[poty]

I'm trying to help a guy to choose a phono preamp. Knowing JH's products very well I tried him to consider Trumpet Reference or Classic, but got to some trouble of explaining several inconsistencies in the spec.
First question came from the following sentence from the site's description:
" The circuitry is fully-balanced from input to output, including the choke regulated power supplies. Each gain stage operates in class A and employs no feedback, for a more pure and accurate sound. "
How come the Trumpets are fully balanced if they have single ended inputs and outputs? (I can prove this from Classic schematic; the Reference schematic has been hidden now, but I remembered that the amplifier part is essentially the same + added "like Piccolo" MC input with modified tails)
Relating "no feedback" - does it means that there are no global feedback (because each stage definitely has local feedback) or that the equalization is in the direct path (not in feedback)?

[hagtech]

The circuitry is entirely balanced.  The I/O connections are single-ended.  You could run XLR connectors, but they are not very popular, so i don't sell the TRUMPET in that configuration anymore.

There is neither local or global negative feedback.  The tubes run wide open.  Equalization is passive between gain stages, not inside a feedback loop.  This is the difficult way to do things, but the payback in sound quality is worth it. 

jh

[poty]

The circuitry is entirely balanced.  The I/O connections are single-ended.  You could run XLR connectors, but they are not very popular, so i don't sell the TRUMPET in that configuration anymore.
There is neither local or global negative feedback.

You are not quite right. The matter is not about connectors (while I still have my own IMHO at this matter), the matter is about the first and the last stages entirely. The first stage (I speak now about Classic version, but in the Reference there are the same things) receives the signal on the grid of V1A and the ground. How could you explain the V1B driven? It is driven through the "tail" circuit (doing sort of CCS here) halving the total signal amplitude the grid of V1A get - is it not feedback? The first stage runs as the single end-to-balanced convertor. It can't be seen as "balanced". To be balanced it should receive the signal to both grids. The last stage (as pictured by schematic) is a cathode follower. It is 100% feedback circuit! And also... Where the signal is taken from? Just from either V3a/V4A or V3B/V4B, depending on the Polarity/Mute switch, not both as should be in the fully balanced circuit! The only fully balanced circuit is the 2nd stage (V2A and V2B) and I can say that ideally it can be seen as "no feedback", but the world is far from ideal. The R26, R31 work here as another CCS which has been put here to linearize the balanced response of the stage through the common tail. It is weak feedback, but still it is here.

The tubes run wide open.

Sorry, cannot understand what it means.

Equalization is passive between gain stages, not inside a feedback loop.  This is the difficult way to do things, but the payback in sound quality is worth it.

It is true. I do not have questions about this sentence.

[hagtech]

The first stage can be connected to the cartridge (via XLR) with 24k resistors on each leg to ground.  That would satisfy your balanced requirements.  The problem with this is that it is relatively easy to pickup hum.  [It works very well, however, connected to a step-up transformer that has a center tap on secondary to ground].  Connecting one leg of the cartridge to ground, you end up with exactly the SAME differential voltage at the grids.  Output differential voltage is the same in both cases.  Nothing is halved.

You are technically correct that the cathode follower employs 100% voltage feedback.  But it is not a voltage gain circuit, it is a current gain one!  And there is no current feedback.  The whole point of the circuit is to provide a high input impedance and low output impedance.  As you said, there is no voltage gain. 

So to be technically correct myself, I should have stated the voltage gain circuits do not have voltage feedback and the current gain circuits do not have current feedback.

The outputs can be connected to XLR using both polarities (V3a & V3b).  The circuit is indeed fully balanced.  And as I was hoping to explain, only the RCA connectors are single-ended.

Also, if I need to be technically articulate, the B+ and B- power supplies are fully balanced too. 

jh

[poty]

I'd like to point out, that I'm not trying to blame you in any way!

The first stage can be connected to the cartridge (via XLR) with 24k resistors on each leg to ground.  That would satisfy your balanced requirements.

Yes, but "can be connected" and "is connected" are different things. The entry stage now is connected as single-end to balanced convertor and is not entirely balanced. But I understand, that this can be mended easily.

Connecting one leg of the cartridge to ground, you end up with exactly the SAME differential voltage at the grids. Output differential voltage is the same in both cases.  Nothing is halved.

I hope I read this correctly that output differential voltage is the same after the first stage. But at the output we have not differential voltage, but the same single ended: the Polarity switch connects one or the other half of the output stage, so at the output we have only half the available signal.
Looking at the schematic - the only common load for balanced signal is C13, R35, R98. Then, the last stage technically is a balanced one, but its load is not balanced at all. Well... I should admit that this can also be easily mended by connecting output to both C10 and C17 instead of either C10 or C17 and the ground as you correctly mentioned:

The outputs can be connected to XLR using both polarities (V3a & V3b).  The circuit is indeed fully balanced.  And as I was hoping to explain, only the RCA connectors are single-ended.

The only thing I cannot agree to: in the current design the load is connected single-ended, not RCA and this makes the balanced good side useless.

Also, if I need to be technically articulate, the B+ and B- power supplies are fully balanced too.

Well... Technically speaking I can't catch what does this mean? Of course it is bipolar (every bipolar power supply is balanced?)! But what good is it to the design?

[hagtech]

Most +/- power supplies are not balanced.  They are typically regulated separately.  In the TRUMPET they track each other.  That is, if the positive supply drifts up a volt, the negative supply drifts downward a volt.  This allowed me to do a very unique trick, in that I match the bias point of the plates on the positive side with a bias point of the same magnitude on the negative side.  You will note the 10uF decoupling capacitors.  What this does is remove to a first order correction any change in AC line voltage.  The DC bias point of the plates do not change!  No other preamp in the world can do this.  It is sort of an inherent passive regulation, only possible because of the differential operation. 

I have proven this using a variac in the lab.  I can crank the AC mains up and down (rather violently) and the bias voltage of the plates on the 12AX7s do not move.  On a single-ended design such as the CORNET they would shift a lot, following the average absolute value of the AC input voltage. 

This is a great trick that really separates the TRUMPET from all others.

jh

[poty]

Sorry, I have not heard the term "the bias point of the plates" yet. Does it mean the voltage on the plate in relation to ground?

[hagtech]

Yes.

jh

[hagtech]

The XLR version used to look like this.  I actually have one left for sale, if you are interested (but without the step-up).

jh

[poty]

I should ask my friend and he'll get in touch with you about that.