pp transformer line output: how much dc current?

Hi,

this is a question about (internal) tube matching, as well as a question about transformers.

When I want to use a double triode like 12AT7 or 12BH7 to build a low power pp output stage with a center tapped transformer to drive 600 Ohm, there will be some magnetisation from a mismatch in DC current thru the two halves of the primary winding. When I look at a transformer like this
http://www.sowter.co.uk/specs/9041.htm for instance, 500uA are allowed.

Is it save to use a dual triode like the ones mentioned above, with auto biasing thru a shared cathode resistor, without any means to trim the current distribution thru the two triodes? Even with ageing? Even when I run the tube with 2 x 20mA ? (I could add a trimmer or a servo loop with opamps if needed, or just use a different transformer, but I just don't have the right "feeling" and experience yet to know if this is necessary or not.)

JH.

If you are asking if you will need an airgapped core for push-pull outputs, the answer is no. Like in PP poweramplifier transformers, the "natural" airgap of EL-core transformers should be able to handle these small current differences.

E.g. the original Pultec output transformer was not airgapped, and was run in push-pull:

http://www.gyraf.dk/gy_pd/pultec/pultech.gif

But you can't use a toroid thansformer because of it's sensitivity to the potential current mismatch..

Jakob E.

[quote author="gyraf"]If you are asking if you will need an airgapped core for push-pull outputs, the answer is no. Like in PP poweramplifier transformers, the "natural" airgap of EL-core transformers should be able to handle these small current differences.

E.g. the original Pultec output transformer was not airgapped, and was run in push-pull:

http://www.gyraf.dk/gy_pd/pultec/pultech.gif

But you can't use a toroid thansformer because of it's sensitivity to the potential current mismatch..

Jakob E.[/quote]


Hi Jakob,

thanks for your reply.
Well, I know that ordinary pp transformers can tolerate "some" DC current, and that untrimmed pp circuits create "some" bias current mismatch, but I want to get a better feeling for the quantities of this.

For instance, by how much percent will the two plate currents of dual triodes differ, when their plates and grids are at the same dc level?
If it's less than 5%, I can run them at 10mA without exceeding the 500uA from the Lundahl data sheet. I just don't know what matching to expect, worst case.

On the transformer side, thanks for pointing out the "natural gap" issue.
Is there a rule of thumb for this stuff (transformer size / AC ratings and maximum DC magnetisation current) that works across various manufacturers? (Always speaking of transformers that don't introduce a gap intentionally)

JH.

Crap, i ment Sowter data sheet.

JH.

You should be OK for a line amp etc, but for something like a vari-mu compressor, you need a gap as the dc gets rather large due to mismatching of tube characteristics thru the whole db compressing range. Just a little dc can really drop the inductance thru the primary, and you will get a huge ac current increase as reactance is a direct function of inductance. I think that is the reason for the "mystery" resistors in series with the innerstage transformer on the RCA BA6a compressor. They added the resistors after they found out how much current was going thru the transformer.
Permalloy and other high nickel cores will disapprove of the dc much more than M6 (regular grade transformer iron).
cj

[quote author="cjenrick"]You should be OK for a line amp etc, but for something like a vari-mu compressor, you need a gap as the dc gets rather large due to mismatching of tube characteristics thru the whole db compressing range. Just a little dc can really drop the inductance thru the primary, and you will get a huge ac current increase as reactance is a direct function of inductance. I think that is the reason for the "mystery" resistors in series with the innerstage transformer on the RCA BA6a compressor. They added the resistors after they found out how much current was going thru the transformer.
Permalloy and other high nickel cores will disapprove of the dc much more than M6 (regular grade transformer iron).
cj[/quote]


Hi CJ, thanks for the infos!

I didn't know about different core materials making a difference. Now I conclude that the "better" transformers might generally tolerate less DC current (unless they have a gap on purpose), when they use better core material.

BTW, the background of my question was not the variable Mu project.
At the moment, I have several tube projects on my mind. PP to 600 Ohm output mostly.

JH.

[quote author="jhaible"]I didn't know about different core materials making a difference. Now I conclude that the "better" transformers might generally tolerate less DC current (unless they have a gap on purpose), when they use better core material.
[/quote]
It is nonsense to use mumetall in transformers with DC current,
with good Si-Fe cores you can have lower distortion.
To obtain smaller gap, you can EI divide to sections. Example for
classical natural gap:
EI EI EI EI EI EI EI EI
two sections is:
EI EI EI EI IE IE IE IE
(bigger inductancy)
four section:
EI EI IE IE EI EI IE IE
(bigger inductancy than upper ones)
and this is eight sections = classical transformer without gap = maximal inductancy:
EI IE EI IE EI IE EI IE

If you do not like Si-Fe, you can use perminvar. it is FeCo alloy and
is very very expansive, but have bigger Bmax .

xvlk

[quote author="xvlk"][quote author="jhaible"]I didn't know about different core materials making a difference. Now I conclude that the "better" transformers might generally tolerate less DC current (unless they have a gap on purpose), when they use better core material.
[/quote]
It is nonsense to use mumetall in transformers with DC current,
with good Si-Fe cores you can have lower distortion.
To obtain smaller gap, you can EI divide to sections. Example for
classical natural gap:
EI EI EI EI EI EI EI EI
two sections is:
EI EI EI EI IE IE IE IE
(bigger inductancy)
four section:
EI EI IE IE EI EI IE IE
(bigger inductancy than upper ones)
and this is eight sections = classical transformer without gap = maximal inductancy:
EI IE EI IE EI IE EI IE

If you do not like Si-Fe, you can use perminvar. it is FeCo alloy and
is very very expansive, but have bigger Bmax .

xvlk[/quote]

Interesting! Thanks!
I had considered the last one the "good" method (for mains transformers), but I see that for audio all the other versions make sense as well.
This audio transformer stuff is really fascinating.

JH.

Lundahl's tube line level output transformers are available with different gaps to accommodate different applications. There are "PP" versions for push-pull applications and these will tolerate a few mA of current imbalance, due to a small intentional gap between the halves of the C-core.

[quote author="kevinc"]Lundahl's tube line level output transformers are available with different gaps to accommodate different applications. There are "PP" versions for push-pull applications and these will tolerate a few mA of current imbalance, due to a small intentional gap between the halves of the C-core.[/quote]

Yes, C cores is for output with DC current best.
This is why Lundahl uses it.
C core is wounded core with anisotropic Si-Fe.
After wounding this core is cutted and if he put halves together,
you can not see any gap. You can insert thick foil between halves and
make very small gaps.
Side effect of C - cores is to use two coils. Transformer have
then sections (better hi freq response) and is astatic (less
sensitive to HOMOGENOUS spurious outer magnetic fields)
The homogenous I typed HOMOGENOUS, because badly
engineered shield can astatics degrade.

xvlk