S216 - multi-pattern tube microphone (E80F//K47//K12 //6.3:1) - fit Gyraf G7 PCB

[ln76d]

[PROJECT FINISHED]

After i finished my S16 microphone i started again digging in my G7, which wasn't true G7 from the beginning.
Since i changed too much it would be stupid still call it G7.

Thanks to Jakob project i learn a lot - really lot!

My original build was really ok, but it wasn't sound as i wanted - it was sibiliant mike, with boosted 3k region and with low end response which i didn't truly loved
After i finish this project i will write more about differences, issues etc.

This microphone is "designed" for specific parts - especially capsule!

Capsule - K47 - (RK47, WGT/ISK K47, MJE-K47 - and any other name for the same chinese k47), of course there's no problem to use M7 and original Neumann capsules - effect will be even better

My final build is using "K12" (like RK12). Schematic for use of this capsule publicated at the bottom of the thread

Tube - E80F (same pinout as EF86 but a lot better tube for microphone use - not related with EF80!)

Output transformer - 6.3:1 - i resigned from 10:1 to make it similar to the 5:1 used in G7 if someone would like to convert his build

Am suggesting to make point to point build but if you want to use G7 PCBs, few posts later is all info needed for how to fit it.

Schematic updated.

----------------------------------------------------------------------------------------------------------------------------------------------------------------
All microphones projects:
https://groupdiy.com/index.php?topic=62423.0
https://groupdiy.com/index.php?topic=62441.0
https://groupdiy.com/index.php?topic=64687.0
https://groupdiy.com/index.php?topic=62601.0
https://groupdiy.com/index.php?topic=63436.0
https://groupdiy.com/index.php?topic=63682.0
https://groupdiy.com/index.php?topic=62262.0
https://groupdiy.com/index.php?topic=62682.0

 

[ln76d]

Aaaaa what the hell!
If multi-pattern, lets make 5 patterns.
Additional hypercardioid and hypocardioid.

Schematic updated for R11 proper value!

 

[ricardo]

Nice.

But R1 needs to be 1G too.

 

[ricardo]

But R1 needs to be 1G too.

Why?

Do you think that there will be significant voltage drop with 66M or it's about impedance difference?

R1 is in parallel with R6.

The reason for these large resistors is only slightly due to LF response.  The real reason is cos a small capacitive source like a capsule needs to see a HUGE resistance else noise will be poor.

Great Guru Baxandall explains this in his nov, dec 1968 Wireless World article.  There's a copy in my Yahoo MicBuilders Files.  You have to join.

 

[ln76d]

Thanks ricardo!

I will need to find my account on  Yahoo MicBuilders, i didn't used it for a long time
For now i found "Noise in Transistor Circuits" article.

http://www.keith-snook.info/wireless-world-magazine/Wireless-World-1968/Noise%20in%20Transistor%20Circuits%20-%20P%20J%20Baxandall.pdf

I have some problems to imagine how the dual backplate system + impedances  works

Should i consider dual backplate capsule, from capacitance point of view, as two capacitors in series?

Let say that R1 (or R1+R2) is parallel to the R6. 
So the grid1 see impedance (resistive) 62M or 31M?

For LF,  C1 in parallel to the capsule, move response fall below 10Hz.

What's relation of R3 to R1 and R6. Should i consider that it's also parallel?
Then it would be 21M or 16M? Then response fall below 20Hz.

Am using same values in S16 microphone:
http://groupdiy.com/index.php?topic=62423.0
There's no signifcant audible difference - overall microphone is really low noise.

 

[ricardo]

For now i found "Noise in Transistor Circuits" article.

http://www.keith-snook.info/wireless-world-magazine/Wireless-World-1968/Noise%20in%20Transistor%20Circuits%20-%20P%20J%20Baxandall.pdf

Yes.  This is the article.  I don't know of any other which explains this topic accurately.

It's his Fig 6 (c) which is important.  For noise, you have to get the LF roll-off MUCH lower than 20Hz.  Hence the use of 1G+ resistors in modern mikes.

Fig 6 (c) gives the TOTAL noise for a given capacitance.  The resistor shifts the noise to lower frequency.  On a good mike, with 1G or larger resistors, it is shifted to subsonic frequencies ... which means noise in audio frequencies is reduced.

There should be a LF roll-off somewhere else to get rid of the subsonic noise.  In these classic circuits, it is usually the capacitor to the transformer and the transformer itself  ...  C4 & TR1 in S16.png

Should i consider dual backplate capsule, from capacitance point of view, as two capacitors in series?

In most of the common dual diaphragm (also backplates) circuits, the 2 'capacitors' are in parallel for noise.

Am using same values in S16 microphone:
http://groupdiy.com/index.php?topic=62423.0

This is a good circuit ... something Bernhard Weingartner, who designed the C12 etc for AKG would have used ... if he could get 1G resistors.  Weingartner was a good friend of Clem Beaumont, my mike mentor at Calrec.

The 33M resistor, R1, is decoupled by C1 so the backplate is at earth.  The capsule only sees R4, the 1G resistor, for noise.  BTW, I would make R1 100n.

S16MP & S16H, in their hyper position, have the same problem as S216.  In that position, R1p is in parallel with R4, 1G, so the resistance seen is slightly less than 33M

In 'S216 v1.png', R2 is decoupled by C2.  But R1, 33M, is not decoupled and is always in parallel with R6, 1G, for noise (via C1, 1n).  R3 is decoupled by C3 so is not seen.

E80F is an excellent choice for a modern valve mike because of its guaranteed tiny grid current.  Some older valves won't like 1G grid resistors and have to be selected ...  especially if NOS.  This may be one reason for the differences when swapping tubes.  It will affect DC conditions and also current noise.

 

[ln76d]

Thank you very much ricardo!

As i stated before somewhere here - you are my hero!

Better can't be explained!

I know, that those are some basics but sometimes really confusing!

For S16H i need to verify these values! There's possibility that i will use lower value

 

[ricardo]

My brain hurts trying to understand these polarizing systems but IIRC, this is what is used in C12 and also some 414 variants.

[edit] R1 needs to be 0R ie shorted out [/edit]

 

[ricardo]

Yes, this topology is well known, but there's a big difference without input capacitor.

Today, with good CGO/NPO ceramics, the 1n input capacitor should not affect the sound in any way.  Besides in your original circuit, there is still a 1n capacitor in the signal path.

In cardioid mode with disconnected backside it sound really good.

This is because in cardioid mode, the backside doesn't do anything.  But its capacitance is across the input and halves the signal if left connected as in the usual multi-pattern mikes.

i will do test with  much lower R1 and R2.

If R1 in your original S216 circuit is small, electrical noise will be MUCH greater.

 

[ln76d]

Today, with good CGO/NPO ceramics, the 1n input capacitor should not affect the sound in any way.  Besides in your original circuit, there is still a 1n capacitor in the signal path.

Usual am testing with good styroflex (sometimes two in parallel), C0G and MKP (those never sound good at the input ).
No matter what capacitor am using - there's always difference (even considering  simple  cardioid - one backside disconnected - topology).  It's not that i'm not fan of any capacitor anywhere It's because i like to make many tests
Yes i know that there's still 1nF in the path and this one is a really bad guy - but i wanted at least one less at the input
In G7 you will have both!

This is because in cardioid mode, the backside doesn't do anything.  But its capacitance is across the input and halves the signal if left connected as in the usual multi-pattern mikes.

Unfortunately yes...

If R1 in your original S216 circuit is small, electrical noise will be MUCH greater.

Yeah i get that for the first time Sometimes it is negligible - it's a matter of test and measurement. For AKG topology there wouldn't be need probably

Next tests probably next week - i need to order new tube socket.
"NO MORE" belton mica sockets!
Even chinese ceramic sockets are better...

 

[ricardo]

Usual am testing with good styroflex (sometimes two in parallel), C0G and MKP (those never sound good at the input ).

One reason why styroflex caps sound 'better' is because they are microphonic.  Guru Scott Wurcer tests this in one of his LN articles for Linear Audio.

If you have to use a capacitor in the signal path, CGO/NPO caps are the best if the value & voltage are good.

No idea what MKP are.

Yes i know that there's still 1nF in the path and this one is a really bad guy - but i wanted at least one less at the input

You can remove that cap if your adjustable voltage for pattern (from PSU) can go negative.  ie

• Backplate directly to grid and its 1G resistor at 0V and NOTHING else.
• One diaphragm to 50V via 33M & 100n
• Other diaphragm from 50V to -50V via 33M & 100n

This also gets rid of one 1G resistor which will DEFINITELY reduce noise & other yucky stuff.

The 100n s are also in the signal path ... as they are in the other circuits.
________________________________

Du.uuh!  Just realised R1 in my S216v3.png should be 0R 

But it should be 1G in your original S216VX though I don't think that's the best way to do Variable Pattern.  In fact, I'm not sure it will work for Multi-Patterns. 
________________________________

Quite often, we forget (or don't understand) why the old gurus did stuff in certain ways.  Today, we have MUCH better components ...  and gurus like Bernhard Weingartner would laugh at some of our fashionable choices.

It's educational to speak to these old guys and find out which modern bits they wished they had to use in their designs.

 

[ricardo]

Do i really need negative voltage here?
......
Doesn't it behave similar as Jakob described on his page for G7 pattern system?

In G7, the backplate is at 1/2 the  Minimum & Maximum voltage.  Yours must have the same.

S216 v4.png has the backplate at 0V via the 1G resistor & the grid.  That's why you need a control voltage which can swing both +ve & -ve

 

[ln76d]

In G7 all components except transformer have to be mounted from the "printside" (tracks side).

For S216:
- all components (except C5) and transformer need to be mounted on the "component side" as usual is made

- C5 and tube need to be mounted from the printside

- tube socket need to be mounted, on the small PCB, from the printside

- both PCBs need to be mounted with the printsides facing each other

- front diaphragm connection place is changed, i suggest to solder capsule wire directly to the grid1 pin at tube socket

- there's need to cut one trace and use two jumpers

 

[ln76d]

R1, R4, R5 - 100M - 1G (i would rather start from 100M but currently used all i have for different builds)
R9 - can be used 3.9k but two resistors in series would be better - 2K+2K, 1K+3K, 3.9K + 100R etc.
R11 - 78k - 68k+10k in series or 80k

C1 - 3.3pF to 33pF - high voltage rating cap, lower value - better. Styroflex or Ceramic NP0/C0G
C5 - 470nF - 680nF.  0.5uF MP cap is the best what you can use here.
C9 - 2200uF or 2x 1000uF

R7,R8,R9 - carbon (not carbon composite) - beyschlag, draloric etc.
Rest of resistors - metal film

TR1 - from 5:1 to 10:1
TR2 - 135V/9V or double transformer as for G7 - in both cases 230V primary

 

[ln76d]

Rest of components same as on G7 layout.
R12-R15 soldered direct to the switch.
C8 soldered from the tracks side direct to the C7 soldering points, same way C11 to C10.
R11 (one or two resistors in series) on the top perpendicularly or on the bottom from the tracks side, horizontally from R10 point to the output point. Can be used also instead of wire to the switch, then need to use jumper instead R11 in the drawing.

 

[shot]

I'm confused with this project!

Is this working design, or (as written in first post) still has errors?

I can't keep track through this thread!!



Luka

 

[ln76d]

It has error - by meaning wrong operation point and polarisation voltage mismatch - Anyway it works, but not as it should be I will update it soon as possible, currently have a little hell in my life.
100M sounds a lot better thn 1G, but if someone qould like to use 1G then R2 from 1M - 100M.

 

[shot]

Hahaha! I'm confused again with your previous post!
First you say it still has errors, and then you throw an updated schematic.
Is that schematic now for good, or it's something in between til you fix errors?

Please don't get me wrong! Big respect for all the good work you do!
And take your time. You don't have to hurry.
Real life always comes first!
Soldering second!

 

[ln76d]

Ok, i will try to explain

Original circuit was really great sounding microphone, with proper polarisation voltages etc.
It was "properly"  working
But!
Murdock build it also and sent me measurements.
And...
There was some unidentified problem in my build, symptom was too high current consumption.
I changed xlr connector inside microphone and resoldered tube socket aa also improved other connections.
Then!
I had similar current consumption (read proper) as Murdock, but different sounding microphone.
So?
Higher current consumption  set different operation point as also give different polaristion voltage values.
I had to change voltage dropping resistor in PSU for back diaphragm polarisation and cathode resistor.
By occasion i improved it a little.
Higher C1, high impedance resistors changed for 100M, polarisation voltage set to 63V.
At the end it sounds close to my first "finished" build, but little bit better.

[Project finished and schematics updated]

 

[ln76d]

Here's a schematic for chinese K12 caapsule.
I finally used it because it fits better my hedbasket (lollipop type).
Only change in the circuit is additional "CF" cap from plate to ground for HF damping.
Sounds good.
R9 can be 3.6K
For low end adjustement (this microphone have lot of lows) you can use pot for R9 and adjust value between 2K and 4.7K and see what happend ;D
Other options for less lows are C4 1uF, C5 330nF, C1 1nF.
I would suggest to test one position with new value not to use all together. C1 change in worst case scenario.