27-11-2019, 11:04 AM
OT
TX and /or Generation is MUCH easier than RX (or video capture):
* DACs are easier to do MUCH cheaper and faster than ADCs. Even resistors and FPGA or PIC output pins can do to generate video up to at least SVGA.
* While oversampling makes output filters simpler, it's less critical than input filters for ADC. The 48 KHz of MP3 doesn't mean 24 KHz or even 22kHz audio. It's to make the antialiasing filtering simpler. A sharp filter just above the required band creates ringing / poor transient performance and pass band ripple. Even FM radio needed filters to limit deviation and also later to prevent interference to the 19kHz pilot for the DSBSC 38KHz L-R channel. The CD Audio 44.1kHz was a compromise for achievable anti-aliasing filters, below the 22.050 KHz, and to suit the digital tape archive based on a Sony helical scan system. Later A/D was done at twice or four times rate to have "better" cheaper analogue filters and then DSP decimation to 44.1 KHz. Probably there shouldn't be much difference in frequency response for FM radio, CD, MP3 or DAB. The differences are in distortion, especially at lower bitrates of DAB and MP3. Low bit rate (digitally created distortion) is worse for people with poorer hearing.
* DDS can be done in SW + DAC, discrete HW, or ADI single chips. You can use DDS with say a 20MHz IF and then upconvert to 50MHz to 1GHz etc. Higher frequencies, 5GHz to 400 GHz may use a second IF in the 450MHz band and a monolithic filter for the unwanted sideband. DDS has been done for DVB-S, DVB-T, DVB-C and DAB. Usually the DAC o/p is in the 50MHz band. Most DVB-S, DVB-T, DVB-C and DAB test gear and feeds for transmitters uses IF, with a filtered conversion to desired band. For my 10.240GHz DVB-C transmission experiments I replaced the modem uplink input (5 to 65MHz band and the uplink for data is normally 0.8, 1.6, 3.2 or 6.4MHz) with the IF out of DVB-T/DVB-C/DVB-S PCI card in the PC, at about 50MHz. The MPEG2 multiplex stream was encoded in advance with MPEG4 / H.264 SD and HD channels, which took all day for a 2 hour loop on a PC. You need seriously expensive gear to encode multiple TV channels in real time.
(26-11-2019, 03:48 PM)ppppenguin Wrote: As for SDR, only the most expensive ones will be doing direct synthesis for TX (or ADC for RX) at much over a few MHz. Where "a few" is not well defined but might be as low as 10MHz or as high as 50MHz. Typically they will work at IF and use up (TX) or down (RX) conversion. not really my field but they may well use quadrature conversion. Mike, as you found, many will be working at very much lower IF.
Frank Cuffe and I both did some work on DDS for Band 1. He got a decent result for 45MHz vision using a sample rate around 200MHz. DACs for this sort of speed are readily available. I think his was 14 bit. I did a rough trial at about 16MHz with about 80MHz clock. My setup (borrowed from another project) wasn't really suitable . for anything higher. It's all written up somewhere here on GVR.
TX and /or Generation is MUCH easier than RX (or video capture):
* DACs are easier to do MUCH cheaper and faster than ADCs. Even resistors and FPGA or PIC output pins can do to generate video up to at least SVGA.
* While oversampling makes output filters simpler, it's less critical than input filters for ADC. The 48 KHz of MP3 doesn't mean 24 KHz or even 22kHz audio. It's to make the antialiasing filtering simpler. A sharp filter just above the required band creates ringing / poor transient performance and pass band ripple. Even FM radio needed filters to limit deviation and also later to prevent interference to the 19kHz pilot for the DSBSC 38KHz L-R channel. The CD Audio 44.1kHz was a compromise for achievable anti-aliasing filters, below the 22.050 KHz, and to suit the digital tape archive based on a Sony helical scan system. Later A/D was done at twice or four times rate to have "better" cheaper analogue filters and then DSP decimation to 44.1 KHz. Probably there shouldn't be much difference in frequency response for FM radio, CD, MP3 or DAB. The differences are in distortion, especially at lower bitrates of DAB and MP3. Low bit rate (digitally created distortion) is worse for people with poorer hearing.
* DDS can be done in SW + DAC, discrete HW, or ADI single chips. You can use DDS with say a 20MHz IF and then upconvert to 50MHz to 1GHz etc. Higher frequencies, 5GHz to 400 GHz may use a second IF in the 450MHz band and a monolithic filter for the unwanted sideband. DDS has been done for DVB-S, DVB-T, DVB-C and DAB. Usually the DAC o/p is in the 50MHz band. Most DVB-S, DVB-T, DVB-C and DAB test gear and feeds for transmitters uses IF, with a filtered conversion to desired band. For my 10.240GHz DVB-C transmission experiments I replaced the modem uplink input (5 to 65MHz band and the uplink for data is normally 0.8, 1.6, 3.2 or 6.4MHz) with the IF out of DVB-T/DVB-C/DVB-S PCI card in the PC, at about 50MHz. The MPEG2 multiplex stream was encoded in advance with MPEG4 / H.264 SD and HD channels, which took all day for a 2 hour loop on a PC. You need seriously expensive gear to encode multiple TV channels in real time.
