开云体育

  1. Tinysa
  2. Topics

Measure bandpass filter

 

Hello guys,?
I am waiting for mt TinySA ultra. Can somebody direct me to how to use the Tiny sa Ultra to measure the 3db skirts of a band pass filter.

Thanks in adv and Happy holidays
Andrea
 

Although the tinySA is not designed for filter measurements you can watch this video for some inspiration.
https://www.youtube.com/watch?v=l0xPmkf_EkI
--
For more info on the tinySA go to https://tinysa.org/wiki/
 

开云体育

Thank you Eric

On 24 Dec 2022, at 09:00, Erik Kaashoek <erik@...> wrote:

Although the tinySA is not designed for filter measurements you can watch this video for some inspiration.

--
For more info on the tinySA go to https://tinysa.org/wiki/
 

Do in menu MEASURE / -3dB WIDTH
 

Andrea,

I don't know how long you have been on this list, but I recently posted my experience using a newly-purchased noise source to test a low-pass filter. The noise source approach would work for a band pass filter as well. My TinySA Ultra measurements compare very favorably with my measurements using a NanoVNA. I also simulated the filter using Elsie software and the results also compared favorably with the measured results until the gain got down to about -45 dB when the effects of real components caused the response to level-out as compared to the simulation that used ideal components. My low pass filter was also constructed on a breadboard with some short length of wire running to SMA connectors which probably affected the measured response.

The TinySA and TinySA Ultra are great tools for a very reasonable cost considering their capability!
--
Ed, K0KC
?
k0kc@...
http://k0kc.us/
 

good info!


On Sat, Dec 24, 2022 at 9:30 AM Ed Wilson via <ed.wilson=[email protected]> wrote:
Andrea,

I don't know how long you have been on this list, but I recently posted my experience using a newly-purchased noise source to test a low-pass filter. The noise source approach would work for a band pass filter as well. My TinySA Ultra measurements compare very favorably with my measurements using a NanoVNA. I also simulated the filter using Elsie software and the results also compared favorably with the measured results until the gain got down to about -45 dB when the effects of real components caused the response to level-out as compared to the simulation that used ideal components. My low pass filter was also constructed on a breadboard with some short length of wire running to SMA connectors which probably affected the measured response.

The TinySA and TinySA Ultra are great tools for a very reasonable cost considering their capability!
--
Ed, K0KC
?
 

Hi Ed,

Thanks for?getting back to me. I am a newbe on this group. Is it a video you posted? If so can you send me the?link please?

Thanks
Andrea

On Sat, 24 Dec 2022, 15:30 Ed Wilson via , <ed.wilson=[email protected]> wrote:
Andrea,

I don't know how long you have been on this list, but I recently posted my experience using a newly-purchased noise source to test a low-pass filter. The noise source approach would work for a band pass filter as well. My TinySA Ultra measurements compare very favorably with my measurements using a NanoVNA. I also simulated the filter using Elsie software and the results also compared favorably with the measured results until the gain got down to about -45 dB when the effects of real components caused the response to level-out as compared to the simulation that used ideal components. My low pass filter was also constructed on a breadboard with some short length of wire running to SMA connectors which probably affected the measured response.

The TinySA and TinySA Ultra are great tools for a very reasonable cost considering their capability!
--
Ed, K0KC
?
 

Sorry, Andrea, I did not post a video but I would be glad to assist you in any way that might be helpful once you receive your TinySA Ultra.

Please keep in mind that I am also a "novice" using a noise source to make filter response measurements. If you are going to do a lot of this type of work, you might consider investment in a NanoVNA which provides much more detailed measurement of filter characteristics.
--
Ed, K0KC
?
k0kc@...
http://k0kc.us/
 

On Sat, Dec 24, 2022 at 06:30 AM, Ed Wilson wrote:
Andrea,

I don't know how long you have been on this list, but I recently posted my experience using a newly-purchased noise source to test a low-pass filter. The noise source approach would work for a band pass filter as well. My TinySA Ultra measurements compare very favorably with my measurements using a NanoVNA. I also simulated the filter using Elsie software and the results also compared favorably with the measured results until the gain got down to about -45 dB when the effects of real components caused the response to level-out as compared to the simulation that used ideal components.
With the noise generator method you have limited range and cannot measure the stopband attenuation much beyond what you observed. ? This is because the noise generator level is not that high compared to the noise floor of the spectrum analyzer.?? If you want to see more measurement range you can use the slow sweep generator method with Trace Calc set to max.? That way you can set the generator to -25 dBm and you are much further above the noise floor.?

Roger
 

Follow up to my last post.? Below is a s21 plot of a Para Dynamics low pass TX filter that was done using a NanoVNA-H4.




One method to measure a filter with a tinySA is to use a Noise Generator (NG) but one has to be careful that the NG output is not too high because it might damage the tinySA.? I have a lab NG that outputs 100 mV RMS into a 50 ohm load? (-6.9 dBm) when set to a specific noise bandwidth. In the plot below? the tinySA measures -88 dBm/Hz with the trace marker set to Noise.? With 120 MHz BW the total noise power is 10*log(120E06) = -88 + 81= -7 dBm which is within the safe limit of the tinySA



There is another method to verify the total noise power with the tinySA and is recommended when the noise spectrum is not reasonably flat. ? The technique requires use of the Channel Power measurement tool.? Below a span of 114 MHz. divided into 3 bands yields measurements of -12.1, -13.2 and -13.8 dBm.? Added together on? power basis this is a total of -8.2 dBm which is very close to the calculations above.?




In the plot below the span is set from 2 MHz. to 120 Mhz. and the Trace Averaging is turned on to get a smooth trace. Note that the power measured by marker 1 is the total power measured with a resolution bandwidth of 600 kHz. The noise has a slight rolloff so the Trace Normalize is activated prior to measuring the filter.?




Next we insert the filter and make the response measurement.? It is clearly not showing the maximum attenuation of the filter that was obtained with the NanoVNA.? There is not enough dynamic range and increasing the NG level has risks of damage to the tinySA.




However if one has a RF or AWG signal generator with sweep capability there is a solution that can be used when the SA does not have a tracking generator.?? For this test a Siglent 2122x is set to sweep over the frequency range of 2 MHz to 120 MHz over 120 seconds.? This is slow enough that the tinySA can make a measurement in Max Hold mode.? After a sweep the normalize function is used to adjust for amplitude vs frequency variation of the generator.?? Then the filter is inserted and the measurement made (reset of Max Hold is required.) The filter measurement is much better and similar to that obtained with the NanoVNA.




Summary - These filter measurements were made with a tinySA Ultra but also applies to the tinySA.? Filter measurement is easier with a NanoVNA? but good results can be obtained with a tinySA using a noise or signal generator in swept frequency mode.

Roger
 

开云体育

Hi Roger

Your calculation is not fully correct as when sung a span of 120MHz (118MHz) your noise power is directed to the mixer via the 800MHz low pass filter so the math is 10*log(800E06) = -88 + 89= +1 dBm and on the brink to be too high. Using Ultra the bandwidth it is even higher

Kind regards

Kurt

?

Fra: [email protected] <[email protected]> P? vegne af Roger Need via groups.io
Sendt: 24. december 2022 22:25
Til: [email protected]
Emne: Re: [tinysa] Measure bandpass filter

?

Follow up to my last post.? Below is a s21 plot of a Para Dynamics low pass TX filter that was done using a NanoVNA-H4.




One method to measure a filter with a tinySA is to use a Noise Generator (NG) but one has to be careful that the NG output is not too high because it might damage the tinySA.? I have a lab NG that outputs 100 mV RMS into a 50 ohm load? (-6.9 dBm) when set to a specific noise bandwidth. In the plot below? the tinySA measures -88 dBm/Hz with the trace marker set to Noise.? With 120 MHz BW the total noise power is 10*log(120E06) = -88 + 81= -7 dBm which is within the safe limit of the tinySA



There is another method to verify the total noise power with the tinySA and is recommended when the noise spectrum is not reasonably flat. ? The technique requires use of the Channel Power measurement tool.? Below a span of 114 MHz. divided into 3 bands yields measurements of -12.1, -13.2 and -13.8 dBm.? Added together on? power basis this is a total of -8.2 dBm which is very close to the calculations above.?




In the plot below the span is set from 2 MHz. to 120 Mhz. and the Trace Averaging is turned on to get a smooth trace. Note that the power measured by marker 1 is the total power measured with a resolution bandwidth of 600 kHz. The noise has a slight rolloff so the Trace Normalize is activated prior to measuring the filter.?




Next we insert the filter and make the response measurement.? It is clearly not showing the maximum attenuation of the filter that was obtained with the NanoVNA.? There is not enough dynamic range and increasing the NG level has risks of damage to the tinySA.




However if one has a RF or AWG signal generator with sweep capability there is a solution that can be used when the SA does not have a tracking generator.?? For this test a Siglent 2122x is set to sweep over the frequency range of 2 MHz to 120 MHz over 120 seconds.? This is slow enough that the tinySA can make a measurement in Max Hold mode.? After a sweep the normalize function is used to adjust for amplitude vs frequency variation of the generator.?? Then the filter is inserted and the measurement made (reset of Max Hold is required.) The filter measurement is much better and similar to that obtained with the NanoVNA.




Summary - These filter measurements were made with a tinySA Ultra but also applies to the tinySA.? Filter measurement is easier with a NanoVNA? but good results can be obtained with a tinySA using a noise or signal generator in swept frequency mode.

Roger
 


On Sun, Dec 25, 2022 at 03:56 AM, Kurt Poulsen wrote:

Hi Roger

Your calculation is not fully correct as when sung a span of 120MHz (118MHz) your noise power is directed to the mixer via the 800MHz low pass filter so the math is 10*log(800E06) = -88 + 89= +1 dBm and on the brink to be too high. Using Ultra the bandwidth it is even higher

Kurt,

Your math would be correct if I was using a very wideband noise generator.

However, in my post I mentioned that my noise generator can be set to a specific bandwidth (it? has a programmable filter in the output).? I did not show the spectrum beyond 120 MHz but the screenshot below shows the noise spectrum when set to cutoff at 120 MHz.? Here is a channel power measurement that is wider than the one in my post.? Total Noise power in 147 MHz span is -11.8 dBm + -13 dBm + -17 dBm? = -8.6 dBm which is a safe level.




When using the noise generator method to measure filters I suggest users place a filter on the output of their wideband noise generator that limits the noise to the band of interest.?? Doing this means that for a given total noise power they can have a higher noise level in dBm/Hz and therefore get more measurement range.


Roger
 

开云体育

Hi Roger

My apology, I did not scroll down to see you excellent description. Sorry once again

Kind regards

Kurt

?

Fra: [email protected] <[email protected]> P? vegne af Roger Need via groups.io
Sendt: 25. december 2022 18:56
Til: [email protected]
Emne: Re: [tinysa] Measure bandpass filter

?


On Sun, Dec 25, 2022 at 03:56 AM, Kurt Poulsen wrote:

Hi Roger

Your calculation is not fully correct as when sung a span of 120MHz (118MHz) your noise power is directed to the mixer via the 800MHz low pass filter so the math is 10*log(800E06) = -88 + 89= +1 dBm and on the brink to be too high. Using Ultra the bandwidth it is even higher

Kurt,

Your math would be correct if I was using a very wideband noise generator.

However, in my post I mentioned that my noise generator can be set to a specific bandwidth (it? has a programmable filter in the output).? I did not show the spectrum beyond 120 MHz but the screenshot below shows the noise spectrum when set to cutoff at 120 MHz.? Here is a channel power measurement that is wider than the one in my post.? Total Noise power in 147 MHz span is -11.8 dBm + -13 dBm + -17 dBm? = -8.6 dBm which is a safe level.




When using the noise generator method to measure filters I suggest users place a filter on the output of their wideband noise generator that limits the noise to the band of interest.?? Doing this means that for a given total noise power they can have a higher noise level in dBm/Hz and therefore get more measurement range.


Roger

1 - 13 of 13