Download Directly-coupled Fee..

Tuesday, January 08, 2002 11:59:52 AM
From:
momitch ([email protected])
Date: 1/7/2002 3:51 PM
Subject:
Concertina Phase Inverters
Greetings all. I'm interested in using a concertina (split-load) PI in my next EL84
homebrew. In Fender designs, the driver stage was RC coupled to the input of the PI. Is
there a way to directly couple the driver stage to the PI in such a design? In other words,
can I have the output of the plate feed the grid of the PI directly? Maybe a tube other
than a 12AX7 would be a better candidate for such an application. I know some of the
Sunn amps used something like this with a pentode driver stage direct coupled to a
triode concertina PI. Thanks in advance!
Morris
From:
kg
Date: 1/7/2002 4:03 PM
Subject:
Re: Concertina Phase Inverters
mo,
check out the front end on this gec amp... you'll see a plate loaded, common cathode
6sn7 directly coupled to a concertina phase splitter.
for a 12ax7, you will be scaling all of the impedance values up. i've noticed biasing of
the direct coupled concertina is easier with a larger plate load on the first stage, say a
220k. then, set the plate voltage on that first stage to be b+/3 by varying the cathode
resistor. if there is no NFB to be injected into the cathode you may bypass it with a cap
for more gain.
the concertina's total load resistance should be on par with the plate impedance, so for a
12ax7, use something like 47k's on top and bottom. you can go higher impedance than
that if you'd like, but don't go lower. the plate current through that 2nd stage will be
approx. b+/3(rk).
williamson was the first one to really adopt this direct coupled topology, which, since it
eliminated a coupling cap, was key to keeping his NFB design working and stable.
hth
ken
From:
momitch ([email protected])
Date: 1/7/2002 5:12 PM
Subject:
1 of 5
Thanks Ken!
Tuesday, January 08, 2002 11:59:52 AM
Ken,
As always, thank you for your help.
Morris
From:
momitch ([email protected])
Date: 1/7/2002 5:26 PM
Subject:
more design questions
Let's say we use a 12AX7. We use a 220k plate resistor on the triode used as a driver.
On the PI triode, we use 47k plate and cathode resistors. The plate of the driver tube is
connected to the grid of the PI tube. To make sure I understand, I need to:
1) Find the proper cathode resistor for the driver tube that gives me a voltage on the
plate that is 1/3 of the B+ feed for that tube. If I do this, will the bias for the PI tube work
out naturally as a result, as the grid voltage of the PI comes from the plate voltage of the
driver tube. What should the grid to cathode voltage difference be for the PI tube in this
configuration?
2) Should I use a grid stopper resistor on the input of the PI tube?
Thanks in advance!
From:
kg
Date: 1/7/2002 7:45 PM
Subject:
Re: more design questions
1) Find the proper cathode resistor for the driver tube that gives me a voltage
on the plate that is 1/3 of the B+ feed for that tube. If I do this, will the bias for
the PI tube work out naturally as a result, as the grid voltage of the PI comes
from the plate voltage of the driver tube. What should the grid to cathode
voltage difference be for the PI tube in this configuration?
the bias for the concertina will be just about perfect if that first plate is at b+/3. by perfect
i mean it will have the maximum undistorted signal swing.
the concertina along with its load resistors should form a voltage divider with equal
voltage dropped across each. assuming a perfect tube (which none are, but just for
illustration) the tube, when fully on, would have 0v across it. that means that the cathode
and plate would be at the same potential, halfway from B+ to ground. if the tube shuts
off completely, you'd have the cathode at 0v and the plate at B+. your splitter would
have a max swing of B+/2 volts peak. now, the imperfection of tube means it cannot turn
COMPLETELY on, so your effective output swing will be somewhat smaller, probably
around B+/3 volts peak, but the operating point stays the same.
2 of 5
Tuesday, January 08, 2002 11:59:52 AM
the splitter's grid to cathode voltage will be very small, probably around 2v. this will not
change very much throughout the cycle.
2) Should I use a grid stopper resistor on the input of the PI tube?
no, there's no need. the gain of the concertina is 1 (well, a little less, but very close to 1)
so there's no real danger of oscillations. in fact, without the grid stopper, you can
actually pull the concertina into grid current when the output of the first tube goes "high,"
and the concertina "turns on."
hth
ken
From:
momitch ([email protected])
Date: 1/8/2002 4:12 AM
Subject:
Is a 12AT7 or a 12AU7 a better
I guess my main goal for this thread is to get closer to finding the perfect PI for driving
EL84's in a guitar amp made for rock. One could argue that this goal has already been
reached by Vox in the AC30's PI design, but the Vox PI really has too much drive for the
application. If you look at Matchless DC30 schematics, a voltage divider is inserted
before the Vox-style LTP PI (after the tone stack) that dumps more than half the input
voltage to the PI.
As for this discussion of direct-coupling between a triode driver stage and a triode wired
as a concertina PI, what commonly available tube is best suited for the job of driving
EL84's in this arrangement? Might a 12AT7 or a 12AU7 be better suited for the job?
Obviously these tubes would provide less gain in the driver stage (before the PI), but
maybe that is a good thing for the EL84 application. Any and all ideas are appreciated!
Morris
From:
Mark Lavelle ([email protected])
Date: 1/7/2002 5:22 PM
Subject:
Huh?
I thought a concertina was the kind using only one triode, with the signal being inverted
on the cathode with respect to the signal coming off the plate...
From:
momitch ([email protected])
Date: 1/7/2002 5:31 PM
Subject:
Re: Huh?
It is. We are talking about how to direct couple the gain atage that precedes the
3 of 5
Tuesday, January 08, 2002 11:59:52 AM
concertina PI to the PI itself (rather than using RC coupling).
From:
PaulC ([email protected])
Date: 1/8/2002 2:09 AM
Subject:
Re: Huh?
Now we're talking about one of my fav circuits! I like to use this setup with the B+ trick in
the princetons. A 12AU7 is great for this also. You'll see this in the Leslie's.
From:
momitch ([email protected])
Date: 1/8/2002 3:14 AM
Subject:
Paul, can you elaborate?
Paul, can you tell me more about the circuit you use? Any and all information is
appreciated. I'm trying to find the ideal PI for driving EL84's. Thanks in advance!
Morris
From:
Mark Lavelle ([email protected])
Date: 1/8/2002 5:21 PM
Subject:
Re: Paul, can you elaborate?
I'd like to know more about that, too (I'm cooking up a p-p EL84 amp and my brother just
sent me his Princeton for sprucing up...)!
Morris: what advantage/effect do you hope to achieve with direct coupling?
From:
momitch ([email protected])
Date: 1/8/2002 6:40 PM
Subject:
Re: Paul, can you elaborate?
I don't know what advantage direct-coupling might provide. I've been doing some
calculations with a 12AX7, and the setup of the driver triode is difficult. Maybe a 12AT7
or a 12AU7 will work better. The RC-coupled setup is certainly more versatile, as you
can play with the gain parameters of the driver triode without affecting the bias of the PI
triode.
From:
Doc ([email protected])
Date: 1/8/2002 7:29 PM
Subject:
4 of 5
Re: Paul, can you elaborate?
Tuesday, January 08, 2002 11:59:52 AM
The direct coupling between the gain stage and the split-load phase inverter stage is
useful when you are applying large doses of loop negative feedback around a few
stages because it leaves out one R-C phase shift element. The number of phase shift
elements can become critical when your feedback would become positive at some
frequency, so you eliminate some phase shift in order to keep it always in the negative
region for stability. The direct coupling also helps with transient reponse, signal fidelity,
and reduces all those undesirable things that capacitors and dielectric absorption
characteristics are noted for. Unless you are concerned about any of these items, I see
no advantage to using direct coupling in a guitar amp except for elegant simplicity. It's a
lttle tougher to design the circuit, because the second tube's bias is dependent on the
plate voltage level of the preceding gain stage. It's a balancing act. If you couple with
capacitors, the two stages are isolated. You are free to run each stage at different
combinations of electrode voltages. Only the AC signal is transferred. Biasing is
independent.
For examples of commercially successful circuits employing a direct coupled gain stage/
PI, you can look at the classic Williamson circuit (6SN7 triodes), any Dynaco amp circuit
(usually 7199 or 6AN8 pentode/triode, but the small p-p 6BQ5 amp used a 12DW7/7247
pair of triodes with the "12AU7-half" as the PI), and some Eico circuits which go by the
classic Mullard circuit using an EF86 & 12AU7 or 6SN7 long tailed pair PI in sequence).
The Williamson had low-mu (6SN7) driver stages between the PI and the output tubes
for increased linearity, but the Dyna and Eico circuits were for affordable low cost kits,
so the PI drove the power tubes directly. The Sunn 2000S bass amp, whose power
section is a direct decendent of Dynaco circuit, employs a direct coupled 7199 pentodetriode.
A low-mu tube like 12AU7 or 6SN7 is ideal for a split-load/ concertina phase inverter
service. However they are rarely seen in commercial guitar amplifiers simply because
the designers don't have much use elsewhere in the circuit for the other half of the tube.
Some early Gibson amp circuits (look at the GA-30RV) employed it as the reverb driver
ahead of the transformer, but they needed 1/2 a 12AX7 for a signal amplifying stage just
ahead of the low gain driver. So most guitar amps simply use a 12AX7 for everything,
including a PI, since the designers are looking generally for high overall circuit gain and
economy in number of tubes mounted to the chassis. Guitar amps usually run much less
than max allowable feedback, and we welcome slight coloration of the sound by
different types of coupling capacitors.
5 of 5