I was looking at an application note from Jenson on the AS016 preamp(see attached pdf) and noticed that the mic impedance switch simply added 68.1 ohm resistors in series with the microphone. Why did they chose that value?
Last edited:
I am just guessing that the values are chosen to provide a theoretical standard 150Ω input impedance when used with their amplifier (which isn't shown and we don't know its input impedance) and when the attenuator is set to 0db.
Yeah I also confused on why they labeled a low impedance label other than 150 ohms. I thought that 600 ohms was the common higher impedance value.
Deane Jensen was a very precise designer.
When you design a transformer you must balance many things. In particular the high-end resonance shape is determined by stray inductance, stray capacitance, and all the circuit resistances (source, windings, loads).
One way to tame resonance is to add resistance. However that reduces noise figure.
Older dynamic, ribbon, and most condenser head-amps had output impedances near 150 ohms (120-250 more or less). Deane relied on this to shape the top-end response.
Some high-performance head-amps, maybe more today than when Deane designed much of the classic Jensen products, have high NFB amplifiers and near-zero output impedance.
The near-zero Zout helps drive long lines, absorbs hum, and improves the bass response of a transformer.
However it spoils the careful high-end shape Deane aimed for.
I'm no Deane, nowhere near. But say I designed a transformer for the upper (green-dot) situation: 2K source. The green plot shows a hi-end rise which could be tamed other ways.
Now change the 2K source to a few ohms, red plot. That is a ZINGER of a hi-end rise, and can't be tamed later in the signal path.
In the hands of an expert, like Deane, the green plot can be held to 1db over the audio band. I think when these transformers are used with OUT a ~~150 ohm source, with a near-zero source, there is a couple db rise at/above the audio band.
> A simulation free zone. Design it, build it, test it.
Yes, SPICE lies. I know I will never see these curves on real iron. I've flogged enough real transformers and windings to know that. But I'm not going to "build it", again, for some casual internet question. Nor even use the better SPICE models with core-loss.
Note that many other mike input transformers, especially the cheap ones, are not as good as Jensen's at their best, but OTOH are too low-strung to get way out of shape when working from zero-Z. You simply won't see this kinda zinger on M6 laminations, the core losses suck it out.
When you design a transformer you must balance many things. In particular the high-end resonance shape is determined by stray inductance, stray capacitance, and all the circuit resistances (source, windings, loads).
One way to tame resonance is to add resistance. However that reduces noise figure.
Older dynamic, ribbon, and most condenser head-amps had output impedances near 150 ohms (120-250 more or less). Deane relied on this to shape the top-end response.
Some high-performance head-amps, maybe more today than when Deane designed much of the classic Jensen products, have high NFB amplifiers and near-zero output impedance.
The near-zero Zout helps drive long lines, absorbs hum, and improves the bass response of a transformer.
However it spoils the careful high-end shape Deane aimed for.
I'm no Deane, nowhere near. But say I designed a transformer for the upper (green-dot) situation: 2K source. The green plot shows a hi-end rise which could be tamed other ways.
Now change the 2K source to a few ohms, red plot. That is a ZINGER of a hi-end rise, and can't be tamed later in the signal path.
In the hands of an expert, like Deane, the green plot can be held to 1db over the audio band. I think when these transformers are used with OUT a ~~150 ohm source, with a near-zero source, there is a couple db rise at/above the audio band.
> A simulation free zone. Design it, build it, test it.
Yes, SPICE lies. I know I will never see these curves on real iron. I've flogged enough real transformers and windings to know that. But I'm not going to "build it", again, for some casual internet question. Nor even use the better SPICE models with core-loss.
Note that many other mike input transformers, especially the cheap ones, are not as good as Jensen's at their best, but OTOH are too low-strung to get way out of shape when working from zero-Z. You simply won't see this kinda zinger on M6 laminations, the core losses suck it out.
Thanks for the plots...
Mics and mic input stages aren't my field of expertise I'm afraid.
What you show makes sense but there is still huge variablity to my simple way of thinking about this, huge in comparison to specifying a 68.1 Ω resistor for example rather than an off the shelf 68 Ω. A mic impedance of 120 to 250 Ω absolutely swamps that. 10 Ω to 2k is a massive alteration in impedance.
Maybe it was designed as a perfectionist example of something that obvioulsy needed all the partnering ancillaries to go with it.
Mics and mic input stages aren't my field of expertise I'm afraid.
What you show makes sense but there is still huge variablity to my simple way of thinking about this, huge in comparison to specifying a 68.1 Ω resistor for example rather than an off the shelf 68 Ω. A mic impedance of 120 to 250 Ω absolutely swamps that. 10 Ω to 2k is a massive alteration in impedance.
Maybe it was designed as a perfectionist example of something that obvioulsy needed all the partnering ancillaries to go with it.
68R1 is a standard 1% value. Nothing mysterious about that last tenth ohm.
PRR, thanks for the graphic explanation.
PRR, thanks for the graphic explanation.
Thanks for the reminder SY 🙂 I tend to think in just E24/48 series values.
I guess it's just what the designer wanted to use or had available. At 1% tolerance it falls mostly within a 68 Ω 1% values anyway.
I guess it's just what the designer wanted to use or had available. At 1% tolerance it falls mostly within a 68 Ω 1% values anyway.
> specifying a 68.1 resistor ... rather than an off the shelf 68
Deane Jensen was precise... _his_ "shelf" probably had only 1% resistors.
"68" is specified as "68.1" in 1% series so that it is much harder for workers who have both 5% and 1% shelves to mistakenly use a 5% part where a 1% part is wanted.
I'm undecided if 1% is truly needed here. Depends on the winding balance.
But certainly in a system with $100++ opamps and $100++ transformers, the cost of possibly "over-specifying" the resistors is negligible.
I'm sure the 68/68.1 value is a dart-toss. Since the design specs say 150 source, 75 would be a reasonable value. Or maybe the problem first surfaced on a mike with 12 ohm output, hence (150-12)/2= 68, yeah makes sense, except there's lower and higher Z mikes. 39 would give nearly as much damping with lower NF. If you have a bucket of good 47 ohm, use that.
> 10 to 2k is a massive alteration in impedance.
Is a hasty-example, with round/default-number reactances, and a too-simple transformer model.
And I did not quantify that "near-zero" impedance of some snazzy mikes. 1 to 200 is the same ratio as 10:2K, and some high-NFB transformerless head-amps must get down to thar zone.
Deane Jensen was precise... _his_ "shelf" probably had only 1% resistors.
"68" is specified as "68.1" in 1% series so that it is much harder for workers who have both 5% and 1% shelves to mistakenly use a 5% part where a 1% part is wanted.
I'm undecided if 1% is truly needed here. Depends on the winding balance.
But certainly in a system with $100++ opamps and $100++ transformers, the cost of possibly "over-specifying" the resistors is negligible.
I'm sure the 68/68.1 value is a dart-toss. Since the design specs say 150 source, 75 would be a reasonable value. Or maybe the problem first surfaced on a mike with 12 ohm output, hence (150-12)/2= 68, yeah makes sense, except there's lower and higher Z mikes. 39 would give nearly as much damping with lower NF. If you have a bucket of good 47 ohm, use that.
> 10 to 2k is a massive alteration in impedance.
Is a hasty-example, with round/default-number reactances, and a too-simple transformer model.
And I did not quantify that "near-zero" impedance of some snazzy mikes. 1 to 200 is the same ratio as 10:2K, and some high-NFB transformerless head-amps must get down to thar zone.
I would think that much of Deane Jensen engineering was done with a slide rule rather than a scientific calculator.
- Status
- Not open for further replies.