In which case the trimpot should have its fixed ends connected to the source of the JFETs, and the wiper to the output / drain of the MOSFETs, I guess.
I would be tempted to use 20R. And the Idss has to go to 9mA to compensate for the degeneration.
Patrick
I would be tempted to use 20R. And the Idss has to go to 9mA to compensate for the degeneration.
Patrick
I think even without the trimpot, you will get lots of feedback, especially if the output is connected to a load to ground.
The load wants to be below 50 ohms, to avoid excessive open-loop gain.
But the DC offset is determined by how good the matching is.
The trimpot should provide a means for DC offset adjustment.
Right ?
Patrick
The load wants to be below 50 ohms, to avoid excessive open-loop gain.
But the DC offset is determined by how good the matching is.
The trimpot should provide a means for DC offset adjustment.
Right ?
Patrick
Yes, that is the idea. To make the current the same in R3 and R4 or adjusting the Fets for same Gm. By just looking at the schematic i do not know where the distortion null is. When it coinsides with the offset null you have luck. Otherwise you can null the offset by trimming the source resistors or develop some kind of bias servo.
The distortion null is when the drain resistors are identical (hence P to N matched mosfets), and the JFETs are Yfs matched (see my published curves at the Pass forum). The K1530/J201 are perfect complementaries, as I published a long while ago at the Pass F5 thread.
That means in reality an extra 5-ohm degeneration resistor at the source of J74. And the rest is all done by matching (in this case not necessarily identical Idss). The trimpot is only necessarily for the last mV offset trim then.
Am not home this week and next, so I cannot spice this for a while.
(Will see Jan next week.....)
Patrick
That means in reality an extra 5-ohm degeneration resistor at the source of J74. And the rest is all done by matching (in this case not necessarily identical Idss). The trimpot is only necessarily for the last mV offset trim then.
Am not home this week and next, so I cannot spice this for a while.
(Will see Jan next week.....)
Patrick
For all of you who are having trouble with dither, here is an even bigger brain-cramp: Compressed sensing prototype demos improved DAQ
Essentially, remove the anti-aliasing filter, add a known modulation and decode more info from the resultant mess. Given who is pursuing it its probably very real in its benefits. I can barely get my head around the concept and i will assiduously avoid the math.
Essentially, remove the anti-aliasing filter, add a known modulation and decode more info from the resultant mess. Given who is pursuing it its probably very real in its benefits. I can barely get my head around the concept and i will assiduously avoid the math.
I believe that is discussing a sub-band coding technique combined with averaging - not readily applicable to audio - there's no reason to believe that modulating the audio to translate to higher frequencies would be more linear that existing audio converters - the technique's benefits would come at much higher signal frequencies where current technology ADC sample rate, bit depth is too low for you
if we had audio ADC with sub lsb linearity but insufficient resolution we could apply uncorrelated (preferably subtractive) dither to the audio at the input of multiple ADC and digitally average for increased resolution and S/N output
but SOTA monolithic ADC for audio linearity, resolution, S/N is substantially beyond any previous recording technology – and industrial/medical instrumentation requirements keep pushing the technology – now well beyond “audio” demands - 20 bit linearity at 1 MHz sample rate is available today
if we had audio ADC with sub lsb linearity but insufficient resolution we could apply uncorrelated (preferably subtractive) dither to the audio at the input of multiple ADC and digitally average for increased resolution and S/N output
but SOTA monolithic ADC for audio linearity, resolution, S/N is substantially beyond any previous recording technology – and industrial/medical instrumentation requirements keep pushing the technology – now well beyond “audio” demands - 20 bit linearity at 1 MHz sample rate is available today
Last edited:
20 bit A/D and 1MS/s would be fine for audio. In this case the digital filter was not required to make up resulting shape of signal.
Once again, listening to the Sony FM tuner, shows great potential, great bass, but a 'grittiness' even with live transcriptions. I think that the coupling caps are the biggest problem, as the Blowtorch should be able to handle the RFI with relative ease.
Dunno about this FM thing... seems rather hopeless as things stand today. For one thing the signal chain on the broadcast end is rather questionable, for the other end that pesky 19kHz left/right switching bit always gave FM a "sound" of its own. To me sort of like that sound that you got from a Phase Linear 400 or 700, and maybe not that good...
John, for your own use and not as a commercial solution, I think I might try some output iron on each channel, after changing out all the caps, tranzeestors and ICs for the better that I could find... the idea is that the iron would do something as an "integrator" of sorts, "smoothing" the grit a bit... easy enough to try?
_-_-bear
John, for your own use and not as a commercial solution, I think I might try some output iron on each channel, after changing out all the caps, tranzeestors and ICs for the better that I could find... the idea is that the iron would do something as an "integrator" of sorts, "smoothing" the grit a bit... easy enough to try?
_-_-bear
Maybe I am missing something, but i do not understand why FM is discussed in a high end thread. Inherently bad dynamics and bad resolution. It may sound "pleasant", but nothing more. It is just impossible to get enough dynamic range from FM.
Pavel,
The last US FM station I did takes the digital output from the CD player or server and stays digital through the mixing engine through the studio to transmitter link and into the transmitter. There it is turned to analog for one carrier and data compressed for the digital carrier.
The stereo generator may even be done in DSP.
Since digital is perfect, as we all know, what more could you want? The uncompressed data stream?
Comparing the Off Air signal to what is being sent out does show differences. The mandatory compressor shows up, as does some degradation.
One listener complained he did not like the announcer's microphone (an RE20) and donated some Sennheiser ones. I prefer the older darker ones the newer ones have a bit too much sibilance for my taste.
But I suspect the broadcast quality is better than what is on many CD's.
ES
PMA, where I am located, I do not find this to be so. In the SF Bay Area, we still have a lot of folk music and jazz played directly from records, or even live. Also, some broadcasts, while necessarily digital in origin, still can sound pretty darn good from these same stations. The best example I know, is 'The Prairie Home Companion' that broadcasts weekly from NPR stations, our community supported ones. Through a quality tube tuner like the Marantz 10, or even a Fisher 200B tube tuner, these special broadcasts can be nearly perfect, exposing me, at least, to a variety of singers, notably female, who might be worth finding additional recordings of.
This little SONY tuner is 1/2 complete. The digital processing seems almost perfect, yet it has a marginal analog output. This is why I am addressing it. Fix that, and who knows?
This little SONY tuner is 1/2 complete. The digital processing seems almost perfect, yet it has a marginal analog output. This is why I am addressing it. Fix that, and who knows?
The best FM sound i heard in Germany is a direct broadcast from The Alte Oper in Frankfurt on HR1. This does not go over tape and is a direct feed from the mixing board.
- Status
- Not open for further replies.