If it don't glow it don't go!
What exactly is wrong with FETs as current sources?
Some DACs have ouputs in the ~<10v range.
(few amps want to see that, for sure) Talking about headroom issues.
For the sat TV iso, the way to go is with a 1:1 balun on the RF input coax line! Then ground the stupid box to the same place the system is grounded, with the dish end of the coax on a *hard* ground, that puts the balun outside, btw. (it needs to be a real rf transformer, not an auto transformer, so that there is no connection through the case, of course)
If there is still a groundiness issue from the dish's box then add an audio transformer between that and ur preamp, imho.
This sort of deal works nicely on cable TV too, where the cable co rarely has a good ground for the incoming coax.
I'd never use a BUF O3 for anything... the device has a horrid sound everytime I've heard it used which leads me to think that the spectra of the apparently low distortion is oddly placed so that it is most annoying!
Ick! FYI replacing it with a much more expensive National hybrid buffer chip in the output of a Theta DAC was *extremely* audible in the positive direction.
The other question I have is why is a tube being used at all? A solid state component in the same hole would need potentially less "care and feeding" to make it work. Are we looking for a sonic benefit from the tube? In which case why isn't the tube type being selected in that light?
As far as the issue of bandwidth... I suppose that depends on your "theory" of audiophilosophy I suppose. In my view it's a good idea to not alter the input signal, assuming it is a high quality signal. In which case bandwidth out to ~250kHz. isn't a bad goal. On the other hand if you actually want to add "euphonia" to a given signal (like a highly processed TV source) then my view is to build specifically designed circuits with that goal in mind....
just thinking out loud...
_-_-bear
Some DACs have ouputs in the ~<10v range.
(few amps want to see that, for sure) Talking about headroom issues.
For the sat TV iso, the way to go is with a 1:1 balun on the RF input coax line! Then ground the stupid box to the same place the system is grounded, with the dish end of the coax on a *hard* ground, that puts the balun outside, btw. (it needs to be a real rf transformer, not an auto transformer, so that there is no connection through the case, of course)
If there is still a groundiness issue from the dish's box then add an audio transformer between that and ur preamp, imho.
This sort of deal works nicely on cable TV too, where the cable co rarely has a good ground for the incoming coax.
I'd never use a BUF O3 for anything... the device has a horrid sound everytime I've heard it used which leads me to think that the spectra of the apparently low distortion is oddly placed so that it is most annoying!
Ick! FYI replacing it with a much more expensive National hybrid buffer chip in the output of a Theta DAC was *extremely* audible in the positive direction. The other question I have is why is a tube being used at all? A solid state component in the same hole would need potentially less "care and feeding" to make it work. Are we looking for a sonic benefit from the tube? In which case why isn't the tube type being selected in that light?
As far as the issue of bandwidth... I suppose that depends on your "theory" of audiophilosophy I suppose. In my view it's a good idea to not alter the input signal, assuming it is a high quality signal. In which case bandwidth out to ~250kHz. isn't a bad goal. On the other hand if you actually want to add "euphonia" to a given signal (like a highly processed TV source) then my view is to build specifically designed circuits with that goal in mind....
just thinking out loud...
_-_-bear
Some good thoughts!
Regarding the use of a tube, you're correct it is irrational, just as with the use of tubes anywhere in a sound reproduction system (with some extremely rare and niche-y exceptions). But I'd rather drive a '57 T-Bird than a '05 Lexus.
Ideally, yes, we would take care of ground and noise issues at the source. But sources change all the time and there's lots of them. I can kill all the problems, present and future, with a single transformer.
FET current sources I used were drifty and had lowish output Z that varied quite a bit with signal. I'm sure a good one can be implemented, but it was so easy to do with bipolars and get very high performance...
In any case, note that the signal is mostly developed across the load, not the CCS.
Regarding the use of a tube, you're correct it is irrational, just as with the use of tubes anywhere in a sound reproduction system (with some extremely rare and niche-y exceptions). But I'd rather drive a '57 T-Bird than a '05 Lexus.
Ideally, yes, we would take care of ground and noise issues at the source. But sources change all the time and there's lots of them. I can kill all the problems, present and future, with a single transformer.
FET current sources I used were drifty and had lowish output Z that varied quite a bit with signal. I'm sure a good one can be implemented, but it was so easy to do with bipolars and get very high performance...
In any case, note that the signal is mostly developed across the load, not the CCS.
Yeah, SY, I've been meaning to mention that your avatar pix seems to indicate a rather intense condition! Could this account for your "irrational" audio choices??
Anyhow, perhaps you will be willing to try using Mosfeters in place of all the bipolar silicon devices in this circuit and see if you measure or hear any difference?? I'd be curious about that...
_-_-bear
Anyhow, perhaps you will be willing to try using Mosfeters in place of all the bipolar silicon devices in this circuit and see if you measure or hear any difference?? I'd be curious about that...
_-_-bear
> For the sat TV iso, the way to go is with a 1:1 balun on the RF input coax line! Then ground the stupid box to the same place the system is grounded, with the dish end of the coax on a *hard* ground, that puts the balun outside, btw.
Strictly speaking, not a "balun" because we don't want to change the unbalanced signal format to balanced. However I agree that all such gizmos end up being called "balun" even when they are really "un-iso-un"s.
And they may not work here. The satellite dish has an amplifier/converter in the dingus on the arm at the focus of the dish. This has to get power from the inside box (unlike cable-TV which powers its repeaters from the pole). While it would be possible to send power up at 50MHz (so it would pass a transformer) and rectify it, that might need a bigger iso-transformer. ANd most folks won't need such a trick. It is a lot simpler to run DC up the cable and use a choke/capacitor splitter to get the up-power and down-signal going the right way. DC won't pass a standard isolation transformer. You could combine an RF transfo with a common-mode choke, but the CM-Choke has to rise to high impedance by 60Hz, which isn't a standard part.
Depending on the source of the buzz, some clamp-on ferrites on the RF wire might help, or might do nothing. Winding the cable 100 times around a very large transformer core might null the audio band without hurting DC or RF, but that's a big lump of work and it may not help.
If it were only the satellite box, the "rational" thing would be to transfo its outputs, or use a diff-amp to take its output. The satellite feed can't be SO good that a couple low-price transfos or a 5532 diff-amp would hurt it, and your other, high-quality, sources could go direct to the grid/pot without iron in the path. But pure rationality seems to be about #11 on the list of goals. And some good iron won't do any harm to the sound; some folks like extra iron in their system.
Strictly speaking, not a "balun" because we don't want to change the unbalanced signal format to balanced. However I agree that all such gizmos end up being called "balun" even when they are really "un-iso-un"s.
And they may not work here. The satellite dish has an amplifier/converter in the dingus on the arm at the focus of the dish. This has to get power from the inside box (unlike cable-TV which powers its repeaters from the pole). While it would be possible to send power up at 50MHz (so it would pass a transformer) and rectify it, that might need a bigger iso-transformer. ANd most folks won't need such a trick. It is a lot simpler to run DC up the cable and use a choke/capacitor splitter to get the up-power and down-signal going the right way. DC won't pass a standard isolation transformer. You could combine an RF transfo with a common-mode choke, but the CM-Choke has to rise to high impedance by 60Hz, which isn't a standard part.
Depending on the source of the buzz, some clamp-on ferrites on the RF wire might help, or might do nothing. Winding the cable 100 times around a very large transformer core might null the audio band without hurting DC or RF, but that's a big lump of work and it may not help.
If it were only the satellite box, the "rational" thing would be to transfo its outputs, or use a diff-amp to take its output. The satellite feed can't be SO good that a couple low-price transfos or a 5532 diff-amp would hurt it, and your other, high-quality, sources could go direct to the grid/pot without iron in the path. But pure rationality seems to be about #11 on the list of goals. And some good iron won't do any harm to the sound; some folks like extra iron in their system.
Yeah but SY what about that apparent glandular problem??
SY you lose the filament transformer, etc... and with only unity gain required, would the differentials you mentioned be a serious issue? <--- for the "gain" part
As far as the regulator and the CCS, it seems to me that I've seen measurements that show that the spectra of noise is nicer with Mosfets over bipolars...
PRR, you are correct of course about the "balun" and the "phantom supply" biz.
_-_-bear
SY you lose the filament transformer, etc... and with only unity gain required, would the differentials you mentioned be a serious issue? <--- for the "gain" part
As far as the regulator and the CCS, it seems to me that I've seen measurements that show that the spectra of noise is nicer with Mosfets over bipolars...
PRR, you are correct of course about the "balun" and the "phantom supply" biz.
_-_-bear
What glandular problem? Do you think I look like someone with a glandular problem? Just today, one of the girls in the office compared me to Mel Gibson.
The 12 volt rails will prove to be useful. Trust me.
CS noise seems to be dominated by modulation of the reference current. Presumably that could be lowered by feeding the reference with yet another current source... "It's turtles all the way down!" But any CS noise is shunted off by the divider ratio of the CS impedance with the Thevenin equivalent of the looking-back impedance at the tube's cathode paralleled with the ultimate load.
The 12 volt rails will prove to be useful. Trust me.
CS noise seems to be dominated by modulation of the reference current. Presumably that could be lowered by feeding the reference with yet another current source... "It's turtles all the way down!" But any CS noise is shunted off by the divider ratio of the CS impedance with the Thevenin equivalent of the looking-back impedance at the tube's cathode paralleled with the ultimate load.
I've dropped four or five candidate tubes into this for measurement purposes. They all worked fine other than the microphonic CCas. Gold grid is a very nice thing to have when you're driving with a very high impedance source.
So far, my favorite is 6KN8/6RHH8.
Which Vreg are you talking about?
So far, my favorite is 6KN8/6RHH8.
Which Vreg are you talking about?
SY said:
Gotta be gold... if it isn't gold ya izn't gonna get that warm golden tone now will ya?? Ya gotzta have golden tone, right?
Which Vreg? Isn't there a Vreg in that schematic??
Thought I saw one upper right... but any Vreg will do...
Ah heck, it doesn't really matter as long as it sounds good!
_-_-bear
PS. Lookin' at dee avatar, if yaz tinks yaz looks like Tommy Cruise, best also check wit un shrink two??
Ah, OK, I think I see. I was trying to get my power supply chapter and schematics up before leaving town (I hit the road in an hour), but some computer problems (@#%$ing Outlook!) and a hyperactive 4 year old took care of that ambition. Looks like Tuesday for that.
Just between you and me, there will be two options presented: an RCRCRC that knocks the rail noise down to the thermal remnant of the Big Bang for those who are satisfied with this version, or an active two-transistor job (bipolars again!) for those who want to explore the interesting world of direct coupling with servos. The +/-12V rail is easily accommodated by 317/337 and some capacitors. Heater load helps knock down supply impedance and noise.
The fundamentals of the design make power supply issues easy to deal with. Signal to noise targets are reached by low tech means. A good time is had by all.
Just between you and me, there will be two options presented: an RCRCRC that knocks the rail noise down to the thermal remnant of the Big Bang for those who are satisfied with this version, or an active two-transistor job (bipolars again!) for those who want to explore the interesting world of direct coupling with servos. The +/-12V rail is easily accommodated by 317/337 and some capacitors. Heater load helps knock down supply impedance and noise.
The fundamentals of the design make power supply issues easy to deal with. Signal to noise targets are reached by low tech means. A good time is had by all.
No chokes in the coffee can. And an RCRCRC following the reservoir cap drops the supply noise to ridiculously low levels without any of the bulk, expense, and field issues of chokes. Besides, some real drop is needed there- with a C input filter, an isolation transformer will get you something like 160V on the rail, 70 of which need to be knocked down.
Choke input filters are popular, and someone willing to nurse that along is welcome to do so. In this application, the additional regulation of L input is not important.
But there are many mansions in my father's house (wait a minute, I used that quote already), and I would not argue with anyone who loves chokes to go that route.
Choke input filters are popular, and someone willing to nurse that along is welcome to do so. In this application, the additional regulation of L input is not important.
But there are many mansions in my father's house (wait a minute, I used that quote already), and I would not argue with anyone who loves chokes to go that route.
Frequency response opinion
SY,
Thanks for this little project.
I have one "bone to pick".
You stated that there is no/little evidence that extending the frequency response past 20 to 25kHz has any benefit.
My experience from my tube projects BUT more particularly from the work I did with Hugh Dean on the AKSA 55 Nivarna Plus option to his SS Amp is that when considering the effect on stereo imaging this NOT the case.
You want to limit the phase shift in the 20Hz to 20kHz audio band to not more than say 5 degrees for optimum stereo imaging. The ear is very sensitive to phase. To achieve that limited phase shift in the audio band the frequency response needs to be extended by about 2 octaves in both directions. Allen Wright in his Preamp Cookbook talks about extending it two decades in each direction BUT that is "over the top".
That is 5Hz to 80kHz is what we should be aiming for.
Any comments? I've painted a big "bulls eye" target on my T-shirt so feel free to take a pot shot.
Cheers,
Ian
SY,
Thanks for this little project.
I have one "bone to pick".
You stated that there is no/little evidence that extending the frequency response past 20 to 25kHz has any benefit.
My experience from my tube projects BUT more particularly from the work I did with Hugh Dean on the AKSA 55 Nivarna Plus option to his SS Amp is that when considering the effect on stereo imaging this NOT the case.
You want to limit the phase shift in the 20Hz to 20kHz audio band to not more than say 5 degrees for optimum stereo imaging. The ear is very sensitive to phase. To achieve that limited phase shift in the audio band the frequency response needs to be extended by about 2 octaves in both directions. Allen Wright in his Preamp Cookbook talks about extending it two decades in each direction BUT that is "over the top".
That is 5Hz to 80kHz is what we should be aiming for.
Any comments? I've painted a big "bulls eye" target on my T-shirt so feel free to take a pot shot.
Cheers,
Ian