R/E/P Community

R/E/P => R/E/P Archives => Klaus Heyne's Mic Lab => Topic started by: soapfoot on January 15, 2011, 04:15:58 PM

Title: U47 schematic, component-by-component
Post by: soapfoot on January 15, 2011, 04:15:58 PM
Here's something I thought might be a cool exercise.  I know there's a lot of knowledge here, and mine is limited, so I hope others can help to fill in the blanks.

I thought it might be educational to go through the U47 schematic component-by-component and explain the purpose of each component one at a time.  The idea would be to learn a little bit more about how tube microphones in general work.

If anyone thinks this is a worthwhile thing, add your insight and I can update this post, hopefully until it's complete.  If any of my initial suppositions are wrong, let me know and I'll correct.

Here's the schematic.

http://img.photobucket.com/albums/v283/jazz347/47-inspired%20mic%20build/u47.gif

Obviously KK47 is the capsule head, and U1 is the BV8 transformer.  The VF14M is obvious as well.  But what are the functions of the other components?

Let's go one at a time.  There are 8 resistors and 3 capacitors-- shouldn't be too overwhelming.  Thanks to Kai, JJ, JohnR, and others for helping fill in the blanks and increase my (our?) understanding of this seemingly simple circuit.  If there are any additional corrections/clarifications, I'll keep updating this post.

R1 - With C1, forms a LPF to remove hum/noise from capsule bias voltage (Kai)

R2 -  negative biasing resistor, works with R3 to determine bias point of the tube. Together with capsule capacitance, forms a high pass filter at about 21 Hz. (Kai)

R3 - allows a positive voltage tapped off the heater supply to develop on the cathode. (Kai called this "positive biasing resistor")

R4 - Dropping resistor to obtain filament voltage from B+ voltage

R5 - 100k plate load resistor for the VF14M, active for AC signal

R6 - With C3, forms low pass filter to filter out noise from plate supply, also combines with R5 to form plate resistance to set the DC working point of the tube's plate. (Kai)

R7 - part of voltage divider to get capsule polarization (JJ)

R8 - part of voltage divider to get capsule polarization (JJ)


C1 -With R1, forms LPF to remove noise from capsule bias voltage & forms AC ground coupling for capsule backplate. (Kai)

C2 - Blocks DC before U1, shapes timbre (OT coupling cap)

C3 - With R6, forms a low pass filter to filter out noise from plate supply (Kai)

Title: Re: U47 schematic, component-by-component
Post by: J.J. Blair on January 15, 2011, 10:20:15 PM
Since nobody of the experts are chiming in, here's what I'll offer (corrections welcome):

R1 is the capsule bias resistor.

R2 is the control grid resistor.  A very high frequency low-pass filter.  Different values sill determine the top end of the mic, which is why you see many 1GΩ resistors in that position in modern mics.

R5 and R6 are part of the plate loading circuit.  I'd love more explanation on this from somebody, please.

R7 and R8 form the voltage divider to you your capsule voltage.  Resistor to ground (3) / the sum of both resistors (2+3=5) = 3/5 x 105 = 63

C2 is the output / transformer coupling cap. I think you have this confused with C3.  
Title: Re: U47 schematic, component-by-component
Post by: soapfoot on January 16, 2011, 12:41:04 AM
nice!  Thanks J.J.  I understand a little more already.  You were right about my typo.

I fixed that and also updated the OP with your additions.

In case you know in greater detail-- what is the function of the capsule bias resistor?  What's its purpose in life and what happens if you move the value up or down?

(According to johnR, the following does not apply--thanks!):

The 60M grid stopper (R2) makes sense now that you mention it.  In guitar amps, these are used to filter out RF, prevent parasitic oscillations, and help prevent the grid from going positive with respect to the cathode.  I assume a similar function here?  Am I safe in assuming that as the resistor increases, the corner frequency of the LPF moves upward... 60M seems very high (high enough to get out of the way of the audible band?)  Interesting that people tweak it higher to get increased HF response.  I wonder what the tradeoff is.

I'm especially curious also about C3.

Thanks for your input!
Title: Re: U47 schematic, component-by-component
Post by: J.J. Blair on January 16, 2011, 02:24:08 AM
Brad, I'm not an electronic engineer or designer.  I'm simply a tinkerer.  I wish I could give you a more informed answer, and I'm hoping that some of the resident geniuses here will fill in the blanks, as well as correct what misinformation I might have.  
Title: Re: U47 schematic, component-by-component
Post by: johnR on January 16, 2011, 08:13:27 AM
There seems to be some confusion here. R2 isn't a grid stopper. It's a grid bias resistor, and as such it just holds the tube's control grid near ground potential. It doesn't actually have anything to do with low pass filtering (although the resistor's parasitic capacitance may have some effect at high frequencies).

R2 does form a high pass filter in conjunction with the capsule capacitance, so increasing its resistance would extend low frequency response at that point in the circuit (whether that's desirable is another matter).
Title: Re: U47 schematic, component-by-component
Post by: J.J. Blair on January 16, 2011, 09:42:59 AM
Thanks, thanks for the clarification.  I've even heard it called a "grid shunt," and when using that term, David Bock corrected me, saying it wasn't shunting anything.  

I didn't say "grid stopper," but I have noticed an extension in the high end on the U47 when increasing the value of R2.  The information I had found said "very high frequency LPF," in explaining the function, which corresponded with my experience.  The only thing I could find describing the circuit called it the control grid resistor.  

I appreciate you're chiming in.  I actually didn't want to be the first to offer anything up, but nobody was saying anything, so I figured I'd get the ball rolling, and you guys could make corrections as we go along.  

Also, I'd love to know about the function of C1 and C3, if anybody can elaborate on that.
Title: Re: U47 schematic, component-by-component
Post by: soapfoot on January 16, 2011, 09:47:39 AM
thanks both.  I've made the correction(s) above.

Would love to learn more!

Here's a question I have.  It might be elementary, but like others I'm a "tinkerer" with no formal training--

The voltage divider created by R7 and R8, given an input voltage of 105v, will output a voltage of 63v.

However, between that point and the capsule backplate, there's a series resistance of 100M.  I was trying my best to do the math (not my strong point) and I'm wondering how there can be +63V on the capsule backplate?  Wouldn't there be some (rather large?) voltage drop across that 100m series resistance?
Title: Re: U47 schematic, component-by-component
Post by: KaiS on January 16, 2011, 12:15:58 PM
soapfoot wrote on Sun, 16 January 2011 08:47

The voltage divider created by R7 and R8, given an input voltage of 105v, will output a voltage of 63v.

However, between that point and the capsule backplate, there's a series resistance of 100M.  I was trying my best to do the math (not my strong point) and I'm wondering how there can be +63V on the capsule backplate?  Wouldn't there be some (rather large?) voltage drop across that 100m series resistance?

No, the load is purely capacitive, this means no DC current does flow through the capsule or C1, except some parasitic due to imperfect isolation.
Therefore the voltage on both sides of R1 is the same.

R1 + C1 do form a LPF to remove hum and noise from the capsule bias voltage.
It is tuned  to 0.17Hz , suppressing hum about 50dB.
C1 2nd function is to create a low impedance AC ground coupling for the capsule backplate.

R6 + C3 do the same for the tube's plate voltage, the filter x-over is about 6Hz, filtering hum by about 20dB.
The filter even reduces DC fluctuations, caused by Line AC change, reaching the signal output.

R5+R6 form the plate resistor, where only R5 is active for signal's AC.
Together they set the DC working point for the tube in the plate branch.

R2 is the negative biasing resistor, R3 the positive biasing resistor (by placing the tube into positive voltage related to ground).
Together they are responsible for the grid/anode working point of the tube.
R3, in theory, acts as a negative feedback resistor too.
In fact this is neglectible, to my measurements there is no gain difference if it is bypassed with a cap.
The resistance of 29 Ohms is simply to small to build up a significant feedback voltage.
It's inductance (it's wire-wound) can, of course, build a RF filter the reduces the possibility to catch up radio interferences.
The capsule capacity together with R2 form a HPF, tuned to 21Hz.
Increasing R2's value would widen the LF response but make the mic more sensitive to sub-sonics like wind and plosives.
The same applies to C2, which does form a 12dB/Okt. HPF together with the output x-former's inductance.
I can't tell the x-over frequency of this one.
I guess (only guess) it's around 30Hz.

R2 does not have any direct influence on the HF response of the mic. Even the parasitic capacity of R2 only dampens the signal over the whole audio range very slightly (less then 0.1dB).
But changing the low end can make us believe the high end does sound different.

Increasing R2 would bring the tube circuit out of the tubes spec's. In fact it's already far of (x200), the spec says no more then 0.5MOhms (for Pentode use, in this quasi-triode circuit it doesn't seem to be such a problem)!
So changing it will change the working point and stability of the circuit.
100MOhms allow some more negative charging of the grid (by parasitic electrons), which adds to the bias caused from R3.
This would mean less current is flowing through the tube, the working point is shifted, distortion can increase.
I've never measured the amount of extra biasing, although it's possible using a high voltage probe with an Ri of e.g. 1GOhm.
One could even do a relative measurement by shorting R2 with a low value resistor and measure the dc change on the plate.
DC on the plate should go down if the high R2 value did change the biasing.

BTW: The U47 circuit is very nice for understanding how a tube condenser mic works, as it's the most basic one possible.
It uses the least number of parts for a working mic.
In theory only the hum filters could be left away or build different.
E.g. you could leave away R1 and increase C1 to 1uF to achieve the same result.
If you have a perfectly regulated PSU you could even leave away R6 and C3, and replace R5 by 130kOhms, but that's all.

Regards
Kai
Title: Re: U47 schematic, component-by-component
Post by: MagnetoSound on January 16, 2011, 02:01:32 PM
KaiS wrote on Sun, 16 January 2011 17:15

In theory only the hum filters could be left away or build different.
E.g. you could leave away R1 and increase C1 to 1uF to achieve the same result.




Doesn't R1 need to be that big to provide the high DC impedance needed for the capsule, as without it the capsule capacitance would need to be impractically huge?

C1 is primarily there to provide the AC path to ground. Without this, there would be no signal (or only a very poor signal) from the capsule and the mic would not work. The filtering action is a nice bonus.


Title: Re: U47 schematic, component-by-component
Post by: KaiS on January 16, 2011, 02:12:23 PM
MagnetoSound wrote on Sun, 16 January 2011 13:01

KaiS wrote on Sun, 16 January 2011 17:15

In theory only the hum filters could be left away or build different.
E.g. you could leave away R1 and increase C1 to 1uF to achieve the same result.

Doesn't R1 need to be that big to provide the high DC impedance needed for the capsule, as without it the capsule capacitance would need to be impractically huge?

C1 is primarily there to provide the AC path to ground. Without this, there would be no signal (or only a very poor signal) from the capsule and the mic would not work. The filtering action is a nice bonus.
Exactly!
Neumann didn't have another chance as to use 60MOhm on R2, as the capsule capacitance is fixed to 80pF by design and cannot be significantly magnified.

Don't underestimate the noise filtering by C1, without it all PSU noise would be coupled into the audio and amplified together with the music.
Therefore R1 is that big to provide sufficient filtering with a relatively small cap.

Regards
Kai
Title: Re: U47 schematic, component-by-component
Post by: MagnetoSound on January 16, 2011, 02:15:32 PM
KaiS wrote on Sun, 16 January 2011 19:12

R1 does not have any influence on the frequency response, it's not loading the capsules output signal against ground, but is connected to the backplate.




No, correct me if I'm wrong, but I meant that since the capsule operates by varying a DC charge, that a high DC source impedance is necessary for such a small capacitance in the capsule to pull against.

You're saying that the only reason for the high value of R1 is to allow such a tiny filter cap value for C1?

Title: Re: U47 schematic, component-by-component
Post by: johnR on January 16, 2011, 03:29:58 PM
KaiS wrote on Sun, 16 January 2011 17:15


Increasing R2 would bring the tube circuit out of the tubes spec's. In fact it's already far of (x200), the spec says no more then 0.5MOhms (for Pentode use, in this quasi-triode circuit it doesn't seem to be such a problem)!
So changing it will change the working point and stability of the circuit.

The resulting change in bias may increase harmonic distortion, which could account for at least some of the perceived increase in high frequencies when R2 is increased.

Edit: I'm just speculating here. See Oliver's (better informed) comment below.
Title: Re: U47 schematic, component-by-component
Post by: soapfoot on January 16, 2011, 06:00:11 PM
This is awesome.  I'm learning a lot.  I've been updating the first post to reflect the clarifications you all are bringing about.  Thanks to Kai, especially.
Title: Re: U47 schematic, component-by-component
Post by: J.J. Blair on January 16, 2011, 08:59:01 PM
Yes.  Thank you Kai and Dan for explaining some things that I've never had properly explained to me, or that I've been unable to find.  Great thread idea, soapfoot.  

BTW, through some trial and error, I'd arrived at using 200MΩ in the R2 position, as being what seemed most pleasant to my ears.  Interesting note about the harmonic distortion, John.  Thanks for that.
Title: Re: U47 schematic, component-by-component
Post by: mad.ax on January 17, 2011, 11:04:01 AM
It is worth noting that the current flowing through the filament is also flowing through R3. Given that this current is much higher (37mA) than the one flowing through the tube, the bias is stiffer, and cannot be affected as it would with other biasing scheme...

Axel
Title: Re: U47 schematic, component-by-component
Post by: Oliver Archut on January 17, 2011, 11:58:33 AM
The resulting change in bias may increase harmonic distortion, which could account for at least some of the perceived increase in high frequencies when R2 is increased.

Sorry guys, there is a lot of speculation on your side, the THD won't increase and the higher the grid leak the better for the low end. The THD is given by the capsule and won't change to much with external components.

The grid leak and polarization cap is a matter of compromise, the higher the grid leak the lower in frequency the capsule will be. But getting higher the noise of the tube will increase, depending on the state of function the VF14, this resistor can be everything from 500M to 1G and in nearly all cases a sound improvement.

The same is valid for the polarization resistor.
The ground reference cap will change the sound most drastically depending on type. Neumann used everything from ceramic to film caps in this position, one of the reasons why U47 can sound so different.

The 29 Ohm with the 1750 Ohm is a voltage divider to keep the bias always at 1.1V, increasing in this voltage will increase the noise of the tube.

Coupling cap can be 05uF for flat response or 1.0uF for 3dB boost at 30Hz.(with BV8s, GN8 have a build in low cut).

Best regards,
Title: Re: U47 schematic, component-by-component
Post by: J.J. Blair on January 17, 2011, 03:31:50 PM
Oliver Archut wrote on Mon, 17 January 2011 08:

The grid leak and polarization cap is a matter of compromise, the higher the grid leak the lower in frequency the capsule will be. But getting higher the noise of the tube will increase, depending on the state of function the VF14, this resistor can be everything from 500M to 1G and in nearly all cases a sound improvement.



Oli, thanks for this info, but can you please clarify: when you say "getting higher," do you mean the grid leak itself, or the value of the resistor in that position?  And should we refer to this resistor then as a "grid leak resistor"?  
Title: Re: U47 schematic, component-by-component
Post by: gk on January 18, 2011, 09:37:26 AM
Hope this lends to the discussion- if anything, my 47 (orig m7)
exhibits a bit of hiss when used for things like bare dialog,  and prob one of the biggest contributors to this artifact would be the noise value of R5.  Anyone know if a high quality metal film might gain a hedge on this?  
Title: Re: U47 schematic, component-by-component
Post by: soapfoot on January 18, 2011, 09:54:58 AM
Have you tried replacing R5 with a new carbon composition resistor before going the metal film route?

From my experience in other types of high voltage tube circuits, replacing plate load resistors with resistors of a different type can sometimes affect subtle changes in the sound of an audio circuit.

Carbon composition resistors can often absorb moisture if improperly stored.  If this happens, they will hiss like crazy.

Apologies if you already knew this.
Title: Re: U47 schematic, component-by-component
Post by: Oliver Archut on January 18, 2011, 10:07:19 AM
Oli, thanks for this info, but can you please clarify: when you say "getting higher," do you mean the grid leak itself, or the value of the resistor in that position? And should we refer to this resistor then as a "grid leak resistor"?

Hello J.J,

getting higher refers to a higher value of R2, during the time the U47 was made there was nothing higher available that 200M and then super expensive.

R2 was chosen at 60M because during the 1950s 40Hz was the norm for recording mics. Changing this resistor to somewere from 500M to 1G will open up the lowest frequency of the mic, but also depending on the state of the tube might increase the tube noise.

Grid leak might not the proper name, it should be bias/ground reference resistor, but the same as the Buffalo is actually a Bison, it might be not correct but it sounds proper after all those years of mis-use.

All resistor should be updated to metal film or carbon composition for better noise specs.

Best regards,
Title: Re: U47 schematic, component-by-component
Post by: MagnetoSound on January 18, 2011, 11:08:51 AM
Oliver Archut wrote on Tue, 18 January 2011 15:07

Grid leak might not the proper name, it should be bias/ground reference resistor, but the same as the Buffalo is actually a Bison, it might be not correct but it sounds proper after all those years of mis-use.




I know this as the grid leak resistor, so named because the grid leak current flows up through this resistor and into the grid.

This is old school amplifier terminology, but certainly still applicable today - and definitely proper.


Title: Re: U47 schematic, component-by-component
Post by: J.J. Blair on January 18, 2011, 01:10:36 PM
I prefer the sound of the carbon comps resistors.  I certainly don't care for those blue Xicon metal films.  PRP resistors are nice, as far as metal film goes.

For the R1 and R2, I have some TRW carbon spiral resistors that have been fantastic.  They might not pass muster with Klaus, but the mics I installed them in, as well as used carbon comps for the rest, sound really fantastic.
Title: Re: U47 schematic, component-by-component
Post by: J.J. Blair on January 18, 2011, 01:12:55 PM
MagnetoSound wrote on Tue, 18 January 2011 08:08

I know this as the grid leak resistor, so named because the grid leak current flows up through this resistor and into the grid.

This is old school amplifier terminology, but certainly still applicable today - and definitely proper.


Thanks.  That's what I'll call it then, from now on, just so people know what I'm talking about.  LOL.
Title: Re: U47 schematic, component-by-component
Post by: Oliver Archut on January 19, 2011, 09:34:19 AM
It might be right that old school techs used grid leak but talking to old tube engineers, they point out if something is leaky it is broken. I am with you grid leak is fine with me, but the proper description is ground reference or bias grid 1 resistor.

We also had a discussion about that about a year ago were other techs point out that grid leak is not the right word.

Title: Re: U47 schematic, component-by-component
Post by: J.J. Blair on January 19, 2011, 11:59:19 AM
There needs to be an agreed upon nomenclature.  Let's put it up to a vote!  
Title: Re: U47 schematic, component-by-component
Post by: MagnetoSound on January 19, 2011, 12:33:19 PM
Oliver Archut wrote on Wed, 19 January 2011 14:34

but talking to old tube engineers, they point out if something is leaky it is broken.



What are they, tube engineers or plumbers?  Very Happy


Wanna go with grid bias, then?

Title: Re: U47 schematic, component-by-component
Post by: KaiS on January 20, 2011, 03:40:54 PM
MagnetoSound wrote on Sun, 16 January 2011 13:15

KaiS wrote on Sun, 16 January 2011 19:12

R1 does not have any influence on the frequency response, it's not loading the capsules output signal against ground, but is connected to the backplate.

No, correct me if I'm wrong, but I meant that since the capsule operates by varying a DC charge, that a high DC source impedance is necessary for such a small capacitance in the capsule to pull against.

You're saying that the only reason for the high value of R1 is to allow such a tiny filter cap value for C1?
R1/C1 is connected to the backplate of the capsule, which is the low impedance side and being held on a steady 60V bias.
It is therefore AC-connected to the ground, while DC-GND-connection is isolated by C1.
This way the capsule can be polarised with +60V through R1.

The diaphragm side is the high impedance signal output.
This is the reason for R2 being a high 60M.

The capsule as capacitor keeps the bias from reaching the tube grid, saving Neumann one extra cap (+one resistor) that they would need if they would bias the diaphragm instead of the backplate.

One more advantage of this biasing scheme:
the diaphragm is on GND potential and therefore doesn't attract dirt as much as a if it was charged by 60V.

Regards
Kai
Title: Re: U47 schematic, component-by-component
Post by: David Bock on January 20, 2011, 04:38:52 PM
Quote:

R1/C1 is connected to the backplate of the capsule, which is the low impedance side and being held on a steady 60V bias.
It is therefore AC-connected to the ground, while DC-GND-connection is isolated by C1.
This way the capsule can be polarised with +60V through R1.
So what happens if you 1)increase 2)decrease 3)remove C1?
Title: Re: U47 schematic, component-by-component
Post by: MagnetoSound on January 20, 2011, 06:13:08 PM
Oliver, I would be grateful if you would clarify what you meant here with regard to the polarization resistor.


Thanks!

Title: Re: U47 schematic, component-by-component
Post by: gk on January 21, 2011, 09:36:58 AM
Great Post-   grid leak works for me,  isn't that what it does?
Such a straightforward circuit, one might think it would be noise free, but hiss can still be an issue with the 47. Is there a best choice for R5? Oliver, might you or someone have experience at a solution for this, thanks.
Title: Re: U47 schematic, component-by-component
Post by: soapfoot on January 21, 2011, 11:10:52 AM
One question I have:

Plate voltage in the VF14 is listed as +34V.

I see a voltage divider between R7+R8 an R5+R6.  However, this wouldn't get you all the way down to 34V... more like 102V.

So where does the rest of that voltage go?  There must be something 'hidden' that I'm not seeing.

Title: Re: U47 schematic, component-by-component
Post by: johnR on January 21, 2011, 11:11:37 AM
gk wrote on Fri, 21 January 2011 14:36

Such a straightforward circuit, one might think it would be noise free

Unfortunately no circuit is noise free unless you cool it to absolute zero (and then it wouldn't work anyway).
Title: Re: U47 schematic, component-by-component
Post by: Pasarski on January 21, 2011, 01:44:55 PM
soapfoot wrote on Fri, 21 January 2011 18:10

One question I have:

Plate voltage in the VF14 is listed as +34V.

I see a voltage divider between R7+R8 an R5+R6.  However, this wouldn't get you all the way down to 34V... more like 102V.

So where does the rest of that voltage go?  There must be something 'hidden' that I'm not seeing.



Ohm's law. The tube draws current and voltage drops across the plate resistor depending of bias.

Pasi Siitonen
Title: Re: U47 schematic, component-by-component
Post by: soapfoot on January 21, 2011, 02:09:25 PM
ah!  of course.  Thank you so much.
Title: Re: U47 schematic, component-by-component
Post by: KaiS on January 22, 2011, 05:10:04 PM
David Bock wrote on Thu, 20 January 2011 15:38

Quote:

R1/C1 is connected to the backplate of the capsule, which is the low impedance side and being held on a steady 60V bias.
It is therefore AC-connected to the ground, while DC-GND-connection is isolated by C1.
This way the capsule can be polarised with +60V through R1.
So what happens if you 1)increase 2)decrease 3)remove C1?

Increase - nothing, only it would take some time to charge the polarization voltage if you'd go to extreme values.
Practically there is a limit, as larger condensers have more leaking current.
Pol. voltage will become lower, depending on the type of C used.
Electrolytics e.g. cannot be used here.

If you decrease it, the mic will start to deliver more hum (filter becomes less efficient), finally if you remove C1 (=maximum decrease) the backplate will no longer be referenced to GND AC-wise.
The AC signal circuit is no longer closed, except for some parasitic capacities of the mount.
The mic will deliver very low audio level.

This happens when C1 breaks.

Regards
Title: Re: U47 schematic, component-by-component
Post by: volki on January 23, 2011, 01:52:44 PM
Oliver Archut wrote on Mon, 17 January 2011 17:58

 The THD is given by the capsule and won't change to much with external components.



Really? To my knowledge, a capsule wouldn't necessarily dominate the generation of nonlinear distortion in comparison with the electronics - except for the modern high headroom designs. An interesting paper by Peus of Neumann (1997) for example reveals that nonlinearities of pressure gradient capsules will mostly dominate (if at all) in the upper third or upper half of the frequency range - depending on capsule type but also head amp topology (tube, single FET, etc.) Among other examples, they provided a comparison of KK67/87 capsules with U67 and U87 electronics.

Now I would be really interested to learn about distortion measurements of M7 or KK47 capsules (Oliver?). On the other hand, the u47 circuitry produces a fair amount of distortion (low order, dominant k2) at relatively low sound pressure - with 0,8% THD quoted for 110dB SPL. So an M7 or KK47 would even exceed these figures?
Title: Re: U47 schematic, component-by-component
Post by: Klaus Heyne on January 23, 2011, 03:17:27 PM
I think distortion measurement protocols for capsules are hard to design.

Suppose you can somehow extract and eliminate THD distortion from all electronic components in the chain, including the mic processor itself, how do you want to trigger (or simulate) and isolate the distortion emanating from a capsule? With a sine wave, fed into a speaker, fed into the capsule?

That is about as far away from a real-world scenario as can be. The absence of complex wave forms, let alone the random air movement associated with a singer's exhaling, makes it really hard to understand what non-linearities come from what component. There is also the electro-mechanical interaction of the mic amp's high impedance stage coupled to the capsule.

Bottom line for me: If I hear distortion in a mic, the first thing I do is replace the capsule with a known good one (i.e. new, if available). If the distortion is no longer audible, by deduction, it was the capsule. if the distortion is still there, by the same deduction, it is something downstream from the capsule. I challenge anyone reading this to design or point to a testing protocol to isolate capsule distortion that is less Luddite.
Title: Re: U47 schematic, component-by-component
Post by: David Bock on January 23, 2011, 07:31:25 PM
Peus used two speakers with two different tones,  not a bad idea.
Distortion referred to as a percentage is still as uninformative as it ever was.
Title: Re: U47 schematic, component-by-component
Post by: J.J. Blair on January 23, 2011, 10:47:54 PM
And is distortion necessarily a bad thing in mics?  Isn't that part of why we like the sound that comes out of mics so much?  A certain level of harmonic distortion and phase shift?
Title: Re: U47 schematic, component-by-component
Post by: Oliver Archut on January 24, 2011, 12:05:34 AM
Really? To my knowledge, a capsule wouldn't necessarily dominate the generation of nonlinear distortion in comparison with the electronics

Distortion measurements of capsule are given by the capsule and will not change unless a nonlinear feedback circuit is designed to compensate.

The capsule of a 47, it does not matter M7 or K47, will start to distort earlier than the buffer amplifier itself, that is one of the most liked features of the 47 design.

Distortion is a funny thing, it is still measured static and yes, we do have intermodulation distortion, but it tells us nothing else than the static relation of the measured frequency at a given pressure level. A chosen set up that really does not reflect real life recording set ups.


Best regards,

Title: Re: U47 schematic, component-by-component
Post by: KaiS on January 24, 2011, 01:27:36 PM
Klaus Heyne wrote on Sun, 23 January 2011 14:17

I think distortion measurement protocols for capsules are hard to design.

Suppose you can somehow extract and eliminate THD distortion from all electronic components in the chain, including the mic processor itself, how do you want to trigger (or simulate) and isolate the distortion emanating from a capsule? With a sine wave, fed into a speaker, fed into the capsule?
A measurement setup isn't too hard to design:

1.You need a very linear buffer amplifier.
No problem, as levels are high when measuring distortion; a JFet Op-Amp can be used, almost zero distortion.

2. As acoustic signal generator take two speakers coupled to an as small as possible volume.
Extremely high SPL can be reached this way (imagine, a plug type headphone in your ear can easily deliver 130+ dBSL).
Both speakers are feed with different tones, e.g. 5000Hz and 6000Hz.
Now IM dist. (intermodulation distortion) is measured.
E.g. 1st order dist. would be 1kHz, far below the 1st harmonic of the generating speakers (10kHz).

Therefore it's not necessary to use dist. free speakers as generator.

Regards
Kai

Title: Re: U47 schematic, component-by-component
Post by: David Bock on January 24, 2011, 03:58:25 PM
Kai,
Seems as though you're just repeating the Peus idea, and not addressing Klaus' comments.
If you are addressing Klaus' comment, I'm curious about how you are connecting your test to his concern. Seem like there is no link between the two.
Title: Re: U47 schematic, component-by-component
Post by: zebra50 on January 24, 2011, 04:51:08 PM
Hi David,

Would you be kind enough to post some references, or some links, to Peus's work with this approach?

Many thanks!

Stewart
Title: Re: U47 schematic, component-by-component
Post by: volki on January 24, 2011, 05:44:47 PM
Stewart,

The Peus document is at  http://http://www.neumann.com/download.php?download=lect0020 .PDF

It includes measurements of difference frequency distortion (DFD) with a setup of two speakers and the complete microphone (electronics and capsule). The basic arrangement is to feed a sweep of two sines with a relative spacing of 80Hz to one speaker each. Thus, the speakers on their own only generate THD, their two signals combine acoustically in front of the mic, and you can be sure that the DFD you measure is actually caused by the mic itself.
DFD yields even and odd order distortion components: d3 (odd order) occur at 2*f1-f2 and 2*f2-f1, rising continuously with the sweep. d2 (even order) occurs at f2-f1 and remains constant. By appropriate filtering, you can exclude the THD components generated by the speakers. What Peus measured is just d2 by cutting everything except 80Hz. The problem of measuring d3 in this setup is that they're only 40Hz apart from the freq's of the input sweep, so they obviously couldn't be filtered adequately (or they didn't bother) and thus were omitted.
Title: Re: U47 schematic, component-by-component
Post by: Klaus Heyne on January 24, 2011, 06:18:25 PM
volki wrote on Mon, 24 January 2011 14:44

(...) Thus, the speakers on their own only generate THD, their two signals combine acoustically in front of the mic, and you can be sure that the DFD you measure is actually caused by the mic itself.

This test does not address the question of how to isolate capsule distortion/artefacts from amp distortion or the interaction of the two in generating unwanted distortion/artefacts.

A previous post mentioned to simply hook up the capsule to a mic amp with (high headroom) op-amp design.
But that is not real-world scenario either:
take the fet47 which uses the same capsule as the U47 tube. So this would be theoretically a good capsule distortion testing device, as the fet47's amp has a higher headroom than the U47 tube by at least 10dB, depending on setting.  

Yet, the fet47 is known for an initial fast transient response with a lag time response to the full signal (a priced quality for kick drum application.) That means, the anomalies or characteristics of a mic amp cannot be easily taken out of the equation, even when, on paper, its static specs look flawless and suitable to isolate the capsule's flaws.

I just think the issue of isolating whether capsule or amp distort first when hit with massive SPLs may be more of a red herring upon deeper investigation, and may just leave us with this:

Testing a capsule's contribution to a mic's distortion may only be possible by substituting either capsule or mic amp with like components (after establishing a distortion base line through testing of one or several healthy specimen.) Then, one would listen for a decrease in audible distortion or other artefacts. If no decrease is audible after capsule or mic amp substitution, one can safely assume the relative level of distortion onset for that model, and, if that level is objectionable, that the mic as a whole may not be suitable for the type of recording envisioned.
Title: Re: U47 schematic, component-by-component
Post by: Eric H. on January 24, 2011, 07:27:19 PM
To this day, I still don't get why we still do audio tests whith sine waves. To me, it is like travelling on a horse.
Measurements with filtered pink noise, by octave or 1/3 octave, burst signals, sweeps and others I don't know about would probably give more info on the complexity of the phenomenons.
Softwares like matlab really make it available to any lab to do very complex comparisons of original and measured signals.
I am sure the real mic labs have all kinds of measurements that they just don't make public.
I'd like to see one day transient measurements in the brochure for exemple.
Title: Re: U47 schematic, component-by-component
Post by: Klaus Heyne on January 24, 2011, 08:22:34 PM
What would transient measurements tell you about how a microphone sounds, or how useful the mic may be for music recording?
Title: Re: U47 schematic, component-by-component
Post by: volki on January 24, 2011, 09:09:08 PM
The fact that I was summing up Peus' paper in my last post doesn't mean I subscribe to the view that distortion measurement with static signals were the ultimate method of evaluationg sonic qualities - even if I do think that it does lend itself to outline a basic direction thereof.

Of course even the envelopes of quasi-stationary signals (let alone transients) will interact with the characteristic linear and nonlinear distortion mechanisms of a given capsule and amp design (where I remember some posts of David Bock in the recent tube biasing thread quite making sense to me), resulting in a more or less unique signature.

I'm also not debating by default that distortion of the capsule may sometimes (or even most oftentimes) be dominant over the head amp. On the contrary, that would explain sonic properties of certain mic's which otherwise would only rather weakly be backed up by measurements of the electronics.

Quote:

 (Oliver)
The capsule of a 47, it does not matter M7 or K47, will start to distort earlier than the buffer amplifier itself,

Could you elaborate as to where this information comes from - listening, measurements, etc.?

Anyway - in this case, one could also argue "the other way round" and attribute negative properties (such as sibilance) in some modern mic's to those capsules producing gross acoustic distortion in the upper midrange - any comments?

Quote:

 (Klaus)
There is also the electro-mechanical interaction of the mic amp's high impedance stage coupled to the capsule.

Klaus, what exactly do you mean? Nonlinearities caused by the electric force due to the membrane's excursion towards the backplate and thus being increasingly attracted by it? Metal contacts between capsule and amplifier...? And then there's of course the nonlinear compression property of the air cushion between membrane and backplate... they all seem to matter somehow.



PS. I once briefly compared my Neumann UM57 to a Neumann/Gefell UM70 head (containing an M7 capsule) combined with an M582 head amp (same tube but different circuit and xformer), but wasn't able to spot a significant difference. Now would that be due to the nonliniear characteristic of that M7 dominating any circuit nonlinearities...?!?  Cool




Title: Re: U47 schematic, component-by-component
Post by: volki on January 24, 2011, 09:20:46 PM
PPS. Maybe I’m mistaken , but it seems to me that  manufacturers used to get their published max. SPL data by simply measuring THD of the electronic part of the mic, assuming that this were the dominant figure.

Now for modern mic’s, with their low THD and high headroom figures,  this method would indeed be valid somehw since the clipping of the circuit - and thus the very audible sudden rise of a lot of harmonics - definitely determine the max. SPL possible. On the other hand, if the capsule has a higher distortion figure than the amp, the above might not hold true for all types of signals - even if the capsule’s distortion products are dominant low order and rising proportionally with level, a property much more tolerable to the ear...?
Title: Re: U47 schematic, component-by-component
Post by: Klaus Heyne on January 24, 2011, 09:27:25 PM
volki wrote on Mon, 24 January 2011 18:20

PPS. Maybe I’m mistaken, but it seems to me that  manufacturers used to get their published max. SPL data by simply measuring THD of the electronic part of the mic, assuming that this were the dominant figure.
Correct. Unless David Josephson has new information to the contrary, amp + dummy load instead of amp + capsule is still the standard for measuring headroom of mics.
Title: Re: U47 schematic, component-by-component
Post by: Oliver Archut on January 24, 2011, 11:59:16 PM
Hello Volker,

the distortion of the capsule is not the dominant one, in a 47 the capsule starts to distort earlier than the electronics, but at one point the electronics will catch up and the final distortion is dominated by the amplifiers THD. My point is that the static value of testing distortion does not take this into account.

I hope that this clears it up.

Best regards,


 
Title: Re: U47 schematic, component-by-component
Post by: zebra50 on January 25, 2011, 03:25:06 AM
Thanks for the link, Volki!
Title: Re: U47 schematic, component-by-component
Post by: Eric H. on January 25, 2011, 09:28:18 AM
Klaus Heyne wrote on Tue, 25 January 2011 01:22

What would transient measurements tell you about how a microphone sounds, or how useful the mic may be for music recording?

Well, the same way we have transient measures for power amp to reproduce music.
Music is the contrary or static and the way any audio device can "track" those micro changes in the music/sound can get a good idea of how much accurate a mic would be in transducing a pressure or gradient point in the sound field.

More accurate transient response is one of the reason people will love condenser for example.
We could do more into detail about this, but it my conviction that when we mike a group/ensemble, the transient response of the different mics is a very important factor in creating our reconstructed sound stage.
Title: Re: U47 schematic, component-by-component
Post by: David Bock on January 25, 2011, 01:00:26 PM
Quote:

Well, the same way we have transient measures for power amp to reproduce music.
Music is the contrary or static and the way any audio device can "track" those micro changes in the music/sound can get a good idea of how much accurate a mic would be in transducing a pressure or gradient point in the sound field.
Sadly, or Greatly, measuring microphones is still far more complex than measuring an amplifier.
There are a number of measurements that can be taken of a microphone beyond fr, but I haven't seen a successful deconvolution into a meaningful characterization that would permit decisive differentiation between mics. And in those measurements, transient response doesn't come up so much. FWIW greater minds than mine have and are addressing this topic, with still, no overwhelming evidence that one could use measurements instead of listening.
Quote:

More accurate transient response is one of the reason people will love condenser for example.
You have proof or evidence to justify this?
Title: Re: U47 schematic, component-by-component
Post by: KaiS on January 25, 2011, 03:56:11 PM
David Bock wrote on Tue, 25 January 2011 12:00

Quote:

Well, the same way we have transient measures for power amp to reproduce music...
Sadly, or Greatly, measuring microphones is still far more complex than measuring an amplifier.
... in those measurements, transient response doesn't come up so much. FWIW greater minds than mine have and are addressing this topic, with still, no overwhelming evidence that one could use measurements instead of listening.
Quote:

More accurate transient response is one of the reason people will love condenser for example.
You have proof or evidence to justify this?
Measuring the transient response of a mic is simple:
Use a gunshot - compressed air pistols works very well delivering an almost perfect pulse of not to high SPL.

When I did comparisons with this type of measurement I found the following:
The B&K reference mic (1/8") recorded @ 96kHz/24bit delivered a sharp pulse of 1-2 Samples with no ringing (as hoped and expected).

The larger the condenser mic the broader the pulse became, but no obvious ringing (may be some in the low level range).
This complies with theory, as those mics have a more limited bandwidth, therefore the pulse must be broader.

Using dynamic mics there wasn't a sharp pulse, but more like a bell shaped form with evidence of pre- and post-ringing.

Does this help us - it just confirms what most of us know from our own listening experience - condenser mics have a better transient response.
As this is in the time domain it's not possible to change a dynamic into a condenser sound and vice versa.

But - I couldn't find any significant difference in the transient responses of obviously different sounding condensers.

My conclusion - listening is the final judge.
Measuring can help to track down problems when constructing a mic.

Regards
Kai
Title: Re: U47 schematic, component-by-component
Post by: Eric H. on January 25, 2011, 04:49:26 PM
Maybe what I wanted to say is that the requisite to have more info on a mic behaviour is to gather dynamic info vs. static info of either sine waves or pink noise. On the path to the measurements of a mic, there must be something something smarter than sweeping the mic with an HP in front.
Maybe scientist should go back to how hour hearing works to get there. To me that would probably be the most rewarding path.
And we know that our hearing is a comparison machine, and not an absolute one(except maybe some highly trained and specialized people). The only parameter we have more ability to judge absolutely (and it is only for a handful of humans) is the frequency.


Note that I don't know much about measurements, but I do know that to achieve something visually palpable is no picnic.

David, if you have some reference to those paper on new ways to measure microphones, I'd love to know about them.
Title: Re: U47 schematic, component-by-component
Post by: volki on January 25, 2011, 04:57:05 PM
Oliver Archut wrote on Tue, 25 January 2011 05:59

 in a 47 the capsule starts to distort earlier than the electronics (...) My point is that the static value of testing distortion does not take this into account.  


Point well taken - but again, is there anything you can back up your statement with?
Title: Re: U47 schematic, component-by-component
Post by: Klaus Heyne on January 25, 2011, 05:31:03 PM
Yes, I too would like to know some kind of method, even if it's just empirical, that Oliver uses to isolate capsule distortion from amp distortion.

My best guess would be: it won't be a linear, easily repeatable, method to reliably pinpoint capsule distortion.

For example, if sound waves hitting the membrane are accompanied by strong capsule exertion caused by exhaled breath, that might easily distort the fairly linear swinging process of a diaphragm* sooner than the amp distorting. On the other end of the application spectrum for a microphone, I cannot quite imagine diaphragm distortion with sound sources at great physical distance from the capsule.

* a conventionally constructed capsule- diaphragm swings against a closely spaced polarized backplate - is inherently incapable of a 100% symmetrical, and therefore linear, distortion-free swinging process. The backplate's electrostatic attraction prevents this.
Title: Re: U47 schematic, component-by-component
Post by: David Bock on January 25, 2011, 07:32:14 PM
Quote:

For example, if sound waves hitting the membrane are accompanied by strong capsule exertion caused by exhaled breath, that might easily distort the fairly linear swinging process of a diaphragm* sooner than the amp distorting.
Are you saying LF modulates mid and HF?
@Kai: I wasn't questioning the superior transient response of condensers over dynamics. That is well documented. I was questioning this statement:
"More accurate transient response is one of the reason people will love condenser for example."
Quote:


David, if you have some reference to those paper on new ways to measure microphones, I'd love to know about them.
Can't think of any books but there have been a few AES papers in the last 30 years addressing the subject, but nothing groundbreaking or conclusive. Besides the Peus paper nothing comes to mind as published "new methods".
Jackie Green wrote one that attempted to correlate measurements with adjectives and other verbal descriptors, but it didn't result in a useful, repeatable language. You should read it to see how far they went, for such a non-result.
Title: Re: U47 schematic, component-by-component
Post by: Oliver Archut on January 25, 2011, 08:44:30 PM
Point well taken - but again, is there anything you can back up your statement with?

Most of the statement are based on my own measurements.

As Klaus pointed it out the polarization voltage is the first problem of nonlinearities with condensers, LDC in particular.
The higher the polarization the earlier the nonlinearities show up. The capsule is bound by design and there are a few parameters that can be changed, but like everything you change one and the rest will change as well.

The membrane material is the 2nd problem, most capsules are not 100% tensioned with reference to dead center of the capsule. There are only a few capsule out of any given series that collapse uniformly when the polarization voltage is pushed up, most capsules need a 5 to 10V window before the entire membrane is collapsed. In my measurements the less uniform the capsule collapses the more the distortion.
Additional the material itself is one factor itself for distortion, as an example a multiple layer membrane will produce less distortion. AKG did a couple of test membranes for Telefunken that were at least 4 individual layers fused/sintered together. The problem this process was to expensive for production so it never reached perfection. At a later point AKG rolled the mylar with a special pattern aside several other production advantages it did reduced the capsule distortion.

Best regards,



Title: Re: U47 schematic, component-by-component
Post by: volki on January 26, 2011, 06:01:42 PM
Oliver Archut wrote on Wed, 26 January 2011 02:44

Most of the statement are based on my own measurements (...)
The membrane material is the 2nd problem, most capsules are not 100% tensioned with reference to dead center of the capsule. There are only a few capsule out of any given series that collapse uniformly when the polarization voltage is pushed up (...) the less uniform the capsule collapses the more the distortion.

Interesting. Could you elaborate on your measurement setup and procedure? I figure that it takes more than just measuring the change of capsule capacitance to evaluate partial collapse. An how did you check for the distortion that you relate the "collapsing properties" to?

Quote:

Additional the material itself is one factor itself for distortion, as an example a multiple layer membrane will produce less distortion (...) AKG rolled the mylar with a special pattern aside several other production advantages it did reduced the capsule distortion.


Interesting detail (BTW could it have been AKG's Karl Peschel who gave you the info?).

A source of distortion that hasn't been mentioned is the nonlinear property of the air volume inside the capsule as it is compressed by the excursion of the diaphragm. There's even research by people like Hibbing (of Sennheiser, e.g. an 1981 AES paper) and Pastill
Title: Re: U47 schematic, component-by-component
Post by: volki on January 26, 2011, 06:06:47 PM
Klaus, would you think that it makes sense to relocate the last portion of posts into a thread of it's own? Capsule distortion is obviously a rarely discussed yet important subject, which mabe shouldn't be sitting hidden in this thread...  Cool
Title: Re: U47 schematic, component-by-component
Post by: Oliver Archut on January 27, 2011, 10:11:49 AM
Hello Volker,

the measuring test were build by myself, for collapsing test I modified a circuit that was suggested by Mr. Sattler/Sennheiser USA for measuring capacitance with active polarization voltage. The set up has a resolution of 0.1 pico farad. With a fine stepped power supply the nonlinearity of polarization vs. capacitance can be viewed while slowly working to the point of collapse.
A 2nd test setup to collapse capsules is a 1 to 80 step up x-former, that injects a square wave into the capsule and a finely stepped power supply, any capsule can be slowly collapsed.

As I said, the capsule that collapse unevenly had the higher distortion.

Measuring distortion, the capsule is hooked up to a Reaktanz circuit, either a tube or a transistor is operated as an electronic blind-resistor, this circuit will amplify the output of the capsule without being active in the circuit. It is technical speaking a parallel amplifier that follows the movement/capacity changes of the capsule.
Technical an amplified copy of the original signal without effecting the capsule signal. THD of this circuit itself is not measurable.

The signal source were electronic stimulated tuning forks.

It is still not the best set up in my view to quantify distortion.

Got a bunch of good info over the years from Karl, but the capsule material data I got out of old Telefunken data that was thrown out when the audio archive was closed.

The distortion papers you mention, they read well from a technical point, but in my view they try to describe chemical composition of wine (I slightly modified Klaus comparison here)
with finer and finer measurments of what is in it, without knowing what needs to be measured to quantify why the wine taste the way it does.

Capsule distortion one thread? More for a technical minded forum in my view.

Best regards,

Title: Re: U47 schematic, component-by-component
Post by: Geoff Emerick de Fake on February 28, 2011, 01:42:47 PM
The way distortion is spec'd by manufacturers has almost no informative value.
It's like mentioning distortion of a loudspeaker at 1kHz, when most of the time the actual problem is distortion at low and low-mid frequencies.
Everybody knows a mic diaphragm's excursion increases with the inverse of frequency, so the accepted 3% limit at 1kHz may become 30% at 100Hz.
In fact the construction of the capsule has an enormous influence on distortion; a center-terminated capsule certainly has a different behaviour than a side-terminated, if only because they don't rely on the same acoustic damping.