Cathode followers..

[raysolinski]

Forgive my newb question...After studying the various tube circuits and enough theory to only be dangerous (a little knowledge...). Why don't more circuits employ a cathode follower? By lowering the impedance this way aren't you preserving more of the circuits electronic gain vs. using a stepdown transformer on the output? I understand that using another toob creates a need for more plate and heater current but what are the drawbacks? Remember, kid gloves here as I am only trying to get a handle on this

Thanks,
Ray

 

[emrr]

search author PRR and cathode follower.  should provide hours of reading. 

 

[raysolinski]

Umm..yeah... thanks for that..I know how to search (I was a reference librarian a few years back and before that ..a teacher) which is something I already did both here and  with just a general google search..your suggestion brought up a boatload of project specific responses and most not even close to my simple questions (not really wild about the search function on this board)..As I stated, after some research I came up with my aforementioned  very simple questions. Why NOT use a cathode follower? Are the drawbacks greater than using a stepdown transformer in YOUR opinion? I have read some pros and cons online but I trust the opinions and designs of you folks more than anonymous web people...

Cheers,
Ray

 

[Wagtbtoobz]

hmmm, what's your need ?

 

[raysolinski]

Nothing specific...What interests me most is sound being a studio owner. I see endless debates on harmonic distortions and voltage swings and RF oscillations etc...what I don't see is what the effect on the audio is..some well know tube mic amps have had cathode followers while other deride the very notion...just curious..

Ray

 

[pstamler]

From the tenor of your questions, I suspect you're interested in the use of a cathode follower vs. an output transformer to drive a low-impedance load, like the 600 ohm input impedance of a classic compressor. The short answer is that a cathode follower, by itself, won't do that.

Here's the story. 600 ohms is low enough that you can't drive it from the plate of a tube; it loads down the tube way too much, the gain drops drastically, and the distortion skyrockets. So you use, say, a 10k:600 ohm tranny, and now the tube's plate sees a load of 10k, which it can handle (if it's a nice hefty tube like a 6SN7). The cost, however, is a 12dB drop in signal level from the stepdown transformer.

So why not just use a cathode follower? After all, it has an output impedance around 600 ohms, right?

Well, not exactly. A cathode follower can be thought of as a voltage amplifier with 100% feedback, so that the gain is unity. (Actually a little less, but we'll ignore that.) While you're bringing the gain down, the output Z also comes down, so the output impedance is lowered as well.

Now load the tube with 600 ohms. The feedback...uhh, WHAT feedback? The 600 ohm load has brought the open-loop gain of the tube down to virtually nothing, leaving no margin for the feedback to reduce the output impedance, clean up distortion, etc.. So a cathode follower feeding a 600 ohm load has a very low clipping point, high distortion and a high output impedance. Bad news.

The good news: a cathode follower does a fine job at the output of a mike preamp if it's feeding reasonable impedances, like 10k. (It's still unbalanced unless you do some maneuvering, but that's doable.) I use one in my Big Mike preamp. But it won't work into 600 ohms. To do that with tubes you need a stepdown transformer; pity, but 'tis true. (For those interested, I cribbed this explanation lock, stock and barrel from Norman Crowhurst's "High Fidelity Circuit Design", which is available in .pdf form at the http://www.pmillett.com/technical_books_online.htm site.)

Even then it only does a fair job. The problem of providing really clean outputs into a 600 ohm load, with high headroom and low distortion, is remarkably difficult. Some things, I hate to say it, are easier to do with solid-state, and this is one of them.

Peace,
Paul

 

[raysolinski]

Excellent Paul, and thank you. That was a great explanation that was easy for even someone like me to understand...and it explains why we use stepdowns... :guiness: for you!

Ray

 

[emrr]

Yeah, the new search feature totally blows.  But PRR has said far more than anyone else here about the misconceptions and negative aspects.  There are just about absolutely zero vintage tube preamps using cathode followers in any situation other than a high-Z load.  Pretty much never when driving an output transformer. 

 

[guavatone]

maybe try google and input:

"topic" PRR  site:www.groupdiy.com/

 

[chr1s]

Another problem (or maybe another perspective of the said feedback reduction) with a cathode follower is that it can only drain the idle current from the load. So you need either a relatively high idle current or you would run into clipping of one half.

Chris

 

[PRR]

> Why don't more circuits employ a cathode follower?

CF has low SMALL-signal impedance.

This is NOT the same as saying it will "drive low-Z load"!

A 7-cent '741 chip with NFB has an output impedance (measured with small signals) near 1 ohm. It will NOT -drive- a 1-ohm or few-ohm load at interesting level. It delivers 5mA clean and 40mA max; the most it can put across 1 ohm is 0.040V.

Champ-Amp has a 6V6 and an OT. It will put 4V and 1 A into a speaker. Scratch the OT, wire 6V6 as cathode follower. The max the 6V6 can put out is ~~100mA, actually 50mA each side of idle. This is far less than we need to drive a 4-ohm, or even 40-ohm, speaker to small-room guitar levels.

As Paul says, the CF is just a high-NFB amp. The tube is REALLY a several-Kohm source. NFB can make the small-signal impedance 200 ohms. But put a low-Z load on it, drive big signal, it is STILL a several-K source.

To replace a Champ-Amp with an OTL amplifier, to put 1 Ampere into a speaker, we'd need four to eight sections of 6080 fat-power-triode, 60W-120W of heater power, 50-100W plate power. Instead we use 4W heater, 15W plate, and a 4-Watt OT. Save about 200W worth of iron.

Yes, the 4-ohm speaker is not essential. We can buy 16 ohm, we can series four 16 to get 64 ohm. Our 4W speaker power then needs only 0.25A rms from OTL tubes. Ten 6V6 "could" do it. Even then, the 25W of added heater-demand costs more in iron than the 4W OT costs. (Moreso because a Champ OT does not have to support 60Hz operation; some of these tradeoffs get less unfavorable if you demand 20Hz audio.)

And many large audio systems "want" floating balanced outputs. Push-pull CF is expensive, doubles your large-signal loss and small-signal impedance, and has very limited (or no) common-mode range. Transformer is perfect for this need.

Transformers and tubes GO TOGETHER. Tubes would be far less useful without transformers. Tubes are medium-high internal resistance, and most copper-snarl loads (speakers, motors, long-lines) are low impedance.

The argument is weaker with transistors, because they CAN squirt BIG current. A saturated 6L6 is about 1K internal resistance, a saturated 2N3055 is 1 ohm internal resistance.

I do think a lot of small-studio work could be done without line OTs. Unbalanced low-Z drive into unbalanced hi-Z inputs "can" work well in small situations. Awful lot of hi-fi gets done that way.

 

[PRR]

> google ... PRR  site:www.groupdiy.com/

That pulls up an awful lot of crap.

 

[jackies]

;D
All hail PRR!
He sure provides all the best reading on this site.
Very good style, normally electronics theory tend to be rather dry.
Should publish a magazine or something.

 

[raysolinski]

> google ... PRR  site:www.groupdiy.com/

That pulls up an awful lot of crap.

Thanks PRR..You are correct..the search pulls up a bunch of information ( I wouldn't call your writing crap!)..thanks for taking the time to condense it into a short, concise treatise..along with the other answers I have my head wrapped around this now. Didn't curiosity kill the cat ?

Ray

 

[pstamler]

Another problem (or maybe another perspective of the said feedback reduction) with a cathode follower is that it can only drain the idle current from the load. So you need either a relatively high idle current or you would run into clipping of one half.

Of course, that's equally true of a voltage amplifier tube...or anything that's not being run in Class-B. In a Class-A amplifier you always need to set idle current a bit higher than the maximum possible current to be driven into the load.

Peace,
Paul

 

[PRR]

> it can only drain the idle current from the load.

That's true of all SE audio amplifiers. I neglected the factor-of-2 correction because....

Vacuum is a -poor- conductor. Tubes are several-K ohm devices. Most happy wire-wound loads are several Ohms. It is like wiring your house with damp salty macaroni: large voltage-drop. This is a 100:1 mis-match. So any 2:1 correction is a small detail until we find a way to get large power across a 100:1 mismatch.

A 100:1 mismatch this way means, roughly, that to get 1 Watt out you must lose 100 Watts in the source (tube). It is like Income Tax at 99% of gross pay.... pretty heavy. We already must accept 90%-99% loss in a home-size wide-range loud-speaker, we can't afford an amp that bad.

Tubes can get better. Early ones were over 10K, by 1928 a 5K power triode (RCA-10) was possible, by the 1930s 1K and even 250 ohm was possible, albeit at lower gain and difficult to drive. Pentodes can do better; the saturation resistance of 6L6 approaches 250 ohms (160mA at 40V) with easy drive. Still closer to 1K than to 1 ohm. Monsters like 6c33(?), and wide-arrays, can reach lower resistance, but at the cost of huge heater power.

Loudspeaker voice-coil mass must be low, yet fill the magnet gap. There's some other constraints; you often end up with a 2-layer coil as the "obvious" winding. And the resistance tends to be 2 to 20 ohms. Yes, you can scale-up to more turns of smaller wire. 45-ohm used to be available for intercoms. Philips wound some 400 ohm speakers for OTL radios but the very-fine wire tends to corrode; or if well-shellacked you end up with more useless shellac than useful copper.

Only a crazy person would try to drive loudspeakers from tubes without transformers. Fortunately, there are plenty of crazy people in audio.

A short piece of audio cable "looks like" many Meg-ohms, but the capacitance of a mile of cable is 70 ohms at the top of the audio band. (Remember that before recorders, the only way to get audio from here to there, from source to paying-customer, was over a wire.) You don't "see" all the capacitance because there is a mile of inductance too. if you look "into" an infinite length of audio cable with audio signals, it "looks like" 100 to 1,000 ohms. This is also the optimum impedance for shorter lines to get good response with low loss. (And low echo, but that is not a problem for a mile of cable.) So a cable-driver tube has roughly 10:1 mis-match to the cable impedance, if the cable is more than maybe 100 feet long. Since line levels are millWatts and it is not cheaper to build tubes smaller than a Whole Watt, a CF line-driver is not so crazy as a OTL speaker driver; but when your load is in another building on another electric utility power connection, you are likely to get power-interference and transformers are very helpful.

BTW: the maximum power output of a Cathode Follower is the same as the max out of a plate-loaded stage. And the optimum loading is the same. Some practical difference: the plate-loaded may "fail" for high THD where the CF is still clean; CF's huge drive voltage means its driver will usually limit max output long before the CF itself is in trouble.

The CF is unity-gain, the plate-load has voltage gain. If you trade-off all the voltage-gain with NFB, the plate-load stage shows essentially the same Zout as the CF; but you can choose to retain some gain and accept higher Zout in favor of less obnoxious drive voltage.