Non-gapped output-TX with DC... and 'counter-magnetizing' ?!

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

clintrubber

Well-known member
Joined
Jun 3, 2004
Messages
5,982
Location
The Netherlands
Hello,

I'm curious: could a non-gapped output TX with DC running
through its primary (like for BA283) be 'counter-magnetized'
by running a DC-current of suited polarity through an additional
winding ? (like a tertiary - some scaling of that current might be required)

I know this is wasteful w.r.t. current, but OK.

How about it ? Will it work ?
It will probably have been tried already (and abandoned... ?! :? )

Also: the countercurrent could be varied for controlling core-saturation,
as suggested here by I forgot who.


Thanks,

Peter

BTW, how could one easily see when there's adequate compensation ?
We could add fancy servo-like loops to copy the DC of the first winding,
but that might be too much.
I guess this needs some X-Y-scope-juggling, but are there other ways ?
 
There's some good questions there. I need to think.

In the mean time, read these pages:

http://www.ee.surrey.ac.uk/Workshop/advice/coils/BH_measure.html

Jakob E.
 
Thanks for having a look at this thread & also for the link.

I'm curious - it's in fact related to what I was wondering here: does it matter for 'the sound' (...) to have DC/magnetizing going on ?

from
http://www.groupdiy.com/index.php?topic=4318&start=15
Is DC-through-the-primary essential for the sound of circuits like the Neve BA283 ?
Or would it have been 'just as nice' with no DC through a non-gapped TX and using AC-coupling ? (w.rt. 'sound' - so apart from efficiency and or certain smaller side-effects)

Or in other words (and as far as this could be judged separately):
Is it the class A & a sweating TX in general -or- the core becoming saturated asymmetrically ?

Thanks,

Peter
 
from Rafafredd:
In my mind, if it works in a push-pull transformer, why it wouldn´t work in a single ended one?
Right, one would think so. So I'm fairly confident it can be done... (?)

There may be an issue with with providing an exact amount of countermagnetizing but I can't image there's more difference/tolerance between say two primary windings and the two halves of a single centre-tapped primary like you mentioned.

But as long as there's an easy way to determine whether the correct amount counteraction is provided, I figure one could use in principle any separate winding (there's that tertiary winding again) - all within bounds of course.

If this all can work but just isn't done often I guess it's just because of the wasteful use of DC, but I don't mind.
And it would give that saturation-control without disturbing the biascurrents of the (say) BA283 itself.


Bye,

Peter
 
I guess that you could easily balance DC currents with a gadget like a DC servo, comparing voltage across two equal dropping resistors - one in series with primary and one in series with secondary.

----

Another (related) thing that I've been thinking about for a while now, is using HF bias to "break" the magnetic hysteresis of audio transformers..

---

Jakob E.
 
from Jakob:
I guess that you could easily balance DC currents with a gadget like a DC servo, comparing voltage across two equal dropping resistors - one in series with primary and one in series with secondary.
Right, this shouldn't be adding much circuit complexity. I have no idea whether those tertiary windings of those TXs I have around have the same amount of turns but it could all be scaled by means of those resistors once the prim-to-tertiary ratio is measured.

At least I assume what kind of winding is counteracting doesn't matter, as long as the induced prim flux & the counteracting flux are equal, right ?
So something like
DC_prim * turns-ratio = DC_counteracting
(should check if it's as simply as the turns-ratio or that another thing sneaks in here).

And a mismatch could be introduced easily with a servo.
Since this is what I'm still very curious to: how essential is that DC through the 'BA283-primary'?
Some places mention 'it's the input-TX', others point at the
'DC-thru-primary' for explaining the 'sound' (...) of this preamp.

Well, at some point one simply has to start soldering :!:

Another (related) thing that I've been thinking about for a while now, is using HF bias to "break" the magnetic hysteresis of audio transformers..
Thats interesting. Maybe I'm taking too much steps at a time now or am I missing your point altogether, but do you hint at making ungapped transformers suitable for DC by means of HF-biasing ?
Or do you want to do this for something else ? Can't imagine you are after 'cleaning-up' of TX-influences :wink:

(No idea BTW what they exactly do in tape-heads w.r.t. HF, DC, ...)


Bye,

Peter
 
Hi,

Here we have a little drawing, please shoot:

283_nongapped_01.jpg


FWIW, using R7 for the current sensing in the primary avoids
messing with the original circuit.

OK, so it looks like it can work with a non-gapped TX.
But the big question is of course whether the 'sonic signature'
is still there... :roll:

And here we have those Ampex-transformers again :wink:
I'm not too cheap to get me gapped TXs, it's just that I'm curious to where
that Neve-signature is coming from !

TX-out_pic07.jpg



Bye,

Peter
 
[quote author="clintrubber"]I'm not to cheap to get me gapped TXs, it's just that I'm curious to where
that Neve-signature is coming from !

[/quote]

I think it´s a really nice thing if it works. So, teoretically, you can feed DC throught the windings, without a gapped transformer and without wasting any inductance. If it works, it will be very good in many aspects.
 
from Rafa:
If it works, it will be very good in many aspects.

I figure it just works, but (probably apart from the magic saturation-controlTM :grin: ) it won't have advantages in sound w.r.t. the real thing, do you think ?
It's just that gapped TXs can now be used. But Rafa do you see other benefits ?


Hmm, but hey, the output-section probably doesn't matter at all...
and it's the input-TX... :cry: :shock: :sad: :?: :!: :?:

...since it's been reported/claimed those Vintech X73 & X73i preamps sound alike - and based on their descriptions I bet the economy version ('i') doesn't use an output-TX.

Vintech X73 http://www.vintech-audio.com/x73_info.html

Vintech X73i http://www.vintech-audio.com/x73i_info.html

But if it sounds OK it's all fine, and let's not spoil X73i-sales - but let's now jump back again to the circuit above :guinness:

Bye,

Peter
 
Yes, maybe it will miss that DC on the core vibe, but one would have to build and listen to know, if possible comparing with the another transformer with the same winding structure, but gapped. The main good thing about it is that you can avoid coupling caps without wasting primary inductance. You see... Even with a gapped transformer, you waste inductance when you run DC throught it. That´s why gapped transfoprmers are usually bigger. This trick should lead to better low frequency response and distortion figures than the gapped transformer, I guess...

Maybe one could use litz wire at the primary for a split winding and then use one of the windings for DC+audio and the other for counteract-DC only...
 
The problem is that the ac on the primary will be induced into the secondary of your tert winding, which will then be shorted into that servo circuit.
I ran into this problem while trying to put a dc bias on the Neve outputs I was testing. I almost blew up the power supply with all the ac going into it.

What you have in the Neve is an ac signal superimposed on some dc. If you use a dc canceling circuit, you will cancel out your signal. I guess this is why you don't see a dc canceling circuit, best just to use a gapped transformer and forget about it. Or use the Neve Class A/B circuit that uses cap input to the pri of the LO2567.
 
Yes, maybe it will miss that DC on the core vibe, but one would have to build and listen to know, if possible comparing with the another transformer with the same winding structure, but gapped.

You're right, there are a lot of differences going on here at the same time(among other things, I need to characterize that TX a bit further).

That´s why gapped transfoprmers are usually bigger. This trick should lead to better low frequency response and distortion figures than the gapped transformer, I guess...

There we have a benefit - hmm, but lower distortion, this output-stage should give oomph & warmth, right ? :wink:

The principle definitely holds yes, the gap-drawbacks are gone.

In practice though, I've compared the dimensions to those five in the MoMetal-thread and the non-gappedness might just make up for the smaller size of those Ampex-TXs :oops:

Maybe one could use litz wire at the primary for a split winding and then use one of the windings for DC+audio and the other for counteract-DC only...

If I understand this correctly: Yep, you need three windings. Those Ampex-types happen to have that third.

But for a two-winding TX I guess you have to add quite a lot of windings otherwise you have to spend silly amounts of the counteracting current:
it scales by the P to T turns ratio. Those Ampex-TXs have - I think - a ratio of 2:1 (P:T) so I would already need more than 120 mA in the third winding.

Note though that this current does not need to flow from the 24V supply :thumb:
If worth while, a separate low-voltage supply could provide the counteracting current, waisting less power in the 'BD...'-transistor.

Bye !
 
The problem is that the ac on the primary will be induced into the secondary of your tert winding, which will then be shorted into that servo circuit.
I ran into this problem while trying to put a dc bias on the Neve outputs I was testing. I almost blew up the power supply with all the ac going into it.

What you have in the Neve is an ac signal superimposed on some dc. If you use a dc canceling circuit, you will cancel out your signal. I guess this is why you don't see a dc canceling circuit, best just to use a gapped transformer and forget about it. Or use the Neve Class A/B circuit that uses cap input to the pri of the LO2567.

Thanks CJ, there it all stops :evil: :oops: :cry:

Hadn't thought about the induced signal in the tertiary !

Hmm, could it be solved ? What's actually happening ? The tertiary tries to put the signal out and sees the hi-Z load of the BD...-device. So its collector starts jumping, but I figure the DC-current keeps flowing and stays the same. Might very well be wrong here !

Wait, weren't old speakers magnetized as well ? I don't really know them, but weren't there types without permanent magnet ? They required some kind of electromagnet if I'm remembering it OK.

OK, some sleep now, let's see where this all goes in the end.
But either way, thanks for the responses everyone ! :guinness:

Bye,

Peter
 
We will need a current generator that has low impedance at sub-audio frequencies and high impedance at audio frequencies.

An inductor in series with our DC-bias generator output?
 
from Jakob:
We will need a current generator that has low impedance at sub-audio frequencies and high impedance at audio frequencies.

An inductor in series with our DC-bias generator output?

The opamp+transistor is a VCCS, with already a high output impedance
for audio (& could be made higher - if needed).

But would the low impedance at sub-audio frequencies be needed ?

While then having the same number of irons, adding an inductor
here would give the whole structure the benefit w.r.t. parafeed
of not having a coupling cap in the audio path.
(I found the url back but forgot in which thread it was relevant :grin:
http://members.aol.com/sbench/outstru.html )


Bye,

Peter
 
[quote author="clintrubber"]Hi,

Here we have a little drawing, please shoot:

283_nongapped_01.jpg


FWIW, using R7 for the current sensing in the primary avoids
messing with the original circuit.

OK, so it looks like it can work with a non-gapped TX.
But the big question is of course whether the 'sonic signature'
is still there... :roll:
[/quote]

Just run the voltage from the current sensing resistor thru a LPF before it goes to the positive input of the opamp that makes the VCCS.

JH.
 
[quote author="clintrubber"]Hello,

I'm curious: could a non-gapped output TX with DC running
through its primary (like for BA283) be 'counter-magnetized'
by running a DC-current of suited polarity through an additional
winding ? (like a tertiary - some scaling of that current might be required)

I know this is wasteful w.r.t. current, but OK.

How about it ? Will it work ?
It will probably have been tried already (and abandoned... ?! :? )
[/quote]

It's been done in the Urei LA-3A.
There, the output transformer is capacitor coupled, but as the same cap is also used for bootstrapping the driver stage of the output amp, the bias current of the driver stage (not the output stage!) flows thru the transformer.
This is compensated by an extra winding with a DC current. No fancy servo loop, though: Just a resistor in series, and connected to the supply voltage. I guess it's precise enough because they only have to compensate the (rather small) driver stage current.

JH.
 
from JH:
It's been done in the Urei LA-3A.
There, the output transformer is capacitor coupled, but as the same cap is also used for bootstrapping the driver stage of the output amp, the bias current of the driver stage (not the output stage!) flows thru the transformer.
This is compensated by an extra winding with a DC current.
Interesting !

http://www.waltzingbear.com/Schematics/Urei/Urei_LA-3A.JPG

Why would they want to run the driver stage current through the TX ?

No fancy servo loop, though: Just a resistor in series, and connected to the supply voltage. I guess it's precise enough because they only have to compensate the (rather small) driver stage current.
The resistor indeed does it.
The BA283-mod could be simpler as well. Adding the opamp etc makes it automatically tracking the bias-pot setting, that's all... Probably isn't even a 'real' servo - it's just a controlled current source sensing an AC-silent voltage.

Bye,

Peter
 
Probably isn't even a 'real' servo - it's just a controlled current source sensing an AC-silent voltage.

As long as you know the winding ratio (of driver winding and compensation winding), the VCCS will do fine. No closed loop, i.e. no servo, though.

OTOH, if you feel the need to _make_ a servo (unknown winding ratio? but even then once you've measured it you know it, it won't do bad things like drift to another ratio (;->) ), that would be hard, because the entity you want to control is the DC magnetisation current, and you cannot easily measure this. Not without a DC-sensitive device like a hall sensor at least.

So I'd say make sure about this winding ratio, and then make sure that your compensation current will follow all the possible electronic drifts inside your circuit. The VCCS (with added LPF) that measures the actual amplifier DC output current will do this just fine, even if the bias inside the amp, or the supply voltage, drifts.

JH.
 
Back
Top