Changing the load resistor in the cathode follower

[Gaz]

After reading Merlin's article on the DC coupled cathode follower, I thought I might try his suggestion and lower the follower's load from the normal 100k down to 47k, ostensibly creating a more compressed tone.

I didn't favor the results by lowering the load, as it reduced gain and clarity too much, but I did prefer raising it 220K (however, I'm not sure if I liked it just because it was louder).

Can anyone explain what happens when the load is increased? I'm still having trouble understanding Merlin's article, but I can't deny I liked the sound better when quickly switching values with my handy substitution box.

Would it be a good idea to change the plate load in the previous stage as well? I'm really just curious if anyone else has experimented with this.

Thanks.

[loudthud]

That article makes a good point but needs a little work. I ran into this in an amp I built where I wanted more gain out of a 12AX7 and still needed to drive a lowish impedance. To get the gain up, I used a 470K load resistor on the gain stage. Normally, 470K is so high you really can't drive much, a cathode follower is about it. The first mistake I made was copying the Vox AC100 which uses a 56K load on the cathode follower. That really clamped the positive excursion of the gain stage to about 150V. I changed the 56K to 100K and then tweeked the cathode resistor of the gain stage for max headroom. That worked well.

Most amps that use that circuit don't use a 1.5K resistor in the gain stage's cathode. They use 1K or 820 ohms. This biases the gain stage for more headroom when the cathode follower is connected. It's true that reducing the resistance of the cathode follower's load will clip the top part of the output waveform. To compensate, you should reduce the cathode resistor of the gain stage to preserve headroom.

The cathode follower makes it possible to drive a lower impedance load but it does eat into the maximum available peak to peak voltage. Tweeking one or more of the resistors in the circuit will change the gain, headroom and clipping characteristics.

[Gaz]

thanks a lot loudthud for your input and experience.

i currently have the gain stage driving the CF with 2.7k on the cathode with a .68uf bypass, and 220k on the plate. i biased it this way to get a bit more distortion at full volume/less headroom (it's a plexi syle amp). as i mentioned, the load now on the CF is 220k.

when initially experimenting with lowering the load of the CF, i had an 820ohm/100k combo on the stage driving it, which as you said gave increased headroom. because i wanted more distortion, this wasn't what i was looking for, and 'hot rodding' the previous stage seemed to help along with actually raising the load of the CF to 220k.

sorry for rambling, but my real question here is about what's actually happening when i increase the load of the CF. i know it's biasing the tube differently, but that's as far as my understanding goes. to my ear the amp is louder and sounds 'better', but i can't find any other designs that do this, and i figure there has to be a theoretical reason why they're not. i totally believe in trusting my ears, and i know the change won't ruin my amp, of course, but i'm at the point in my autodidactic education where i need to understand what's really going on before moving on

[daz]

I went thru the same kind of thing when i added a bypass cap across it. I couldn't figure out why no one did because no matter how many times i tried the amp w/o that bypass cap it never sounded or felt as good. So why then does no one do it? Well, there are a few that do as i was told like the plexi for one, so i felt in good company there and couldn't then help but wonder if that could be part of the legendary plexi sound. I think it is too because it gives the amp a much spongier and a little harmonically richer sound. but that makes it even harder to understand why so few do it. So for some reason there are indeed some design aspects that give the kind of results that are generally thought to be good sonic qualities yet aren't used for the most part. I can't figure it.

[Gaz]

so you put a bypass cap across the cathode load resistor? i'll have to try that out. yes, sometimes it's hard trusting your ears when they're up against a 'legendary design'.

[daz]

Well, what i was saying is that what i did WAS the same as a legendary design...the plexi. But the point was that is one of very very few amps that use this design and when i did it i didn't even know any amps used thast. i was told in a thread here in fact. But i still was confused because why only those few? In any case, i find it a necassary part of the amps tone. It's never as good w/o it.

[tubeswell]

Originally Posted by Gaz

so you put a bypass cap across the cathode load resistor? i'll have to try that out.

You mean the CF cathode resistor? Hmmm... you'd have to have a care for how big/small the value of the cap was. I can see how smaller caps would only cut highs, but wouldn't larger caps only hinder the functioning of it? Hooking a larger value cap in parallel would work against the voltage swings (and it'd have to be a fairly high-voltage cap too I should imagine)

[Gaz]

You mean the CF cathode resistor? Hmmm... you'd have to have a care for how big/small the value of the cap was. I can see how smaller caps would only cut highs, but wouldn't larger caps only hinder the functioning of it? Hooking a larger value cap in parallel would work against the voltage swings (and it'd have to be a fairly high-voltage cap too I should imagine)

daz's original post had some typos, and i think that's why i was confused. i think he was referring to the gain-stage preceding follower, not the follower itself. perhaps he misunderstood my original post as well...

i realized my misunderstanding only shortly before reading tubewell's post: i jumped in a 400v .68uf cap across the load resistor of the CF only to find that it cut highs and volume. i imagine that tubewell's precisely right in thinking that a large value would render the follower useless.

with all that cleared up, i'm still interested in why increasing the follower's load resistor from 100k to 220k makes the amp louder, and the CF presumably more efficient i'd really like one of you electrical engineers to blow my mind on this one

[tubeswell]

Well my shot at it would be that increasing the size of the load resistor allows a bigger voltage swing to occur. (So that E = I x R, where 'R' increases but 'I' is constant - therefore 'E' ultimately increases).

(Edit - My initial thought was that current may also be reduced through the resistor, but on 2nd-thoughts the electrons are flowing from the ground return to the anode, and the size of the cathode resistor (which just happens to be in series with the tube) wouldn't affect the amount of current in this regard. (At least I don't think it would)

[Gaz]

thanks, i'm looking forward to hearing any more of your ideas, man.

[pdf64]

You get how a diode clipper limits the signal to +/- 0.6V (using 2 back to back diodes in opposite polarity)?
With just 1 diode pointing to ground, only the top half of the signal gets limited.
Merlin's article explains how the hot biassing of the cathode follower creates a tube version of a 1 diode clipper, with how hot it's biassed governing at what level the clipping comes in. 100k a bit clipped, 56k more clipped.
At 220k, I'd guess it's less clipped than standard, like tubeswell says.
As you heard, clipping reduces the signal level, so going to 220k reduced the clipping, making the signal louder.
Louder usually sounds better, really need to normalise the levels to get a true tonal comparison.
The downside of 220k is that the cathode Vdc goes up. The spec for the insulation between cathode and heater is 100V.
With a 100k cathode resistor, the circuit makes it sit at 180-200V, then puts a massive ac signal riding on that, so it may be peaking at 250V (assuming a 450V B+ at rectifier).
At 220k, it will go even higher.
The insulation on most older tubes seemed to be way over spec, but there's reports of many current russian 12AX7s failing when used in regular cathode followers.
It's good practice not to design circuits which exceed the absolute max ratings of the components.
When modding a standard circuit which actually does this, then best not to make the situation worse by pushing it further over the limit.
But it's not going to hurt anyone, so do what sounds good to you, bearing the above in mind (and avoiding russian tubes for that application). Peter.

[tubeswell]

What pdf64 said - maybe you want to think about elevating the heaters to about 90V, as well as using older tubes in that slot.

(Having said that the 'Bassman+' I finished recently has a few CF stages with high voltages, including the CF after the trem LFO stage, which idles at 213VDC with a 470k load resistor, and the heaters are only elevated to 40VDC, and there don't seem to be any problems with it so far - touch wood. I gather the Vibroverb 6G16 LFO CF stage operates in a similar ballpark without elevated heaters)

[daz]

daz's original post had some typos, and i think that's why i was confused. i think he was referring to the gain-stage preceding follower, not the follower itself.

yep, sorry.

[Gaz]

You get how a diode clipper limits the signal to +/- 0.6V (using 2 back to back diodes in opposite polarity)?
With just 1 diode pointing to ground, only the top half of the signal gets limited.
Merlin's article explains how the hot biassing of the cathode follower creates a tube version of a 1 diode clipper, with how hot it's biassed governing at what level the clipping comes in. 100k a bit clipped, 56k more clipped.
At 220k, I'd guess it's less clipped than standard, like tubeswell says.
As you heard, clipping reduces the signal level, so going to 220k reduced the clipping, making the signal louder.
Louder usually sounds better, really need to normalise the levels to get a true tonal comparison.
The downside of 220k is that the cathode Vdc goes up. The spec for the insulation between cathode and heater is 100V.
With a 100k cathode resistor, the circuit makes it sit at 180-200V, then puts a massive ac signal riding on that, so it may be peaking at 250V (assuming a 450V B+ at rectifier).
At 220k, it will go even higher.
The insulation on most older tubes seemed to be way over spec, but there's reports of many current russian 12AX7s failing when used in regular cathode followers.
It's good practice not to design circuits which exceed the absolute max ratings of the components.
When modding a standard circuit which actually does this, then best not to make the situation worse by pushing it further over the limit.
But it's not going to hurt anyone, so do what sounds good to you, bearing the above in mind (and avoiding russian tubes for that application). Peter.

Thanks for clarifying that. I'd rather just use the 100k, and be running the tube within its limits. And as you said, I'd have to normalize levels before making judgments on tone anyway - I have to do that all the time in recording with mic placement. I tried, actually, but it was too hard to tell... Thanks again for the explanation.