Attenuator between common cathode stage and cathode follower.

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

Kingston

Well-known member
Joined
Nov 1, 2005
Messages
3,716
Location
Helsinki, Finland
I wish I had some other people to pester with my constant stream of questions, but I don't know a better place to ask...

which of the following is the best solution for an attenuator between a common cathode stage and a cathode follower?

index.php


Gut feeling says number 2. It doesn't need as large capacitor, and needs only one of them. Drawback is that pot (or MBB rotary switch) is referenced to a relatively high voltage, but at least it'll work. Notice I compensated plate resistor so total parallel resistance stays at 27k with the 100k pot.

Number 3 I was not sure of, would it even work? It also has the problem of being referenced to a relatively high voltage.

Or maybe there is an even better option I could not think of?

Thanks,
Mike
 

Attachments

  • atten-choice.png
    atten-choice.png
    18.6 KB · Views: 402
I really would go for solution #1 or 3
#2 is the worst because each time you move the pot, there's a huge DC variation that creates an unpleasant noise and a risk of destroying speakers.
#3 is not a bad solution, saves one cap. In addition, this cap can be much smaller because the input Z of the second stage is very high because of the bootstrapping effect of the cathode res.
As you said, #1 has no DC on the pot. DC on pots is not a big problem, but may become one when the pot accumulates dirt. Parasitic conduction between pot and ground is a source of noise.
 
abbey road d enfer said:
#3 is not a bad solution, saves one cap. In addition, this cap can be much smaller because the input Z of the second stage is very high because of the bootstrapping effect of the cathode res.

I see, bootstrapped cathode grid leak resistor, and the input impedance to the stage skyrockets. Didn't know what to look for, but now that I know the term I found this:

Zin = Rg / (1 - Av * (Rl / (Rl + Rb)))

(PS. the valve wizard pages have a typo of "Rg / (1 - Av * (Rl / Rl + Rb)" everywhere on their site, pretty serious error)

Zin = (470000 | 100000) / (1 - 0.96 * (18000 / (18000 + 1800)))

Zin = 650k

But when calculating the high-pass cutoff, doesn't the cap just see the paralleled 100k pot with 470k, ie. 82k? Or does the cap actually see the whole 650k?
 
The non-bootstrapped input Z varies between 80k and 100k depending on the position.
The actual bootstrapped Z is about 8x higher. You must take the lower value in consideration so, yes, about 650k. Anyway, unless you want to use it as an HPF, good practice is to use 3-20x larger value for extended bass response.
BTW, with #3, you don't need R5. The pot is enough for grid leak.
 
R5 safety makes sense in some places, though in practice, I have never found one implemented in any piece of commercial equipment.

I don't think I've ever seen your plan(s) done.  Every example I can think of has the two stages at fixed gain, with a 3rd stage in front of it, gain control being at that junction. 
 
Plan 3 won't cut more than about 40dB.

But big-picture: WHY smack V1A with max signal THEN turn-down? As Doug says, sane design would usually be controlled before V1A grid. V1A is not strained. And then you can lose all the interstage crap, DC-couple V1B grid to V1A plate. Adjust the 18K to get your target current back.
 
PRR said:
But big-picture: WHY smack V1A with max signal THEN turn-down?

It's the first stage of a preamp. I have a U-pad before input transformer (for line-in duties), but V1A is the first active stage and I don't want to gamble with an attenuator there. I've thought that's generally a bad and noisy idea. I've never seen it done anywhere either.

 
Kingston said:
PRR said:
But big-picture: WHY smack V1A with max signal THEN turn-down?

It's the first stage of a preamp. I have a U-pad before input transformer (for line-in duties), but V1A is the first active stage and I don't want to gamble with an attenuator there. I've thought that's generally a bad and noisy idea. I've never seen it done anywhere either.


For line only duties it seems to make more sense to use a pot( or stepped atten) - cleanest with no input transformer but still ok ran as a termination on the transformer secondary ( just like many tube limiters).

U pad really just a necessary evil to match 2 differing impedances, so mic only.  For line attenuation via fixed pads - balanced H can give a wider range of the often critical smaller attenuation values as can a T-pad.  This of course will give control over core saturation if needed.

As usual the mic situation is the trickiest - esp if you need exactly 5db of (non coloring)attenuation to get the perfect gain for a lead voc track.
 
We're not stating the obvious, which is that you're putting an attenuator after all of the gain, and before an output stage with negative gain.  The 1st stage here is traditionally seen as the driver for the follower, with the two DC coupled. 

Manley puts an attenuator on the sec of their preamp input transformer.  I'm pretty sure lots of other modernists do as well.  Many vintage program amps do this too, and are equally suited for preamp use so long as reflected Z to mic is okay. 

Look at it another way; why throw away gain with a U pad to counteract transformer step-up gain?  Maybe because line levels overdrive the transformer?  Okay, but if not, the better noise condition will be attenuating after the (not saturating) input trans, before the possibly overdriven tube.  Lead lengths short and well laid out, of course. 
 
emrr said:
Look at it another way; why throw away gain with a U pad to counteract transformer step-up gain?   Maybe because line levels overdrive the transformer?

That's it, input transformer headroom is designed for mics, but U (or H) pad will take care of that.

The first stage is never going to distort and it has ample headroom. If not, I will increase B+. I will not need any kind of gain control before that other than to prevent input transformer clipping with line level. Plan 3 cutting about 40dB is ideal.

I guess I should have given more parameters so there would be less guessing.

V2A above is not an output stage. It's driving a tone stack. A variant of this:

Amp-Tone-5-A-B.gif


After that I have V2B, and after that in fact a whole another dual triode for output stage, which in fact has it's own last step of gain control.

Lot's of stages, lots of points for controlled distortion. Lot's of knobs and switches on this one with preamp gain, tone, and master gain controls. Guitar amp ideas fused into a mic and line amp.

The plan is to have the tone stack feedback stage (as above) drive the stage after that to slight distortion (boost bass for better round overdrive etc.). A good control point for before the tone stack amp stages. I am also able to separately overdrive the output stage with one last stage of gain control. a cream-out instead of fart-down.
 
emrr said:
The 1st stage here is traditionally seen as the driver for the follower, with the two DC coupled.

You mean this? http://www.freewebs.com/valvewizard1/dccf.htm

Is it possibly to balance that one so there won't be distortion?

If yes, would it be possible to implement an interstage gain control there?
 
Kingston said:
That's it, input transformer headroom is designed for mics, but U (or H) pad will take care of that.
I will not need any kind of gain control before that other than to prevent input transformer clipping with line level.

Why not use a multi-step U pad on the front, that switches varying amounts of the two series resistors?  I think typical 3p4t will get you 3 steps and bypass.  I've done it before, but not recently. 

As to the DC coupled guitar amp circuit, I can't say, I haven't delved into it very far.  Thinking about it raises another question; why have V1A at all?  Do you need the gain at that point in the path?  If you are always using this at line level, I think you probably don't.  Run transformer to V2A grid.  Probably still fine even if bypassing the U pad and using it with mics, there's still transformer gain. 

Or, put in a switch that takes V1A in and out of circuit depending on gain/headroom needs.  Switches transformer sec between V1A and V2A grids, while also switching appropriate terminations. 

Or maybe I'm missing a point, and you really do want fully variable input gain. 
 
emrr said:
Thinking about it raises another question; why have V1A at all?  Do you need the gain at that point in the path?  

Or maybe I'm missing a point, and you really do want fully variable input gain.

I need the gain and the control both. Fine grained control for distortion is the name of the game. Too much gain is better here, but a control with even 5dB steps is too coarse already. There goes the loudest part of vocal take, when at 3dB down it would have been perfect etc. U-pad's and T-attenuators with fine grain is a chore I don't want to partake in, hence simple 100k logs (or maybe rotary switches). No more dual deck balanced attenuator madness for me, even if performance suffers.

The way I have envisioned the whole thing is that it will have two points where you can overdrive it with fine grain. One point is a stage right after the tone stack above. But I do not want to overdrive the tone stack that is absolutely drenched in feedback, it's gonna sound Sh*te, hence the gain control before that. But I also need some gain before the tone stack since a mic signal alone is too low at that point, that would be just asking for noise.

The second point is a separately overdriven (and controlled) output stage.
 
I guess I'm not seeing the point of fine control on the input, if one goal is to never overdrive the EQ section.  Seems variable input pad would suffice, and you'd have several fine controls after the EQ stage. 

But also sounds like you want the EQ to drive the next stage directly. 

This brings me back to pot on input trans sec.  Consider all the people who use an LA-2A as a mic pre. 
 
Back
Top