Call Search

New to Ham Radio?
My Profile

Friends Remembered
Survey Question

DX Cluster Spots

Ham Exams
Ham Links
List Archives
News Articles
Product Reviews
QSL Managers

Site Info
eHam Help (FAQ)
Support the site
The eHam Team
Advertising Info
Vision Statement

donate to eham
   Home   Help Search  
Pages: [1]   Go Down
Author Topic: Filters made by Piezo Technology  (Read 7817 times)

Posts: 4

« on: November 04, 2012, 09:38:33 AM »

I have couple of crystal filters made by Piezo Technology Inc. and I would like to use them in homebrew receivers, but I don't know this filter's termination impedances.  I'm pretty sure this filter was used in a commercial/military receiver, maybe a RACAL unit, maybe some other receiver.  The receiver's schematic would reveal the impedances, so if you know what receiver used a Piezo Technology Inc. Model 1815, center freq 1750 kHz, bandwidth 500 Hz filter, let me know.  The filter is also stamped with what looks like the receiver manufacturer's in-house part number:  1920-0603.  Happy holidays to everyone, tnx, 73.....Larry  KB2EE

Posts: 800


« Reply #1 on: November 04, 2012, 11:01:40 AM »

Have you tried contacting PTI?

Alternatively, if you have a network analyzer or access to one, you could try some different turns ration toroidal transformers and  see what results in the lowest insertion loss.

Dale W4OP

Posts: 4

« Reply #2 on: November 04, 2012, 11:13:15 AM »

Hi Dale,
I think most filters termination impedances are complex, not simple and pure resistances.  Lacking a network analyzer, I lean towards seeing what an end-user manufacturer did in applying the filter, then confirming or fiddling with as necessary, the response as seen on my spectrum analyzer.

Yeah, I will see if I can contact PTI......figured they would ignore a junk box inquiry from a ham, but worth a shot.  Tnx.....LC

Posts: 800


« Reply #3 on: November 04, 2012, 05:15:58 PM »

Hi Larry,
re PTI: Maybe not. I have had them build  VHF crystal filters in very small quantities. Maybe you will get ahold of a guy who has been there a long time and recalls your filter.

The Zin/Zout may well be complex for a perfect conjugate match, but I think many mfg just specify a terminating resistance.

I have worked backwards from the schematic as long as I can locate the small signal paramters of the active device the filter looks into. In my experience, most are bilateral- so if you can calculate one side, you have the other.

Dale W4OP

Posts: 1560

« Reply #4 on: November 04, 2012, 07:15:43 PM »

Depnding on frequency the 5-11 mhz filters are usually around 200-500ohms, the lower HF
filters are typically 500 to as high as 5000 ohms.

Its possible to sweep a filter with a oscillator and a power detector (diode) with variable terminations.
tends to be a very tedious but otherwise simple process.  When you get close the filter shape will
match the known data for bandwidth and ripple.  Generally there is about a 2:1 range of error on
that with minimal ill effects.


Posts: 800


« Reply #5 on: November 04, 2012, 07:37:00 PM »

For the same crystals, the  in/out impedance goes down quickly as the filter BW narrows- so I would think it would be difficult to judge Rin and Rout solely based on frequency.

As an example, a  6 pole filter I built from 5MHz rocks has  Rin and out of just under 1K. The same crystals configured as a 500Hz filter has Rin  and out of 85 Ohms. Different topologies  will likely yield different results- but I believe the trend remains the same.

Dale W4OP

Posts: 165

« Reply #6 on: November 04, 2012, 10:40:12 PM »

I use a 1kOhm series cermet pot between sweeper and filter and the same between filter and detector - I use a spectrum analyser. Then adjust the pots for minimum ripple. Add a 30 pf Trimmer across the filter in and aout and adjust again for minimum ripple. Then measure pots and capacitors.

Where the filters were diode switched, it was common to use 50 ohm filters: otherwise 1 kohm and 15pF in shunt was a common termination.

Posts: 4710

« Reply #7 on: November 07, 2012, 10:37:29 AM »

Would you be interested in selling or trading one or more?

My thought is to build an 80/40 band-imaging receiver with an LO that tunes 5.25 - 5.375 MHz. Would cover 3.5 to 3.625 and 7.0 to 7.125 MHz simply by retuning the input tuned circuits.

73 de Jim, N2EY

Posts: 46

« Reply #8 on: November 07, 2012, 11:47:08 AM »


If you get stuck and can't get the termination Z figured out, you can send one my way and I will match it up for best response, document the necessary termination impedance, and plot the response.


Posts: 325

« Reply #9 on: November 07, 2012, 04:57:44 PM »

Grab a couple of 5 or 10 K pots.
Connect the pots in series with
the filter's input and output.
Attach a 51 ohm resistor to the
pot on the side towards the
generator and connect the other
end of the resistor to ground.
Do the same on the output side
of the filter. Connect the 51 ohm
resistor to the pot on the side away
from the filter. Now sweep the filter
adjusting the pot on input and output
for minimum passband ripple. Remove
the filter from the test set and measure
the dc resistance of each pot. You will
have the value of the input and output
filter impedance.

Are they the ones that look kind of like
an individual crystal or do they look
like a one inch deep drawn metal
case maybe one inch wide and 1.5
inches long. (Approximate dimensions)

My guess is the filters are 21.4 MHz
or 10.7 MHz. Usually Piezo was pretty
good about marking the center frequency
on the side of the metal case. I only
remember 10.7 MHz, 21.4 MHz, 30 MHz
and 70 MHz models. PTI sold a boatload of
these filters back in the late 70's and
through the 80's. Typical insertion
loss with a good input and output
matching network should be less than
1.5 to 2 dB.

The filter ripple when properly matched
should exhibit maybe 0.5 dB insertion loss. Well designed
LC matching networks will provide a nice
low ripple level (less than a 0.25 dB) and
low insertion loss. The pots will
yield the flatest response but the insertion
loss with the pots is likely to be on the
order of 40 dB or so.

If memory serves me correct the nominal
input and output frequency of the filters
will be around 1500 ohms if they are for
21.4 MHz. Typical BW's were six KHz, 15 KHz,
25 KHz and 50 KHz at 21.4 MHz depending on the
model. They were popular in COMINT receivers
used by the intel and surveillance community.

Once you hit the requirement for 100 KHz you
switched over to LC Lumped element filters.
Normally the next step up from a receiver with
a 50 KHz filter was 500 KHz (lumped element)
then 1 Mhz and 2 MHz. Several of the boxes
offered during that time generally used 1, 2 and
4 MHz lumped element filters.

Hope this helps. Let me know if futher info will
help. I might still have the Piezo catalog around.

As an aside the same styles were made by both
Piezo and PTI (Piezo Technology Inc). They were
two separate outfits, one was in the State College,
PA area and the other in Florida. I thought they
were the same company but during a conversation
with the PTI rep one day he advised they were
different outfits.

Chuck WD4HXG

Posts: 325

« Reply #10 on: November 07, 2012, 05:10:23 PM »

For those looking for crystal filters they
frequently show up on e-Bay. Very popular
are the 21.4 MHz center frequency and 15
KHz bandwidth.

For a matching network use a circuit tuned
to the center frequency on the input and
output. Split the capacitor using two variable
caps to provide a 50 ohm match. Alternately
adjust the the two caps on input and output
for minimum insertion loss and minimum ripple.

I remember the cap values were relatively small.
25 pf or so max.


Chuck WD4HXG

Posts: 165

« Reply #11 on: November 08, 2012, 12:45:05 AM »

The OP said that they were 1750kHz centre frequency. So forget 10.7 or 21.4MHz.

Posts: 119

« Reply #12 on: November 17, 2012, 03:21:43 AM »


According to the information I've got on 1750 kHz filters, the termination that's most often used is 600 ohms in parallel with 10 pF. However, Racal in the UK seem to prefer 1000 ohms in parallel with 75 pF.

I've collected information on commercial crystals filters over the years because of an amateur interest in filters, but have no idea in which bits of equipment they're used. If PTI manufactured filters for Racal in the US, there is always the possibility they might have required them to have different termination impedances to the ones used by the UK Racal since they would have been designing different bits of kit for different markets. Anyway, it's worth trying the two terminations I've mentioned to see if either of them seem to be the right one.

Whatever you do, don't use the criterion of minimum insertion loss to establish the right terminations for data or CW filters. They are usually terminated to give the lowest differential group delay and the amplitude response and loss are subordinate to achieving that. Linear phase, Bessel or Guassian-Transitional filter responses are preferred for CW and data. I usually go for linear phase myself. It's hard to beat overall.

Jim, N2EY,
If Larry doesn't want to part with either of his 1750 kHz CW filters, I can let you have a bunch of high Q 1748 kHz crystals to build your own CW filter, if you want. I can also let you have a 6-pole linear phase design for 1748 kHz crystals which I did for Patrick, GW0VMR, and myself a few years back. We re-designed the filter for the SimpleX Super and restricted the coverage to 100 kHz at the CW end of 80 and 40.

73, Dave. 
Pages: [1]   Go Up
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.11 | SMF © 2006-2009, Simple Machines LLC Valid XHTML 1.0! Valid CSS!