Exactly, Francis, as I've stated previously one important thing is to let the input of the ADC see a correct *and constant* source impedance. Software is the most complex part of the AGC in this scheme.
Depending on how the ADC is used and the way the resulting samples are processed (say decimation and/or down conversion) the data flow will be of considerable magnitude. This "self decides" the architecture of the following stages. If the data flow is relevant it maybe impossible to make the PC to perform the successive process. For example a USB connection should be too slow to transport the data due to the fact that the only isochronous standard available under USB has been foreseen for slow audio applications only. BTW I have got a look at Firewire but it is a really complex way to realize the transfer even if the ideal one. If memory helps there is a device-side IEEE1394 implementation for FPGAs but I should investigate further for final costs of such a solution. Anyway lets go back to the AGC. Tendency is to use the ADC in a oversampling concept using a double or more sampling rate as the highest frequency to convert. As an example to convert a 30MHz signal it should sampled at least at 60Msps but 130 or 170Msps are better (and actually feasible). This has the consequence to multiply by some factors the rate of flowing converted data from the ADC stage! At this point a DSP (and maybe some reduction through the use of a FPGA or other high speed components) has to be used. You have mentioned the AGC control signal generation using a DAC approach. Another possible (and old known) solution is to use an I/O pin as output for a PWM signal followed by an integrator. This solution is more cost effective. ;-) Maybe other guys of this list will have more lighting lamps... vy 73s de ik2wqi - Andreas FRANCIS CARCIA wrote: > I suppose it would be easy to put an analog loop around the front end that > reduces the gain as the saturation point of the A/D is approached. A variable > gain amplifier or passive pin attenuator would work well. Another approach > would to use the computer to generate a D/A output based on the noise floor > of the frequency of interest and maximum signal coming into the A/D. Flex > does this with an amplifier and a pad or none to get 3 input levels so the > dynamic range can be shifted in 3 10 dB steps by the operator. Frank WA1GFZ > ------------------------ Yahoo! Groups Sponsor --------------------~--> Check out the new improvements in Yahoo! Groups email. http://us.click.yahoo.com/6pRQfA/fOaOAA/yQLSAA/ELTolB/TM --------------------------------------------------------------------~-> Yahoo! Groups Links <*> To visit your group on the web, go to: http://groups.yahoo.com/group/soft_radio/ <*> To unsubscribe from this group, send an email to: [EMAIL PROTECTED] <*> Your use of Yahoo! Groups is subject to: http://docs.yahoo.com/info/terms/
