How to add gain adjustment of first triode

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

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

BluegrassDan

Well-known member
GDIY Supporter
Joined
Mar 17, 2009
Messages
507
Location
Elizabethton, TN
Is it possible...?

Can a stepped attenuator be implemeted somehow so that the first gain stage alone can be varied? I'd like something other than just a -20dB PAD on the front end. I want to be able to reduce the first tube's gain when it sees medium-hot signals so it's not always on the verge of crapping out, and it can be dialed in with more precision.

For example, with this configuarion, singing at a normal level into a U87Ai distorts the first tube.

As another example, I see how the Gyraf G9 is designed for some first-stage attenuation for rotary switch positions 1-4.

Thoughts?

(FYI - there is a 10k stepped attenuator between stages not shown in the schematic)

 

Attachments

  • 3-triode, Mar 29, 2020.pdf
    65.9 KB · Views: 53
The bias of the first stage looks a little odd to me. The 6072 data sheet recommends, with a 300V supply and a 240K plate resistor, that you use a cathode resistor of 2700 ohms. You 7500 seems rather high. I suspect you r plate voltage is too low and you are getting early clipping on one half of the waveform.

Cheers

Ian
 
Sure. Just change your 1M input grid resistor to a 1M stepped attenuator. But having lots of steps probably isn't that terribly useful for this specific scenario. It might be more appropriate to just make 3 steps with a 3 position toggle.

Note that if you use an ON-ON-ON toggle (yes there is such a thing, the center position is actually staggered between the two poles) you can make the switch positions work in order like -20, -10, 0 instead of the usual -20, 0, -10 sort of goofy switching you have to do with the more common ON-OFF-ON toggle.
 
Input transformer with secondary taps.

Use multiple primary taps. 

Multiple input transformer options.

Local 1st stage feedback. 

I’d expect 1M pot approach to lose highs rapidly.  Test to prove usefulness.
 
I haven't looked at the data sheet but Ian's catch concerning the cathode resistor value sounds reasonable to me.     
I agree with Doug's suggestions too and would add to the reasoning behind *not*using a 1M attenuator that it will be quite noticeably noisier at the -6dB position than full up (off).  So, noisier and less high end    ???

What about returning a little feedback from after you 2nd stage follower's blocking cap (so the same point as your 10K fader) to the bottom of your CMMI-5C's secondary?  If you're only looking for an amount of attenuation that's less than your 20dB pad then, 10dB of feedback probably won't do too much harm to your current open loop "tone". :)

Alternatively,  something like the  EMI 6 position, 6dB per step input pad as used on the REDD desks. 
 
I have been looking at and trying to understand the Redd 47 pad.

Perhaps I am simply undermistanding what is possible with this topology.

I was discouraged last night using my Seventh Circle C84 (Cohen-style). It sounds SO good compared to my tube preamp. If I could only figure out a way to build a tube preamp that sounds/operates like that, with just a bit of tube/transformer color...

And it’s how that first gain stage is handled. It’s integral to the potentiometer. Didn’t need a pad for a U87, like I do with my tube pre.
 
I have been looking at and trying to understand the Redd 47 pad.

The pad I'm referring to isn't part of the REDD.47 schematic, it was a desk control surface function rather than an amp feature,  but I did post it somewhere on here some years ago.


Perhaps I am simply undermistanding what is possible with this topology.

Dunno but I've never heard that before  - undermistanding -  Haha  :D I like it.

I was discouraged last night using my Seventh Circle C84 (Cohen-style). It sounds SO good compared to my tube preamp. If I could only figure out a way to build a tube preamp that sounds/operates like that, with just a bit of tube/transformer color...

A tube pre that resembles a Cohen in terms of gain adjust  is a far cry from what you have, it'd take more heavy lifting in terms of getting your poles and zero's in order to be able to enclose a tube gain stage to operate over such a large gain range without instability. 

If the only thing that was letting your tube pre down was a matter of reducing the 1st tube gain by 10dB then feedback would do that.  If you liked the tighter, more controlled aspect of the Cohen then... ? 
Why not think about grafting a tube line stage on the back end of a Cohen? 

P.S.  What's your pre like with your U87's pad engaged?
 
BluegrassDan said:
I have been looking at and trying to understand the Redd 47 pad.

doesn't have a pad that I see, just adjustable feedback.  You can do that around your first stage if you want, probably need an iso series resistor between transformer and grid like the OP-6, otherwise you're driving the input transformer backwards and changing input Z. 

BluegrassDan said:
Didn’t need a pad for a U87, like I do with my tube pre.

20dB+ transformer, fact of life, use a hot mic into it you might need closer to unity from a 20dB pad, and that's fine with plenty of gain later.   

Something like a 150:5K  transformer  (or less!)  is probably better matched to many condenser mics if you want to avoid a 20dB pad.  Then it won't work well with ribbons or other dynamics. 

The indisputable answer to this problem, given the assumption of a transformer, is always the Hardy 990 circuit:  low transformer step-up with an opamp that matches to a 600 ohm winding as perfectly as anything does.  Never needs a pad with a microphone.  Untrue with most anything else with a transformer front end, because of the gain. 


Winston got here before I finished:  yeah, chase a 990 stage with a tube line amp, why not. 

 
A hybrid design is an interesting idea.

With the preamp's -20dB pad engaged it sounds just fine. I just worry about predicting when the input stage is gonna overdrive, versus a topology that can be driven harder inherently. To me, having the mid-stage attenuator set to 25 dB with a condenser mic should not have gross distortion that I'm getting. I understand why it does, and how to remedy it, but I want it to be "not this way."

Most of my frustration is that I'm still green at all this and wish I had the requisite knowledge to know how to do what I wanna hear. My ears are good, and I wanna put good-sounding sound into them.
 
I found a copy on the site of the variable input pad I've used in the past so uploading it here too.

In regard to feedback, again, I would probably wrap it after your follower and return it to a resistor underneath your transformer.  That way you can keep the R FB shunt to a decently low value to not add appreciable resistance  & noise on the grid and have enough current to drive the suitable R FB series from the follower.  Doing it this way won't adversely affect your transformer loading etc.

Another vote from me for the Jensen dual 990 design being a good one

Edit:  Hey Dan, only just noticed a couple of things
1/ Your 1st stage is de-coupled on the cathode yet your 2nd isn't, is that schema correct?
2/ Not related to gain but - You're feeding your output White follower with a cathode follower,  is that correct?
 

Attachments

  • Hinson stepped att Redd47.gif
    Hinson stepped att Redd47.gif
    24.6 KB · Views: 38
BluegrassDan said:
With the preamp's -20dB pad engaged it sounds just fine. I just worry about predicting when the input stage is gonna overdrive, versus a topology that can be driven harder inherently.


Think of it this way:  It's the input transformer that's driving it hard, not the mic.  The tube with a 1:1 transformer connection may take more than the C84, which is a 1:1 connection to the mic.

Example 2:  I put my U67 directly into my converter input with no preamp if there's anything loud in front of it; plenty of level. 
 
Winston O'Boogie said:
Edit:  Hey Dan, only just noticed a couple of things
1/ Your 1st stage is de-coupled on the cathode yet your 2nd isn't, is that schema correct?
2/ Not related to gain but - You're feeding your output White follower with a cathode follower,  is that correct?

Yes, and those were all intentional after many comparisons. It's been a few months, so let me try to recall the reasonings (in reverse order of your questions)...

2. I went with the first 6072 follwer to get the impedance down in order to use a 10k pot instead of 200k. The 200k pot was causing Miller rolloff. I followed the same scheme for the 2nd 6072, but the CF alone wasn't enough to drive the output. Added the White after that, and now it will easily drive 1:1.

1. Without bypass caps the pre has about 57dB of gain. Bypassing one added something like 7.5dB of gain, putting up around 65db. I remember exactly why I chose the first stage over the second - might have been something to do with how the lower impedance the bypassed cathode presented to the first CF helped with high frequency response.

This current configuaration has very good specs out far past 20k. The 1:5 ratio does play very well, but the hot condensers need padding in most cases.

I could remove the cathode bypass to drop the level and provide some local feedback, but I'll be back under 57dB max gain.
 

Attachments

  • 3 Triode, Sweeps, FR.png
    3 Triode, Sweeps, FR.png
    165.3 KB · Views: 18
I ran your first stage through LTspice. Stage gain is 29dB as is. With the 20dB from the transformer that's a total of 49dB before the first pot. Fortunately tubes can take a lot of level before they fall apart. I simulated yours with 0dBu on the grid and its +29dBu output has about 1% distortion at 1KHz which is a bit of tube tone but not noticeably nasty. This means you should be able to input -20dBu before things start to get into serious distortion.

Cheers

Ian
 
BluegrassDan said:
2. I went with the first 6072 follwer to get the impedance down in order to use a 10k pot instead of 200k. The 200k pot was causing Miller rolloff. I followed the same scheme for the 2nd 6072, but the CF alone wasn't enough to drive the output. Added the White after that, and now it will easily drive 1:1.

1. Without bypass caps the pre has about 57dB of gain. Bypassing one added something like 7.5dB of gain, putting up around 65db. I remember exactly why I chose the first stage over the second - might have been something to do with how the lower impedance the bypassed cathode presented to the first CF helped with high frequency response.

This current configuaration has very good specs out far past 20k. The 1:5 ratio does play very well, but the hot condensers need padding in most cases.

I could remove the cathode bypass to drop the level and provide some local feedback, but I'll be back under 57dB max gain.


Thanks Dan.
I was just curious on the 1st stage bypass cap since it is raising the gain and that (possibly)  having too much there is giving you grief. 

Yep, the 1st stage's follower is great.

On the double follower output, I didn't doubt that you needed the 'White' to drive the outside world, just was questioning whether you needed the follower beforehand as well.  Although that's a diversion from your issue and,  if it works - don't fix etc  ;)

I didn't mean to be second guessing your design, I can imagine it's after lots of back & forth testing on your part,  I was mostly just asking questions really and, whatever works is OK with me.

If you just need a cleaner throughput  for hot signal's, maybe  check the biasing as per Ian's suggestion and then try a variant of the variable pad I uploaded.  That schema, and off-shoots has worked for me.

Edit:  Ian did some real work for ya.  So, maybe you're dealing with signals closer to 0dBu which isn't impossible.  I say, again, try a variable input pad for those signals.  Or your U87's pad.
 
You can add some negative feedback by splitting the 56K cathode load of the CF stage. For instance you could split it into a 50K top section and a 6K bottom section. If you now feed your input transformer to the first grid via a capacitor you can then connect the grounded leg of the input transformer to the junction of the 50K and 6K resistors. This drops the gain down to 14dB and also drops the distortion by 15dB or so.

Cheers

Ian

 
ruffrecords said:
  If you now feed your input transformer to the first grid via a capacitor you can then connect the grounded leg of the input transformer to the junction of the 50K and 6K resistors. 

Ian, does  simulation show any noise increase with the 6K shunt?
Can it be bettered by taking feedback after the follower's cap and downscaling the values?
 
ruffrecords said:
I ran your first stage through LTspice. Stage gain is 29dB as is. With the 20dB from the transformer that's a total of 49dB before the first pot. Fortunately tubes can take a lot of level before they fall apart. I simulated yours with 0dBu on the grid and its +29dBu output has about 1% distortion at 1KHz which is a bit of tube tone but not noticeably nasty. This means you should be able to input -20dBu before things start to get into serious distortion.

Cheers

Ian

And the U87Ai has a max output -6 dBu. In other words, the mic simply needs to be padded down.

Thanks for taking the time to run this, Ian. You guys have certainly helped me learn a lot the past few years.
 
I just realised the pot after the CF is 10K rather than the 100K I had assumed.

it is the 10K pot loading the CF stage that is causing your premature distortion. The CF cannot output much more than +14dBu into the 10K pot. With your 49dB of gain that means the maximum input signal before distortion is only -35dBu so no wonder it is easy to distort.

Cheers

Ian
 
ruffrecords said:
it is the 10K pot loading the CF stage that is causing your premature distortion. The CF cannot output much more than +14dBu into the 10K pot. With your 49dB of gain that means the maximum input signal before distortion is only -35dBu so no wonder it is easy to distort.

The plot thickens.  Ah-Ha, didn't think of that.  Nice one.

 
Winston O'Boogie said:
Ian, does  simulation show any noise increase with the 6K shunt?
Can it be bettered by taking feedback after the follower's cap and downscaling the values?

I have not checked that but there will be some increase in noise. If you see my other post you will see the 10K pot is already causing the CF problems so adding a low value pot divider after the CF output cap will only make that worse.

Cheers

Ian
 

Latest posts

Back
Top