AKG Perception P220 to Neumann u87 5 min mod ( p200, p100, p400, p420? )

[ln76d]

• • It would cost nothing to the manufacturer to change this cap; then, why don't they do it? I guess there are enough people who like the added sizzle...

  
  Because it's typical chinese factory circuit, modified only to use dc converter. No one bothers how it sounds
  
  

  2. In this case they just followed blindly u87 schematics, even with some blank spots. Like Piotr said, didnt even bother to bias the fet, so even with 220pF doesnt work as it should.

  
  Matija, to be clear - one thing is that unbiased fet makes a nasty repsonse - other thing is that the whole bias circuit don't fit deemphasis network - that's why it only works with higher capacitance.
  If you will change biasing to the same as in U87, then it will work with 220pF.
  Left - chinese biasing, right - U87
  

 

[Khron]

Ok, feedback values affect the response differently between the two arrangements, but what about the max undistorted output?

I recall the Schoeps JFET gate biasing being, in essence, the equivalent of a "split source resistor", and adjusted to get the max unclipped level. The KM84 / U87 one has the 1G straight to ground, and you just mess with the source resistor.

 

[KrIVIUM2323]

You mess with the source resistor or you define a stric idss range for the fet, select them fix the resistor value and all is good.
But, this is time consumming as jfet are all over the map (even for a given range) and time is money, easier to load an amno box or a reel and let robot do his work.

 

[kingkorg]

Have to try it out. Thanks Piotr.

 

[wirelesssapien]

Hi, KingKorg!

Thanks for doing this research, and making it available. I have an AKG P220, and I would really like to do this 680pF capacitor switch.  I have done a lot of repair soldering on musical electronics, but have never modded a microphone before.  Could you give me complete specs on the capacitor you used? Or perhaps a link to a good one?

I can see that this might become addicting......

Thanks again, wirelesssapien

 

[Khron]

A film cap (or styroflex, if you wish) should be plenty. Mounting it can be a bit interesting, since it can get a little bit cramped in the area.

 

[kingkorg]

Well it is extremely addicting, especially once you hear it side by side with original.  I used NP0/C0G couse that is that's what i had. Wima FKP or MKP might be a better choice, or what Khron sugested. You just have to be careful with size, though there seems to be plenty of space if i remember well.  But i am sure everybody has own preferences, this was a quick mod for a friend, and he wasn't demanding.

If it's first time removing of SMD component, apply a extra drop of solder to each side of capacitor, so that you can lay soldering pen tip horizontally touching both soldering points of the capacitor.  It should come right off, and stick to soldering pen tip.

 

[Khron]

I often use that method - a wide tip helps a lot. Although i like to "make sure" / maintain control of the component with a pair of tweezers, as well

If it's first time removing of SMD component, apply a extra drop of solder to each side of capacitor, so that you can lay soldering pen tip horizontally touching both soldering points of the capacitor.  It should come right off, and stick to soldering pen tip.

 

[MS Vienna]

In this case it is not absolutely necessary to remove the original capacitor. It might be easier to add a second cap in parallel like this:

I did various experiments with different generation Perception microphones. It was worth the effort. Thanks kingkorg for the suggestion!

The only original Neumann U87 I have yet compared it to is my very early specimen - clear plastic domed U67 switch housing, 7-pin DIN connector and all... My modded Perceptions are somewhat similar in timbre but more veiled and distant, "smaller" sounding. Still great tools for putting instruments more in the back of a mix to create depth of field.

 

[ln76d]

Much better effect you will get biasing fet correctly with Neumann schematic. Then 220pF-270pF is enough value and the audible effect is much better. Note one thing! Perception have full low end response vs. U87
Another good option is to remove input cap and replace output cap with film type.

 

[abbey road d enfer]

Much better effect you will get biasing fet correctly with Neumann schematic.

What is the theory behind this statement? Biasing certainly affects gain and the value of the cap depends on this stage's gain. The choice made by the AKG designers has been guided by the FET's characteristics, that would result in improper biasing without the bottom source resistor. If you want to get rid of this resistor, you need to change the FET too!

Note one thing! Perception have full low end response vs. U87

Probably because this added source resistor bootstraps the bias resistor; if you remove it, LF response will be reduced as well, but the operating point will be shifted significantly, and probably reducing the max SPL capability.

Another good option is to remove input cap

So you would apply the polarization voltage to the FET's gate?

 

[ln76d]

What is the theory behind this statement? Biasing certainly affects gain and the value of the cap depends on this stage's gain. The choice made by the AKG designers has been guided by the FET's characteristics, that would result in improper biasing without the bottom source resistor. If you want to get rid of this resistor, you need to change the FET too!

There's no theory, just clear practice.
In different topologies, different biasing type can cause really different effects.
Rather chinese designers. Same problem is with all the mxl deemphasis circuits, since they use the same  biasing. Then 220pF isn't enough. Probably becuase 1G resistor isn't directly grounded.

Ok - so i will repeat your question
"What is the theory behind this statement?"

Why do you think the FET should be changed?
Here's 2SK30 - to get the same level as in original and bias it properly (original totally isn't) you need just use proper value of pot/resistor - nothing more nothing less. With properly biased FET and grounded 1G resistor you will get same level but properly working deemphasis network where Neumann capacitor values works as it should.

Probably because this added source resistor bootstraps the bias resistor; if you remove it, LF response will be reduced as well, but the operating point will be shifted significantly, and probably reducing the max SPL capability.

No - totally not
Just because deemphasis network use 560nF instead of Neumann 10n-33nF capacitor.

So you would apply the polarization voltage to the FET's gate?

No, because polarisation voltage is on the different node (*this isn't valid for P420)

 

[ln76d]

I will try explain a little.

U87 type of deemphasis network is very simple circuit only for the first look.
I experimented a lot with it in FET and tube microphones.
I'm not theorist, totally, so i will describe it as i can.
First rule - high ohm resistor on the gate (or grid) need to be grounded!
Second rule - lowering value of that resistor below 1G will cause the need of higher capacitance in HF eq part.
See Neumann U77 schematic  - for 400M, Neumann used 2.2nF and this is relative to  ca 2.4k not 7.5k (becuse it's easier to set the value of HF cap).
I tried optimised circuits for different values like 100M, 200M, 470M. The lower resistance at the gate/grid the harder is set proper capacitor value and with 100M there's need to use pretty high values - which cause really worse response (phase and other factors).

 

[Khron]

Which cap is that?

Attached is Henry Spragens' drawing of the P220 circuit (same as P200 just with a larger pad cap), where i then filled in the values of the ceramic caps (measured with my DER-EE DE-5000 LCR meter after removal from the circuit).

Just because deemphasis network use 560nF instead of Neumann 10n-33nF capacitor.

2024 edit: This thread motivated / gave me the excuse to identify the four remaining "anonymous" ceramic capacitors in Henry's schematic, and i'm attaching the updated version here, "just in case". Two of the caps showed so-and-so values, but arguably still within a normal-ish +/-10% tolerance, so there you go.

 

[abbey road d enfer]

There's no theory, just clear practice.

That's the issue here. If you think you are abetter designer than teh AKG/Chinese guys, you must sustain that with some theory. Ther is "some" science involved there.

In different topologies, different biasing type can cause really different effects.

Such as?

Why do you think the FET should be changed?

The FET in the U87 is a selected type, that can operate with 0V gate voltage. The FET in the P220 is very likely not capable of operating satisfactorily at 0V gate voltage (if it was, I doubt the designers would have added a resistor that increases the BOM).

Here's 2SK30

Is it? 2SK30 is a very different animal than the one in the U87. If it was used with 0V Vgs, it would draw 3mA, which is not correct for this circuit. That's why teh designers had to add the bottom resistor.

to get the same level as in original and bias it properly (original totally isn't) you need just use proper value of pot/resistor - nothing more nothing less.

What evidence do you have that it's not properly biased?

No, because polarisation voltage is on the different node (*this isn't valid for P420)

In heh absence of a proven schematic, I can only refer to the fragmented schemo that circulates on the internet, and I see that the polarization resistor is between teh capsule and the coupling cap. If you know different, please post a schemo.

 

[ln76d]

Which cap is that?

Attached is Henry Spragens' drawing of the P220 circuit (same as P200 just with a larger pad cap), where i then filled in the values of the ceramic caps (measured with my DER-EE DE-5000 LCR meter after removal from the circuit).

This one which you have described as 47nF
Also instead of 470nF was the same 560nF - in two P100.
Don't remember exactly what values the rest had - but at this position i didn't found nothing low as Neumann values. P100 was the last one which i dig in it.

 

[abbey road d enfer]

First rule - high ohm resistor on the gate (or grid) need to be grounded!

Simply not true; Schoeps headamp is a perfect example of this resistor not being grounded, but rather bootstrapped from intermediate node in source resistors.

Second rule - lowering value of that resistor below 1G will cause the need of higher capacitance in HF eq part.

Gate/grid resistor value affects LF response and noise, taht's all, doesn't change HF response in anyway.

See Neumann U77 schematic  - for 400M, Neumann used 2.2nF and this is relative to  ca 2.4k not 7.5k (becuse it's easier to set the value of HF cap).

This capacitor's value must be considered in comparison with all the other elements in teh network (not only C106, but R105, R107, C104) and the differences in capsule and stage gain.

 

[ln76d]

Simply not true; Schoeps headamp is a perfect example of this resistor not being grounded, but rather bootstrapped from intermediate node in source resistors.

Gate/grid resistor value affects LF response and noise, taht's all, doesn't change HF response in anyway.
This capacitor's value must be considered in comparison with all the other elements in teh network (not only C106, but R105, R107, C104) and the differences in capsule and stage gain.

Abbey - this is only about U87 deemphasis network circuits - not the biasing of every circuit. So the schoeps is out of consideration here.
Again gate/grid resistor in standard topology - yes, but i'm writing here about deemphasis circuit - and it matters here for HF response capacitor value.
And your description for U77 is totally false. When i modified U77 with 1G instead of 400M, all the standard values from U87 gave proper response.

Instead of writing - just take any circuit with deemphasis and do few simple tests - you will see

 

[ln76d]

That's the issue here. If you think you are abetter designer than teh AKG/Chinese guys, you must sustain that with some theory. Ther is "some" science involved there.

I'm not any publication writer or a teacher so my scientific approach is thankfully  limited to measurements and tests
I don't have a time and need to think  how to dress it nicely in words

  Such as?

Such as Perception and other chinese deemphasis circuits, such as Oktava mk-219/319 (this is main cause of ugly sound of these microphones) or such as every different biasing of the same tube circuit - which give you differnt response for different biasing type. I'm not talking about change of gain. Usual when i'm designing custom tube microphone, i'm starting from donor body, capsule, tube, transformer etc. Then usual i'm testing three different type of biasing to choose the best which fit the needs - try - worth

The FET in the U87 is a selected type, that can operate with 0V gate voltage. The FET in the P220 is very likely not capable of operating satisfactorily at 0V gate voltage (if it was, I doubt the designers would have added a resistor that increases the BOM).

Usual i found 2n3819, sometimes there are FET without any marking (or removed).

Is it? 2SK30 is a very different animal than the one in the U87. If it was used with 0V Vgs, it would draw 3mA, which is not correct for this circuit. That's why teh designers had to add the bottom resistor.

Works normally and there's no significant voltage drop due to higher current.
Only difference is that value of pot/resistor need to be at least half of that you will  find in real U87.
Don't remember exactly but probably i used 5k pot. I will check in P420 from the pic when i wil get in my hands again.

What evidence do you have that it's not properly biased?

Is there any case? Who is the defender and who is the prosecution ??

Inject signal to the gate or "measurement input" and check on scope - nothing more simple.

In heh absence of a proven schematic, I can only refer to the fragmented schemo that circulates on the internet, and I see that the polarization resistor is between teh capsule and the coupling cap. If you know different, please post a schemo.

Maybe, truly don't remember i didn't make perception from a really long time. If so, then i probably change the capsule connection as a mod.

Sorry for different order of replying, but i didn't saw the earlier answer.

 

[joelgurule]

Hi people,

As we all know Perception circuits are pure ripoff of u87, but don't sound like one because of the capsule. A fellow musician sent me vintage u87 to clean super dirty capsule, and two AKG P220 to try to match the sound to Neumann.

I found out by comparing and measuring the two, for some reason AKG has weaker HF attenuation, and brighter capsule. I just wanted to report that Neumann u87's C6 capacitor responsible for HF de emphasis network is super easy to reach and change.

By measuring frequency responses of u87 and P220 i came to conclusion that it needs to be changed to an 680pF capacitor, instead of original 220pF.  You get incredible replica in minutes. It's just that Perception has better bass response , because i t's bass roll off is weaker as well. I didn't want to touch that. I sent files of the shootout to the owner, he picked wrong one to be Neumann. Probably due to AKG's better bass response. Akg is louder, and has less noise. Modded even less due to more HF attenuation.  By googling i found that people report this cap to have up to 870pF in some Neumanns. This value gave me u67 kind of response.

No need to disassemble any boards, it's right in front of you when you open the mic. This should work for other Perception mics as well. Might be tricky for newbies to remove smd cap. 

One thing to add, AKG is built like a tank!

I'll upload audio files and frequency response graphs tomorrow.
Here is the cap we are talking about.

So where would the smd cap be located on a perception 400 version? please help,