• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

balanced tube preamp

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.

I want to build a balanced tube preamp (my first tube project). But I haven't find any schematics for something like this, except for this. http://home.c2i.net/semfielec/tubepre.htm

Have you got any other schematics for a balanced tube preamp?
I want the preamp to work like a BoSoZ, which means that it should be able to convert a singel-ended input to balanced output.... And I would also like the preamp to have as low output impedance as possible (<1000Ohm).

A balanced tube section will make an excellent line stage. Two comments:
1) To get that low an output impedance, you'll need either cathode followers (no problem, there) or to massively parallel the tubes in the differential (such as Conrad Johnson do in their ART preamp). Cathode followers are simple and effective. Paralleling the tubes has the fringe benefit of reducing noise etc. but is more expensive. Your choice.
2) To get good results from a differential being used as a phase splitter (conversion from single-ended to balanced) you'll need a lot of resistance under the cathodes. Again, you have two choices. You can use a really high rail--on the order of 4-500V--and use a very large value resistor under the cathodes, or you can use a current source. I'm on record as not liking current sources, but this is one of those cases where I have to admit that they earn their keep.

Current sinks

The easiest way to get good balance is to cheap and use solid-state current sources. I know it's not purist, but to me the benefit of using tubes as amplifiers is retained with this approach.

The dirtiest (but working) circuit I've used in this position is a 317 IC voltage regulator, with one resistor setting the current. Transistor circuits also work OK, as do n-channel enhancement FET's. But the best results for low current is the LM334 current source IC. These are good to 10mA, so ideal for preamps with hi-mu valves.
I dare say you can't beat the solid-state circuits performance without using a pentode or cascode triode circuit, and either way you get a great deal of added complexity and expense to your circuit.

As far as the Zout of the preamp is concerned, I'dd have to agree that using a cathode follower would be a good option, and you could direct-couple it to the amp. stage. A *mu-follower* with a pentode active load would also work well.

This is my first post on this forum. I hope no one flames me for being of a "hybrid" opinion when it comes to tubes...erm, valves.

[Edited by MuFollower on 11-22-2001 at 06:26 PM]
The current sources (sinks, for the nomenclature purists) I use in the front end of my tube amps are discrete solid state. I tried chips, but they sounded simply ghastly. Not only do they have a maximum current, but a minimum as well, below which they get extremely nonlinear--soft and wooly sounding. I tried several options down that line and gave it up. I believe I saw somewhere that they're no good above about 4kHz. I can well believe it. I never was able to get reasonable high frequency response out of them.
Discrete circuits for this low a current can be built easily, and for only about a dollar more, if that much. The one I'm using now is actually fairly elaborate, but simple ones work well and only need two transistors and two resistors (another option: 2 small diodes or 1 Zener, 1 transistor, and two resistor). It's just not that difficult to do.

Every time I have made prototypes with tube differential stage the sound is always kind of bad.
This assuming that one side is used for input and the other grounded (trough a small resistor to prevent oscillation).

The best performance I have got have been with parallelled 6922 loaded with a output transformer with balanced output.
This way you also get rid of the output cap.
Until today I have not heard any cap that beats a top level output transformer.

Just a note about Conrad Johnson Art, LS16 and LS17 preamps:
All use parallelled singel-ended 6922, not in differential mode.
A good friend of mine have LS17 and it sounds absolutely marvelous for a commercial tube preamp.

If you want to be balanced, please keep the signal balanced all the way. From source to loudspeaker. Converting single-end to balanced is better done in a transformer.

As I said above, a differential makes an excellent phase splitter <i>if</i> it has enough resistance under the tail. Low resistance will result in very poor performance from the 'backside' of the differential--lopsided gain between the positive and negative halves of the signal, which amounts to very high distortion down the line. Think in terms of 5-10k minimum for a cathode resistor, and higher would be better still, as the goal is infinite resistance. At this point you'll be needing to cap couple the grid etc.
It's much simpler to use a current source under the cathode, as a well executed current source gives virtually perfect balance between the two halves of the signal, and need not sound bad (avoid current sources based on voltage regulator chips). Yes, it's possible to arrange a current source based on a tube (Atma-Sphere uses one in their OTL front ends), but the rails get high very quickly indeed.
Note that all of the above applies to solid state differential phase splitters, too.


I can't argue with you regarding the cathode resistance.
I think you are absolutely correct about it.
I was talking about a series grid resistor of around 100 ohm or so to prevent oscillation because of Miller capacitances. A grid connected directly to ground will mostly make the circuit oscillate. Grounding through a cap can also oscillate and will be unlinear. It's better to use a linear resistor here.
This is also the case when parallelling tubes.
You can never connect several grids together. You must separate them with a series resistor to each grid.

Regarding current sources the best I have used is based on tubes, resistors or J-fets. I have not been able to get as good result with bipolar transistors or mosfets.

Regarding regulators as current sources ?
Haven't even crossed my mind.

Thanks for the replies...

What do you think of the circuit I posted a link to?

It seems pretty good to me.. cheap, the tubes are easy to get, the output impededance is low enough. There's only one change I would like to do to it: increase the cathode resistors a bit, cause I don't need the gain anyway.

By the way, as volume controll, what do you think of using of using a 'relay ladder circuit' (controlled by an optical encoder) which ties/unties resistors from the output to ground to reduce the gain, similar to the circuit used in Aleph P?

And what kind of powersupply do you recommend me to use to for the tube preamp?

Basiclly I like the circuit if you have balanced signal in.
If you have single-ended signal at the input you better keep it single-ended. I have tried a very similar circuit with bad result. If you use 470 ohm in series with each grid instead of 100 for both you can remove the 47 ohm resistors at the plate and the 100 ohm resistors at the cathode. It is better to modifiy the attenuator instead of using cathode resistors to lower the gain.
With cathode resistors you will more than tripple the output impedance which is low enough at 1800 ohms.
The plate resistance should be raised to around 24000 ohms to get the lowest distortion. The powersupply should then be at around 250 Volts to keep platevoltage at 125 Volts. Run around 5mA trough each 1/2 tube. Increasing the plate resistance will not ruin the low output impedance of a parallelled 6922.

For powersupply I would use a solid state rectifier followed by 150 uF metalized polypropene cap.
Couopled trough a 10ohm/5W resistor parallelled with a 1N4007 diode for resistor safety reason into another 150 uF metalized cap. After that a shuntregulated mosfet (IRF720)
followed by a 2,2 uF foil/polypropylene cap.

What ever you do. Don't use feedback anywhere in the powersupply.

This will give you a preamp that can compete with most of the top models from the best brands.

Active regulators

Hedlund asks:

"Allen, do you prefer actively regulated powersupplies too ??? "

A whole chapter on this in the TubePreamp CookBook.

I hate series regulators and find they do much more sonic harm than good. A good passive supply has always been preferred by me to any series regulator.

But I love shunt regulators, especially when fed via a very good current source. This combination I call a "SuperRegulator". I find this combination much superior to any passive supply - even if it has a lot of internal negative feedback!

I am working to make my shunt regs use much less NFB but unless you just use a chain of zeners or gas regulator tubes - any active shunt has NFB.

Allen (VSE)

I just checked your website and the circuit you use yourself.
That's the way I meant that a differential amplifier should be used. Balanced in and balanced out.
If you use only one half of the differential amplifier for input and output you better use the single-ended version of your preamp for the best sound.

Have you tried parallelled single-ended stages instead ?

Regarding series regulators:
Yes, they can sound bad, but can (I have found a way) be better than any passive power supply.
NFB in any form? No thanks, not for me.

>>>Here comes the requested schema.<<<

Thanks. I understand clearly now - schematics allow an engineer to go beyond language!

What you have used is not a shunt regulator - it is a series regulator, but as you say - it does not have any NFB.

I have tried many series regulators, with and without NFB - but found that a shunt reg fed from a current source always sounded better - even when using a lot of NFB. My experience across 25 years.

>>>The key is to present a linear impedance over the whole audio range and to prevent any voltage shifts between drain and gate<<<


I'm sending you a PDF of such a current sourced shunt reg - this is not my design but it shows you the concept. It is VERY complex and this is not good IMO.

In your design, if you were to forget the MOSFET and take the power from the junction of the 22R and the top of the zener string - you would have a current sourced shunt reg - without NFB! But without the large current possible from the MOSFET...

I prefer the SuperReg current sourced shunt reg topology because:

1/ It's bi-directional - accepting current reflected back from the load as well as supplying current.

2/ It maintains VERY precise powersupply line currents in the ground and B+ wiring - minimising crosstalk and general confusion.

Regards, Allen (VSE)
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.