• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Cathode Follower? Yay or nay?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I am building the circuit below (Angela easy 2A3), and it sounds quite good, certainly better than SRPP or simply 1/2 a 6SL7. Sounds very neutral and not too euphonic. Nice to listen to however. Bass is tight and power is up.

However there seems to be a prejudice against cathode followers, people really don't like them. Anyone tell me why?

In the below circuit, I would also like to try a 6SN7 as the driver tube (i.e. half 6SN7 per channel), how would I do this? Is it as easy as changing the 100K cathode R to a 47K?
 

Attachments

  • cf-2a3.gif
    cf-2a3.gif
    5.1 KB · Views: 1,493
Sure, or even 22k. I'd remove the cathode bias from the 2A3, run the CF ground return to a negative (-100V or better) supply, move the coupling cap to before the CF and apply fixed bias to its grid. That way you can run a smidge of grid current, making for more graceful clipping when at full power, important with only 5W to spend.

As for people hating CF's, well, people are stupid. :devilr:

Tim
 
ShiFtY said:
However there seems to be a prejudice against cathode followers, people really don't like them. Anyone tell me why?

Aside from personal tastes and preferences, the argument is that cathode followers (or emitter followers or source followers for that matter) employ 100% negative feedback and of course as everyone knows, negative feedback is bad. :)

It's a bit ironic really. Those who don't care for the CF because of negative feedback tend to prefer triodes over pentodes, even though pentodes have less inherent negative feedback than triodes.

Go figure.

se
 
diyAudio Senior Member
Joined 2002
Hi,

Aside from personal tastes and preferences, the argument is that cathode followers (or emitter followers or source followers for that matter) employ 100% negative feedback and of course as everyone knows, negative feedback is bad.

Is it?

The more advanced argument should be:

Some CFs are bad sounding, the one in the schem in post #1 being a typical example, not because of the local NFB but because it's just a lousy circuit.
See Tubecad for a more indepth explanation.

Overuse of global NFB loops is also a no-no in good audio design, local FB loops as used in well designed CFs and triodes is generally preferred.

Penthodes: well, isn't that a thermionic device with a local feedback inhibiting grid?
And what is used to linearise that entire circuit? You guessed it, overall NFB, the more the merrier.

That's just scratching the surface though, suffice it to say that believing that any amount of feedback is a cure all is a sad case of selfdelusion...

First of all try to understand how a circuit works, than see what FB can do and can't do to help it along if it can at all...

Cheers,;)
 
fdegrove said:

Depends on the individual it would seem. And my comment was intended sarcastically. However some do in fact wholly reject negative feedback in all its forms. Or at least those forms that they're aware of. :)

The more advanced argument should be:

Some CFs are bad sounding, the one in the schem in post #1 being a typical example, not because of the local NFB but because it's just a lousy circuit.
See Tubecad for a more indepth explanation.

Overuse of global NFB loops is also a no-no in good audio design, local FB loops as used in well designed CFs and triodes is generally preferred.

Penthodes: well, isn't that a thermionic device with a local feedback inhibiting grid?
And what is used to linearise that entire circuit? You guessed it, overall NFB, the more the merrier.

That's just scratching the surface though, suffice it to say that believing that any amount of feedback is a cure all is a sad case of selfdelusion...

First of all try to understand how a circuit works, than see what FB can do and can't do to help it along if it can at all...

I couldn't agree more. Very well put, Frank.

se
 
diyAudio Senior Member
Joined 2002
Hi,

And my comment was intended sarcastically.

Sure, I know you well enough for that...:D

Fortunately there other ways than FB to reduce distortion if that's the holy grail.

Unfortunately, to achieve low Zo from a tube circuit the means are rather limited; other than a good old buffer I see...Ah, xformers.
The good news is that at least those are getting better as years go by....
Not something I can wholeheartedly say about tubes even though the situation isn't all doom either.

Just another thought, some of the best tube amps I've ever heard, with the exception of OTLs, often used CF windings on the OPT and very low amounts of global FB.
This could well be a coincidence although somehow I doubt it.

Cheers,;)

P.S. Steve, I finally got a hold of Dr. WDC. I'll try to summarise what he told me in an e-mail to you asap.
 
fdegrove said:
Sure, I know you well enough for that...:D

Jes' makin' sure. :)

Fortunately there other ways than FB to reduce distortion if that's the holy grail.

Unfortunately, to achieve low Zo from a tube circuit the means are rather limited; other than a good old buffer I see...Ah, xformers.

True. And with transformers, rather like feedback, your lower Zo comes at the expense of signal gain. So you need a certain amount of excess gain in either case.

The good news is that at least those are getting better as years go by....

Yes. And one nice thing about them, they tend to have a much longer in-use lifespan than tubes or solid state so we'll have them around for quite some time yet.

Not something I can wholeheartedly say about tubes even though the situation isn't all doom either.

Tubes are your department so I'll just have to take your word on that. :)

Just another thought, some of the best tube amps I've ever heard, with the exception of OTLs, often used CF windings on the OPT and very low amounts of global FB.
This could well be a coincidence although somehow I doubt it.

No idea.

Though it would be interesting to discover how much of the anti-feedback sentiment out there is driven purely by prejudice which has a nasty habit of coloring our subjective experiences.

P.S. Steve, I finally got a hold of Dr. WDC. I'll try to summarise what he told me in an e-mail to you asap.

Thanks, Frank. I'll keep an eye out for it.

se
 
diyAudio Senior Member
Joined 2002
Hi,

Care to share any specific examples, especially schematics?

Sure, but I'm bound to forget some:

McIntosh (maybe not everyones' cupper but alot of tweek potential), same goes for the Quad II, more recently: Jadis, Audiomat....

None of the diagrams however reveal the exact amount / percentage of FB to the cathodes,~ 20% is my guess.

As for schematics, the Quad II is on the net, as is most of Frank McIntosh's work.
Audiomats' older amps I know by heart but I shouldn't divulge the schematics, same goes for the Jadis.Sorry.

Let's not forget either that the ferryman needs to be paid; what you gain in linearity has to be paid in drive voltage and gain in the preceding stage(s).

So, unless you have a competent winder as your best friend, you're still stuck...

Come to think of it, I can't think of any DIYer having such an amp on the net...

On the bright side, I vaguely recall at least one source for replacement Mac iron...
So, if you don't mind using the 6L6 family, KT88 and relatives, you should at least have one source of suitable iron for a project.

Cheers, ;)
 
Frank said:
As for schematics, the Quad II is on the net, as is most of Frank McIntosh's work.
Could you name a few specific McIntosh models that you think would be good subjects? I have found a couple of schematics, and they looked insanely complex to me.

So, unless you have a competent winder as your best friend, you're still stuck...
Come to think of it, I can't think of any DIYer having such an amp on the net...
On the bright side, I vaguely recall at least one source for replacement Mac iron...
So, if you don't mind using the 6L6 family, KT88 and relatives, you should at least have one source of suitable iron for a project.

Would you mind eyeballing these transformers, and telling me which, if any, of your favorite designs they might suit? The last one even talks about the McIntosh design... :hbeat:

Best,

George Ferguson
 
diyAudio Senior Member
Joined 2002
Hi,

Care to share any of your thoughts on tweaking MACs?

Sure.
Basically what we did, provided it wasn't a collectors' item, is to upgrade the PS, passive components, selected NOS tubes, wiring etc.

The schematic itself I wouldn't touch.
It's all very much interdependent and utterly clever.

Could you name a few specific McIntosh models that you think would be good subjects?

MC240, MC275.

I have found a couple of schematics, and they looked insanely complex to me.

They are impressive at first glance indeed.

Would you mind eyeballing these transformers, and telling me which, if any, of your favorite designs they might suit?

Will do.

Cheers,;)
 
diyAudio Senior Member
Joined 2002
Will do.

Hi,

Done....

Pretty impressive specs as I'd expect from someone as brilliant as Menno VdV....

This is one guy that actually built at least one OTL amp BEFORE he started speccing his OPTs. Recpect that...

If only the price were on a per pair basis, right?...Sigh....

What can I say, if this is your thing and your pockets are deep enough, it's hard to imagine any better quality than this.

Put it another way, if I wouldn't have an OTL already and I could chose, this would by my choice.

Cheers, ;)
 
Hey-Hey!!!,
No wonder they have a HF roll off. a 12AX7 driving a pair of UL rigged power valves with each plate?

The Jadis OPT is an expansion of the Acro TO-350 and Dynaco A440/441. The originals have a 16% tertiary winding and Jadis adds another( unknown %-age) so both screen and cathodes have a winding to ride.

I am considering an interim progression with my current amps and use the A441's for pentode rigged 813's with the filaments riding the tertiary winding. B+ to be ~600. All I'll need are a pair of good, low capacitance( to keep the capacitive load on the cathode down...) filament tx's.
regards,
Douglas
 
diyAudio Senior Member
Joined 2002
Hi,

No wonder they have a HF roll off. a 12AX7 driving a pair of UL rigged power valves with each plate?

Yeah...I know, I know.
Discussed that with the designers ages ago and they insist on stick with their recipe claiming it's part and parcel of the "magic" of the amp.

I can live with it, for awhile, as these amps make you forget you even HAVE loudspeakers in the room.

Between you and me, I think it must be possible to have both treats...
That would be a real winner and that's what put Audiomat on the map anyway.

Cheers, ;)
 
Not sure if this follows from the thread, but let's discuss CFs.

Since grids should be shaken not stirred it does make sense to use a CF since Zout is very low.

However, a CF is not the best option if the load is hard to drive, as load current seems to severely affect its linearity, particularly as grid current can be significant (up to 25% of plate current) and the CF stage current is usually around 5-15mA. And while setting the B- point of the CF is a convenient way of implementing bias control, slow turn on, and direct coupling the output tube grid, the unity gain is an issue. To stay with just two voltage gain tubes, you need all the gain you can get with both SE and PP. To further complicate this choice, I've found the best operating point for a CF is difficult to find compared to a PLT.

On this basis, I normally go for a grunty triode, such as a 6BX7, running around 30mA in plate loaded configuration. With its 1.3K rp this gives more than adequate drive, and because plate current is high, any grid current is usually not such a problem. The one disadvantage is cap coupling, but this isolates fault conditions and in any case I normally use cathode biasing.

I honestly believe the reason most amps use this topology, in fact, is due to the phenomenon of convergence which SY referred to some time back. Of all the choices available, this one delivers the optimal cost/benefit

Cheers,

Hugh
 
AKSA said:
However, a CF is not the best option if the load is hard to drive


Huh?

as load current seems to severely affect its linearity, particularly as grid current can be significant (up to 25% of plate current) and the CF stage current is usually around 5-15mA.

And you suggest driving grids with what??? Transformer coupling where there's NO corrective NFB applied? Maybe not...

And while setting the B- point of the CF is a convenient way of implementing bias control, slow turn on, and direct coupling the output tube grid, the unity gain is an issue.

Howso? You need the 100 gain anyway. And now it can be provided by a single 6AV6 rather than a cascaded 6SN7. (If they had more gain (in both sections), a dissimilar dual triode would be perfect.)

To stay with just two voltage gain tubes

What's this matter? I'll assume you are talking in ZNFB terms so there's no loop to cause phase shift oscillations in.

you need all the gain you can get with both SE and PP. To further complicate this choice, I've found the best operating point for a CF is difficult to find compared to a PLT.

Huh? Oh, more of that... Dude, the best operating point is the one which suits the circuit electrically. You can't hear a CF anyway, remember? :rolleyes:

On this basis, I normally go for a grunty triode, such as a 6BX7, running around 30mA in plate loaded configuration. With its 1.3K rp this gives more than adequate drive, and because plate current is high, any grid current is usually not such a problem.

Hmm, above you said grid current can easily be up in the 100mA range ... so um, 30mA plate... but... uh. . . hey, I think 70 miliamperes just disappeared. That must be it.

The one disadvantage is cap coupling, but this isolates fault conditions and in any case I normally use cathode biasing.

If you mean while direct-coupled to an output tube, this is almost as bad as direct-coupling to the driver tube. :whazzat: With fixed bias, the driver's drift has NO effect, the CF's will shift 1V absolute maximum through its own drift and output tube current can be corrected as it ages.
With cathode bias on the output, you waste 50 useful volts of course, and the CF needs to be fixed at zero volts. But that's impossible, so either you use a negative supply for it (and if you're going to do that you might as well fixed-bias it) or you have to cap-couple it as in the original schematic above. But that defeats the purpose of using a CF in the very first place because you can't drive the output any better than a plate follower 6SN7.

Tim
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.