Electron tube pre output stages shootout

[Walter66]

Hello,
planning my new tube pre, I wonder what output stage type with 600 Ohm output is most promising to build?
After studying, I see two general parafeed types that could deliver excellent results in combination with the output trannie (Peerless S-217D):
-anode output type in combination with anode choke (type TAB V76 for example). That was the standard type of parafeed output in most high quality german studio gear.
-cathode follower output type (Marantz 7 in combination with output trans). This is what Shindo san uses today in his upscale models so it's high quality, too.
What design will show the best specs and soundwise has the most potential ?

 

[abbey road d enfer]

Cathode follower does not require wound components of as high quality as parafeed.
For the cost of a plate xfmr plus a choke, you can get a pretty good cathode xfmr.
Making good parafeed chokes is some kind of a lost art.
Cathode follower has inherently lesser THD than common-cath, and because it drives the xfmr with quite a low impedance, the LF THD of the xfmr is reduced.
The only drawback is that it has no gain.
Because a cath-follower has low drive impedance, many think it's a good reason for using a lesser step-down ratio, e.g. 4:1 instead of 10:1, but it increases the tube distortion by twisting the load line.
IMO, the best option is a low-gain triode followed by a cath-follower. One dual triode or a triode-pentode for the output stage.
Now, do you plan on using global NFB?

 

[Walter66]

Yes, both would have NFB, the anode follower would have the same topology as TAB V76, but there will be a low impedance, low gain triode employd in both.
OPT is 12K: 600 Ohm, turns ratio 22,2:1, I have a Lundahl LL1667 270H/ 15mA anode choke. Is this a good one with regard of the lost art of parafeed choke winding?
I saw that Shindo labo uses unusual low impedances in the cathode follower. From my experience, this leads to interesting effects on the sound.
But is it the truth? On the other hand, an anode choke fed tube uses extreme high impedances as load for the tube. Both variations seems opposite, but both can sound very good. In one case, the tube works under heavy load conditions, the other case has nearly a lightweight load for the tube.

The other thing is, the complexity of the PSU of both varieties differ much. With the CF version, one needs heavy filtering in the PSU where with the anode choke version, 270H right in front of the anode makes the PSU design an easy task. Look at the V76 PSU, its as simple as could be and no noise at all.

 

[abbey road d enfer]

OPT is 12K: 600 Ohm, turns ratio 22,2:1

? 12k:600 makes a turn ratio of 4.5:1. One of your figures is wrong.

, I have a Lundahl LL1667 270H/ 15mA anode choke. Is this a good one with regard of the lost art of parafeed choke winding?

I don't know. The proof is in the pudding. Make the output stage, test it, and then you'll know if the choke is good.

Both variations seems opposite, but both can sound very good. In one case, the tube works under heavy load conditions, the other case has nearly a lightweight load for the tube.

A given load, applied to the anode or the cathode, is the same for the tube. For a system under global NFB, for a given global gain, the non-linearities end up being similar.

The other thing is, the complexity of the PSU of both varieties differ much.

Does it?

With the CF version, one needs heavy filtering in the PSU where with the anode choke version, 270H right in front of the anode makes the PSU design an easy task.

You fail to take into account the fact that a common-cathode stage has about zero PSRR, when a cath-follower has very high PSRR.

Look at the V76 PSU, its as simple as could be and no noise at all.

The immunity of V76 to power supply ripple comes from the global NFB. You say it's "simple", but it's not. The choke in parafeed is what makes it not so simple.

 

[ruffrecords]

One problem with many classic tube pre designs is that a lot of gain is lost in the output transformer. 4:1 is typical and often higher ratios are used so at least 12dB of gain is lost. It may not seem a lot but it reduces the amount of gain that can be used to reduce distortion. The reason such a high ratio output transformer is necessary is that the output stages are mostly single ended types which means their peak output current is equal to their quiescent current. For +20dBu into a 600 ohm load you need close to 100mW of power and a peak current of over 18mA. With a 4:1 transformer you need at least 18/4 = 4.5mA quiescent current in a single ended output stage and the stage will be dissipating 250 x 4.5 mW = 1.125 watts. And that is an absolute minimum.

On the other hand, if you use a push pull output stage, the peak output current is twice the quiescent current which means you get 4 times the power for any given quiescent current when compared to a single ended stage running at the same current. But push pull circuits are complicated requiring a phase splitter and a transformer capable of handling the quiescent current imbalance. A possible answer is to use an SRPP output stage which is inherently push pull. An SRPP stage running at 5mA from 250V will easily pump +20dBu into a 600 ohm load with only a 2:1 transformer in between them thus saving 6dB of gain and eliminating the transformer's dc requirement and the phase splitter. Sounds wonderful but the downside is the SRPP stage produces a lot more distortion than most single ended stages so it needs some more NFB. This is the basis of my twin line amp design (schemo attached).

Cheers

Ian

 

[Walter66]

Thank you both for your input. The transformer impedances are correct, it is a 4,5:1 type.

While designing the schemos, for a while I thought about the possibility to go PP, but too many parts count for my boards and the space is limited.
But the idea of more efficiency and greater energy (power) output is attractive. Its just that after some years of search I want to go the SET route instead of PP, SRPP or other varieties.

Understood, that for a given loadline, its not important of having the same load on the anode or cathode, but I think the difference is clearly audible.

In one case, the circuit has 100% negative feedback, in the other it has no feedback at all. And for the examples given, the difference is even greater. Because I wrote about an extreme high anode load or an extreme low cathode load. How would the tube behave in terms of sound quality?
Even when external negative feedback is applied to the anode choke circuit, it still is a gain stage, so feedback will never be 100%.
The proof should be in the practice, think I have to breadboard the circuit and listen to the differences.

 

[abbey road d enfer]

the downside is the SRPP stage produces a lot more distortion than most single ended stages

But that's because the load is much heavier with a 4:1 than a 10-12:1. For a given load, there should be no significant difference between the different topologies, as long as gain is regulated by NFB and identical.
The real perversion comes from the fact that the lower impedance provided by cath-follower or SRPP leads people to use heavier loads. One should remember that the output impedance has almost nothing to see with the drive capability, which depends mainly on idle current, excluding push-pull of course.

 

[abbey road d enfer]

Understood, that for a given loadline, its not important of having the same load on the anode or cathode,

I rather think any performance comparison should be done with identical loads.

In one case, the circuit has 100% negative feedback, in the other it has no feedback at all.

Does cath-follower have no NFB at all?

And for the examples given, the difference is even greater. Because I wrote about an extreme high anode load or an extreme low cathode load.

In one case the non-linearities come mainly from the anode voltage variations, and in the other from the 3/2 exponent in the transconductance (Langmuir-Child equation), but in practice, in a holistic approach, for a given gain and output capability of a complete system, the difference is not so clear-cut. That probably explains the variety of topologies used by designers, with comparable results.
I think the choice is individual, choose your own dose of poison.
And of course, economics put a heavy weight on the decision. A xfmr for a cath-follower is easier to make than for an anode load, even if they have the same nominal impedance.

Even when external negative feedback is applied to the anode choke circuit, it still is a gain stage, so feedback will never be 100%.

One could argue that a common cathode stage with a gain of 30 and a 30:1 xfmr has the same gain as a cath-follower with a 1:1, and a 100% difference in NFB. Of course these are extreme examples.

 

[abbey road d enfer]

Understood, that for a given loadline, its not important of having the same load on the anode or cathode, but I think the difference is clearly audible.

Incidentally, it's not fair to compare a parafeed common-cath with a resistor-loaded cath-follower.
The drive capability of a good parafeed is twice that of a resistor-loaded one. In all fairnes, it should be compared with a choke-loaded cath-follower.

 

[k brown]

What are your thoughts on the Pultec MB-1 topology?

 

[abbey road d enfer]

What are your thoughts on the Pultec MB-1 topology?

66dB Open Loop Gain, 10mA idle for the output stage, 15k output xfmr. It works.

 

[Walter66]

Please comment on this schemo for a CF linestage (CV391=807). Thank you.

 

[abbey road d enfer]

Please comment on this schemo for a CF linestage (CV391=807). Thank you.

Brute force attack!

 

[ruffrecords]

But that's because the load is much heavier with a 4:1 than a 10-12:1. For a given load, there should be no significant difference between the different topologies, as long as gain is regulated by NFB and identical.

In practice the differences are significant. Even unloaded, the distortion from a single ended and an SRPP running at the same idle current are significantly different mainly due to the load the rop tube places on the lower tube in the SRPP configuration.

The real perversion comes from the fact that the lower impedance provided by cath-follower or SRPP leads people to use heavier loads. One should remember that the output impedance has almost nothing to see with the drive capability, which depends mainly on idle current, excluding push-pull of course.

Agreed. As I have said many times before, output impedance is no indicator of drive capability. SRPP of course is push pull.

Cheers

Ian

 

[abbey road d enfer]

In practice the differences are significant. Even unloaded, the distortion from a single ended and an SRPP running at the same idle current are significantly different mainly due to the load the rop tube places on the lower tube in the SRPP configuration.

SRPP can have a THD performance similar to an SE stage, maybe even better, but this requests tuning teh SRPP for the specific load impedance and vor a certain gain, where the transconductance non-linearity is partially compensated by the anode characteristics.
In practice, I don't think there is much subjective difference between a well-designed SE and a decent SRPP. Of course, figures will differ.

SRPP of course is push pull.

Indeed. However it doesn't have the same drive capability as a class B push-pull.

 

[analag]

Here's something I designed for an EQP1a clone a decade plus ago when I was into plate followers.

 

[Walter66]

Thanks for sharing. SRPP?

 

[Egor Reshetnikov]

idk, im goin for 6n6p srpp unbalanced output without transformer (may be impedance balanced?) If it is supposed to drive headphones then 10k line input would be easy job for it.

but wonder what is the differens in distortion content with and without cathode bypass cap in srpp?

 

[abbey road d enfer]

idk, im goin for 6n6p srpp unbalanced output without transformer (may be impedance balanced?) If it is supposed to drive headphones then 10k line input would be easy job for it.

From The Tube CAD Journal:SRPP vs. White Cathode Follower

"It is often misused as a line stage amplifier (where the load impedance is not predefined)"

 

[ruffrecords]

SRPP can have a THD performance similar to an SE stage, maybe even better, but this requests tuning teh SRPP for the specific load impedance and vor a certain gain, where the transconductance non-linearity is partially compensated by the anode characteristics.

I suspect that is what was done in the original application which was for transmitters. There is definitely a lot of discussion in audio circles about optimization of an SRPP for minimum distortion into a specific load but that does no interest me. The distortion is certainly worse away from the optimum but not so bad that a little NFB can't fix it.

In practice, I don't think there is much subjective difference between a well-designed SE and a decent SRPP. Of course, figures will differ.

Subjectively, the SRPP can sound 'nice'

Indeed. However it doesn't have the same drive capability as a class B push-pull.

Because it is class A all the way. I never quite understood the fixation with class B output stages. A class A push pull stage can theoretically beup to 50% efficient and a class B can be up to 63%.

Chweers

Ian