Speaker Line Out Isolation Transformer question

[Soulchief]

Speaker Line Out Isolation Transformer question:

Does anybody know what Type/Spec 'Isolation Transformer' Suhr or Bradshaw uses for their Speaker Line Out Boxes? Correct resistances?

Want to attempt to make one.


I see these resistance specs options but don't know what is needed:
— 600:600
— 10k:10k
— 10k:600 ?

— Xicon, Jenson, Edcor ?


Anybody have any ideas or knowledge on this?


Thanks for any help

[Rob Strand]

The idea behind those device is very simple: 
   resistive attenuator --> transformer --> possible dummy load resistor

Which is essentially the same as,
   resistive attenuator --> off the shelf isolating DI box

The thing about those type of boxes is they aren't dummy loads and they aren't speaker sims.   Given they are targeted for guitar you might find some do actually roll-off the highs - I don't know.   (The Suhr unit makes a point of saying it doesn't.)

Even the Behringer GI100 DI has an isolation transformer but I can't rememeber if you can bypass the speaker sim part.

More than likely the transformer output impedance is 600 ohm.   The input side is up to you.   A 10k:600 will give some voltage attenuation already.


There's an example here, figure 2,
https://sound-au.com/project35.htm

[merlinb]

I'd opt for a 600:600 followed by a 1k level pot.

[Rob Strand]

I'd opt for a 600:600 followed by a 1k level pot.

A 600 ohm transformer at speaker level (50V pk or so) is going to be unnecessarily large.

[PRR]

I'd opt for a 600:600 followed by a 1k level pot.

A 600 ohm transformer at speaker level (50V pk or so) is going to be unnecessarily large.

I think you know what Merlin meant. Attenuate before the transformer.

Also 50V peak in 8r is a very powerful stage-amp. Even so, 35Vrms in 600r is "only" 2 Watts.

[Rob Strand]

I think you know what Merlin meant. Attenuate before the transformer.

Not at all because he just said "I'd opt for a 600:600 followed by a 1k level pot." but I see what you are mean and that idea looks fine to me.
Putting the tx after the divider,  like the ESP example,  maintains a true differential output.    If you put that
aside a post tx pot will work.

Also 50V peak in 8r is a very powerful stage-amp. Even so, 35Vrms in 600r is "only" 2 Watts.

Well I allowed for 16 ohm, since it's not uncommon on guitar amps. 
It's not really a matter of power of the 600R, it's more a matter of finding an off the shelf
600 ohm transformer which will handle such voltages.  It also makes the tx harder to design.

On the other side of the coin some of the small cheap transformers aren't so great either.   If you drive
the Tx with a low impedance you can extend the LF response (with the caveat of keeping away
from saturation). In other words no need to stick to a 600 ohm divider impedance at the input
of the tx to "match" the tx.  RG has mentioned that a number of times.

The ESP link I posted also links to some Tx info,
https://sound-au.com/articles/audio-xfmrs.htm

The iron-core example the core is 35 x 30 x 13mm and goes out to 7Vrms (article says 3Vrms).
(I'd say that's a modest size common in these types of devices.)

It doesn't add to the basic idea but here's another example with a schematic (no level control),
page has the user manual linked,

https://musicstudio.bigredroo.com.au/Recording_Gear_Dual_Direct_Inject_AM403_DI_Box.html

[merlinb]

Putting the tx after the divider,  like the ESP example,  maintains a true differential output. 

The output is still differential, it doesn't matter which side of the transformer the pot is on. Putting it on the secondary has the advantage of keeping the drive impedance to the transformer low (or at least constant).

It's not really a matter of power of the 600R, it's more a matter of finding an off the shelf 600 ohm transformer which will handle such voltages.  It also makes the tx harder to design.

Fair point

In other words no need to stick to a 600 ohm divider impedance at the input of the tx to "match" the tx.

Not sure what you mean by that. Cheap or not, the source impedance should be as low as possible for best audio performance. It's the secondary loading that is important for matching, not the source impedance.

[merlinb]

It doesn't add to the basic idea but here's another example with a schematic (no level control),
page has the user manual linked,
https://musicstudio.bigredroo.com.au/Recording_Gear_Dual_Direct_Inject_AM403_DI_Box.html

There are some serious errors of understanding on that page He shows the amplifier plugged into >100k impedance (totally unecessary), the guitar plugged into <1k impedance (good luck), and a grounded centre tap which destroys the benefits of a balanced output. Yikes!

[Rob Strand]

The output is still differential, it doesn't matter which side of the transformer the pot is on. Putting it on the secondary has the advantage of keeping the drive impedance to the transformer low (or at least constant).

It's not technically full differential.  Bill Whitlock (from Jensen) has written about it  number of times.  See pages 38, 39.  You want to have balance impedances to ground at every point - looking forward or backward.   Looking back into the terminals the simple divider isn't 100% compliant.
https://www.aes-media.org/sections/pnw/pnwrecaps/2005/whitlock/whitlock_pnw05.pdf

That's not to say a simple divider isn't OK.    On page 2, figure 2.2 vs figure 2.3.
https://www.jensen-transformers.com/wp-content/uploads/2014/08/an003.pdf

It's just that if you want high CMRR over the widest frequency range the "true different" is better.  Something in-between figures 2.2 and figure 2.3 is a U-pad attenuator, which maintains the impedances.
https://diystrat.blogspot.com/2013/08/diy-attenuators.html
A variable U-pad divider would replace R2 with a pot that opens and shorts.

To gain all the benefits you need a transformer with balanced capacitances feeding a a differential input. If you connect to a single-ended input or uses transformers that aren't really balanced then it all fall in a heap.

All I'm saying here is there is a fully differential version.   The simplified case like figure 2.2 is usually fine.

Not sure what you mean by that. Cheap or not, the source impedance should be as low as possible for best audio performance. It's the secondary loading that is important for matching, not the source impedance.

No doubt about it.

All I was implying is good transformers will have a good response with a drive impedance equal to the rated impedance (that's usually how they are spec'd) but the small transformer could probably need some help from a low-impedance.    Your 100R divider is good.  Whereas the ESP circuit with 1k pot might not be so great for those small transformers.

With posts like this it's hard to know where to set the quality bar.   I suspect Jensen transformers would be out but it's also possible those small Xircon's are pushing their luck as well.

There are some serious errors of understanding on that page He shows the amplifier plugged into >100k impedance (totally unecessary), the guitar plugged into <1k impedance (good luck), and a grounded centre tap which destroys the benefits of a balanced output. Yikes!

Good catch, I didn't read it all.  I was only hunting down a commercial example with a schematic.
[FYI: one of the pic does show Speaker or Dummy Load]

[merlinb]

It's not technically full differential. You want to have balance impedances to ground at every point

There is no ground connection, so it is fully differential and impedance balanced by definition. The article you cited is talking about electronic balancers, not a floating transformer winding.

[Rob Strand]

There is no ground connection, so it is fully differential and impedance balanced by definition. The article you cited is talking about electronic balancers, not a floating transformer winding.

That's just it, it's not technically balanced.   What Whitlock says about "it's all about impedances" is correct regardless of the situation.   That's the point he is trying to make.

For the L-pad:  looking back into the transformer one output lead feeds sees R1 to one transformer winding and R2 to the other transformer winding.  The other output lead feed directly into the transformer winding and via R1+R2 to the other.    So the output leads see different impedances looking back into the transformer.

The U-pad preserves the impedances looking back into the transformer.  Each lead feeds R1 into one transformer lead and R1+R2 into the other.

In terms of definitions that's the correct way to look at.   As for performance, the L-pad is OK for audio because the effects are small - hence why Whitlock gives it the thumbs up.

FWIW, it's not the audio signal that's the problem.  It's how external noise sources feed into the system.  Unbalanced impedances cause different noise levels to appear on the differential inputs and that degrades the noise rejection.

[merlinb]

That's just it, it's not technically balanced.   What Whitlock says about "it's all about impedances" is correct regardless of the situation.  In terms of definitions that's the correct way to look at. 

You are confusing different issues. What you say is true at the downstream end of the cable, not the source end. Pads are normally added at the input to preamps (i.e. downstream end) which is what Bill discusses, but that's not the case in this thread. With a floating transformer secondary it doesn't matter what sort of pad you use at the source end, the common-mode impedances will always be identical. It's easier to see if you draw the circuit rather than try to describe it in words.

[Rob Strand]

I think this thread is getting to esoteric for people.

This is the simplest example I can can put together.


It essentially follows figure 2 of this article,
https://sound-au.com/articles/balanced-2.htm

The changes I've made are the primary secondary capacitances of the transformer, addition of the unbalanced attenuator, and my common-mode impedances are done with capacitances instead of resistors.

(A differential load will reduce the levels but won't entirely removed the effect.)

[merlinb]

The changes I've made are the primary secondary capacitances of the transformer, addition of the unbalanced attenuator, and my common-mode impedances are done with capacitances instead of resistors.

Yes what you say is true for stray capacitance, but that is irrelevant to this thread.

[Rob Strand]

Yes what you say is true for stray capacitance, but that is irrelevant to this thread.

Guitar buzz comes from capacitively coupled noise.  That's the type of noise you are trying to get rid of with balanced system.

[merlinb]

Guitar buzz comes from capacitively coupled noise.  That's the type of noise you are trying to get rid of with balanced system.

It would take unrealistic levels of stray capacitance for a mere 1k pad to degrade the background hum coupling at power line frequencies. The guitar and the rest of the chain are going to be orders of magnitude more 'hummy' than the speaker line out box, even if it wasn't balanced at all. Something about seeing the wood for the trees.

[Rob Strand]

It would take unrealistic levels of stray capacitance for a mere 1k pad to degrade the background hum coupling at power line frequencies. The guitar and the rest of the chain are going to be orders of magnitude more 'hummy' than the speaker line out box, even if it wasn't balanced at all. Something about seeing the wood for the trees.

It was already mentioned a few times in the thread that the L-pad works fine. I even cited the Whitlock reference.

You started off by saying:
https://www.diystompboxes.com/smfforum/index.php?topic=130339.msg1264226#msg1264226

There is no ground connection, so it is fully differential and impedance balanced by definition. The article you cited is talking about electronic balancers, not a floating transformer winding.

That's not true.  The simulation shows that it's not the case, even for the floating transformer.