Here is the data from a batch of 4 crystals previously match by another method.
ID - crystal identification
-3dbm - bandwidth at -3dbm
Fs - serial resonance freq.
Rs - series resistance
Cm - motional capacitance
Lm - motional inductance
Q - Quality factor
ID Fs hz -3dbm BW hz Rs Cm Lm Q
x2 4,913,469 102 8.76 0.0199 52.68 185,601
x4 4,913,475 108 9.99 0.0203 51.56 159,350
x1 4,913,487 105 9.43 0.0201 52.19 170,851
x3 4,913,487 120 14.17 0.0202 51.95 113,192
x2 4,913,469 102 8.76 0.0199 52.68 185,601
x4 4,913,475 108 9.99 0.0203 51.56 159,350
x1 4,913,487 105 9.43 0.0201 52.19 170,851
x3 4,913,487 120 14.17 0.0202 51.95 113,192
Found the match for Fs pretty good. Although the numbers for x3 appear to be bit of an outlier and I wonder if this explains why there is a bit to much of a ripple in the pass band. My previous post shows about a 5dbm ripple and I believe this should be around 1dbm max. I only have 4 of the crystals on hand so I can't easily test this theory without more crystals on hand. So instead I grabbed a bunch of different crystals and matched those.
I used a breadboard to mount the crystals and built a filter with the best ones and made some measurements. After which I put in a couple of the outliers and did see a substantial increase in the ripple. I tried changing capacitor values as well, but in the end I gave up as the breadboard is not really a good test bed for this experiment, at least as far as making definitive judgement on the values of components. It did serve to prove however, as to what effect changes in capacitors and crystals can have on the characteristics of the filter. I will need to design a board to facilitate in making component changes easily and quickly and allow me to have confidence in measurements.
Its amazing the amount of reading there is on this subject for such a small circuit.
73, Peter
Did you do any checking on the caps? Impedance can cause these waveforms as well, maybe trimmers added in series/parallel at the in/out can smooth it out.
ReplyDelete73, ve3gkn
I did play around somewhat with the caps. Added capacitance, took away etc. It didn't take to long before I realized that soldering and unsoldering so many combinations that it was not going to end up well. So I need to think about a test jig to support this kind of activity. I did see a lot of changes in bw. and between that and some reading on the subject, I am questioning the values of the caps that are being used. Specifically the middle one and the 2 caps leading in and out of the filter.
Delete1, why is the middle cap not the same as the other 2?
2, the 2 39pf caps, it might be worth seeing a lower value here.
3, i did see the importance of matching. Which of course brings into question the toroids.
4, from what I've read, the importance of crystal matching is very important to the ripple. That's why I wondered about the other parameters I measured bedsides the frequency. The frequency match alone can't get much better, but the crystal x3 does have I think significant differences in bw, Rs, and Q. Is this why the ripple is around 5dbm? Won't know for sure until I build a test jig and get a few more crystals from you. hihi.
So, in order to make sense of this, build the test jig and do further testing. I will need some more crystals though and hopefully find one close to the other parameters. This would show for sure the effect they have on the ripple and of course anything else. May work on a board tonight over a beer.
Need the 06:00 again Gord.
73, Peter