Baxandall puzzle

[abbey road d enfer]

In most cases, the biggest load will not be the EQ components, it will be the actual load to which the circuit is connected. That's what really motivates the existence of the output CF.

I am not 100% sure of this. Referring to the last schematic I posted with a 50K pot and 5K resistor in series at the cut side; when maximum cut is applied, the feedback impedance tends to 5K. Assume the reactive part of the feedback is also 5K. If the virtual earth is working then the load on the output is 10K. At maximum boost it drops to 55K or so. A CF will be fairly happy with these changes in load. A CC 12AX7 will not.  I know I have simplified the issue a lot but I think it is clear the load the stage sees varies a lot. Hence my comment about scaling component values. I'll do some sims to check this out.

Cheers

Ian

I made a sim with the values published by peter J.. 50k pots is too low for vacuum tubes.

I simulated some variants of the original.
The center-tapped HF pot does not seem to be nice. In fact it makes the variation faster around neutral, which may counter the idea of making the neutral position flat.
See attached: graph with center-tapped pot

 

[abbey road d enfer]

Graph LF with standard pot (not center-tapped)

 

[abbey road d enfer]

The next graphs depict the difference between the original arrangement with two caps across the LF pot, joined at the wiper.
Graph shows response with 2x4.7nF as presented in the original paper.
One can see that the corner frequency varies very significantly with the amount of boost/cut.

 

[abbey road d enfer]

This graph shows the response with only one cap across the LF pot.
The corner frequency doesn't change much but the slope does.
The corner frequency is too high compared with most of the previous curves.

 

[abbey road d enfer]

This one matches better the original curves at moderate boost/cut. The single cap across the LF pot has been doubled 10nF)

 

[ruffrecords]

@abbey

Thanks for doing all those sims with original values. In the text of the article, PJ said you could use sources with an impedance up to 10K. Can you check your sims and see how it is affected by changes in source impedance between 0 and 10K?

Cheers

Ian

 

[ricardo]

The center-tapped HF pot does not seem to be nice. In fact it makes the variation faster around neutral, which may counter the idea of making the neutral position flat.
See attached: graph with center-tapped pot

The centre-tappped HF pot isn't to make neutral position 'flatter'.

It's to convert a shelving EQ to 'sliding turnover' which GG Baxanadall considered superior.  He shows curves for both in his original Fig 8.

IMveryHO, shelving is more useful .. especially for EQ
____________

But I second his recommendation for EF37A instead of da new-fangled EF86 and ECC83/12AX7 rubbish.  ;D

 

[abbey road d enfer]

The center-tapped HF pot does not seem to be nice. In fact it makes the variation faster around neutral, which may counter the idea of making the neutral position flat.
See attached: graph with center-tapped pot

The centre-tappped HF pot isn't to make neutral position 'flatter'.

It's to convert a shelving EQ to 'sliding turnover' which GG Baxanadall considered superior.  He shows curves for both in his original Fig 8.

These dotted curves are quite difficult to discern and I must admit the notion escaped me, probably due to language.
I understand the motivation in an era of discs with surface noise and AM broadcast.
But I think the alternative implementation with two caps, one at each end of the Treble pot, is even better since it has higher slope and does not require an exotic pot.

IMveryHO, shelving is more useful .. especially for EQ

I totally agree.

 

[abbey road d enfer]

@abbey

Thanks for doing all those sims with original values. In the text of the article, PJ said you could use sources with an impedance up to 10K. Can you check your sims and see how it is affected by changes in source impedance between 0 and 10K?

Cheers

Ian

0.3dB variation for 1k, 1dB for 10k source.
I've posted the Tube baxendall_12AX7.asc file to your directory on the LTspice group, so you can play all your content

 

[ruffrecords]

@abbey

Thanks for doing all those sims with original values. In the text of the article, PJ said you could use sources with an impedance up to 10K. Can you check your sims and see how it is affected by changes in source impedance between 0 and 10K?

Cheers

Ian

0.3dB variation for 1k, 1dB for 10k source.
I've posted the Tube baxendall_12AX7.asc file to your directory on the LTspice group, so you can play all your content

Many thanks

Cheers

Ian

 

[ricardo]

I've posted the Tube baxendall_12AX7.asc file to your directory on the LTspice group, so you can play all your content

Abbey & Ian, where is this treasure trove? 

 

[abbey road d enfer]

I've posted the Tube baxendall_12AX7.asc file to your directory on the LTspice group, so you can play all your content

Abbey & Ian, where is this treasure trove? 

https://groups.yahoo.com/neo/groups/LTspice/info
You have to register to access the Files directory, then Temp where there's a RuffRecords file.

 

[PRR]

Beating a dead horse because I happened to trip over a Clue:

> with zero input impedance all is fine. ... again with the source impedance set to 2500 ohms. Result disaster
> started designing tube mic pres over ten years ago

OK, I thought you had more depth. Anyway there's always lurkers.

Here is the horse's mouth. An extract from Baxandall's Jan 1955 article. See Attachment.

The core control was disclosed in WW Oct 1952 but without context.

So on to the 1955 which is a complete preamp.

Note the serious symmetry between V3 and V4. Identical pentodes with identical components, both under serious negative feedback. 68K plate load so Zout would be say 63K, but feedback around V3 causes noise-gain about 11 with open-loop gain near 100, about 9:1 reduction, 63K/9 is about 7K effective node impedance. V4 works much the same, more feedback, possibly <1K node impedance flat, 10K boosted. Since network values are generally >100K, from 10K to 1K is "negligible".

When nodes are non-negligible, response is skewed, much harmony is lost. The 1952 article does point-out "The output impedance of the circuit feeding the tone-control circuit should be reasonably low, preferably not more than about 10K if the tone-control elements have the values given in Fig. 6."

Both issues of WW may be found at:
http://www.americanradiohistory.com/Wireless_World_Magazine.htm

Historical note: the 1955 article is 60 years old but only 40 years since the availability of electronic amplifiers. For more than half our history the Bax has dominated basic boost/cut tone controls.

 

[abbey road d enfer]

The simplified Baxandall equations are those of an inverting amplifier. Indeed, the discrepancy between source and output Z will be more noticeable at either end of the pot (full boost or full cut).

 

[Tubetec]

Thanks for all the replies on the stabilised screen supply post .
I thought it best to revive an old post with some great contributions to continue the Bax discussion here.

So whiIe trying to reduce residual noise in my Bax curcuit I tried  including the smoothing choke in the anode load circuit,initially with the screen supply tapped off before the choke ,bypassed to cathode in the usual way ,then I saw the possibility of UL and  triode mode as well  .Maybe a  variable screen bias regulator like  in Merlins valve curve tracer could also be used with ac coupling of the signal voltages on triode and UL connection or coupling cap connected to cathode in the case of pentode mode. That might also prevent any loud thumps and bumps compared to switching the contacts at dc as well as allowing the screen volts to be adjusted in either triode, UL or pentode modes  , what I dont understand is how I might allow ac feedback via a cap, say from plate to screen without the bias regulator eating up the signal .

Aside from the frequency dependent nature of the feedback caused by lack of inductance of the choke its still goes some way to increasing the impedence seen at high frequencies and theres an increase in the available boost as a result .Choke loading of the output of the C3g also permits higher output voltage swings and lower effective output impedences. Should I expect a frequency response that slowly rises in the bass due to less feedback due to the drop in inductive reactance as frequency falls or will the very high over all feedback keep everything linear  anyway ?  What would be a more appropriate value of inductance ?

Thanks for the 1955 version of the bax you posted before  Prr, I knew I had seen it before but hadnt located it in years .

Apologies for the poor contrast in the sketch below ,hopefully it gets the idea across .

           



The choke in the load even though just 1Hy does add to the Ac load seen by the valve especially at higher frequencies ,

 

[abbey road d enfer]

What would be a more appropriate value of inductance ?

It depends on what you expect.
With the actual values, the choke's inductance starts to kick in at about 1.5kHz; if you want to shift that lower you have to significantly increase the inductance. Typically, low-level plate chokes are 100+H. They're difficult to make, with sectioned windings for minimizing stray capacitance, thus expensive.

 

[Tubetec]

I see suitable chokes on Sowters site , yeah costly  :'(

I dont need the open loop gain increase the choke action gives the circuit on the low end per se ,I can see how a superior component would do a better job though.If it means a futher reduction in distortion  at higher frequencies and an increase in available output voltage swing  I'll take that gladly and give the little millitary surplus 1Hy choke a reason to live .In the low, the full boost of 20+db isnt often called for, maybe we can afford a bit more distortion in the low  to bring out subtle harmonics on bass heavy material ,the balance between 2nd-3rd-and higher harmonics could be changed by two mode switches,one switch setting  either self biased dc coupled screen or  regulated  adjustable screen bias with ac coupling to the other electrodes ,then another switch  for  triode,UL,pentode mode, you could add a range of flavours of distortion to the sound  ,as subtle or full-on as you like.

I'd like to try roll my own choke if possible , I found some small amorphous c cores with dual long narrow bobbins ,doesnt look like a major job to wind two or more sections side by side on each bobbin and use it either as a centre tapped choke in push pull or all sections  in series for single ended ,in conjunction with the correct gap .Probably all do-able with my handcranked coil winder, I can grab a Sowter ac coupled mumetal line output transformer later if needs be

 

[guavatone]

Interesting tune the c3g postal tube

https://www.diyaudio.com/forums/attachments/tubes-valves/399482d1392241034-c3g-tubes-c3g-c3m-info-pdf

https://www.diyaudio.com/forums/tubes-valves/251095-c3g-tubes.html

To follow from the other post. I read that keeping screen voltages at a moderate level would help with extending tube life. Since these babies are rare, I would definitely do Zener on the screens.


I think you’re asking for trouble to combine choke load and triode-pentode-UL switch. Unless it’s for experimentation. Pentodes want a high inductance choke because of internal resistance is 300K and triode IIRC is like 10Kmfor that tube maybe less.  But some depletion MOSFETS May do well for the  Plate.

 

[merlin]

I'd like to try roll my own choke if possible

You could try using a small power transformer primary as a choke (wired for 120V, all secondaries left open circuit). As long as the DC current is not more than a couple of milliamps, you'll get 20H or more inductance at 50Hz. Who know what at higher frequencies, but it's a fun and cheap experiment.

 

[Tubetec]

Thanks Guavatone , Ill check out those links. The C3g is a high spec 10,000 hours+ tube , I am running it at more or less maximum current ,plate and screen volts ,still it should last many many  years

Good idea Merlin , I have dismantled a couple of small mains transformers just for this purpose , in fact what I was able to do is get two identical primary winding sections 2000 ohms each an mount them both on the same transformer core ,so I have a dual bobbin choke to play around with , Im guessing in this instance butt stacked lams are best with some small air gap between E and I sections I can get around 50 HY , only issue is these small transformers dont have a metal frame to hold it back together or mount it , still as you say a great way to experiment with choke loaded amplifiers without any real cost associated . Even with small chokes not purpose built for anode load  in combination with resistor at the anode end  I can increase the loading on many small signal stages , and they seem to sound better as a  result ,more gain/headroom, less distortion etc.Im drawing around 16 ma in the c3g anode tube so a larger core would probably be needed to get tens of Hy's without saturating .