'Tube Gain Controlled Amp' mastering box(s) project

Hi All

I'm just starting up my next tube limiter project - it's based on the 'vacuum tube leveling amplifier' designed by E J Jurich.

The designer himself posted the schematic and supporting information some years back - can't seem to find it here tho'. It surely was archived somewhere. Received very little notice at the time as I recall.

The document is now available for sale from Amazon, so I'll just put up schems of the  circuit blocks as I adapt them to my 'vision thing'.

Back then I didn't appreciate the novel twists of the design but I am much better able to understand it all now 

.. mostly  or  'somewhat'.  Anyway - it's pretty groovy.

SO - seeing as I have some nice iron coming courtesy of member 'emrr', I wanted a to do another tube limiter to stretch myself goodly-like.

------

It's an all-tube stereo compressor/leveller, intended for radio broadcast duties.  It is not, as the designer says, a 'brick wall limiter' but is very effective in its intended role - low (ish) distortion, forgiving in usage and reportedly very nice of a thing to listen to !

I think its an inspired piece - well thought thru, designed, implemented, built and photographed! Now, many years later, I am ready and excited to bring it to life at 'my place' 

ps  my other finalizer is a finalizer!

Running them into el34 pp tube amps and some funky 70s speakers  8)
( as well as some proper Tannoy 8" full range boxes and  ss power amps I use for creetical listneng  and what not)

------

Apart from interesting implementation details, the design is notable for it's use of two time-constant dynamic compression characteristic.

One time-constant is fast and drives a vari-gm stage, one time-constant is slow and drives a vactrol opto shunt stage.

Now each of the fast and slow gain cells has individual detectors, and are true stereo audio paths.

One detector, the vari-gm side chain, has a kind of 'true power summing' thingo to prevent skewing of the stereo image where one channel is markedly different from the other.

It's all implemented  in the feed-back style (rather than feed-foward).

On top of that, the vari-gm stage uses cathode follower fed topology (CV to grids)  rather than and plate loaded output (kind of like the australian Trimax limiting amp, except in feed-back mode and without the 'nasty' delay line in the audio path).

Summarising the topology :

- all tube and quite minimal audio path

-  all tube detector paths with power boost

- fully independent Left, Right audio channels in a stereo package

- single ended internal gain staging  ie. not using balanced gain cell

-> to be more precise, it uses diff push-pull style amps but across L-R signals.  Not as a classic balanced gain cell config but as unbalanced,  symmetrical over L-R.  The schematic does it better than my words 

- fast  tube  compressor stage with cathode-fed vari-gm gain control cell, 'power summing of left-right' in the detector

- slow vactrol levelling stage with tube driven vactrol shunt-voltage gain control cell, independent left-right detector

- input amplifier  features a passive tonestack style with HF/LF trim (-/-5dB) with gain makeup


My personal twists are :

- transformer input debalancing stage

- transformer balanced output, cap coupled parafeed style (class A operation)

- adapted eq to active bax style. range  +/- 10dB, possibly a switched break point option

- different tube types, different vactrol type

- change vari-gm gain cell to cathode following outputs for possible lower distortion

- adaption to the vari-gm time constant network for more flexibility

- additions to the psu for HV regulation and utility rails

- additions and changes to the GR indication circuit and meter

plus

- point-point wiring (like the original) but with low-cost racking

- 'tubes and transformers on top' construction style with knobs 'on top and in front'

- standalone power supply with 2m connecting supply cable

- DOA based balanced parallel 'line out'  and a hardware bypass    ..... for good measure! 

AND of course

- a pair of beautiful  'true-wonder'  ballistics enhanced  'old timer'  vu meters with chip buffers

RIGHT then - tally ho!

This planned 'Tube Gain Controlled Amp'  mastering box build is the 'next exciting installment' in my 'tube, traffo and doa' builds of the 'tubes on  top'  variety  with external power supplies.

The first installment, I'm working on now; psu unit with two identical halves; 7-pin cables-and-connector sets.
Each around 100mA of HV, 2.5A of heater and a strong +/- rails, 48V phantom and utility rails.

Fully independent of each other, it will  serve two units,  the first of which is .....

a  dual-mono line-amp-with-THD box

plus  a HPF/LPF network with single-ended traffo balanced output as well as DOA parallel out

Here's the start-up pics; 

- a pair of 1RU cheap racks (pcb assemblies inside)

and the preamp_eq unit as planned ....

Also amonsgt the things that are cool about this build are the front panel DI input jack and amp circuit.

It's a DI transformer coupled input to  high-power-opamp and traffo output stage (this mostly for THD).

Tube-mixed with the (traffo debalanced)  line input, fed to the EQ and then to the output amp(s).

Telefunken SE traffo at the output (this mostly for balanced output THD   ) with a parallel  opamp quasi-bal output (uber-convenience).

Tubes AND chips are so much fun     THIS one uses 

- a nice pentode+triode 6JW8  for the  eq amp (nfb bax affair)

- lovely little tiny pentode 5654 for the di+line mixer in a plate follower config

- skintanium armoured output triode 6S4A (25mA!) for the se telefunken traffo
          (from a nice old german desk top radio apparently  )

- ti lme49860 chip family (hidef into 600ohm loads  with EASE! +-22V supply max)

Awesome.

I've recently used all of these in other builds and are again my favs.

The telefunken output traffo is rated for around 3W (optimistic!)  in a 12K:4ohm package, se at around 25mA before Champ style globs of THD trouble set in.

I plan to use it at at 20mA or so in the primary and with a switched dummy load/speaker arrangement, as I do in my Champ and Deluxe style amp heads. With a moderate amount of variable nfb to clean up somewhat.

At  6ohms of secondary loading, primary  approaches 20K  on the triode, which is around 3K7 internal plate resistance.

The idea is when no speaker is connected, the output is variable padded and available for 600 ohms  at a reduced power level (say 1.5W dissipation down from 2.5W) - plenty to distribute a little if required.

When a speaker is connected, it allows a rudimentary 'monitor' facility  (think good quality radio!). 

That's the plan 

NOW if all goes well, this is the first time in a major piece that I will be free of the dreaded nearby-PSU horrors.

Get that thing away! 2m seems good. It will pay for itself in extra noise margin especially when using lots of transformers all over the place.

Never again will I suffer the indignity of a close proximity psu in a busy, tube'n'traffo based unit.  Hurrah!

Life's just way too short for that,  unless you have the nice materials and skill to engineer the issue way down.
Or just nfb-ing the living *$%#@ out of it to drop the hum to acceptable levels.

So - I'll keep you posted   

Hopefully by writing about it, I will motivate myself to push my boundaries somewhat.

AND Imma going to try keep this one neat and tidy like!

Also in use are  ...  'Mitsos All Stars'  high class amplet pcbs 

I'm really pleased to use them in these builds, which are meant to combine the versatility and hi-def capabilities of  modern opamp interfacing with the awesomeness of tubes and transformers.

Really interesting build.

I was contemplating a compressor much like this just this day,
only exception maybe is desired 1:3 ratio vs. limiter.
For use on masterbus, with possible tracking.

But, I'm not nearly as skilled as you,
so, can't jump in

Will watch closely though.

Yep - it's a must have, I think. 

Other units similar in nature would be the avalon unit. I think Gyraf Audio has something *high end* in that space 

Not too many other single piece stereo units  that come to mind, in the analog realm.

The finalizer of course is digital and fully comprehensive across the board.

-------

Here, each functional block is modest in it's impact but all together becomes very useful.

-  simple but effective EQ trims
-  low range of dB GR in each of the fast and slow gain cells
-  some program dependence on the ratio and attack

- simple controls on front, deeper controls on top
- use of local negative feedback at input and output stages

and of course, all the love you get from good old tubes and honest-to-goodness old transformers 

-----

Actually, my skills as a builder are actually fairly dismal     

I am more of a 'frankenstein's physician' type - ever ready to sacrifice quality and craftsmanship to satisfy my tinker's craze ..  for stitching together wild confabulations 

As prototypes, one off pieces, they make a kind of sense however!

I don't  sacrifice basic audio performance though, at all.  I measure a lot, and I use them in a variety of rooms once finished a fair bit too.

Each successive one has improved - hence the external power supply. I'm adamant I want that couple of dBs hum performance *back*!!!.

This will be #30 or there abouts. I still use #3,4 and 5 most often, now more than 12yrs old - also much functionality crammmed into single boxes. Based on the classics and reworked a time or two to get the most from them 

As I said, it is a mashup of parts of the 'Vacuum tube leveling amplifier' designed by E J Jurich and also the australian Trimax 'A30 Limiting Amplifier'.

Just to clarify the intended gain control cells incorporating my changes :

->  fast response vari-gm tube stage

sidechain
- feed-back style of control amp  ie. audio is sourced from the output stage
- two triode booster stages for each L-R sidechain signal
- half-wave tube diode detector for each L-R signal, input to anodes (positive going CV pulse)
- a form of 'true power sum' across L-R channels to feed the time constant net
- variable attack/release with a 'somewhat' program dependent release setting
- metering of CV signal amplitude to provide a 'fast GR' indication

gain amp
- single ended per channel but operated across L-R in a pair of triode stages
  (the phases are the left and right channels, not + and -  signal split style)

- cathode follower using principle of vari-gm to alter cathode impedances  (a la Trimax)

  ( the gr amp looks like a classic 'common plate mixing config' or a 'diff' amp or a 'push pull' amp)
  (BUT signal fed to cathodes, CV to grid. Additional resistive divider in cathode circuit to provide output signal  )
  (CV pulse is 'positive going'    )


-> slow response vactrol opto tube stage

sidechain
- feed-back style of control amp  ie. audio is sourced from the output stage, post booster amps
- half-wave tube diode detector for each L-R signal, input to anodes (positive going CV pulse)
- variable attack control
- cathode follower vactrol led driver stage for each independent L-R sidechain signal
- vactrol ldr configured as a variable resistance shunt across each L-R input signal
- metering of vactrol ldr to provide a 'slow GR'  indication

-------------------

Now all of this is a bit of an experiment - there is a lot of stuff here I've not tried before :

- unbalanced signal GR cell topology ?  wtf?  check
- cathode follower resistive divider network vari-gm amp? wtf? check
- half-wave detections with positive pulses  wtf? check
- weird time constant networks?  wtf?  check

I *have* done single-ended compression with half wave detection - in a champ style amp!
I was impressed though - a pair of 6BJ6 in series, each with  simple CV to grid and low dB gain reductions
It  sounded a treat - with care on the setup,  didn't have any real difficulty with CV punch thru.

So this will abound in (for me) off-the-beaten-track bits. I will prototype these sections before committing to metal !

If it goes a bit awry, I will revert to the balanced push pull CF in the vari-gm gain cell with full wave detection.
That will cost an extra two diode tubes and two  more dual triodes. Getting near to my heater budget limit!
Or think laterally - at some more ss 

-------------------

Another benefit of external psu is that one can get all that working first and use it for prototyping the signal box.

Then you can just concentrate on getting the best audio  'onto metal' without any other distractions.

Kind of a gradual modular approach to 'one off' building.

A previous nfg attempt's top plate is now in regular use as a prototyping plate  

That build beat me because of noise from psu into the limiter interstage - all close to a 6v6 pp output stage

SOunded fantastic, but the hum performance had me start again from scratch.
Not before learning a lot of new lessons, though.

One thing I have done for 'mastering' applications,  is used my own AlexC Audio  Dual RCA Style Tube Limiter  feeding my build of the  Analag/Silent Arts Dual Opto Tube Compressor

Fast vari-gm tube box feeding slow opto tube box - the results are very, very satisfying.

The RCA style (pentodes + utc  interstage with 6V6 pp+big edcor output ) I think stands out as the most luzurious and syruppy whilst still being fast and with a lot of detail. It's where you want to be when on your own time.

The Dual Opto style (triodes with jensens) is very transparent, very clean, very relaxed.
Very much the biz  when on someone else's time.

I'm going for characteristics of BOTH in this here project. 

Along with my build of the  Gyraf Dual Pultec Style Tube EQ I can do some lovely sweetening of the top end and boost the bottom end a little. All that is needed.

Now that is 5RU + 3RU + 3RU -> 11RU total for those  8)    (and maybe 30kg , 180VA)

Overall noise floor is pretty damn good considering all the stuff going on - very usable for sure.


------

This box is going to 'set that in metal'  ... 5RU total  (12kg and 80VA ..... excluding PSU)  ;D ;D 

AND improve the noise floor further by virtue of the seperated psu

AS WELL AS  generally benefiting from a simpler and  (more) minimal audio path

Pretty exciting to DIY 'without a net' and push my limits!

Now to be drawing the circuits please.

First up, the input amplifier ........


Input will be from one of my 'sum bus amp' boxes, so  something like 10Vpp (typical) to 20Vpp (max typical)

- transformer debalancing of each L-R to single ended with some utc  5K/10K transformers - nice and big
- dual rotary 10K (or so)  'input level'  for controlling L-R 'input level'  together.
- single gang potentiometer 'fine trim' for each, defeatable on a switch

- cathode follower stage for each L-R channel to buffer to low Z eq feed
- shunt ldr connection at grid of the input CF for the slow vactrol gain cell
- one half of a dual triode stage for each L-R channel
- 1x 6BZ7 tube 9pin

- simple baxandall active high and low band trims  in typical hifi amp style for each L-R channel
- cap switched setting of two break frequencies
- dual rotary hi gain to control both channels 'low gain' together, similarly for the 'hi gain'
- switchable defeat of eq for both channels together
- pentode+triode tube for each  L-R channel for low Z buffered output;  pentode for  bax nfb stage, triode for CF output
- 2x 6JW8 9pin

So all that will provide a fairly simple but nice eq'd, buffered, lowZ, unity-ish gain variable down -10dB with opto GR.

......  to feed the vari-gm dual CFs in the GR amp. 

I'm just firming up my view of the gain structure.

I like keep the gain changes as low as necessary (up or down) to get the job done. I'll know more when I get there!

Hence my deep respect for the opto GR principle. Especially in a first stage.

I'm going for cathode followers and unity-gain all over as a guiding principle     
(grain of salt required here)

Even flirting with the idea of a that vca block in there somewhere  ;D ;D ;D

rev1 circuit for the input amp to follow ....

Crazy project, really keen to see how this turns out. You say its based on a broadcast style processor, does this mean it incorporates some AGC style gain processing or is it more akin to compressor only?

Based on *two* broadcast style compressor/limiters.

Yep it has agc going on    Slow and fast. Together, in series like. You provide the 'smarts' however ;

I'm going to call it 'Esquilax' - a compressor born with the head of a limiter 

( or maybe the other way around ...)

So a slight change ....  going to use balanced vari-gm gain cell after all 

I bowed to my inner critic on this and have decided to 'damn the torpedoes' and go all the way. It is, after all, a 'fine-alogizer'

SO the project is now .. with rough gain staging indicated. Rough like.


Input Amp - overall gain staging is unity -ish on the max rotary attenuation and max vactrol GR

- bal xlr to 5K/10K input traffo debalancer  (+2.5dB)
- simple single ended 10K rotary attenuator  (0 .. -12dB)
- triode CF with vactrol ldr shunt gain reduction (from EJJurich) (0 .. -10dB)
- active Bax eq trim (from Angefire EQ) (unity gain when flat)
- single ended gain stage (+22dB)
- push-pull CF driver (from the Doac)  (-1dB)
- cap coupled to primary of first interstage 15K/15K  traffo (-0.5dB)

Gain Cell VariGm - overall gain staging is unity -ish with no GR down to -10dB max on the vari-gm GR

- balanced CF with resistive divider providing cathode outputs (from the Trimax)  (unity .. -10dB)
- plate loads providing side-chain voltage gain (from EJ Jurich)  ie. sidechains' balanced send from here*
- balanced drive primary of second interstage 15K/15K traffo (-0.5dB)

* sidechain feed arrangement may change!

Output Amp  - overall gain  +6dB down to unity on the rotary attenuator

- single ended debalancer  (-0.5dB)
- simple rotary attenuator (0 .. -6dB)
- gain stage (+23dB)
- single ended triode driver (+5dB)
- cap coupled  to primary 20K/600 traffo to balanced xlr  (-15dB)

SideChain - tbd .  gone fishin' 

- balanced, with tube voltage amp (for vactrol  and  vari-gm control)
- balanced current amps * (for each of vactrol and vari-gm control)
- fullwave detector #1 for vactrol led, with threshold, attack control of TC  (from EJ Jurich) -- may be ss diodes
- fullwave detector #2 for vari-gm,  with threshold, attack, release control of TC  (from Trimax) -- good ol' 6AL5
- metering #1, #2 (adapted from EJ  Jurich as well as the Trimax    )

* sidechain current stage (for feeding the vari-gm gain cell control amp time constant network) may be tube or mosfet

Phew!  So that's more conventional whilst retaining ideas from all round!

Now to make the power budgets work    IF POSSIBLE!

Here's my thinking for the input amp

.. all pretty straight forward really 


- resistor values etc are ball park and not yet calculated
- nfb for the bax may not be quite right either

The vari-gm gain cell is simple and uses a balanced config

- signal to cathodes resistive divider circuit, vari-gm CV to grids
- output from cathodes
- plates have some loading to provide some voltage gain for sidechain

Now the gain reduction principle  is this :

- idle current is set low    ie. grid-cathode voltage difference is larg-ish  negative
- CV pulse is 0 .. +15Vdc or so
- varying  CV  drives the current from low (min GR) to high (max GR)
- resistive divider in the cathode circuit is in series with tube Zout provides the gain-reduced output voltage signal

Now for a CF,  Zout approx = 1/gm  so  as gm varies from high  (high current, low Zout) to low (low current, high Zout) the 'potentiometer action' of the resistances  in the cathode circuit vary the Vout down from Vin.

Because the cathode circuit resistances are generally  low and the change in Zout of each CF is also low, the 'freq' distortion (as Trimax puts it) is  .. 'very'  ..  low. They go on to say (that) is also  'the problem with most potentiometers'.

Not sure what they mean here - I think it means that because all the impedances are low in the CF , the variation in the Vout swings are also low,  so the tolerance of the resistances (1/2W types, 5K and 300R each leg) in play here do not exacerbate
(or amplify if you will)  to significance, the 'errors' in the resulting output voltage waveforms (relative to the input signal!).

ie. imbalances at the cathodes are less than with  large plate resistor loaded anodes.

Hence the claim to 'largely distortion free'.

For sure, CFs are regarded as being  'inherently low distortion'.  More or less.

Cool!

The downside is apparently, limited GR range. As to the rest, only prototyping will tell.  8)

------

Hence  the gain change here is going to be modest!  -10dB or so.

The orginal Trimax A30 uses a 6SN7 tube (similar AU7).
This has a gm range around 3400 umhos to 100 umhos or so.

That is roughly a low Zout of 300ohms and a high Zout of 10Kohms or so.
The cathode circuit resistances are 5K+300R each leg and a small 33R 'long tail' resistor

In effect, the cathodes' circuit is a  sort of a 'bridge arrangement' with a reference voltage at the bottom and the cathodes' voltage at the top.

With fixed resistances (300R) across the bottom and a divider across the top.  The top divider has a fixed part (5K) and a varying part (the cathode). Inputs, outputs across the phases.

------

I plan to (again!) use a pair 6BJ6 - a nice little remote pentode, low heater, big current, big gm range (from 3600umhos down 3 decades).

The CF outputs of the gain cell are going to feed the low-ish power rated  15K/15K interstage traffo directly at what I consider to be a decent idle current of 5mA each side to maybe 15dBm max and down.

CF outputs are fairly low-ish Zout so I hope they can drive adequately the 15K primary with good bottom end.

The proposition is ..  most of the time, the Zout of the gain cell is low-ish, so the bass response will be best then.
Some of the time .. not all ...  it will rise. Quite a bit really! and the freq response will suffer a bit.  When compressing.
It's the opposite proposition to the regular vari-gm plate loaded gain cells.

The cell has at least one differential stage to absorb the inevitable (hopefully less dramatic) common mode whump - the transformer primary.  My first traffo choice here is not too big  :'(

So the whumpus to be swallowed needs to be significantly smaller than my 'other' limiters can  (and do!) handle without hiccup.

ps - I should note that the sidechain signal will have a fixed high-order filter to mitigate bass heavy program material from dominating the vari-gm sidechain action. 

----

Other side  of the (first interstage) traffo is likely to be single ended  recovery amp, driving a line amp with cap-coupled stepdown traffo.

CV pulse  0 ..+15V will slam up the current from 5mA to 18mA each side - down the cathode followers' towards' ground.

So cell idle goes to max GR . Gm goes from 3.6umhos to 3600umhos.
CF cathode Zout is the inverse, so 278K ohms to  278ohms

Good news for gain-cells that sit idling a lot of their lives! Low current for liesurely living
High current huffing only on the loudest passages!

In practice I expect it to burble along at a max of 10dB, so the Zout range could be 2K8 to 28K or so.

Even less,  typically - perhaps 6dB - maybe 5K to 25K ohms.

Whatever the (second) interstage traffo can handle without walking out.
My first pick is a little wimpy - but it may surprise on the upside. Else goodly $$ for better!

(as falls witchita falls so witchita falls falls).

Schemo of the gain cell to follow  ....

Here is the schem for the vari-gm gaincell.

It is an adaption of the Trimax A30 and EJ Jurich gain cells.

A  bit like a pair of split phase inverters (like in the Deluxe amp), but in 'grounded grid' config. 


- I haven't calculated it all yet but it's basically there.

I might be able to get away without the input and output caps, depending on how the traffo goes.

It may be necessary to add a differential  gain stage before the output transformer, depending on how much common mode whump I get.

The output amp will need quite a bit of gain, to recover the level after the vari-gm gain cell as well needing 15dB for the stepdown in the 20K:600 output traffo.

I'm using cap coupling to the traffo to keep dc out of the primary (she don't like that!).

Looks like a grounded cathode gain stage on a high u triode,  followed by another on a low u triode for around
With a little local nfb I think. Probably 1/2 of a AV7 type + 6S4A power triode.

Other possibilities are a u-follower or srpp with perhaps  a 6U8A or even 6GW8 triode+power pentode tube - just to be different!

Schematic to follow ...

Now to do the heater budget - all 6.3V and I'd like to keep a max of 6A all up.

So the heaters are likely to be

- input amp :                0.45A  (6JW8) + 0.3A (AT7) -> 0.75A x2 -> 1.5A total

- vari-gm gain cell :  0.15A (6BJ8) x2 -> 0.3Ax2 -> 0.6A total

- output amp :            1/2 x 0.3A (AY7) + 0.6A (6S4A) -> 0.75A x 2 -> 1.5A total

Thats 3.6A all up excluding side chain.

sidechain vactrol led driver is an AU7 shared across channels, so 0.3A there.
sidechain voltage gain is AT7 per channel, so 0.3A x 2 -> 0.6A there
sidechain full wave detector is 6AL5 per channel, so 0.3A x 2 -> 0.6A there
sidechain current gain is AU7 per channel, so 0.3A x 2 -> 0.6A

Total sidechain is 2.1A

All up is 1.5A (input amp), 1.5A (output amp), 0.6A (gain cell), 2.1A (sidechain) -> 5.7A

That's my starting point - I'd like to try to use only 5A if possible
(so now to think on some heater current reducing tweaks.

HV should be something like 15mA (input amp), 20mA (gain cell), 25mA (output amp) -> 60mA x2 -> 120mA for signal amps.
And around 25mA x2 -> 50mA for sidechain. Total HV around 170mA.

Now to scope out some Edcor PSU iron. I'm going for SS regulation of the HV B+, using a series pass mosfet, zener reference  and a BJT based error amp. Hence no rectifier tube or supply choke needed here. All that of course, in a seperate power supply unit connected by cable.

I like the Edcor that I've used a number of times now, XPWR063 : 250V @ 200mA, 9V@6A, 30V CT @ 1A.
I may decide that I can't drop the heater to 5A and so select a different one.

Possibilities are to use 12.6V sidechain tubes and run off each half of the 30V ct winding at around 500mA each.
That would cut the 6.3V heater requirement to 3.6A

PS - oh yeah! Forgot! Since this one is using cathode followers (CF) all over the place, it means ac heaters.
          Needs to have elevated heaters

OR select tubes with  biggest heater-cath voltage rating, finagle the voltages to minimise cathode voltage as far as possible AND hope for the best.

So - looks like ac heaters. I have used them that way many a time with a very low noise floor so not really a problem.

Also remembered that here is a good opportunity to use a mu-follower stage as the output amp line driver.

Some good gain available (approx = u    ) quite a low output impedance for driving the 20K/600 output traffo.
Pentode top (highish current and gm), triode bottom (mid range u, bit less power here)

Sounds perfect for a pentode+triode tube     I like the 6u8a here at at 3.0W for the pentode max and 2.5 W triode.

I used it in my big 15" box+head, parallel SE tube amp driving a humble xsm15K:600 to provide a traffo bal Line Out.
Sound is sourced at the MAIN traffo outs and uses a 2 position 'sensitivity' switch (for when it gets too LOUD!)

Absolutely glorious there, 'hifi with some crunch' - super quiet  8) 8) 8) 
It's my studio darling amp not afraid of the big jobs.  Now thats a  pentode+triode cathode follower.

So - first time use of a mu stage (based on Alan Kimmel paper) for the main outs 

6u8a it is then.  Alright!

Looking again at the vari-gm grounded grid stage, there is likely to be a problem with the input impedance.

Grounded grid gain stage input impedance is quite low.  For a triode stage, with cathode resistor  'terminated to ground'

Zin = [ (Ra+rp) / (u + 1)  ] || Rk      With  a 6SN7 in a textbook config, that calculates to around    640ohms || 5K83  -> 576ohms

According to Broskie of TubeCAD :

- if the cathode terminates into the signal input instead of ground, then it becomes  Zin = [ (Ra + rp) / (u+1) ] + Rk 

I'm not really sure - I'll need to experiment with the prototype.  I'll be using a pentode, in pentode-mode , so it will differ.but I think I will need a low Z transformer secondary.

EJ Jurich uses his version driven from the plate of an AX7 (with local nfb) + 47K in series with the signal input to the caths, which are 1K to ground each leg.  Now that is bound to be a high-ish source impedance.

The Trimax uses a source transformer (which I don't have data for) plus a series delay line  (common mode choke + cross-fed caps) to drive the cathodes directly. Who knows what that source Z is ? I would think high-ish again.

So,  I'll need to experiment on the prototype!

-----------------

SO I have what I think is ideal for the vari-gm CFs gain cell : Jensen JT-6110K-B 

It has primary of around 37K5 impedance (750ohms dc), and primary 2K34 impedance (50ohms dc).
It's a chunky sucker, very satisfying to hold. As one would expect, it is said to have exceptionally low THD at +24dB for 20Hz and so on.

It can be used with a 2K21 resistor loading to present around 1K as a source impedance (apparently 87ohms dc? as the data sheet states) - a good match for driving an opamp or a low Zin gain cell.

I have a bunch I use as 1:4 stepups in my la2a and doac builds. I've been meaning to rework the dual la2a for lower line amp gain, so that will be a good excuse to replace with some jensen 10K/10K line inputs and rejig for AY7 in the amp.

)

SO - game on (again)!  At present, the transformer choices stand as :

- utc  LH519                  5K/10K        (1:1.4 inputs )
- jensen JT-6110KB  37K5/2K4  (4:1 first interstages)
- altec 15335A              15K/15K      (1:1 second interstages)
- rca MI-11712            20K/600      (5.8:1 outputs)

That's my thinking so far.