[time-nuts] 3GHz prescaler for Pendulum counters

GandalfG8 at aol.com GandalfG8 at aol.com
Wed Oct 14 17:36:01 EDT 2015

A couple of years ago, my oh my doesn't time rush by when you're having  
fun?:-), I posted a recommendation for 3GHz prescalers from Pawel Witkowski, 
an  Ebay seller in Poland, for the HP 53131 and 53132 counters.
I'm now very happy to report that Pawel has also produced a similar  
prescaler for the Pendulum range of counters, his Ebay ID is pa-fko and a  typical 
auction is 111794690250.
Pawel actually lists these as being for the Philips/Fluke PM6685,  this 
being the counter he designed them for, and with a comment  that they also work 
with the PM6680, but those are rebadged Pendulum  counters and a look at 
the service manuals suggested that  the original  Pendulum prescalers were 
likely to be a one module fits all solution, and  this has turned out to be the 
For the past few weeks I've been using a pair of these modules,  one fitted 
to a CNT90 and the other to a Tekktronix FCA3100, which is a  rebadged 
CNT91, and I'm very happy with the results.
Once installed the counter automatically recognises that it now has an  
input "C", and that's all there is to it.
The built in statistical plotting options work for input C just as they do  
for inputs A and B, but note that the level measuring functions of inputs A 
and  B are hardware dependent and aren't available for input C, this also 
applies to  the original Pendulum prescalers.
There's a graph of measured sensitivity and "application area" included  
with the auction photos and I'm seeing results that agree closely with that. 
Up  to 1GHz I've run tests using a Marconi 2022E signal generator and above 
that  I've been playing with a Chinese clone of the Anolog ADF4351 evaluation 
kit. I  don't have any independent means of checking the levels from the 
ADF4351 kit  above 1.5GHz but I've no reason to believe it's not performing as 
it should,  certainly the frequency accuracy is as expected.
The service manuals indicate that later models have extra pins on  the 
prescaler connector but the functions of the rear sixteen pins have  remained 
unchanged, and ensuring that the prescaler is mounted to the rear  of the 
connector, regardless of the number of pins that leaves spare, is all  that's 
A couple of points to note, the mounting  clips shown in the auction photos 
are already fitted to the PM6685 but  were not fitted to my CNT90 or 
FCA3100, so these had to be supplied with the  prescaler. Anyone ordering these 
should make it clear what counter they will  fitted to.
There are two predrilled 2.5mm diameter holes in the  front panel for the N 
connector that will accept M2.5 screws but the shape of  the casting does 
not allow any nuts to be fitted.
The prescaler board is provided with self tapping  screws, although I chose 
to run an M3 tap through the holes and use  standard screws, but I suspect 
the best solution would  probably be to use M3 Taptite screws.

For anyone who might be interested I've uploaded a file containing a set of 
 similar photos for each counter showing stages of the installation, and 
also  some photos of the counters under test and the ADF4351 module......
Apologies for the messy looking link, it's nothing  sinister, just that the 
site encrypts all files and most of that link is the  decryption key.
For these tests, both counters and both frequency sources  were referenced 
to the same Trimble Thunderbolt.
At 1GHz, using the Marconi 2022 signal generator,  I checked sensitivity 
down to -40dBm, possibly could have gone lower but  didn't see too much need 
for that:-)

The ADF4351 evaluation kit only has selectable output levels of +5, +2, -1, 
 and -4 dBm, so for higher frequencies I took a different approach  and 
tested for the maximum frequency that would display properly on the  counter at 
levels of -4dBm, as indicated by the ADF4351 software, and also  at -7dBm 
and -10dBm by adding attenuators.
The module was mounted a very short distance from the  counter using 
semi-rigid coax  to an SMA to N adapter, again  this can be seen in the photos,  
and 23GHz rated Aeroflex  3dB SMA attenuators were added directly between the 
synthesiser module and  the counter input, although this is not shown.
Both counters worked fine at 3.00 GHz with the -4dBm signal and whilst  
there was a difference between the two counters as the frequency increased  
it's obvious that both would have worked fine at 3.00GHz with the -7dBm  signal 
also and very close at -10dBm.
It might be interesting to swap the modules between counters, to  see if 
that has any effect on the overall sensitivity at higher  frequencies but it's 
not something I've tried so far, and I don't see it as  a problem either 
With the FCA3100 (CNT91), the limits I measured were.....
at -4dBm  ----  3.500GHz
at -7dBm  ----  3.400GHz
at -10dBm  ---  3.325 GHz
With the CNT90, the limits I measured were......
at -4dBm  ----  3.200GHz
at -7dBm  ----  3.025GHz
at -10dBm  ---  2.925GHz
The usual disclaimers apply, I paid full price for both my units and  I'm 
just a very happy customer with no vested commercial interest  in this item 
whatsoever, but believing it to be a great product  and excellent value I 
felt that others here might be interested  too.
I've mentioned the Pendulum service manuals, and will comment further on  
that in a follow up post.



More information about the time-nuts mailing list