Low noise BM800 mod

[abbey road d enfer]

Just wondering what is the real noise source for BM800 circuit. When playing with the circuit I removed the FET, replaced the following PNP with 2SK170 using 1G gate resistor and simple bias point for it (FET source  =>2.2k => bias point => 1.5k => GND).

Beware that the PNP noise without the FET in place is much higher; bipolar transistors noise depends very much on the impedance their input sees.
You may find that, when properly loaded, there is no significant difference.

It's much better now,

Have you actually measured a difference?

 

[VictorQ]

A simple 'fix' would be to put two lower voltage zeners in series.. But the better plan is to get rid of the entire circuit and fit something simpler (and impedance balanced) like an Oktava circuit variant.
I've ordered a couple of these (<$10 delivered!), looking forward to seeing what can be made with them.

 

[Khron]

Where did you find them for THAT cheap? :O But i'm guessing those come with plastic shockmounts?

I've ordered a couple of these (<$10 delivered!), looking forward to seeing what can be made with them.

[edit] Wow, my 1000th post ;D Kinda crazy...

 

[analogguru]

Where did you find them for THAT cheap? :O But i'm guessing those come with plastic shockmounts?
....

...and probably only with an el cheapo 9,7mm capsule - especially when there is written "Neewer" or "Transhine" on it.

 

[Khron]

I'd only be interested in those as body-donors, anyway ;D But i prefer metal shockmounts.

...and probably only with an el cheapo 9,7mm capsule - especially when there is written "Neewer" or "Transhine" on it.

 

[analogguru]

I'd only be interested in those as body-donors, anyway ;D But i prefer metal shockmounts.

Then I would look to find a "Mugig - M-1":
https://www.amazon.de/Mugig-Professionell-Live-%C3%9Cbertragung-Studio-Aufnahme-Broadcasting/dp/B01N8SXCS9

 

[rogs]

...and probably only with an el cheapo 9,7mm capsule - especially when there is written "Neewer" or "Transhine" on it.

...and those that do come with the cheaper capsule - like the Neewer NW700 for example -  tend to have single mesh head baskets, which need to have a second mesh added if they are to be used with an external FET capsule..

 

[mhelin]

...and those that do come with the cheaper capsule - like the Neewer NW700 for example -  tend to have single mesh head baskets, which need to have a second mesh added if they are to be used with an external FET capsule..

Looking for BM700  (= BM800 with '87 style grille), cheapest were sold by Wish but the shipping was not free.

Anyway, you can either modify the included PCB to a point until it becomes messy (add a resistor between the zener and cap/tranny which requires cutting two traces, do some other mods like remove the other 1uf filter cap, change FET etc.) or build an all new board.  I modified single mic to an almost standard Schoeps one - it was nice but a little bit quiet and had still some noise (vs. transformer output U87-clone with the 3U audio super smooth 87 cardioid capsule and GZT-87 transformer I just finished). The other BM-800 mic modded had the single resistor added and the filtering (between FET and the phase splitter PNP) removed. It was  clean and loud but didn't sound as nice (boxier sound, more distortion), changing the FET didn't help either. It might be the capsules vary in quality quite a lot.

 

[Khron]

I don't suppose you've tried the "charge amplifier" configurations that Henry Spragens tested, have you?

http://audioimprov.com/AudioImprov/Mics/Entries/2015/12/20_Modding_a_BM-800_Mic.html

"The large capsule is more sensitive than the small one which shipped in the mic, and it will overload the input FET unless measures are taken. The K596 FET which is usually used in these mics has diodes protecting the gate, and they will begin to conduct at high signal levels. So C12 is added, converting the circuit to a “charge amplifier”. C12 feeds back a charge opposing that coming from the capsule, and keeps the FET gate at ground potential."

Looking for BM700  (= BM800 with '87 style grille), cheapest were sold by Wish but the shipping was not free.

Anyway, you can either modify the included PCB to a point until it becomes messy (add a resistor between the zener and cap/tranny which requires cutting two traces, do some other mods like remove the other 1uf filter cap, change FET etc.) or build an all new board.  I modified single mic to an almost standard Schoeps one - it was nice but a little bit quiet and had still some noise (vs. transformer output U87-clone with the 3U audio super smooth 87 cardioid capsule and GZT-87 transformer I just finished). The other BM-800 mic modded had the single resistor added and the filtering (between FET and the phase splitter PNP) removed. It was  clean and loud but didn't sound as nice (boxier sound, more distortion), changing the FET didn't help either. It might be the capsules vary in quality quite a lot.

 

[rogs]

..... It might be the capsules vary in quality quite a lot.

I've found they do - and it seems sometimes as it they try and disguise that...
This image..

shows BM800 and NW700 capsules that look the same from the front....
Viewed from the back the BM800 has a TSB165a (or similar?) and the NW700 has a 'cheapo' electret with an internal FET  - which is how they get away with the single mesh head basket..

I had to add a second mesh to make  the cheap NW700  I bought usable with a 34mm capsule...(it hummed like mad without it!)

 

[analogguru]

.....
The other BM-800 mic modded had the single resistor added and the filtering (between FET and the phase splitter PNP) removed. It was  clean and loud but didn't sound as nice (boxier sound, more distortion), changing the FET didn't help either. It might be the capsules vary in quality quite a lot.

Did you verifiy that the zener-voltage is 9,1V (and not only 3,1V) ?

I don't suppose you've tried the "charge amplifier" configurations that Henry Spragens tested, have you?
.....

As far as I understand it, mhelin is using the original 16mm "small" capsule, which is an electret capsule (TSB-160A and clones).

http://audioimprov.com/AudioImprov/Mics/Entries/2015/12/20_Modding_a_BM-800_Mic.html

"The large capsule is more sensitive than the small one which shipped in the mic, and it will overload the input FET unless measures are taken. The K596 FET which is usually used in these mics has diodes protecting the gate, and they will begin to conduct at high signal levels....."

This guy obviously doesn't know what he is talking about.  First of all I doubt that a capsule is able to deliver 1,3Vss under normal conditions and then this guy doesn't know anything about the BM-800 circuit.

The diodes are not a speciality of the 2SK596, they can be found in every FET designed for electret-capsules since more than 30 years.

If you look at the datasheet (and/or take measurements) you will know that the FET will draw about 200µA idle current.  200µA across the drain-resistor of 2k2 means a voltage drop of only about 440 mV.  The gate diodes will conduct at 650mV.  That means the FET's drain will see the power supply limit long before the diodes will start conducting.  Solution: increase the value of the drain resistor.

"So C12 is added, converting the circuit to a “charge amplifier”....."

This so called "charge amplifer" is nothing new and is used by Sennheiser e.g. in their ME80 since at least 33 years.

C12 feeds back a charge opposing that coming from the capsule, and keeps the FET gate at ground potential."

Interesting, he doesn't explain how this is achieved when the drain is at 7,8V.  The 2SK596 has already a built-in-resistor in parallel to the diodes between gate and source.  As long as the source is connected to ground, the gate already will be at ground level too, via this resistor and without any feedback capacitor.

Yes, this resistor is only 25MOhm.  Obviously this low value is sufficient for an electret capsule, but I leave it open to everyone to make his own conclusions how this loading will affect a polarized LDC.
 

 

[analogguru]

Here is the excerpt from the datasheet of the 2SK596:

 

[mhelin]

Did you verifiy that the zener-voltage is 9,1V (and not only 3,1V) ?

I measured 8.8V or so.  Now I got rid of the 1 uF caps between the FET and phase splitter, and use 2SK170 FET, biased about 4V on drain. It's not bad. There is still some noise I don't like about. Guess it's the amount of electrons going through the circuit, Got to test without the capsule.

The cap next to zener on BM-800 is only 22uF though the schematic shows 47uF, maybe it doesn't matter but might replace that with a little bit larger one anyway, and change the 1 uF ceramics to film caps.

 

[rodaffleck]

Where did you find them for THAT cheap? :O But i'm guessing those come with plastic shockmounts?

Maybe VictorQ can verify, but I assumed he was only talking about an Oktava style pcb and not a complete microphone. For example this: http://russelltechnologies.co.uk/rt-012-a.html

(note, no affiliation with the guy in the link, never used his products, just trying to add some clarification)

 

[mhelin]

Maybe VictorQ can verify, but I assumed he was only talking about an Oktava style pcb and not a complete microphone. For example this: http://russelltechnologies.co.uk/rt-012-a.html

(note, no affiliation with the guy in the link, never used his products, just trying to add some clarification)

Yeah, the PCB*s are <$10. Sure there are cheaper PCB's available. Regarding the Oktava PCB and circuit I think if you leave out the resistor R8 (15k) and use like 10k R10 you could get a little bit higher capsule voltage, maybe 41V->48V depending on mic pre, some use 52V phantom, why not to go to 60V via a switch in pre if you have built your own pre. Just use 63V - 100V caps everywhere.  Also the capsule could be connected directly to the FET if you use U47 like capsule powering scheme.

 

[rogs]

Yes, this resistor is only 25MOhm.  Obviously this low value is sufficient for an electret capsule, but I leave it open to everyone to make his own conclusions how this loading will affect a polarized LDC.
 

I hadn't realised that the 2SK596 had a 25MOhm gate source resistance in parallel with the diode!
....As you say, that makes it all but useless for use with most LDC capsules.... -6dB at 100hz - or maybe even worse?

Not a problem with electrets with internal FETs, but surely a capsule like the TSB160A -  which only has a pre-polarised element and no internal FET - is likely to be of a similar capacitive value, and would suffer the same LF frequency attenuation problem?

If so, then why do folk bother with modding these boards at all.?
OK, you can help reduce the zener noise with a filter, but if the LF response is still so poor, why bother?....

...Or have I missed something about the TSB160A type electret capsule that makes it perform better then 'proper' LDC capsules would with this FET?

 

[analogguru]

It is not the 2SK596 alone, ALL of those "electret"-FETs have such a "low" input impedance:

2SK156  : 20MOhm
2SK334  : 20MOhm
2SK377  : 18MOhm
2SK596  : 25MOhm
2SK2219: 25MOhm

and they are used (in zillions) as INTERNAL FET's inside the capsules too - 2SK596 is the most common of them.  I don't know why and how, but obviously it works.

 

[abbey road d enfer]

-6dB at 100hz - or maybe even worse?

That is not the worse effect. Actually most cardioid mic have serious LF roll-off, which is dominant over the electrical Hi-Pass effect. This is inherent to the cardioid construction.
I would be more concerned with the increase of electric noise, due to the noise energy spectrum being shifted more into audible territory.

 

[mhelin]

Guess no serious modder leaves the FET there.
Now as the hiss is gone I'm wondering how the fake large capsule mount affects to sound acoustically. Maybe modding the mount by removing excess material would help a little bit.
If you can see the image it's getting messy.

 

[rogs]

If you can see the image it's getting messy.

That's what I  discovered -- so I  had a go with a  simple Schoeps type circuit and built it onto a piece of stripboard ..
(image attached- I hope!)
I've been quite pleased with the results so far....

(Some extra notes here:  http://www.jp137.com/lts/LDCX2.notes.pdf )