Fine tuning an EF86 with trimmer resistors for Ra and Rg2

[PharmRock]

Currently using Ra=220K and Rg2=1M as most all Vox-inspired EF86 layouts use, with a typical ratio of 1:5 for Ra:Rg2.   In reading Merlin's chapter on designing an EF86 preamp stage (much of which is not on his website), he discusses the process by which one selects the desired amount of gain, calculates Ra and Rg2 while considering HT, and arriving at certain values. 
On another forum, I was reading a post from over 15 years ago where it was suggested a Rg2 value equal to Ra be used, but put a 1M pot in series with the fixed Rg2 to "dial in" the actual screen grid resistance in order to zero in on the desired tone/gain/compression, then "set it and forget it".  So if Ra=220K, Rg2 could be anywhere between 220K and 1.22M.
I was thinking of taking this approach, as well as plscing a trimmer pot in series with Ra, where for example there might be a fixed 50K Ra resistor in series with a 250K trimmer pot.  Using the dual-trimmer pot approach I could (theoretically at least) dial in the desired amount of gain etc. Its likely I would arrive at 220K+1M, 100K+470K, or maybe something that doesn't reflect the typical 1:5 Ra:Rg2 ratio but sounds a lot better than stock values.

Here's an example of a trimmer pot I could use: https://www.mouser.com/ProductDetail/BI-Technologies-TT-Electronics/37SR500KLF?qs=1qzrMKUcZ2gWLELO%2F%252B9UoQ%3D%3D
It is rated for 250V, 0.5W.  Since my HT at that stage is 300V, I figured if I placed the trimmer pot after the "fixed" resistor dropped voltage I should be ok. 

Anyways...this is a bunch of rambling....hope this makes sense.  Is this something worth looking into?  Has anyone else tried this variable Rg2 approach with any success?  Or should I stick with the 220K/1M set-up and spend my time on more proven approaches looking for other ways to tweak the circuit?

Thanks!

[sluckey]

Go for it. Tweakers love to tweak. When dialed in you may end up real close to Ra=220K and Rg2=1M.   

[PharmRock]

  that's exactly what I was thinking.  But it will be fun trying it out. I'm going to figure out the MOSFET SF this weekend and this will be next on the to-do list.

[HotBluePlates]

... Rg2 to "dial in" the actual screen grid resistance in order to zero in on the desired tone/gain/compression, then "set it and forget it".  ...  Has anyone else tried this variable Rg2 approach with any success?  ...

Look closely at the Vibrato Amplitude control in the Vox AC10.  It adjusts the screen voltage of the EF86.

The best feature of my 1965 AC10 is its ability to adjust the EF86 for more grit by turning off the tremolo, and turning the Amplitude control to max.

[kagliostro]

Just to talk



Franco

[PharmRock]

Franco, thanks for posting that image/table.  Very interesting values on the last 3 using the 330K Ra and 2.2M Rg2.  I came across this link of a circuit analysis of the DC30 EF86 and found it very interesting and insightful: https://www.ampbooks.com/mobile/classic-circuits/matchless-pentode/

Look closely at the Vibrato Amplitude control in the Vox AC10.  It adjusts the screen voltage of the EF86.

The best feature of my 1965 AC10 is its ability to adjust the EF86 for more grit by turning off the tremolo, and turning the Amplitude control to max.

Thanks HBP...you had mentioned this before when I was exploring ideas for the Pentode-Triode Morph Control.  Is there an advantage with how Vox did this using a voltage divider (if I'm interpreting the schematic correctly) vs. the fixed + variable resistor in series prior to the screen?  I have the morph control currently installed, and may temporarily convert that to a variable screen voltage control by moving a couple of wires around.  The morph control is a 1M linear pot...I could replace the 1M Rg2 on the board with a 220K (or so). 

[kagliostro]

Many thanks for the link


Franco

[HotBluePlates]

Look closely at the Vibrato Amplitude control in the Vox AC10.  It adjusts the screen voltage of the EF86.

The best feature of my 1965 AC10 is its ability to adjust the EF86 for more grit by turning off the tremolo, and turning the Amplitude control to max.

Thanks HBP...you had mentioned this before when I was exploring ideas for the Pentode-Triode Morph Control.  Is there an advantage with how Vox did this using a voltage divider ...

The voltage divider changes the screen voltage.

A different screen voltage changes the gain of the EF86.

I've found turning the Amplitude control to max brings the screen to up to ~70v, pulls the plate down to ~70v, and gives the most grit & gain.

Because the screen volts went up, this means the wiper of the Amplitude control went to ground leaving a voltage divider of Power Supply --> 750kΩ ---> Screen ---> 650kΩ total resistance.  I suggest it is very-worth trying out, though I wouldn't try to drive any tone stack directly from it.  Additionally, note the 470kΩ plate resistor.

[_Steve]

Because the screen volts went up, this means the wiper of the Amplitude control went to ground leaving a voltage divider of Power Supply --> 750kΩ ---> Screen ---> 650kΩ total resistance.  I suggest it is very-worth trying out, though I wouldn't try to drive any tone stack directly from it.  Additionally, note the 470kΩ plate resistor.

Learning a lot from this thread thanks! However can you elaborate a little on why you wouldn't try to drive a tone stack from it? That was exactly what I was planning on doing lol

*EDIT* oops looks like this was explained in the other thread about impedance :)

[tubeswell]

Because the screen volts went up, this means the wiper of the Amplitude control went to ground leaving a voltage divider of Power Supply --> 750kΩ ---> Screen ---> 650kΩ total resistance.  I suggest it is very-worth trying out, though I wouldn't try to drive any tone stack directly from it.  Additionally, note the 470kΩ plate resistor.

Learning a lot from this thread thanks! However can you elaborate a little on why you wouldn't try to drive a tone stack from it? That was exactly what I was planning on doing lol

*EDIT* oops looks like this was explained in the other thread about impedance :)

Output Impedance of a pentode:
Zout = Ra.ra/(Ra + ra),
and ra is 2M5 in an EF86: https://frank.pocnet.net/sheets/020/e/EF86.pdf,


so for a 220k plate resistor
Zout = 220,000 x 2,500,000/(220,000 + 2,500,000) = 202k


So to preserve a reasonable modicum of signal, the following load (incl Tone Stack) wants to be a minimum of 1M, which is possible with a simple Tone Stack like a treble cut (e.g. 2n2 cap in series with 1M pot).

[_Steve]

Output Impedance of a pentode:
Zout = Ra.ra/(Ra + ra),
and ra is 2M5 in an EF86: https://frank.pocnet.net/sheets/020/e/EF86.pdf,


so for a 220k plate resistor
Zout = 220,000 x 2,500,000/(220,000 + 2,500,000) = 202k


So to preserve a reasonable modicum of signal, the following load (incl Tone Stack) wants to be a minimum of 1M, which is possible with a simple Tone Stack like a treble cut (e.g. 2n2 cap in series with 1M pot).

Thanks! I guess it really needs a buffer between the EF86 and tone stack? Im going to re-read Merlins chapter on CFs. Maybe I could even make it DC-coupled EF86->12AX7   

Anyway, dont mean to hijack the thread!

[kagliostro]

.... I guess it really needs a buffer between the EF86 and tone stack?

To have a sure response give a look to ALL the schematic (of guitar amp) that you find with EF86 as V1

How many  has a Tone Stack that follows it ?

I mean Tone Stack, not a simple Tone Control

Franco

[PharmRock]

Attached is the EF86 channel on the 18-watt clone I just built.  The volume control wiper goes to the PI.
With both volumes on "10", it seems I have less overall volume on the EF86 channel compared to the tremolo channel, which uses a 12AX7 with a 500K volume with single 500K tone control.

the tone control as shown on the EF86 is lifted from a Carmen Ghia.  Is there anything in particular with this tone circuit that makes it a bad match for an EF86?  Should I be using 1M pots? While not labelled, the morph control is a 1M linear pot. 

Sluckey has recommended that I simply bypass the tone control for diagnostic purposes, which I plan to do but have not yet done.  I'm just wondering if there is something about this combination of EF86 + Carmen Ghia tone control, from a design perspective, that makes you think "huge signal loss" just by looking at it.

[tubeswell]

Attached is the EF86 channel on the 18-watt clone I just built.  The volume control wiper goes to the PI. ... the tone control as shown on the EF86 is lifted from a Carmen Ghia.  Is there anything in particular with this tone circuit that makes it a bad match for an EF86?  Should I be using 1M pots? While not labelled, the morph control is a 1M linear pot.  ... from a design perspective, that makes you think "huge signal loss" just by looking at it.

So assuming the Morph control is at minimum rotation and the Tone and Vol Controls are at full rotation, and ignoring the frequency filtering of the caps in the circuit, the 202k output impedance from the EF86 will be driving a load comprised of the 1M Morph control pot, which is in parallel with the 500k Tone pot in parallel with the 500k Vol Pot in parallel with the 180k||180k, so

1/1000k + 1/500k + 1/500k + 1/180k + 1/180k = 1/62k

So Vout (at the wiper of the Vol Pot) with Source of 202k and Load of 62k = 62k/(202k + 62k) = 23.5% of Vin (at the EF86 plate) with all controls set to allow maximum signal

The picture is a little more complicated that this because of the filtering effect of the various capacitors in the circuit, but overall attenuation is key to understanding.


Of course, you can 'even the odds' by using a current buffer (like a tube CF stage, or a FET source follower) between the EF86 and the following load. This would dramatically lower the source impedance for driving the load

[kagliostro]

Alternatives you can give a try

A further triode as recovery stage

A RAW control on the Tone Control (a pot connected as rehostat between the 0.01 cap of the Tone Control and ground) this permit to decrease the effect of the Tone Control as signal loss (but decreases the effect of It on tone, something between bypass the Tone Control and maintain It)

To me the advantage of the CF Is you can use a Source Follower (It maintains the "Tone" of a Tube without the necessity to feed further heaters), obviously also a recovery gain stage can be arranged in a SS way, but usually this Is avoided

Franco

[Williamblake]

I was messing around with an EF86 and having variable bypas for the screens gave some nice gain and tone conrol. I ditched the idea but planned on experimenting with this some more.

[PharmRock]

Ok this is starting to make sense to me now.  Thanks Tubeswell.

I found a "reverse engineered" layout/schematic of a Dr. Z Stangray, which uses an EF86 as the input matched to a Carmen Ghia tone control, except there are some changes that may make it work better in this application.  Assuming I removed the morph control from my circuit, here's what I'd have in my circuit using Tubeswell's example:
1/500K + 1/500K + 1/180K + 1/180K = 1/66K
66k/(202K+66K) = 25%

In addition to some changes in the capacitor values, the StangRay uses a 1M Tone and 1M Volume, and in place of the 180K resistors, it uses 330K:
1/1000 1/1000 + 1/330 + 1/330 = 1/124K
124K/(202K+124K) = 38%

This would be an increase of about 50% over what I currently have (minus Morph).

I have the parts for the MOSFET cathode follower but didn't get around to it this past weekend.  Will be out of town a couple of weeks but plan on implementing that at some point. 

Thanks Franco for the suggestion of the Raw control.  I think I've seen something similar on TMB circuits but have never tried it.  I believe it was a switch that lifts the Mid pot from ground. 

[kagliostro]

Using a switch will allow to exclude the Tone Control


Using a rehostat give the ability to exclude the Tone Control only partially, in % respect an abruptly ON/OFF


Franco

[tubeswell]

suggestion of the Raw control.  I think I've seen something similar on TMB circuits but have never tried it.  I believe it was a switch that lifts the Mid pot from ground.

Another possibility is a 1M resistor between the tone stack and the ground return, which you short with a SPST (to make the TS work). 1M is sufficient for effectively eliminating the tone stack load. The shorting switch minimises switch pop.

[kagliostro]

Just to complete my previous post



Franco