Cathode follower in a preamplifier

[maxhifi]

Want to build a Super simple line level preamp. Am thinking one stage of 6SN7 resistance capacitance coupled amplifier per channel with a simple power supply using a single 5Y3.

Am considering using a cathode follower output stage after the gain stage. I am curious what consensus here is regarding the sonic merit of cathode followers. It would lower the output impedance, which would ensure good high frequency performance regardless of what interconnects are used, but I've heard a lot of negative talk about them on the web regarding their impact on sound quality. Would like to hear some shared experience.

 

[Palustris]

You will have a coupling cap after the first stage going to the cathode follower. Install two RCAs; one for direct output and one for the cathode follower. Try both and see which one you prefer.

 

[santa]

Google "octal ccda preamplifier"
Thats a good sounding circuit IMHO

 

[natger]

I find this a amusing read, there are other amps with cathodefollowers too, DC coupled too if you want to avoid a extra capacitor in signalpath.
http://syclotron.com/?page_id=2380

 

[thorpej]

The sonic benefit is a low impedance output -- you'll be able to drive a much wider variety of downstream equipment, and it will maintain frequency response in longish cable runs. A well-designed-and-built cathode follower circuit isn't going to have any negative impact whatsoever. There is a small reduction in overall gain (it's not quite unity, but generally 0.9 to 0.95), and in general they are extremely linear. Use a DC-coupled design to eliminate the extra capacitor in the signal path. If done correctly, for all intents and purposes, it's transparent.

 

[kward]

You'll have a polarity reversed output, which can be compensated for at the power amplifier speaker outputs. Other than that, if designed correctly, it can be a design that performs well and sounds great. Several commercial line level preamps use the circuit you describe (meaning a common cathode gain stage either direct coupled or AC coupled to a cathode follower). Two I've owned that use this approach are the Conrad Johnson PV12, and the Audio Electronic Supply AE3. The CJ used a 12AU7 per channel and the AES used a 6SN7 per channel. The CJ is a bit more "high brow" you might say with a full solid state regulated power supply for both high voltage and filaments, but I actually prefer the sound of the AES because to me it sounds a bit more neutral.

Numerous years ago, I built a preamp borrowing the best design elements from these two preamps. Mine was a 6SN7 gain stage direct coupled to a 6H30 cathode follower. I used the 6H30 specifically because of its high transconductance and new production availability, so that I could drive headphones with it as well as use it as a preamp to drive my power amps. I used it with my Sennheiser HD600's (with 300 ohms impedance), and it works very well. (Wouldn't work so well with lower impedance phones.) I made mine with three selectable inputs, a mute switch, phones selector switch (with LED indicators when the respective circuit is activated), and an iPod front panel input jack. It's dead quiet, even with headphones. I went to some length to get it quiet, such as placing the power supply as far away from the audio circuit as I could get, and using regulated DC on the heaters. It's built on a 17" x 14" x 3" chassis.

Here's a pic:

 

[rothwellaudio]

A well-designed-and-built cathode follower circuit isn't going to have any negative impact whatsoever.

... if designed correctly, it can be a design that performs well and sounds great.

I happened to be reading a rather in-depth analysis of cathode followers only last night, written in 1955 - yes, I'm way behind with my reading
Anyway, the usual assumption that cathode followers are simple circuits that have 100% negative feedback and therefore extremely low distortion isn't entirely true. Cathode followers have to be designed well or there will be considerable distortion, which probably accounts for the bad reputation they have in some quarters.
Basically - as with most circuits - well designed ones sound good, poorly designed ones sound bad.

 

[GordonW]

By "designed well"- the biggest thing is to use a tube with enough transconductance to actually be able to drive a low-impedance load- if that's what's planned (i.e, to drive the input of solid-state amps and such).

Also, the same rules about operating points (where the tube is lowest distortion) are roughly the same as with a standard grounded-cathode circuit, in terms of current through the tube and B+ points. You do get the benefit of effective massive feedback (the distortion through the CF will be lower than the normal distortion through the tube in a circuit where it has gain)- but nothing is absolute, in terms of cancelling distortion, with a CF- so starting with the proper operating point is important...

Follow those, and you will get a net benefit from the CF. I've used them many times, to good results.

Regards,
Gordon.

 

[thorpej]

Note that the standard plate curves for a tube aren't going to tell you much about the most linear region of a cathode follower. The curves for a tube in CF mode look entirely different. Merlin has a chapter in one of his books that describes how to calculate and draw the curves -- I'll try to remember to provide a citation when I get home from work tonight.

 

[BinaryMike]

That would be chapter 7 in Designing High-Fidelity Tube Preamps. Merlin makes it abundantly clear that cathode follower performance cannot be reduced to a simple good/bad call. The discussion on distortion cancellation and the load-dependent distortion plot in figure 7-21 are worth the price of admission all by themselves.

 

[maxhifi]

You'll have a polarity reversed output, which can be compensated for at the power amplifier speaker outputs. Other than that, if designed correctly, it can be a design that performs well and sounds great. Several commercial line level preamps use the circuit you describe (meaning a common cathode gain stage either direct coupled or AC coupled to a cathode follower). Two I've owned that use this approach are the Conrad Johnson PV12, and the Audio Electronic Supply AE3. The CJ used a 12AU7 per channel and the AES used a 6SN7 per channel. The CJ is a bit more "high brow" you might say with a full solid state regulated power supply for both high voltage and filaments, but I actually prefer the sound of the AES because to me it sounds a bit more neutral.

Numerous years ago, I built a preamp borrowing the best design elements from these two preamps. Mine was a 6SN7 gain stage direct coupled to a 6H30 cathode follower. I used the 6H30 specifically because of its high transconductance and new production availability, so that I could drive headphones with it as well as use it as a preamp to drive my power amps. I used it with my Sennheiser HD600's (with 300 ohms impedance), and it works very well. (Wouldn't work so well with lower impedance phones.) I made mine with three selectable inputs, a mute switch, phones selector switch (with LED indicators when the respective circuit is activated), and an iPod front panel input jack. It's dead quiet, even with headphones. I went to some length to get it quiet, such as placing the power supply as far away from the audio circuit as I could get, and using regulated DC on the heaters. It's built on a 17" x 14" x 3" chassis.

Here's a pic:

View attachment 817098

This is very much similar to what I had in mind. Thanks for posting the photo and experience.

 

[maxhifi]

That would be chapter 7 in Designing High-Fidelity Tube Preamps. Merlin makes it abundantly clear that cathode follower performance cannot be reduced to a simple good/bad call. The discussion on distortion cancellation and the load-dependent distortion plot in figure 7-21 are worth the price of admission all by themselves.

Note that the standard plate curves for a tube aren't going to tell you much about the most linear region of a cathode follower. The curves for a tube in CF mode look entirely different. Merlin has a chapter in one of his books that describes how to calculate and draw the curves -- I'll try to remember to provide a citation when I get home from work tonight.

I've noticed that most of the "classic" preamps do indeed use a CF output, and only the more eccentric designs avoid it. Jury still out though, I figured the HF pole with my current interconnect and the output impedance of, say, half an RC coupled 5687 or 6SN7 will be well above my output transformers capability.

 

[maxhifi]

I happened to be reading a rather in-depth analysis of cathode followers only last night, written in 1955 - yes, I'm way behind with my reading
Anyway, the usual assumption that cathode followers are simple circuits that have 100% negative feedback and therefore extremely low distortion isn't entirely true. Cathode followers have to be designed well or there will be considerable distortion, which probably accounts for the bad reputation they have in some quarters.
Basically - as with most circuits - well designed ones sound good, poorly designed ones sound bad.

Have you got a reference for this article? Would be interested to read it myself. I'm not above copying someone else's design, or at least referencing it as a starting point but I do like to be a bit original when it's possible.

 

[rothwellaudio]

Have you got a reference for this article? Would be interested to read it myself. I'm not above copying someone else's design, or at least referencing it as a starting point but I do like to be a bit original when it's possible.

It was an article published in Wireless World and written by "Cathode Ray", which was the psuedonym used by M.G.Scroggie. My copy is in a hardback bound collection and not easy to scan but I'll look for it online to see if it exists. It isn't a design which can be copied anyway, it's an analysis of the curves and a study of the different ways to achieve correct bias.

 

[maxhifi]

Google "octal ccda preamplifier"
Thats a good sounding circuit IMHO

It looks like a really smartly designed circuit.

 

[GordonW]

Was just re-examining the math- the real trick to making a cathode follower work correctly, is having a LOW ENOUGH PLATE IMPEDANCE tube, so that you don't run out of "drive". A high impedance tube (i.e, the worst case condition of driving a 600 ohm line from a 12AX7), means that a lot of the available plate voltage is going to be used up inside the tube- leaving very little left to drive the load. Remember, a cathode follower has the load IN SERIES, in effect, with the plate impedance of the tube. The available drive voltage maximum is related to the ratio of the load impedance to the plate impedance of the tube:

Vout-max = B+ * Rload/(Rplate + Rload) (note that this calculates PEAK TO PEAK voltage- you have to divide by 2.88 to get RMS voltage).

For the 12AX7 above- assuming you have 300v B+, and a 600 ohm load: Vout-max= 300 * 600/(60000+600) = 2.97v. That equates to just over 1V RMS.

However- take a 12AU7 under the same load: Vout-max = 300 * 600/(12000 + 600) = 14.28v, which equates to about 4.96v RMS.

This is why using cathode followers for say, a headphone amp- is a more daunting task than it first appears... and why tubes such as the 6AS7 (only 280 ohms plate impedance!) are used:

Let's assume 100 ohm headphones-

Vout-max = = 150v * 100/(280+100) = 39v, or 13.7v RMS. That's 1.88 watts into a 100 ohm load.

For a more typical max loading for a preamplifier- say, a bi-amped system with two solid-state amps each with 10K input impedance, in parallel (giving 5K load)- for the 12AX7:

Vout-max = 300 * 5000/(60000+5000) = 23.08v, or 8.01v RMS. Not much headroom, when you consider some amps need up to 5v!

For the 12AU7:

Vout-max = 300 * 5000/(12000+5000) = 88.2v, or 30.64v RMS. You can plainly see why that would be better- enough headroom, that you don't get near clipping...

(BTW: Do be aware- this math is an approximation. To do it right, you'd also have to include the loading of the cathode resistor in the cathode follower- that would change the magnitudes of the numbers a bit- but it would NOT change the relationships and rankings above...)

Now, low plate impedance does frequently go along with high transconductance- so what we were discussing earlier, is somewhat analogous. But, you really DO need to look directly at Rp, before deciding if you cathode follower is stout enough...

Regards,
Gordon.

 

[maxhifi]

This is all very reassuring. I am driving a tube amplifier which has a 470k input impedance, in parallel with about 400 pf combined miller and interconnect capacitance. The main reason for wanting a CF, is to make sure that the high frequency roll off and phase shift associated with the low pass filter formed by the output impedance of my preamp and capacitance of the interconnect plus miller capacitance at the input of my amplifier, are well away from the audio band.

I've had an idea to change the direction a little bit... maybe I just want to do something slightly different. How about this idea... using one 6GH8 per channel, with the pentode direct coupled to the triode, triode configured as a cathode follower and pentode configured for maximum gain. Then, to take negative feedback from the cathode of the triode section back to the grid of the pentode section. This could give me, say 20dB of negative feedback, which would vastly reduce distortion and output impedance. I think the triode section of a 6GH8A is well suited for using as a cathode follower, and the pentode would be capable of lots of gain. Also, there shouldn't be much in the way of a stability concern since it's possible to keep capacitors outside of the feedback loop entirely. Hmmm.

 

[analog addict]

I just finished building a 27 based preamp without a cathode follower, similar to Dennis Had's LP-27 series of preamps. He was very complimentary of the sound, and also measured the frequency roll off using 12' cables measured into a 10K ohm load. The -3db roll off freq. was somewhere around 50KHz, way above the audible range. Does it sound better than his preamp with the cathode follower? I can't really tell, but I have always been a fan of simpler is better with respect to the signal path.

 

[BinaryMike]

How about this idea... using one 6GH8 per channel, with the pentode direct coupled to the triode, triode configured as a cathode follower and pentode configured for maximum gain. Then, to take negative feedback from the cathode of the triode section back to the grid of the pentode section.

I would call that an 'augmented anode follower,' and it's likely to work very well indeed. You should be able to achieve total feedback of 43~46dB, which will yield vanishingly low distortion at typical preamp output levels. Do be sure to include a little series resistance in the output line, to isolate load capacitance.

 

[maxhifi]

I would call that an 'augmented anode follower,' and it's likely to work very well indeed. You should be able to achieve total feedback of 43~46dB, which will yield vanishingly low distortion at typical preamp output levels. Do be sure to include a little series resistance in the output line, to isolate load capacitance.

thanks for the vote of confidence, I will proceed to design in this direction and see what I come up with. I want to try and get the overall gain to about 15, and to eliminate as many capacitors as possible.