After delivery of my filter modulator board from the BATC shop (thank you!), I am about to start construction of my Portsdown transmitter. I was wondering, if anybody is directly feeding the 5.2V for the Raspberry Pi to the filter modulator board? I checked the schematic diagram and I believe the board should be able to handle it. Or is there another good reason to keep the supplies separate?
73s Achim DH2VA
Powering the Filter Modulator Board from 5.2V directly?
Re: Powering the Filter Modulator Board from 5.2V directly?
The general problem is the PI USB supply loses too much voltage so I would run it from the same supply, but not through the PI. Having said that, in an attempt to keep phase noise down, I keep the supplies separate.
Mike
Mike
Re: Powering the Filter Modulator Board from 5.2V directly?
Hi Mike,
as I have now a complete stable test setup with some 80+ dB attenuation between Portsdown and Minitiouner, I could start serious measurements.
I do reach the given MER values (slightly above 21-22dB) at 23cm and I do get the 4-dot pattern for each state in the constellation diagram as seen in other posts and here:
https://wiki.batc.org.uk/Filter_Modulator_ver2
This pattern is not related to the common power supply (supplied 5.2V separately without any changes) and also not related to the LO (tried to feed from signal generator with no improvement.
I know that the 4-dot pattern is purely academic as the Portsdown will perform nicely like it is, but I would like to give it a try:
I have the feeling that the group of the Nyquist filter is not as flat as it could be leading to this symptom. Simulations in LTSpice show this kind of behaviour which 'could' be tweaked.
Is it correct, that the Nyquist filters are a compromise between power outside our main signal (the 'shoulders') and minimizing inter-symbol inteference ISI? If so, what was the rationale for choosing the dimensioning for the current Nyquist filters?
73s Achim
as I have now a complete stable test setup with some 80+ dB attenuation between Portsdown and Minitiouner, I could start serious measurements.
I do reach the given MER values (slightly above 21-22dB) at 23cm and I do get the 4-dot pattern for each state in the constellation diagram as seen in other posts and here:
https://wiki.batc.org.uk/Filter_Modulator_ver2
This pattern is not related to the common power supply (supplied 5.2V separately without any changes) and also not related to the LO (tried to feed from signal generator with no improvement.
I know that the 4-dot pattern is purely academic as the Portsdown will perform nicely like it is, but I would like to give it a try:
I have the feeling that the group of the Nyquist filter is not as flat as it could be leading to this symptom. Simulations in LTSpice show this kind of behaviour which 'could' be tweaked.
Is it correct, that the Nyquist filters are a compromise between power outside our main signal (the 'shoulders') and minimizing inter-symbol inteference ISI? If so, what was the rationale for choosing the dimensioning for the current Nyquist filters?
73s Achim
Re: Powering the Filter Modulator Board from 5.2V directly?
Hi Achim
The rationale was to have something easy to build, not too expensive, that worked. We used a switched capacitor filter at 500 KS and below; that gave us the good brick-wall cutoff that we needed to suppress out of band transmissions, but is a bit sharp and causes some distortion.
The switched capacitor filter was not fast enough for 1 MS and above, and the band edges generally further away, so we went with good-enough LC filters.
If we had done anything more complex, we would have started to approach the price point of SDRs and the DATV Express, which do give much nicer constellations and out-of-band suppression.
As ever, many design compromises, and perhaps we could have done better with more optimisation. However, on strong signals it doesn't matter, and the impact of the modulation errors is minimal (< 1dB) on weak signals because the error is nearly drowned out by noise.
John, who did the LC filter design, might be able to tell you the criteria that he used.
Dave
The rationale was to have something easy to build, not too expensive, that worked. We used a switched capacitor filter at 500 KS and below; that gave us the good brick-wall cutoff that we needed to suppress out of band transmissions, but is a bit sharp and causes some distortion.
The switched capacitor filter was not fast enough for 1 MS and above, and the band edges generally further away, so we went with good-enough LC filters.
If we had done anything more complex, we would have started to approach the price point of SDRs and the DATV Express, which do give much nicer constellations and out-of-band suppression.
As ever, many design compromises, and perhaps we could have done better with more optimisation. However, on strong signals it doesn't matter, and the impact of the modulation errors is minimal (< 1dB) on weak signals because the error is nearly drowned out by noise.
John, who did the LC filter design, might be able to tell you the criteria that he used.
Dave
Re: Powering the Filter Modulator Board from 5.2V directly?
No idea, you will need to ask the designers. I think the original Portsdown filtering was a little too tight but that has been resolved and it will be OK for DVB-S2. It is a compromise between ISI and bandwidth.
I am currently testing with a lime mini,1080p H264 and the results at 4 Ms/s are stunning. I think we need to improve our video codecs and get away from the standard ones to improve now.
Mike
I am currently testing with a lime mini,1080p H264 and the results at 4 Ms/s are stunning. I think we need to improve our video codecs and get away from the standard ones to improve now.
Mike
Re: Powering the Filter Modulator Board from 5.2V directly?
Thank you Mike and Dave!
So I believe I am on the right track.. I am willing to try and play around with the component values of the existing Nyquist filters and keep the architecture. It will be interesting to see how MER and bandwidth are related
Achim
So I believe I am on the right track.. I am willing to try and play around with the component values of the existing Nyquist filters and keep the architecture. It will be interesting to see how MER and bandwidth are related
Achim