EMRFD Message Archive 6701

Message Date From Subject
6701 2011-09-23 08:34:30 ad7zu crystal filter basic questions
Some questions about crystal filter drive levels ..assume a 4 pole lattice filter

How does the drive level affect the Q and response? is there an optimum drive level?

Does the drive level have an impact on spurious resonances?



Randy
AD7ZU
6702 2011-09-23 10:11:20 Tim Re: crystal filter basic questions
Garden variety microprocessor crystals above a few MHz have rated drive power of a few mW. (Tuning fork crystals are way way lower, in the microwatts).

ECS app notes for their monolithic crystal filters hint that +10dBm levels can cause physical damage, and that IMD increases measurably between -60dBm and -10dBm.

At the same time I look at high IMD receiver designs and know that crystal filters get hit with out-of-band frequencies at levels up to several volts without any obvious to me problems. Many of those of course also have post-mixer diplexers.

Tim N3QE

6703 2011-09-24 08:58:01 kb1gmx Re: crystal filter basic questions
6704 2011-09-24 09:20:43 Dan Mills Re: crystal filter basic questions
On Sat, 2011-09-24 at 15:57 +0000, kb1gmx wrote:

> The big problem is near the band edges the filter input has remarkable
> reactance and phase changes that the driving stage has to handle
> gracefully or that stages distortion figures are adversely affected.

Yea, using a quadrature hybrid filter helps, but at the cost of doubling
the amount of quartz required, I don't know of any off the shelf crystal
filters that take this approach, but there are a few crystal roofing
filter designs out there that do this.

Worse, the filter sometimes gets placed right after the mixer without
any intermediate amplifier, which causes the transition band impedance
changes to appear as the mixer termination, a diplexer helps here, but
even so it can be a problem. I have seen
6705 2011-09-24 10:23:48 AD7ZU Re: crystal filter basic questions
Thanks to all for the  informative tutorial.
 
I am in the midst of modifying a kit that sorely needs some re-design of the mixer / xtal filter interface. 
I am tempted to reduce the gain ahead of the filter in both RX and TX paths and may attempt to add a simple diplexer as suggested along with a pad following both rx and tx mixers (diode rings).
to my ear and with limited lab equipment the filter seems too narrow but i doubt that is the case, more likely there is significant ripple or spurs in the passband.
 
Randy

6706 2011-09-26 06:04:59 Tim Re: crystal filter basic questions
Radios can sound bad for many reasons.

The most common reason for spurs are not the fault of any one component, but is a poor choice of LO/IF frequency combinations, a fundamental system (not component or subsystem) design issue.

As part of that system design, the realization that the crystal filter will let through odd harmonics (or near-odd-harmonics) of its fundamental frequency has to be take into account too. I think the traditional mixer-dB-level charts fail to take this into account with enough severity. (Possibly because the traditional Collins spur chart was probably designed around mechanical filters? Do mechanical filters let odd harmonics through the same way crystal filters do? I would guess the end resonators are low-passes above the fundamental but maybe not.)

In any event L-pads favoring low pass characteristics to match the crystal filters helps a lot with those odd harmonic spurs. In other words, use transformers to match the crystal filters if you want to be able to see the spurs on your test equipment. Use L-matches if you want to keep them out of your radio :-)

A poorly terminated crystal filter can indeed have a lot of ripple, and a pad will help there. But even more than just ripple in the passband, it can also really make the inherent group delay properties far worse. This can result in everything sounding really bad without any obvious numerical (dB) factor visible from single-signal sweep testing.