fdegrove said:
Hi fdegrove
Gotta say, definitely NOT recommended. As a side comment, no wonder that CFs gets a bad name if used in ways such as that.
This raises the recurring question (one that has appeared already on this thread) why Cathode Followers gets such a bad name. For a start, CFs must have either very stiff power supply IF same supply is used for other stages, SS or tubes. Or they must have independent supplies altogether.
Should we discuss further the merits and/or downsides of Cathode Followers? It's a good topic!
Quadruple!
I am certainly no expert on semiconductors but in a valved design that one wouldn't fly for sure...
The similarity doesn't exist? Oh yes it does:
Take TWO tube amps, feed the identical signals but out-off-phase to each, Left and Right if a stereo amp, then connect the speaker across the + off both secondaries, and provided the 'G' of the secondaries are really grounded... what result would we get?
More power!
So it can be done with tubes. Indeed a number of older stereo tube designs had exactly this facility built-in. Dynaco ST-70?
No it didn't quadruple the power, largely because the impedance matching didn't allow it, but I seem to recall that doubling power was claimed.
But in SS, assuming we are talking about a near theoretical 'voltage' amp (implying infinite current capability) the double voltage swing into the load will also double the current (like 40V peak into 8 Ohm results in peak 5A, but 10A if 4 Ohm - Ohm's Law), then double x double = FOUR.
If you then RMS that, then you will realise 40V peak (or 80V peak-to-peak) gives 100W into 8 Ohm and 200W into 4 Ohm.
Now how can we swing 80V p/p to give 100W when our power supply is only 25V plus 25V.
We need to swing +/-40V or 80V p/p - not possible when we have only +/-25V available.
How is it done? Use TWO amplifiers. As one amplifier non-inverting goes to a max of +25V the inverting one is going -25V. We now have a plus 50V peak. When the reverse happens, the results is minus 50V peak; or 100vp/p.
So they criss-cross the voltage across the load.
If this example is into 8 Ohm speaker, then it is seeing 100V p/p. Now RMS that to 156W, and 312W into 4 Ohm.
Put differently, if the inherent losses were the same, then a balanced output, TWO amplifiers into the same load, then half the rail voltages are required.
This is why so many car amplifiers have balanced out and uses 4 Ohm standard. The 4 Ohm gives twice the potential power, the balanced output (using two amps intead of one) doubles the voltage swing even though limited to a puny 12V. So the max potential power is eight times that normally available from 12V power supply into 8 Ohm.
I digress: The paralelled gainclone chips merely furnish current capability. The circuit would also work with two rather than six ICs. But the inbuilt current limiting would negate the effort. We now need to make it up by using four. I went a bit further and went for six. Indeed so does Rowland in Model 10 and 12. I haven't seen his circuits so I am not copying something I haven't seen.
JR
Kathode Follower
Konnichiwa Rasmussen San,
The whole cathode follower topic is quite counter intuitve, until you consider the most generic name for the Toplogy.
A triode (or indeed Transistor/Fet) can be operated in three modes:
1) Common Cathode (Emitter/Source) - some people erronously call this "Anode Follower", in principle this is the most common amplification stage in ANY sort of circuit. The signal input is between Cathode (Emitter/Source) and Grid (Base/Gate) with the Cathode (Emitter/Source) being the "reference" (connected to AC ground). Input Impedance is high, as is the Voltage amplification, output impednace is lowish.
2) Common Grid (Base/Gate) - this is rare in audio, common in RF but has recently surfaced in the various "emitter input" active I/V converters discussed over in Digital, including the OPA660 one. Input is between Cathode (Emitter/Source) and Grid (Base/Gate) with the Grid (Base/Gate) being "reference". Input impedance is very low, Voltage amplification is high, current amplification is unity, hence also the sometimes used term "current conveyer", output impedance is high.
As special case, the cascode combines a common Cathode (Emitter/Source) section with a Common Grid (Base/Gate) section for excellent linearity and bandwidth but still a high output impedance.
3) Common Anode (Collector/Drain) - here we have the Follower device, be it a Valve cathode follower, a BJT emitter follower or FET source follower bipolar "single ended push-pull" or just "single ended". Input is (some will have guessed) between Grid (Base/Gate) and Anode (Collector/Drain) with the Anode (Collector/Drain) being "reference".
Yes, our input signal for the follower is referenced to the SUPPLY LINE!!!!
Input impedance is high, Voltage amplification is below unity, current amplification is high, output impedance is low to very low.
So, once we realise that the Follower is a COMMON ANODE topolgy it is clear why it is a bad idea to give a cathode follower a noisy and signal modulated supply.
Sayonara
Konnichiwa Rasmussen San,
Joe Rasmussen said:
The whole cathode follower topic is quite counter intuitve, until you consider the most generic name for the Toplogy.
A triode (or indeed Transistor/Fet) can be operated in three modes:
1) Common Cathode (Emitter/Source) - some people erronously call this "Anode Follower", in principle this is the most common amplification stage in ANY sort of circuit. The signal input is between Cathode (Emitter/Source) and Grid (Base/Gate) with the Cathode (Emitter/Source) being the "reference" (connected to AC ground). Input Impedance is high, as is the Voltage amplification, output impednace is lowish.
2) Common Grid (Base/Gate) - this is rare in audio, common in RF but has recently surfaced in the various "emitter input" active I/V converters discussed over in Digital, including the OPA660 one. Input is between Cathode (Emitter/Source) and Grid (Base/Gate) with the Grid (Base/Gate) being "reference". Input impedance is very low, Voltage amplification is high, current amplification is unity, hence also the sometimes used term "current conveyer", output impedance is high.
As special case, the cascode combines a common Cathode (Emitter/Source) section with a Common Grid (Base/Gate) section for excellent linearity and bandwidth but still a high output impedance.
3) Common Anode (Collector/Drain) - here we have the Follower device, be it a Valve cathode follower, a BJT emitter follower or FET source follower bipolar "single ended push-pull" or just "single ended". Input is (some will have guessed) between Grid (Base/Gate) and Anode (Collector/Drain) with the Anode (Collector/Drain) being "reference".
Yes, our input signal for the follower is referenced to the SUPPLY LINE!!!!
Input impedance is high, Voltage amplification is below unity, current amplification is high, output impedance is low to very low.
So, once we realise that the Follower is a COMMON ANODE topolgy it is clear why it is a bad idea to give a cathode follower a noisy and signal modulated supply.
Sayonara
Re: Kathode Follower
Yep, that's it!
So there are a coupla fixes/improvements:
1. Get the power supply right. Low Noise and clean. You are right about the 'signal modulated supply' where the follower is highish current but driven from a parallel gain source (like a simple triode) where the current is going opposite and then wonder why it sounds a bit flat?
2. Could use CS for better PS isolatation.
3. Reference the Anode to something else, like the signal itself. Now that is a neat trick. I believe you know about this one already.
4. Define the load and make sure you can swing enough volts (and hence current) - why are so many Cathode Folowers X7s ? - and don't get fooled by the supposed Lo-Z spec.
As a friend said to me (Hi Hugh), to use a Cathode Follower you need to know its phone number!
Hey, do that and they sound pretty good!
JR
Kuei Yang Wang said:
Yep, that's it!
So there are a coupla fixes/improvements:
1. Get the power supply right. Low Noise and clean. You are right about the 'signal modulated supply' where the follower is highish current but driven from a parallel gain source (like a simple triode) where the current is going opposite and then wonder why it sounds a bit flat?
2. Could use CS for better PS isolatation.
3. Reference the Anode to something else, like the signal itself. Now that is a neat trick. I believe you know about this one already.
4. Define the load and make sure you can swing enough volts (and hence current) - why are so many Cathode Folowers X7s ? - and don't get fooled by the supposed Lo-Z spec.
As a friend said to me (Hi Hugh), to use a Cathode Follower you need to know its phone number!
Hey, do that and they sound pretty good!
JR
Re: Kathode Follower
Naive question(s) for Kuei and Joe.
Given that a CF is referenced to B+, and B+ is powering the entire circuit, something like Alan Wright's SLCF circuit is there for the purpose of isolating the anode from the B+. Must this be a voltage regulator, or would choke loading accomplish similar results?
I currently use a CCS(ink) on the cathode of my CF, and have been thinking about implementing Alan's SLCF. If tossing a choke in there got me most of the way there, I might lean that way.
...or would VR tubes be another alternative?
Kuei Yang Wang said:
Naive question(s) for Kuei and Joe.
Given that a CF is referenced to B+, and B+ is powering the entire circuit, something like Alan Wright's SLCF circuit is there for the purpose of isolating the anode from the B+. Must this be a voltage regulator, or would choke loading accomplish similar results?
I currently use a CCS(ink) on the cathode of my CF, and have been thinking about implementing Alan's SLCF. If tossing a choke in there got me most of the way there, I might lean that way.
...or would VR tubes be another alternative?
Member
Joined 2002
Member
Joined 2002
Member
Joined 2002
Member
Joined 2002
Konnichiwa,
Okay folks. Now I don't want give Joe troubles and it is not per se my aim to better the JLIT either....
HOWEVER, if you WHERE to ask me what I would do if I wanted to build this using a Follower (not that I want), here it is:
1) Buy a separate Torroid to drive to the Tube frontend. You can get some with around 110V CT even at fairly low power levels.
2) Use the 110V with a Voltage Doubler to get +/-150V (appx.) and use decent Valve style CLC filtering, DON'T SKIMP. Make sure the final pair of PSU Capacitors is of the Film Type, with a Resistor filter the final supply.
3) Cascode the Follower Valve and use another whole Valve as Cascoded Current Source in the "tail" of the Cathode follower. You can see a suitable schematic here:
http://www.vacuumstate.com/
Look at the schematics section and the various cathode follower implementations, I'm sure someone can draw it out....
Try it like this, but use a Valve instead of the lowest FET in the CF, plus dual rail supply instead of single rail....
I'd probably use the 5687 or 6H30PE as Valve in all positions, it is one of the better sounding options for followers. Alternatively the 6SN7 can also work okay, octals look prettier.
Anyway, I would also (of course) hardwire the whole shooting match....
Sayonara
JasonL said:
Okay folks. Now I don't want give Joe troubles and it is not per se my aim to better the JLIT either....
HOWEVER, if you WHERE to ask me what I would do if I wanted to build this using a Follower (not that I want), here it is:
1) Buy a separate Torroid to drive to the Tube frontend. You can get some with around 110V CT even at fairly low power levels.
2) Use the 110V with a Voltage Doubler to get +/-150V (appx.) and use decent Valve style CLC filtering, DON'T SKIMP. Make sure the final pair of PSU Capacitors is of the Film Type, with a Resistor filter the final supply.
3) Cascode the Follower Valve and use another whole Valve as Cascoded Current Source in the "tail" of the Cathode follower. You can see a suitable schematic here:
http://www.vacuumstate.com/
Look at the schematics section and the various cathode follower implementations, I'm sure someone can draw it out....
An externally hosted image should be here but it was not working when we last tested it.
Try it like this, but use a Valve instead of the lowest FET in the CF, plus dual rail supply instead of single rail....
I'd probably use the 5687 or 6H30PE as Valve in all positions, it is one of the better sounding options for followers. Alternatively the 6SN7 can also work okay, octals look prettier.
Anyway, I would also (of course) hardwire the whole shooting match....
Sayonara
Hi,
Joe,
I see where the confusion comes from now...
You're talking bridging stereoamps while I'm talking of using a balanced inputstage for the schematic as you posted.
Which is one channel of a PP amp that I (mistakenly as it turns out) took for a single channel amplifier.
By Jove, it even looks like one...
By doing so it would render it balanced from front to end since the PP stage is by its very nature a balanced symmetrical stage.
In my book a bridged amp isn't necessarilly a balanced amp as it often still driven in SE mode.
Now if we take the venerable Dyna once again and bridge it for mono operation we would essentially have a 140W Class Ab1 amp from what was once a 70W/channel stereoamp.
Nothing's really limiting it but you could never squeeze out 280W out of that amp regardless of the OPT you'd use for it into an 8Ohm load.
Back to the schematic you posted and assuming the same 70W of output for one channel and it's still a monoblock would it magically turn into a 140W amp by making it balanced?
And I mean balanced without adding any further output modules to it which wasn't implied anywhere anyway.
Of course it wouldn't.
Regarding CF: please don't shoot the messenger (in casu me) and nobody implied at any point that the rails would be shared between the CF and the output, the question was if the same voltage could used.
My philosophy about CF implementation is pretty straightforward:
use them only if you absolutely need them...I'll refrain from further comment.
Cheers,
Joe,
I see where the confusion comes from now...
You're talking bridging stereoamps while I'm talking of using a balanced inputstage for the schematic as you posted.
Which is one channel of a PP amp that I (mistakenly as it turns out) took for a single channel amplifier.
By Jove, it even looks like one...
By doing so it would render it balanced from front to end since the PP stage is by its very nature a balanced symmetrical stage.
In my book a bridged amp isn't necessarilly a balanced amp as it often still driven in SE mode.
Now if we take the venerable Dyna once again and bridge it for mono operation we would essentially have a 140W Class Ab1 amp from what was once a 70W/channel stereoamp.
Nothing's really limiting it but you could never squeeze out 280W out of that amp regardless of the OPT you'd use for it into an 8Ohm load.
Back to the schematic you posted and assuming the same 70W of output for one channel and it's still a monoblock would it magically turn into a 140W amp by making it balanced?
And I mean balanced without adding any further output modules to it which wasn't implied anywhere anyway.
Of course it wouldn't.
Regarding CF: please don't shoot the messenger (in casu me) and nobody implied at any point that the rails would be shared between the CF and the output, the question was if the same voltage could used.
My philosophy about CF implementation is pretty straightforward:
use them only if you absolutely need them...I'll refrain from further comment.
Cheers,
I bought a bunch of the 25V xformers RS is selling for $1 to use.
If I dedicate 4 to the tube PS...what combo would likely sound better: 2 xfrmrs/channel making a separate +/- 35V PS for each channel, or grouping 2 xfrmrs in series to make one +/- 70V for both channels (and thereby being to put a much bigger R in the Pi filter)?
Member
Joined 2002
Kuei Yang Wang said:
Now I know you wouldn't give me troubles and I do welcome your input...
As far as building a better JLTi, I could do that too!!! Give me some $$$. But what've presented here is not the full JLTi, a more basic version that actually does an admirable job when you listen to it. Note the [suggested] separate power supply for each CF. A few extra pennies well spent. Now it could be regulated, externally and internally (CS). If they are capable of figuring that out, and I don't mind pointing them down that road, then go for it!
What you describe is a super Super-Linear Cathode Follower!
HEY, THAT MY PREAMP!
Now, you do realise that I have had not a little to do with this design even though it bears Allen Wright's moniker at the lower right corner. Indeed I did a major selling point on this buffer to Allen and he saw it my way. I think the first time we did the boostrapped device on top of the anode was mid-80's.
Yep this is a high performance buffer, and complex at that (that won't go down well with everybody) and can it be improved upon? As a thought process... I think you answered that.
But, tell me honestly, just where do you stand when it comes to Cathode Followers (or any follower?). Do you prefer to avoid it altogether... as some do. Nothwithstanding that it can be made to sound better than its reputation that seems to precede it, and maybe some of it unfairly?
Are they just a necessary evil that we have to debug as much as possible? And it seems that takes added complexity - not running away from that. Then the fact that most SS outputs are followers... ???
A penny for your thoughts.
JR
Hi fdegrove
Ah... what's in a word? Bridging, of course. It's such a long time I used that word and in this case it is more specific.
Yep, I see where you are coming from now. I do believe in balanced interfaces, indeed I prefer the term differential.
That sounds about right.
In SS speak, the power quadruples. Tubes double. It's the nature of the beast(s).
Not usre exactly what you mean. Two active SS amplifiers driven by a diff source (or phase splitter) with the load across the two active outputs. This is same as bridging, as you say. IF the balanced input already exists, then maybe all that is needed is to buffer it before the inverted circuit. Provided you have those two compatible identical signals the principle applies. It's just that SS can near double the current while swinging twice as much.
Probably not far from my sentiments as well. It's just that a good CF is not half bad at all. In fact the best are indeed complex but also quite superb. The one I supplied here is rather basic. It seems oh so simple but is actually a deceptively good design that can be improved upon even more.
JR
fdegrove said:
Ah... what's in a word? Bridging, of course. It's such a long time I used that word and in this case it is more specific.
Yep, I see where you are coming from now. I do believe in balanced interfaces, indeed I prefer the term differential.
That sounds about right.
In SS speak, the power quadruples. Tubes double. It's the nature of the beast(s).
Not usre exactly what you mean. Two active SS amplifiers driven by a diff source (or phase splitter) with the load across the two active outputs. This is same as bridging, as you say. IF the balanced input already exists, then maybe all that is needed is to buffer it before the inverted circuit. Provided you have those two compatible identical signals the principle applies. It's just that SS can near double the current while swinging twice as much.
Regarding CF: please don't shoot the messenger (in casu me) and nobody implied at any point that the rails would be shared between the CF and the output, the question was if the same voltage could used.
My philosophy about CF implementation is pretty straightforward:
use them only if you absolutely need them...I'll refrain from further comment.
Probably not far from my sentiments as well. It's just that a good CF is not half bad at all. In fact the best are indeed complex but also quite superb. The one I supplied here is rather basic. It seems oh so simple but is actually a deceptively good design that can be improved upon even more.
JR
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.