Hiwatt PI

[Torquil]

Hi guys, I have a question on Hiwatt PI circuits. I'm in the process of modifying a Ceriatone "Hey What 103" (a Hiwatt "clone") into a full-fledged dual-channel preamp.

Instead of combining the channels after the first tube, they are combined closer to the PI (I also added a preamp tube). I'm not experienced with PP amps or phase inverters since I have only built a single-ended amp before.

Some background: According to the hiwatt.org schematics for the "Early 70s Four-Input Preamp" and "Mid 70s Four-Input Preamp" preamps, the stage V3A in front of the PI changed from type A to type B:

A) V3A is a cathode follower:
Ref: http://www.hiwatt.org/Schematics/DR_Pre4Input_v1a.pdf

B) V3A is taken out of the signal chain, and instead used to create a voltage reference for the PI
Ref: http://www.hiwatt.org/Schematics/DR_Pre4Input_v2.pdf

Incidentally, my Ceriatone Hey What originally used type A. Here are my questions in an ordered format:

Q1: Which of these types are able to generate the largest voltage swing on the PI grids? Type A should have an advantage due to its low output impedance, right? Or does the construction in type B also lead to an effective low output impedance for the signal injected at the PI grid?

Q2: What are the advantages of type B?

Q3: Do these Hiwatt circuits generate output tube distortion before PI distortion, or the opposite? My amp has 4xEL34, although I usually run it with only the inner pair, while compensating the resulting impedance mismatch at the speaker impedance switch.

Thanks!
Torquil

[JD0x0]

Q3: Do these Hiwatt circuits generate output tube distortion before PI distortion, or the opposite? My amp has 4xEL34, although I usually run it with only the inner pair, while compensating the resulting impedance mismatch at the speaker impedance switch.

My Hiwatt gets output tube distortion before PI. Mainly because the PI doesnt distort much on these amps, I believe the PI is a copy of a hifi PI which is meant to stay clean.

[morcey2]

The PI's in both of them are slightly different takes on a DC-coupled long-tail pair. The first one is truly DC-coupled w/ the 1M resistors feeding the CF setting the quiescent grid voltage for the PI. The second one isn't truly DC coupled because of the 22n coupling cap, but the isolated triode sets the DC voltage for the PI.

http://www.freewebs.com/valvewizard/dcltp.html

Q1: Type B in theory would impart a larger signal to the PI because Type A attenuates the signal w/ the 1M resistors prior to the cathode-follower. The output impedance benefit of the CF for type A isn't really much of a benefit for feeding this type of PI because the input impedance of the PI is relatively high as compared to the output impedance of a normal gain stage (approx 70k-ish in the schematic).

Q2: Only more signal to the PI, that I can see. And possibly less noise because of fewer 1M resistors in the signal path.

Q3. See JD0x0's post. I've never had a hiwatt or clone. (yet.)

Matt.

[Torquil]

Thanks guys! I'm going to stick with type B unless I have a good reason to change it. It is more compatible with the way I wanted to implement switching at the end of the preamp.

Q1: Type B in theory would impart a larger signal to the PI because Type A attenuates the signal w/ the 1M resistors prior to the cathode-follower. The output impedance benefit of the CF for type A isn't really much of a benefit for feeding this type of PI because the input impedance of the PI is relatively high as compared to the output impedance of a normal gain stage (approx 70k-ish in the schematic)..

Isn't the input impedance of the PI here effectively around 100k, since the top of the 220k cathode resistor of the "unused" triode in type B is effectively a fixed voltage reference? Please correct me if this is wrong.

And in the true DC coupled schematic, the 1M8 / 1M voltage divider is bypassed by a 22nF which should pass most audible frequencies (the 22nF impedance is lower than the 1M8 at frequencies above 4Hz).

If these arguments are correct, I would expect type A to drive the PI harder, since the input impedance of the PI in that case would be around 100k while the output impedance of the cathode follower is very low in the few thousands. Also, there would be no significant loss due to the voltage divider in front of the CF being completely bypassed by the 22nF cap, and due to the unit gain of the CF?

On the other hand, in type B the input impedance of the cathode follower is comparable to the output impedance of the previous gain stage, so there would be around a 50% loss?

Of course, my arguments might be incorrect...