Question about passive attenuation

[johnheath]

Hi all...

According to my attached picture I would like to ask how a passive attenuator on the input transformer primary affect the input impedance.

In A there is just a pic of transformer and a resistor and if I understand it correctly the impedance in will be 1k ohm in that specific case if nothing else is interfering.

I am not sure if Ohm's law is useful to calculate the value of the attenuator... making it a parallel resistance. If so the value of the primary impedance won't go higher that the lowest of resistances. ( I am not sure if this is how it works with impedances ... ohm's law is understood)

If it works then you should be able to use higher values of total resistance in the attenuator since the Zin is 1k and the total value of Zin will not exceed 1k???

If NOT working... How does the total value of an attenuator affect the impedance in? I am also curious about the shown attenuators how they affect the impedance since I have seen them her e and there.


I am just brainstorming with myself here so please be gentle


Best regards

/John

 

[johnheath]

and the forgotten picture

 

[joaquins]

  Here just Ohm's law, all resistances, no reactive impedances. Even with Reactances you can apply Ohm's law using complex numbers.

The B circuit is Zi=2*R1+{(R2+a*R1)||1k}
'a' being the ratio the pot is set at,( 0 is bottom, max att, 1 is top, min att)

For the C circuit you have Zi=(1-a)*R1+{(a*R1)||1k}
'a' means the same as the B circuit)

JS

 

[PRR]

B, C: you will almost always be better taking loss between transformer and tube, than before transformer.

Transformer needs to see a low and fairly stable impedance on one side, conventionally the input.

Tube grid will tolerate low to medium impedance OK.

Exception will be huge signals into small iron. Even then a fixed pre-pad is preferable to a variable impedance.

Constant impedance resistor networks exist but need too many switch-points to use routinely (in today's economy).

 

[johnheath]

  Here just Ohm's law, all resistances, no reactive impedances. Even with Reactances you can apply Ohm's law using complex numbers.

The B circuit is Zi=2*R1+{(R2+a*R1)||1k}
'a' being the ratio the pot is set at,( 0 is bottom, max att, 1 is top, min att)

For the C circuit you have Zi=(1-a)*R1+{(a*R1)||1k}
'a' means the same as the B circuit)

JS

Thank you sir...

I will have a look at some math later on and also consider suitable attenuation levels.

Best regards

/John

 

[johnheath]

B, C: you will almost always be better taking loss between transformer and tube, than before transformer.

Transformer needs to see a low and fairly stable impedance on one side, conventionally the input.

Tube grid will tolerate low to medium impedance OK.

Exception will be huge signals into small iron. Even then a fixed pre-pad is preferable to a variable impedance.

Constant impedance resistor networks exist but need too many switch-points to use routinely (in today's economy).

Thank you sir

It is mostly for understanding it all a little better.

It is mostly in different vari-mu compressors that I have seen an attenuator before input transformer (and often together with an attenuator after the output transformer). I am just curious how the affect the impedances.

Just a another thought... if you can determine the impedance by using a fixed resistor on the secondary... like in this case a 100k to make the primary impedance 1k. What happens if you use the 100k on the secondary and terminate the primary with lets say a 3k6 resistor? (Just an example here) ... Is it just a bad mixup?


Best regards

/John

 

[abbey road d enfer]

I am just curious how the affect the impedances.

There is a factor you have neglected, and which PRR hinted at. The impedance seen by the xfmr's primary is very significant in terms of LF response and LF distortion. Worst case is at about 6dB attenuation.
Ideally, you should make sure that the output impedance of the attenuator is significantly lower than the nominal impedance of the primary.

 

[PRR]

> in different vari-mu compressors

It is more-awkward to put a pad into the push-pull grids.

These boxes typically use a "constant impedance" pad on the transformer primary.

Yes, in non-precision work (most music sweetening) a 1K pot may be constant-enough for a 600r winding.

 

[emrr]

For the sake of your mental exercise, consider the Langevin 117A input loading.  It's a 600:65K.

 

[abbey road d enfer]

For the sake of your mental exercise, consider the Langevin 117A input loading.  It's a 600:65K.

Not exactly. It's a 150+150:50-60k CT.  Since the secondary is loaded with 100k, the reflected impedance at the primary is about double than nominal; that's why they added the 620r resistors across each half-primary. I believe the OP is already enough confused without adding the complexity of split primaries and combined loads.

 

[emrr]

EXACTLY.  Overload is the mother of education.  This was one of my biggest breakthrough moments. 

AND

I told you what the ratio was, explicitly, yet you theorize about it, throwing the student further off the trail. 

 

[abbey road d enfer]

I told you what the ratio was, explicitly, yet you theorize about it, throwing the student further off the trail.

The thing is, the math doesn't agree with your statement. If the ratio was 600:65k, loaded with 102k, the reflected primary Z would be 923 ohms. With the input shunt of 1240 ohms, that would result in an actual input Z of 529 ohms. That is a difference of nearly 20%, that can't be  can't be chalked up to manufacturer's sloppiness. I would think the 600 ohms specified impedance is unquestionable, so something else is off the mark; I would first question the transformer's ratio and its losses. Defining a transformer with an impedance ratio is what the customer wants, but is very inaccurate: you can have two xfmrs with identical impedances and significantly differing voltage ratio, because of the DC resistance.

 

[emrr]

I come up with 535 ohms on primary considering only stated impedances and load resistances.  500-600 considered interchangeable at the time, also forces slightly higher bandwidth, I judge as purposeful decision to maximize system bandwidth.  Winding resistance is 48R:6K2. 

Take the resistances out and reconnect for microphone use.  Gain goes up 12dB from combination of ratio and loading changes. 

 

[johnheath]

> in different vari-mu compressors

It is more-awkward to put a pad into the push-pull grids.

These boxes typically use a "constant impedance" pad on the transformer primary.

Yes, in non-precision work (most music sweetening) a 1K pot may be constant-enough for a 600r winding.

Thank you sir

Yes, I have seen those fixed pads but also variable versions as well.

What would you say... Is there a "good" version of variable pad or should you just try to stay with a fixed for simplicity?


Best regards

/John

 

[johnheath]

Not exactly. It's a 150+150:50-60k CT.  Since the secondary is loaded with 100k, the reflected impedance at the primary is about double than nominal; that's why they added the 620r resistors across each half-primary. I believe the OP is already enough confused without adding the complexity of split primaries and combined loads.

Thank you sir... not so much confused as "not educated"

Best regards

/John

 

[johnheath]

EXACTLY.  Overload is the mother of education.  This was one of my biggest breakthrough moments. 

AND

I told you what the ratio was, explicitly, yet you theorize about it, throwing the student further off the trail.

Thank you sir... for having a feeling of how I would like to understand it all... I want a breakthrough as well .

Anyhow... I found the schematic of the Langevin 117A and it has a wide variety of windings that is understood.

I see those two 620R resistors in series and in parallel with the reflected impedance they make up a "new" actual impedance... I see that you and "Abbey" here are not exactly agreeing on the exact numbers. If I calculate from 600R:65K I also get 535R as actual input impedance but I will not argue about the specific transformer ratio.

Best regards

/John

 

[johnheath]

So gentlemen

To summarize it a bit... If I understand it correctly Resistors in parallel with the reflected input impedance "creates" an actual impedance.

If using a pad of any variety, before the input transformer, its total resistance should be significantly lower than the nominal input impedance of the input transformer. Just to clarify some language barriers..."nominal value here is the value stated of the transformer? Like a 200:20k transformer that nominal input impedance is 200R?

Input transformer have better performance with a fixed pad instead of a variable one.


So finally a question: Are there any useful variable pad that is appropriate to have before the input transformer... for use when it is a bit tricky to add an attenuator inside a circuit like a vari-mu comp?


Best regards

/John

 

[emrr]

Look at mic pads; usually match the transformer nominal.  Some transformers (modern, generally) in some situations require a very low source Z to act properly, most old transformers are made to deliver quoted spec with a matching source.  Nominal 200 ohm mic in to nominal 200 ohm primary, or pad with 200 ohm shunt.  No clear answer, you need to measure and observe. 

Input transformer may behave differently at different points on a variable pad, not necessarily out of spec. 

As said, classic was to put a matching value ladder attenuator before the input.  Bean-counters later spent less money to put a dual pot on the secondary side instead, which may force use of a fixed pad on primary to avoid transformer overload.  Depends on transformer. 

 

[johnheath]

Look at mic pads; usually match the transformer nominal.  Some transformers (modern, generally) in some situations require a very low source Z to act properly, most old transformers are made to deliver quoted spec with a matching source.  Nominal 200 ohm mic in to nominal 200 ohm primary, or pad with 200 ohm shunt.  No clear answer, you need to measure and observe. 

Input transformer may behave differently at different points on a variable pad, not necessarily out of spec. 

As said, classic was to put a matching value ladder attenuator before the input.  Bean-counters later spent less money to put a dual pot on the secondary side instead, which may force use of a fixed pad on primary to avoid transformer overload.  Depends on transformer.

Thank you sir

I have seen circuits using a dual pots but it seems hard to find one with somewhat matching values... I am not sure how that would affect the performance on the following grids making the matching uneven?  So in those cases the fixed pads on the primary is "fixing" the reflected impedance on the secondary?

Please correct me if I am wrong when saying that the total resistance of a dual pot (lets say 2 x 50k) is still seen as 100k even though you alter the pot?

Best regards

/John

 

[abbey road d enfer]

If I understand it correctly Resistors in parallel with the reflected input impedance "creates" an actual impedance.

That is correct. Reflected impedance is in fact somewhat reactive, but not so much, so we allow ourselves to consider them like resistances (a physics teacher would shame us, be an EE would just look elsewhere).

If using a pad of any variety, before the input transformer, its total resistance should be significantly lower than the nominal input impedance of the input transformer.

Not the total resistance, the resistance seen when probing its output. It involves analysing the network including the source impedance, the series resistance (which is actually two resistors in the case of a balanced attenuator)  and the shunt resistor; you must work your Kirchoff theorem.

Just to clarify some language barriers..."nominal value here is the value stated of the transformer? Like a 200:20k transformer that nominal input impedance is 200R?

Correct. Note that this nominal impedance has actually no physical existence. It is a target value that is used at the design stage for calculations.

Input transformer have better performance with a fixed pad instead of a variable one.

Not exactly. With a fixed attenuator, the performance is fixed, with a variable attenuator, the performance is ... variable.

So finally a question: Are there any useful variable pad that is appropriate to have before the input transformer... for use when it is a bit tricky to add an attenuator inside a circuit like a vari-mu comp?

There is no single answer to that. Mic inputs and line inputs have different requirements. For mics, a U-pad provides the lowest noise, for lines, an H-pad may be necessary in order to maintain proper frequency response.