Thanks for the clear and informative video. One possibility why you are not seeing the theoretical filter response is that you may be overdriving the receiver with the noise signal. If you are then intermod products generated in the receiver will mix back into what should be the filter stopband and "fill" in the filter response. Try it again with some attenuation in front of the receiver and you should see an improved display of the filter response.
Thanks for the suggestion. That would make sense why I would not see what I expect. What would be the best/simplest way to do that? Put a low value resistor or potentiometer in-line with the coax?
As it's matched to 50 ohms you need to put an attenuator in series rather than just a resistor or potentiometer. With that noise source and receiver I'd start with a 20 dB or 30 dB attenuator. You can always build a 50 ohm attenuator with three resistors. There are plenty of attenuator calculators on the net or as phone apps to work out the resistor values.
The SDR Sharp method is much better than using the oscilloscope FFT function. I was able to characterize the frequency response of my RF PA by using your method. However, this requires lots of attenuators, even if the output level doesn't overwhelm the input of the SDR device. Definitely you have to pay attention to compression, if using that method.
Hi, did you manage to account for the difference in theoretical filter response to actual? The 2 suggestions below seemed to make sense. I want to try experimenting with Elsie myself.
I computed the basic plan with Elsie and chose the caps to get close to the values in Elsie. Then once I got all the real-world caps, I measured them and plugged the measured values back in. I still don't think the measured response in SDR Sharp matches the calculated values as well as I would like. Might be my construction/soldering techniques.
@@kugellagers This is and older video etc. However, If you look up dynamic range of a spectrum analyzer, noise sources and channel power you wil understand why your filter appeared to have less attenuation than it acutally does. Enjoy...
Thanks! The SDR is just a very basic RTL SDR setup. You can see my build of it here: ua-cam.com/video/4zWsDq-Bmvg/v-deo.html The splitter is just a very cheap TV antenna switch.
I also notice the expected deep nulls in the response are not present. I would first suspect leakage around the filter. Try testing the filter with only a direct path, noise generator->filter-> analyzer rather than with the TV grade switches at each end of the filter. (or remove the direct cable between the switches) For high performance filters, it may be necessary to add shields between the filter sections, as shown here ua-cam.com/video/1sq8Cvju2Oo/v-deo.html
Thanks for the clear and informative video. One possibility why you are not seeing the theoretical filter response is that you may be overdriving the receiver with the noise signal. If you are then intermod products generated in the receiver will mix back into what should be the filter stopband and "fill" in the filter response. Try it again with some attenuation in front of the receiver and you should see an improved display of the filter response.
Thanks for the suggestion. That would make sense why I would not see what I expect. What would be the best/simplest way to do that? Put a low value resistor or potentiometer in-line with the coax?
As it's matched to 50 ohms you need to put an attenuator in series rather than just a resistor or potentiometer. With that noise source and receiver I'd start with a 20 dB or 30 dB attenuator. You can always build a 50 ohm attenuator with three resistors. There are plenty of attenuator calculators on the net or as phone apps to work out the resistor values.
The SDR Sharp method is much better than using the oscilloscope FFT function. I was able to characterize the frequency response of my RF PA by using your method. However, this requires lots of attenuators, even if the output level doesn't overwhelm the input of the SDR device. Definitely you have to pay attention to compression, if using that method.
What causes the compression? Is it related to the sample rate?
Hi, did you manage to account for the difference in theoretical filter response to actual? The 2 suggestions below seemed to make sense. I want to try experimenting with Elsie myself.
I computed the basic plan with Elsie and chose the caps to get close to the values in Elsie. Then once I got all the real-world caps, I measured them and plugged the measured values back in. I still don't think the measured response in SDR Sharp matches the calculated values as well as I would like. Might be my construction/soldering techniques.
@@kugellagers
This is and older video etc. However, If you look up dynamic range of a spectrum analyzer, noise sources and channel power you wil understand why your filter appeared to have less attenuation than it acutally does.
Enjoy...
Great video.. What type of SDR and splitter are they thanks
Thanks! The SDR is just a very basic RTL SDR setup. You can see my build of it here: ua-cam.com/video/4zWsDq-Bmvg/v-deo.html The splitter is just a very cheap TV antenna switch.
But can noise characterize the phase response?
That's beyond my pay grade and dedication to what I use it for. I'm just a casual experimenter. :)
No. For that you need a vector network analyzer. The inexpensive nanoVNA is one option that many people will be able to afford to add to their bench.
I also notice the expected deep nulls in the response are not present. I would first suspect leakage around the filter. Try testing the filter with only a direct path, noise generator->filter-> analyzer rather than with the TV grade switches at each end of the filter. (or remove the direct cable between the switches) For high performance filters, it may be necessary to add shields between the filter sections, as shown here ua-cam.com/video/1sq8Cvju2Oo/v-deo.html
Thanks! I will give those all a try and see how the response changes.