Flipping Phase with Input Transformer?

[Mr. Lime]

Hi,

I'm thinking of converting my Hoso56 Amp into a two channel amp with paraphase PI instead of the LTPI.
Because channel mixing can't be done with the two inputs of the LTPI anymore I have two flip the phase of either the EF86 or the other channel before they meet each other at the paraphase gain stage.
I won't add another gain stage or transistor and would like to keep it simple.
How about small signal transformers at the input of one channel?
There are 1:1 windings with 10k each side.
Would this fit the bill or what are the disadvantages of such an approach?
I read that noise issues could be a problem..
What transformer would you use for example and how would you connect it?

https://www.banzaimusic.com/TM018-Transformer.html


Thanks a lot!

[R.G.]

The little TM series transformers have limited low frequency response. If I remember correctly, they're only specified down to 300hz. In practice, they do a little better than that, but 82Hz low E strings are likely to be suffering a bit. I worked out a way to extend that down to 60Hz, but it used an opamp, and you're already saying that you won't use another gain stage or transistor to keep it simple.

I have a slightly dissenting opinion: for all the effort and expense you're going to do to get a transformer to work well, I think using a transistor is simpler, but that's just me. One LND150 or other high voltage MOSFET with equal source and drain resistors, two bias resistors and two capacitors seems pretty simple to me.

You'll probably need to look up a better transformer. Seems like a company named EdCor makes something workable. How big is the signal in volts where you're going to drive the transformer, and what low end do you need to get out of it? (Hint: if you intend to run a bass into it, you need to get the low end down to about 30 Hz somehow.)

[Mr. Lime]

Thanks for your reply!

I thought about a LND150 gain stage with local feedback to archive unity gain right behind the EF86 pentode but I guess this could cause other issues with mixing very different output impedance of the two channels..

Really I wouldn't spend too much time with it as I have no idea how promising channel mixing is, so a simple transformer for trying would be nice.

I just want to put my Les Paul directly into the amp, I guess about 300mV are going to drive the transformer..

[R.G.]

I thought about a LND150 gain stage with local feedback to archive unity gain right behind the EF86 pentode but I guess this could cause other issues with mixing very different output impedance of the two channels..

Hmmmm... a MOSFET with equal drain and source resistors will have exactly unity gain on both drain and source. The impedance on the source is "lowish" as long as it's lightly loaded, and the impedance at the drain can be considered to be the drain resistance. So unity gain is pretty much assured. Passive mixers generally don't have problems as long as the input impedance of the mixing stage is bigger than about 10x the inputs being mixed if they're mostly equal. One can stick a series resistor between the source and [whatever] to artificially raise the source signals impedance. But you're wanting the drain output anyway, and that's just the drain resistance. You get to pick that one.

Really I wouldn't spend too much time with it as I have no idea how promising channel mixing is, so a simple transformer for trying would be nice.

I just want to put my Les Paul directly into the amp, I guess about 300mV are going to drive the transformer..

Again, hmmm... a Les Paul can put out as much as a volt or so on hard strums. Is this transformer going before the first tube(s) or after? A typical 12AX7 stage can give you 20-30x gains, so you could see 20V of signal off the plate of the first tube.

I just hate to see your experimentation ruined by avoidable transformer issues covering up whatever it is you would otherwise find.

[pdf64]

...How about small signal transformers at the input of one channel?
There are 1:1 windings with 10k each side...

Doesn't 10k strike you as kinda low for a guitar amp input?
OK a 1M grid leak should mean that's what the guitar 'sees'. But the transformer's winding capacitance etc will only be intended for 10k, so the performance at 1M may be worse than expected.

[Xander8280]

A high voltage FET or BJT set up as a follower at the input would be the easiest thing. If you use a N type DFET (like the LND150) the input doesn't need any DC bias, it can be DC coupled like the typical tube input so just an series gate resistor and a source resistor (100kOhm or so).

[sluckey]

A high voltage FET or BJT set up as a follower at the input would be the easiest thing.

But a follower will not flip the phase.

[Xander8280]

A high voltage FET or BJT set up as a follower at the input would be the easiest thing.

But a follower will not flip the phase.

you're right, my bad. Could do 100kOhm source and drain resistors and have an inverting buffer. Could get really exotic with that stage.

[R.G.]

Again, high voltage N--channel MOSFET, equal source and drain resistors, and two bias resistors. The two bias resistors let you pick the voltage on the source resistor, and when you choose the source resistor, that fixes the MOSFET drain current. The drain resistor conducts the same current as the source, as the gate is an insulator. So equal and opposite voltage signals appear at the drain and source. You can pick either one you like.

Suggested bias is 1/4 of the drain supply voltage across the source resistor, gives maximum voltage swing on both drain and source. Bias resistors can be high value, but using a three-resistor "noiseless biasing" setup is recommended.

You >could< use this as the first device the incoming signal hits, but then you'd have the noise contribution of the MOSFET. That may not be a big deal. You >could< use a resistor and a zener to run the drain current to make the MOSFET drain supply quiet and avoid getting B+ noise into the MOSFET.

It's probably more useful to put this after the first tube in, for noise reasons, and make the first tube be common to both preamp channels. You >could< even feed the second channel with one of the splitter outputs internally.

It's a veritable playground or opportunities.

[Mr. Lime]

Thanks for all the suggestions.

Making the first triode common to both channels would also solve the phase issue but I would loose the option of feeding different effects to the inputs..

[R.G.]

True. I didn't realize that the "different effects" was an objective.

[Mr. Lime]

Sorry I missed to mention that I want two inputs for flexibility.
Yesterday another idea came to my mind.
If I keep the LTPI but add passive fx loops (like Matchless does) on both channels, I can bridge the channels in two ways.
The EF86 could be the first or last preamp tube and so I would save the triode I considered for the channel mixing.
For now I have Matchless values in the EF86 cahnnel and I guess that would cause too much gain in series with the other channel.

How about a cold clipper EF86? Cathode resistor and cap maybe switchable for parallel channel operation.
Has anyone ever tried something like a cold clipping EF86?

Here's what came to my mind:

Amp Conversion K19_EF86.PNG