Message	Date	From	Subject
8920	2013-08-08 14:35:27	Wes	More on crystal Qu Measurement
Hi all, Just the other day I had measured a bunch of 10 MHz crystals for use in a narrow filter. The original plan had called for a filter at 11 MHz, but upon measurement of the crystals I had on hand, I decided to move to 10 MHz where the crystals in my junk box had higher Qu. To be sure about crystal parameters, I repeated the measurements. The characterizations were, by now, many years old. All of the filter crystals I picked had Qu>200,000. I did the measurements with the N2PK vector network analyzer. After writing a note this morning regarding the use of a single crystal filter as a method for crystal Qu determination, it seemed appropriate to build such a filter and do the measurement on a crystal that had just been measured with other means. The crystal I picked from the batch had Cp of 3.5 pF, which was measured directly with an AADE L/C meter, and a motional inductance of .0198 H, measured with both a G3UUR test oscillator and with the VNA. A crystal filter was built with a scrap of circuit board material, a pair of shunt 650 pF mica capacitors, and the selected crystal. A cable was run from a DDS signal source and was attached to a 10 dB 50 Ohm pad at the crystal filter end. This provided an available power to the filter of -8 dBm, which is high enough for good measurements, but low enough to not cause crystal problems. This signal was applied to the filter. The other end of the filter was attached to a short piece of coax that then ran to a 50 Ohm terminated digital storage scope. The signal source was tuned for maximum output at the scope. The scope level was measured and recorded. The filter was then replaced with a coaxal through barrel and the level again recorded. This allowed the insertion loss to be calculated at 2.029 dB. The filter bandwidth was somewhere around 100 Hz, although this value was not used in this measurement. The shunt coupling capacitors were checked with the AADE L/C meter just to be sure that I had not used poor capacitors; they came in at 652 pF. They were marked as 1% parts, so all is consistent. OK, we now have a measured result and can use computer modeling to extract a Q value. The filter was "built" in LADPAC using the program "ladbuild08.exe." The resulting circuit was saved and then analyzed with the general purpose ladder analysis, GPLA08.exe. The Qu value was selected (it's parameter #103) and tuned. The sweep was set up to go to a max attenuation of 4 dB and the dB/step was set at 2.029. This gave a well defined "target". The tune function was then used to tune on the Qu value until we had a good fit. The result of this measurement was a crystal Qu of 249,000. The same crystal had been measured at Qu=255,000 with the VNA. Clearly, I cheated by using the computer programs to extract a Qu value from a measurement of insertion loss. A little algebra would provide an exact formula. There is nothing exotic about any of these measurements. It's just a matter of thinking about fundamentals. This single element filter method should provide as much information as we need when designing a crystal filter. The only place we use Qu is to estimate the filter insertion loss. Have fun. 73, Wes w7zoi
8923	2013-08-08 18:51:44	William Carver	Re: More on crystal Qu Measurement
The message that I heard was at the end: two very close values of Q from two totally different measurement techniques. Of course numbers derived from properly thinking things out, and carefully performing experiments. W7AAZ On Thu, 2013-08-08 at 21:35 +0000, Wes wrote: > > Hi all, > > Just the other day I had measured a bunch of 10 MHz crystals for use > in a narrow filter. The original plan had called for a filter at 11 > MHz, but upon measurement of the crystals I had on hand, I decided to > move to 10 MHz where the crystals in my junk box had higher Qu. To be > sure about crystal parameters, I repeated the measurements.

Hi all,

Just the other day I had measured a bunch of 10 MHz crystals for use in a narrow filter. The original plan had called for a filter at 11 MHz, but upon measurement of the crystals I had on hand, I decided to move to 10 MHz where the crystals in my junk box had higher Qu. To be sure about crystal parameters, I repeated the measurements. The characterizations were, by now, many years old. All of the filter crystals I picked had Qu>200,000. I did the measurements with the N2PK vector network analyzer.

After writing a note this morning regarding the use of a single crystal filter as a method for crystal Qu determination, it seemed appropriate to build such a filter and do the measurement on a crystal that had just been measured with other means. The crystal I picked from the batch had Cp of 3.5 pF, which was measured directly with an AADE L/C meter, and a motional inductance of .0198 H, measured with both a G3UUR test oscillator and with the VNA. A crystal filter was built with a scrap of circuit board material, a pair of shunt 650 pF mica capacitors, and the selected crystal. A cable was run from a DDS signal source and was attached to a 10 dB 50 Ohm pad at the crystal filter end. This provided an available power to the filter of -8 dBm, which is high enough for good measurements, but low enough to not cause crystal problems. This signal was applied to the filter. The other end of the filter was attached to a short piece of coax that then ran to a 50 Ohm terminated digital storage scope. The signal source was tuned for maximum output at the scope. The scope level was measured and recorded. The filter was then replaced with a coaxal through barrel and the level again recorded. This allowed the insertion loss to be calculated at 2.029 dB. The filter bandwidth was somewhere around 100 Hz, although this value was not used in this measurement. The shunt coupling capacitors were checked with the AADE L/C meter just to be sure that I had not used poor capacitors; they came in at 652 pF. They were marked as 1% parts, so all is consistent.

OK, we now have a measured result and can use computer modeling to extract a Q value. The filter was "built" in LADPAC using the program "ladbuild08.exe." The resulting circuit was saved and then analyzed with the general purpose ladder analysis, GPLA08.exe. The Qu value was selected (it's parameter #103) and tuned. The sweep was set up to go to a max attenuation of 4 dB and the dB/step was set at 2.029. This gave a well defined "target". The tune function was then used to tune on the Qu value until we had a good fit. The result of this measurement was a crystal Qu of 249,000. The same crystal had been measured at Qu=255,000 with the VNA.

Clearly, I cheated by using the computer programs to extract a Qu value from a measurement of insertion loss. A little algebra would provide an exact formula.

There is nothing exotic about any of these measurements. It's just a matter of thinking about fundamentals. This single element filter method should provide as much information as we need when designing a crystal filter. The only place we use Qu is to estimate the filter insertion loss.

Have fun.

73, Wes
w7zoi