toggle quoted messageShow quoted text
Those crystals in that big bag may have anything from footprints to tire track and dents on them :) Some of the processes I have looked at in the past for selecting crystals involve equipment many of us don't have. I have yet to read the article you referenced (but will). There are reasons why the commercial, type accepted, "high performance" radios have four and five digit price tags.
On the other hand many of us have had useful improvements in performance with more casual approaches. We might also consider costs in part of our performance evaluation. I have a three-xtal, CW filter using television color burst xtals that used the first three xtals I picked out of the bin at the store. It has a passband less than 200 Hz! lumps and bumps in the passband of a CW filter are features and not problems :) Others have experienced usable results by frequency matching a handful of xtals in SSB bandwidths. Somewhat less than perfect? Probably. Expensive? No way. Fun? It's all fun.
Now to read the rest of the messages in the thread and then look at that article.
On 10/23/2018 12:03 PM, Jerry Gaffke via Groups.Io wrote:
Here's a nice classic article on crystal ladder filters:
Lots of other stuff out on the web describing other aspects of this.
Also in EMRFD.
They recommend that crystals be matched in frequency to within +-2% of
your filter bandwidth.
So for a 2500hz SSB bandwidth, the crystals should be within +-50hz.
If you buy well spec'd crystals from a reputable distributor,
they might be held to within 20ppm, or within 12mhz*20ppm = +-240 hz.
You will need quite a few of them sorted by frequency
to get a bunch within 50hz of some center frequency.
I don't know what's in that bag of cheap crystals.
Perhaps they are a surplus run of really good 20ppm crystals.
Perhaps they were rejected because they were well beyond 100ppm off
Perhaps they were not active enough, and would thus make for a very
Mouser and Digikey sell 10ppm crystals of known provenance at 2 for the
20ppm and 50ppm for quite a bit cheaper.
It's not just frequency that should be matched, though that's the most
Read the article.
Though if not too worried about passband shape and loss, and with some luck,
you can just try some crystals matched in frequency and see what you get.
It may be necessary to try different cap values and termination impedances
till the filter works well enough.
Or you may find out it's best to simply go through the steps outlined in
On Tue, Oct 23, 2018 at 08:00 AM, Bill Cromwell wrote:
I am thinking that your 50 percent is on the low side and of the
other 50 percent many of those would match well enough as an
additional set of filters.
bark less - wag more