conical inductors are available that are effectively resonance free to 40GHz 
but the largest value is around 10uH. In principle one could wind one's own 
conical inductor with a larger value, However an iron powder (carbonyl iron - 
available from Ukraine at least via ebay) and epoxy mixture. A cone angle of 
about 15 degrees appears to be suitable.

Failing that, the classical method is to use a series string of inductors of 
increasing value. Even then the various resonances need to be damped.
Lossy Ferrites and resistors can be useful, however one has to be careful not 
to increase the noise at frequencies of interest.

Bruce

On Wednesday, March 30, 2016 06:11:51 PM Oleg Skydan wrote:
> --------------------------------------------------
> From: "Bruce Griffiths" <bruce.griffiths at xtra.co.nz>
> Sent: Wednesday, March 30, 2016 7:29 AM
> To: "Discussion of precise time and frequency measurement"
> <time-nuts at febo.com>
> Subject: Re: [time-nuts] Oleg' s PN test Re:  A new member & PN test set
> 
> > One hidden issue you don't address is that operation of the 40uH inductor
> > at
> > frequencies above its parallel resonance may allow substantial RF at the
> > sum
> > of the LO and RF frequencies to appear at the opamp input.
> > 120MHz at the 797 input will likely lead to RF rectification effects in
> > the
> > opamp input stage. The resultant offset will create a number of issues
> > including operation away from the quadrature point.
> > Unless you use something like a series string of inductors and/or a
> > conical
> > inductor the first parallel resonance of the 40uH inductor is likely to be
> > somewhat below 120MHz.
> 
> Ohhh... I do not like words like "substantial", "much more" and etc. I like
> numbers and tests. ;)
> 
> So I looked at the Murrata inductors datasheets, and it appeared 40uH
> inductor will have SRF in 10MHz region. But it does not mean that pi-LPF
> will not work at the higher frequencies. Actually it mean that our LPF will
> have response similar to the elliptic filters.
> 
> So let's draw the model with the inductor with self resonance at 10MHz and
> well at 120MHz and 1MHz to see how bad the response is:
> 1MHz: http://skydan.in.ua/PNTestSet/1MSRF.png
> 10MHz: http://skydan.in.ua/PNTestSet/10MSRF.png
> 120MHz: http://skydan.in.ua/PNTestSet/120MSRF.png
> 
> As we can see the RF+LO product will be attenuated more than 60dB in all
> cases. So, your comments? Would you like me to measure the RF voltage at the
> AD797 input in the real "test set"?
> 
> > However 50 ohms to ground at the LC filter output shouldn't be necessary.
> > A somewhat larger value should suffice.
> 
> I made some experiments trying to find the optimal value of the resistor at
> the LC filter output. The phase detector gain grew along with the resistor
> value, but so did the harmonics level. So I needed to apply more attenuation
> to the input signal to stay in the linear region. The resulting "test set"
> noise floor was almost identical for 50..300Ohm values (300Ohm was a bit
> better at offsets grater then 2kHz and a bit worse closer). Large values
> noticeably degraded the performance.
> 
> I suppose the noise floor can be lowered only if better LNA will be used
> (currently the LNA noise dominates the PD noise), or if the levels on the
> mixer will be increased (this will require higher level mixer and/or new
> calibration routine if the mixer will not be in a linear region).
> 
> All the best!
> Oleg
> 
> 
> _______________________________________________
> 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.



More information about the time-nuts mailing list