PA with el84 conversion to el84 + 6v6 - PI resistor values - I'm confused about

[kagliostro]

I'm confused about the values to give to the resistors of the PI

I'm planning to build a Vox AC30 based amp but want to use 2 x el84 + 2 x 6v6 tubes, instead of 4 x el84 tubes

I doubt about the values of the resistors in the PI

as can see there are different values in amps like MesaBoogie Blue Angel - Vox AC30- 36W

If I'm not wrong el84 has more sensibility than their cousins 6v6, about that I've just read this about a 18w 6v6 conversion

A couple of resistor values need to be changed in the PI to get the 6V6s singing.

=========

I'm not just saying this with my own experience, but of the many others who have also modded
their 18Ws to run 6V6s. The stock 18W PI was designed for the higher gain of EL84s, and
fails to fully drive the 6V6s.

in the BlueAngel the values are:

anode n.1 82k
grid to connection point between cathode resistors 1M
anode n.2 100k
grid to connection point between cathode resistors 1M
cathode 1.2k + 10k + 4.7k


in the AC30 the values are:

anode n.1 100k
grid to connection point brtween cathode resistors 1M
anode n.2 100k
grid to connection point brtween cathode resistors 1M
cathode 1.2k + 47k


in the 36w the values are:

anode n.1 68k
grid to connection point between cathode resistors 330k
anode n.2 82k
grid to connection point between cathode resistors 330k
cathode 820 + 56k


Which will be the right values for the PI resistors in a mixed el84 + 6v6 amp ?

the standard AC30 values are to be changed as to push more the 6v6 tubes ?


In the Rebel20 mix circuit there is a resistor in series with the pot that acts as PPIMV

are those resistors in series (R1 - R2 - R3 - R4) probably of different values for el84 and
6v6 branchs ?

and are there to compensate the difference in gain between tubes ???

Kagliostro

p.s.: The other schematics are on the next post

[kagliostro]

The other schematics

Kagliostro

[kagliostro]

the last schematic

Kagliostro

[jjasilli]

My memory is telling me that el84's like a different signal voltage level than 6V6's.  Personally, unless you're prepared to use spice & sim programs and do the breadboarding,  I would plagiarize from the production amps that are already using EL84's / 6V6's.  They probably have decent compromises worked-out.

[kagliostro]

Thanks Jjasilli

I'm not able to use Spice or other simulation programs, the way I usually take is to look to many schematics and plagiarize

unfortunately the Rebel20 schematic I've is without components values

Plexy50 give me (already tested) indications about his experience that is to use values as in Fender Deluxe

anode n.1 82k
grid to connection point between cathode resistors 1M
anode n.2 100k
grid to connection point between cathode resistors 1M
cathode 470 + 22k

1.5k for 6v6 tube grid
4.7k for el84 tube grid

at Trinyty Amps forum I've seen their v6 amp schematic that use 2 x el84 + 2 x 6v6 tubes swapping between them using the cathode switch mode (like Mesa do)

the values for that amp PI are:

anode n.1 100k
grid to connection point brtween cathode resistors 470k
anode n.2 100k
grid to connection point brtween cathode resistors 470k
cathode 820 + 56k

they also use

1.5k for 6v6 tube grid
8.2k for el84 tube grid

I would like to have also indications about what you think and/or what you have experimented

also it will be very nice to know something about the theory behind all this

Thanks

Kagliostro

[VMS]

also it will be very nice to know something about the theory behind all this

Here's some theory and calculators for the PI:

http://ampbooks.com/home/amplifier-calculators/long-tailed-pair/

http://ampbooks.com/home/amplifier-calculators/bias-excursion/

http://ampbooks.com/home/amplifier-calculators/pi-pa-bass-response/

...and for the PA one example with 6V6 tubes and one with EL84 tubes:

http://ampbooks.com/home/classic-circuits/fender-deluxe-5e3-power-amp/

http://ampbooks.com/home/classic-circuits/vox-ac30-power-amp/

The input signal amplitude needed to drive the power amp to full power is equal to the DC grid bias

So if you look at the Blue Angel schematic you'll notice that the 6V6 tubes need 20V signal and EL84 tubes need 10V for full power.

[DummyLoad]

LTPI with a larger tail resistor will have a better balance of gain between the two halves, but typically swing less Vout. the two stages are cathode coupled - the higher the potential above ground, (lager tail r) the better the balance between the two halves of the stage. note that when i say stage, i mean both halves of the LTPI. 

the difference in the values of the andode resistors makes up for the difference of the gain between the the two halves if you're using smaller resistor ( typically < 20K) values in the tail. output swing will be better, but balance worse.

the EL84 is high slope tube and requires less grid input swing to make full power. the 6V6 on the other hand requires more grid input swing to make full power and therefore needs a LTPI with greater output swing capability.

with your plan, you will need to find a median point that works well with both tubes. another thing that could be done is to devise some sort of voltage divider for the El84s, alternatively, you could use 2 LTPI stages, one for the EL84 pair, the other for the 6V6 pair. personally, i'd try and find the happy spot for both, but then again no two women a built alike so...

--DL

[jjasilli]

Yes, it depends on how and where you want to compormise.  E.g., many people think el84's sound best with 100K grid leak resistors.  None of your examples show that; and 6V6's would not be at their best with that low a grid leak value.  Both sets of tubes could be satisfied with a proper switching arrangement, possibly using relays.  But now the compromise is in added complexity.

[VMS]

with your plan, you will need to find a median point that works well with both tubes. another thing that could be done is to devise some sort of voltage divider for the El84s, alternatively, you could use 2 LTPI stages, one for the EL84 pair, the other for the 6V6 pair. personally, i'd try and find the happy spot for both, but then again no two women a built alike so...

--DL

I was thinking that one could split the plate resistors of the PI and take the signal for 6V6 from the plate and signal for EL84 between the splitted plate resistors.

Thoughts?

[kagliostro]

Thanks Jjasilli - VMS - DummyLoad

The links to documentation posted by VMS are very interesting, I must study it more

Note from DummyLoad and Jjasilli point in the same direction = I need a compromise

Firstly the idea for a compromise was given by the thread posted by Jeff and the very interesting link posted by Inca Roads

(in this moment VMS had the same idea and posted it)

http://www.el34world.com/Forum/index.php?topic=11495.0

http://www.blueguitar.org/new/articles/blue_gtr/amps/bp010629.pdf

(I've read about the plate resistive divider sometime ago, but was relative to a different aspect of the matter)

after this idea I go to the link posted by VMS

then I try to insert the values of the PI resistors in the calculator of the link posted by VMS

http://ampbooks.com/home/amplifier-calculators/long-tailed-pair/

and obtained the results shown down

so I had the idea to change the value of the grid resistors for the power tubes, before 220k+220k for both kind of tubes now 220k+220k for el84 and 470k+470k for 6v6

What do you think about this way to flay the cat ? (I'll never really flay a cat  )

What will happen with this architecture ? are separated grid resistors permitted ?????

As DummyLoad told it is also possible to put on two different PI one for each pair of tubes, but why do it if I can find a good compromise ......... must think about this .........

Kagliostro

p.s.: the RGpa resistors you insert in the calculator are the power tubes grid resistor values, not the RG resistors of the PI

[kagliostro]

This is the redraw schematic

do you think that to insert some disconnecting cap between the two "stages" of final tubes will be of some help as to prevent the resistors to be in parallel for DC in the signal path ???

Kagliostro

[PRR]

> My memory is telling me that el84s like a different signal voltage level than 6V6s

EL84 like half the drive of 6V6; EL84 has nearly twice the sensitivity of most other audio power tubes.

You need enough drive; if you have too much drive then overload pushes the output past cut-off, grid-blocking.

It is not easy for the driver to grid-block 6V6; EL84 drivers are sometimes "choked" or they would grid-block too easy.

> grid to connection point between cathode resistors
> grid to connection point between cathode resistors
> anode n.1
> anode n.2
> cathode 820 + 56k

The grid resistors don't affect how hard the output tubes are driven.

The plate/anode resistors for cathode-bias power stage are usually near 100K-200K (lower for fixed-bias), and either equal or 10%-20% different (the "far" side of the long-tail-pair has lower output so a higher resistor balances the drive).

The long-tail-pair ~~1K bias resistor sets the triodes at a good operating point.

The big difference is the "large cathode resistor", 14.7K 47K 56K in your examples. A larger resistor gives better balance but less maximum drive. A smaller resistor wastes less B+ voltage and allows larger drive but is less balanced.

For driving 6V6, you won't go wrong stealing the exact values from Fender 5F6A. You might like slightly smaller capacitors since 5F6A was supposed to be a bass-amp and that may be too muddy for guitar. The resistor values and ratios are a good fit for 12AX7 and 6L6/EL34/6V6.

If you go back to EL84 you might raise the 5F6A's 15K tail to more like 50K, to waste more B+ so the sensitive EL84s don't get over-over-driven.

Alternatively you can reduce the B+ to the driver. Usually driver is fed slightly less than power tube screen voltage. With sensitive EL84 you can use much lower B+ on this stage.

[DummyLoad]

with your plan, you will need to find a median point that works well with both tubes. another thing that could be done is to devise some sort of voltage divider for the El84s, alternatively, you could use 2 LTPI stages, one for the EL84 pair, the other for the 6V6 pair. personally, i'd try and find the happy spot for both, but then again no two women a built alike so...

--DL

I was thinking that one could split the plate resistors of the PI and take the signal for 6V6 from the plate and signal for EL84 between the splitted plate resistors.

Thoughts?

that could work. i was thinking of actually splitting the load. please see schematic attached. don't know if it would work, could possibly even be a total waste of time. seems like it would though. there may be HF roll-off on the divider - i'm thinking that may be of concern.
i used tube-cad so the voltages shown are from the tabulated results.   

--DL

[tubeswell]

What DummyLoad said.

Raising the value of the tail resistor in an LTP means there is less plate-to-cathode voltage available, because all other things being equal (i.e. the HT voltage remaining the same), you end up dropping more voltage across the tail resistor so you have to give up this voltage from somewhere else. The 'plus' is that the voltage swing across each triode (between the 1st stage and the follower stage of the LTP) is much more evenly-matched, so you can get away with having the same value plate resistors on both sides (e.g.: as the tail resistor gets to around 50k in a 12AX7) without compromising the signal strength balance. But because the plate-to-cathode voltage is now less than what it otherwise would be with a smaller (say 10k) tail resistor, there is less gain overall from the LTP.

[kagliostro]

Thanks PRR - DummyLoad - Tubeswell

As I've told firstly I was thinking to use a splitted load resistor

The way of a divider in the output or the way of a divider as load resistor ?  

I'm sorry I can't breadboard it  

Kagliostro

[VMS]

Ok, here's my suggestion.

If I were to build this amp I'd use these as starting values and then tweak them by ear.

If there was only one pre-PI master volume then I'd try the split plate resistors.

[kagliostro]

Thanks VMS

Thinking more about what DummyLoad draw I've a consideration

ultimately also ppimv is a voltage divider, so it is likely that this solution works, someone can explain what would happen in terms of load impedance seen from the phase inverter ?

Kagliostro

[DummyLoad]

load impedance seen from the phase inverter

since the grid load is such a large impedance it is near insignificant, boiler-housing numbers 220K+220K = 440K || 430K or ~~217.47K. the EL84s see a 220K grid-leak circuit resistance and the 6V6s see 217K as grid leak ckt.resistance.

--DL

[kagliostro]

Thanks DummyLoad

since the grid load is such a large impedance it is near insignificant, boiler-housing numbers 220K+220K = 440K || 430K or ~~217.47K. the EL84s see a 220K grid-leak circuit resistance and the 6V6s see 217K as grid leak ckt.resistance.

OK !

Kagliostro

[DummyLoad]

EL84s see a 220K grid-leak circuit resistance and the 6V6s see 217K 165K as grid leak ckt.resistance.

sorry - but i believe i mislead you. EL84 see 220K in || with 220K+430K or ~165K as grid circuit resistance. 6V6 is still ~217K grid circuit resistance.

my apologies. 

--DL

[kagliostro]

sorry - but i believe i mislead you. EL84 see 220K in || with 220K+430K or ~165K as grid circuit resistance. 6V6 is still ~217K grid circuit resistance.

my apologies.

Many Thanks DummyLoad for your help and patience in giving

I apologize for the large amount of questions that I always do

Thanks again to all that give help with their answers

Kagliostro