cathode follower output impedance

[dramadisease]

Im finishing up a simple 3 stage mic pre based on 2 6sl7 triode stages with a 6sn7 doing a cathode follower on the output for an unbalanced output, and im expiermenting with output impedance and trying to figure out what i should have on the output.

the natural output impedance is usually quite low on a standard cathode follower - do i need a resistor going to ground after the output cap to drop the imp any further? with a 47k cathode resistor into a 2.2uf cap - i can get an unloaded 5vac (roughly) - i loaded it with a 2.2k series resistor and a 560ohm resistor to ground and get a max of about 1vac. Im asuming i lowered my impedance with the obvious loss in gain - but did i lower it too much? am i sacrificing volume with this?

i just want to make sure my output stage can handle pretty much any gear i hook it up to and not change the sound too much (volume/tone)

totally lame question - sorry
thanks for the help!
-bryan

 

[pstamler]

[quote author="dramadisease"]Im finishing up a simple 3 stage mic pre based on 2 6sl7 triode stages with a 6sn7 doing a cathode follower on the output for an unbalanced output, and im expiermenting with output impedance and trying to figure out what i should have on the output.

the natural output impedance is usually quite low on a standard cathode follower - do i need a resistor going to ground after the output cap to drop the imp any further? with a 47k cathode resistor into a 2.2uf cap - i can get an unloaded 5vac (roughly) - i loaded it with a 2.2k series resistor and a 560ohm resistor to ground and get a max of about 1vac. Im asuming i lowered my impedance with the obvious loss in gain - but did i lower it too much? am i sacrificing volume with this?[/quote]

Yes. What you basically did is create a voltage divider whose output will be 560 / (560 + 2200) of its input. So if you put 5V into it you'll indeed get 1V out. You're not getting a particularly lower output impedance, either, as the 6SN7 is probably about 500-600 ohms out anyway (the equation is in the other room).

The good news is that you don't need anything on the output except perhaps a 1M bleeder resistor going to ground after the capacitor, so the cap doesn't float when there's nothing plugged in. The 6SN7 cathode follower should, if properly designed, handle a decent-sized signal (5-6Vac) into 5k or higher. It won't drive 600 ohms without a transformer, though.

Peace,
Paul

 

[dramadisease]

The good news is that you don't need anything on the output except perhaps a 1M bleeder resistor going to ground after the capacitor, so the cap doesn't float when there's nothing plugged in.

awesome - thanks for the info!

The 6SN7 cathode follower should, if properly designed, handle a decent-sized signal (5-6Vac) into 5k or higher. It won't drive 600 ohms without a transformer, though.

i guess i would say thats acceptable - cutting it close though.. is there any way of easily lowering the impedance of the output?

i did notice that i have some edcor 10k/600 transformers laying around - would the primary impedance on those be an ok value for this circuit?

thanks again for the help!
-bryan

 

[bcarso]

Use a higher gm tube. Or, use a hybrid stage or a multi-tube stage with feedback, like a White follower. You can also parallel sections or tubes per se.

A 6C45 will give you about a 30 ohm or less output Z as a follower iirc. It will need a big grid stopper R though (>300 ohms). I had a multistage circuit with those tubes before I knew how prone to oscillation they were. I should have known better. I promptly blew the grid of a stage to smithereens before I knew what happened :-(

 

[PRR]

You seem to be looking at small-signal output impedance, when you really need to know the large signal parameters.

You neglected to mention any power supply voltage or current. Or what load(s) you drive.

However, if you had a clue when you picked 2.2uFd, then you don't plan to drive less than 5K loads with full ~20Hz response. i.e. you don't plan on finding any 600Ω loads (which a simple 6SN7 without transformer won't drive well anyway).

> with a 47k cathode resistor into a 2.2uf cap - i can get an unloaded 5vac

6SN7 with 200V B+ and 100V on the grid, 47K cathode resistor, will pass 100/47K= 2mA. Unloaded and large distortion, it should swing down almost to zero and up to about 180V. After the coupling cap, this would look like +80V/-90V, or around 60VAC. If you are distorting at 5V, something is wrong.

> i loaded it with a 2.2k series resistor and a 560ohm resistor to ground and get a max of about 1vac.

SO the 6SN7 is swinging 7V peak in 2,760Ω, or 2.5mA peak. I deduce you must be putting at least 120VDC across the 47K cathode resistor. 6SN7 saturated plate resistance is around 5K or 10K, or roughly 3 times the 2K8 load, so you must have at least 4 times 7Vpk or 30V static plate-cathode voltage.

> make sure my output stage can handle pretty much any gear i hook it up to

If you really want to drive 600Ω loads to high level, +4dBm nominal (+20dBm full scale), you need 11V peak and 11/600= 18mA peak. With a single tube, it must be biased to 18mA idle and able to swing to 36mA peak. A 6SN7 will just-barely pass 18mA at zero grid bias without exceeding plate dissipation rating, but can't swing higher. Even if it could, it can't swing to zero current without large distortion: we'd want maybe 5mA minimum current. And rounding-up the signal current to 20mA peak, that suggests a 25mA standing current swinging from 5mA to 45mA. 6SN7 won't do that. Not even paralleled or push-pull.

Driving 600Ω directly needs a BIG bottle. 2A3, 6V6/6L6, or one of the TV V-sweep power tubes. (6EM7 would be a natural.) Even then, you don't get a balanced or floating output...

> i have some edcor 10k/600 transformers

That's a 4:1 transformation. If the 600Ω side must swing 11V 18mA peak, the 10K side must swing 44V 4.5mA peak. This is more like what a 6SN7 can do. Since the Edcors are not rated for DC in the primary, we look at resistance-coupled transformer output. It turns out to be a tight design. Assuming B+ is 300V, we can set up about 175V across the 6SN7, about 120V on its grid giving about 125V across the cathode resistor, about 10mA idle current leading to a 12K cathode resistor. Cap-couple this (the 2uFd is maybe adequate) to the 10K:600 iron. Gain is about 1/8 (1/2 in the heavily-loaded CF, 1/4 in the iron). With a 1:10 input transformer, two 6SL7 stages, and no feedback, you have gain to the CF grid of over 10,000, so you should be OK with a 1/8 voltage loss in the CF and iron.

And the true floating output simplifies interfacing.

> the 6SN7 is probably about 500-600 ohms out anyway (the equation is in the other room).

1/Gm, or for a triode, Rp/Mu.

For 6SN7 at 150V 2mA, Gm is around 2.5mA/2V, so 1/Gm is 2V/2.5mA or 800 ohms. At book conditions (90V and 250V at 10mA) Rp is 6K7 or 7K7, Gm is 3,000 or 2,600umhos, Mu is 20. Output impedance is in the 335Ω-385Ω range. The 500-600 guesstimate implies Gm of about 1800umhos, which for 6SN7 happens at 2mA to 4mA which is indeed a popular operating range.

But note that a cathode follower will NOT drive high levels into a load anywhere near its output impedance. Wiring as CF does not change the large-signal relations. Large-signal, we still have plate resistance of ~7K. The electrons have to flow through ~7K! Because of DC current, the maximum-power load is about twice that or 14K. We see that the best-power load for a triode CF is roughly 2*Mu times its small-signal output impedance, or for 6SN7 about 40 times Zout!

It is a broad peak so 5K and 20K can take large power too. But 600Ω will be far down the mis-match slope. Also distortion falls fast on the hi-Z side of the peak, but is high on the low-Z side (one nice thing: THD below clipping does not rise as impedance drops much below Rp, we just clip at lower power).

 

[Learner]

Hey that is some great stuff PRR!!! :grin: :thumb:

Can never get enough revision on tube lessons. :thumb:

 

[dramadisease]

wow - thanks for the explanation PRR - after reading it about 5 times i went back and tried a bunch of combinations on the output and im not totally sure of my findings.

heres a schematic of what i did ->

http://www.imagecabin.com/imagehost/pic.php?u=336lmAAY&i=2630 in case it doesnt show

i get 59vac before clipping into the CF and 57vac out - looks good - seems like its working well. with the 250k resistor i get a grid bias of -4.7vdc with no signal, with respect to cathode. if i look at my recieving manual - 4.7 bias with 310 plate should be really quite high in terms of plate current but when i measure it it says about 2-3ma.

anything under 47k cathode resistor seems to pull down the plate voltage and doesnt allow signal to pass. i used a 250k pot and found that when it was at 250k it seemed to work well. symetricaly (spelling?) clipping and strong signal.

the circuit works fine but am i doing something wrong? am i really drawing all that current in the CF?
thanks again!
-bryan

 

[dramadisease]

so i realized that when i was reading the chart in the recieving manual i wasnt refering to cathode with my plate voltage so i actually was drawing 2-3ma.

i replaced the cathode resistor with 100k and got the tube to idle at around 5ma plate current. it seemed to oscillate a bit at more current and would stop puting out signal. it seemed like a smaller resistor would just suck to much voltage off the B+ line.

with 100k when i put a triangle wave through, put it into clipping, and look at it with a scope it clips most symetricaly @ 100k. but is this too low current to idle at? i plugged the transformer output into my digi001 and played guitar through the pre and it worked really well. drove the +4 input into clipping before the pre clipped!

im assuming that the low current isnt *too low*
am i wrong in this assumption?

im really starting to grasp how much biasing really makes a difference in preamp tubes. before i could only relate to it on a power tube level.

 

[PRR]

Your IMG tag doesn't work because this board's software does not believe there can be a "?" in an image URL. (It is perfectly legal, just so uncommon that the board programmers forgot to handle it.) Also there was some glitch with your server before (pile of PHP errors). So I'm taking a liberty of re-posting the image:

> i replaced the cathode resistor with 100k and got the tube to idle at around 5ma

That does not add up. 5mA in 100K implies 500V on the cathode, and you only have a (???) 310V supply.

Assuming the 2nd stage plate sits about half supply voltage, grid of 6SN7 is around 150V, cathode is near there, so 150V/100K is 1.5mA.

Peak output is around 60V 1.5mA on the primary of the 10K:600 iron. At the secondary it is 15V 6mA. It may be in trouble for loads less than 2.5K. The sound card or mixer is probably 10K, OK. But the transformer itself drops to 10K:600 at the bottom of the audio band, maybe 20Hz. So you may get "only" 4Vpk at 20Hz, 8Vpk at 40Hz, and not be able to output 15Vpk (+20dBu) until almost 80Hz. Good enough for guitar and voice, a bit short-bottom for strong bass (though it should pass bass well at levels lower than meter-peg).

 

[dramadisease]

thanks for the repost PRR

im going to see if i cant get that stage to idle with a bit more current - so far this pre sounds great as a line amp.

im going to try increasing the B+ to the tube and dropping that resistor down a bit
ill make a post later in the day

thanks again!
-bryan

 

[dramadisease]

thanks PRR - im finally getting what your post said

i was letting my B+ drop way too low - that was my problem

settled on 15k - running around 8mA

the low end performs a bit better but its not much of a difference - atleast on my 001

thanks for the help!

-bryan