Low noise BM800 mod

[mhelin]

That looks neat. Anyone familiar with paper-PCBs? They are PCB designs with the layout printed on a paper which is then glued on top of a usual stripboard on the component side showing also all jumper wires. That would be very handy way to share prototype PCB designs. Se for an example this synth VCO paper-PCB design page:

http://paperpcb.dernulleffekt.de/doku.php?id=paper_synths110_vco
Paper-PCB: http://paperpcb.dernulleffekt.de/lib/exe/detail.php?id=paper_synth%3Aps110_vco&media=paper_synths110a---voltage-controled-oscillator.jpg

Regarding the Schoeps circuit on BM-800 I think it is basically a good idea except the poor FET with built-in  HP, using the FET stage with gain helps a bit. I've been wondering if it's possible to build the 3-device Schoeps using NPN output transistors with some gain, maybe that would be the way to go with lesser output capsules?

 

[abbey road d enfer]

I've been wondering if it's possible to build the 3-device Schoeps using NPN output transistors with some gain, maybe that would be the way to go with lesser output capsules?

In the Schoeps circuit, the output transistors are in emitter-follower mode, which provides a very low output impedance (essential for good CMRR), but does not allow any voltage gain.
An alternative solution has been presented by Sergei Stechenko at micbuilders.com, using Szlikai pairs (a clever NPN/PNP combo), that can be easily modded to provide some gain.

 

[rogs]

That looks neat.

Thanks.

Anyone familiar with paper-PCBs? They are PCB designs with the layout printed on a paper which is then glued on top of a usual stripboard on the component side showing also all jumper wires. That would be very handy way to share prototype PCB designs.

As you can see from my notes I linked to earlier (see here: http://www.jp137.com/lts/LDCX2.notes.pdf ) I adopted a slightly different approach, by taking the stripboard and adding the track cuts, links and components sequentially - with  detailed notes for each stage.  Only takes about an hour to construct....
Not a good idea for mass production of course, but ideal for 'one off' experiments --- and it's not difficult to try out alternative circuitry.... without needing to create  a 'birds nest' !

Regarding the Schoeps circuit on BM-800 I think it is basically a good idea except the poor FET with built-in  HP, using the FET stage with gain helps a bit. I've been wondering if it's possible to build the 3-device Schoeps using NPN output transistors with some gain, maybe that would be the way to go with lesser output capsules?

I was disappointed with the TSB160A type capsules that came with my BM800s... and other Transsound capsules are very expensive here in the UK (shipping costs from the US).
I've found the cheap 34mm K67 type clones from China to be very good value (A bit 'spiky' HF wise, but that is 'fixable' - especially for vocal use).
And using a simple voltage multiplier based on the ideas from Rory Holmes' 1983 ETI magazine article -
(see here:  http://www.gyraf.dk/schematics/Voltage_multipliers_with_CMOS_gates.pdf )
- is  quite a useful way to create 'noise free gain'.....
Supplying the 34mm capsule with a 60V DC bias voltage will of course increase the sensitivity by 6dB over a 30V bias

If you're coupling into a reasonably low noise preamp (and some of the cheaper audio interfaces - like Behringer -  have some pretty  good mic preamps these days) then trying to get more gain from the mic circuitry itself is probably not really worth the effort ... easier to let the 'pros' do the low noise gain....

I started out using some 2N5457 FETs and 2SA872 low noise PNPs that I had lying around in my initial experiments, but they're not so easy to find these days....So  I find the J113 FET and the low noise BC560 PNPs  work pretty much the same.

I found the IDSS for the 2N5457 and the J113  are slightly different,  so it's useful to have a bias control pot to set bias to minimise the second harmonic distortion... but both types work well with around 1mA flowing through the FET.

I chose a 9V zener to allow for the voltage multiplier to create a 60V DC bias output, and also to allow for a passive low pass filter after the zener - to reduce the zener noise - and still  leave around 8.4V to power the 1mA FET bias network.

I find using stripboard - and not committing to a PCB design early on - allows you to have more leeway to try things out, without incurring too much cost .....
But we all have different ideas on the 'best way' to experiment of course !   

 

[mhelin]

I think this was my last attempt that was working OK (using the original TSB-160A capsule), though I didn't had the 47uH inductors and the PNP's were what were there already (2N5401 smd versions) - also the zener is 9V1 one actually.

Let's hope the chinese manufacturers will find the below schematics and change their PCB's, and forget the PC computer use. and bundle the mic with a proper mic cable with XLR-3's on both ends - you can't actually just change the mini jack to XLR on the cable currently bundled as it's only single core lead.

 

[Khron]

Interesting choice... Is the gain boost from that cap bypassing the source resistor, really necessary?

Because if not, especially if equal drain & source resistors are used, that extra PNP phase-splitter is sort of unnecessary (imho).

Or if you really "must" keep the JFET working basically single-ended like that, you don't need to use same-value resistors there, and optimize those for whatever balance of gain vs. distortion (or headroom) you want / need.

I think this was my last attempt that was working OK (using the original TSB-160A capsule), though I didn't had the 47uH inductors and the PNP's were what were there already (2N5401 smd versions) - also the zener is 9V1 one actually.

 

[mhelin]

Yeah, I also modded one mic that way but there wasn't enough gain for me at least for ac. guitar and such, maybe it would be fine with drums though. There are sure some place for optimisations.

 

[kingkorg]

What is actually the purpose of the 10k r7 and r8 in the original circuit.

Also, why use Q5 and not go with the FET as phase splitter. Behrinher B2 uses also bipolar in the phase splitter stage, but not the 10k resistors.

 

[Khron]

The 10k's don't have much point there without any base-emitter caps for HF rolloff, that's true. But Q5 is there as a sallen-key 2nd order high-pass filter, which was unfeasible with just the JFET.

 

[RuudNL]

The 10 K's will only increase the self noise of the microphone.
(We discussed this before and for the people who don't believe it: try it yourself and you will hear the difference!

 

[kingkorg]

Ok, i wasn't aware of this Sallen-key feature. Thanks for that one. @RuudNL I have tested this, it indeed increases noise. Not sure why they put them there in the first place.

 

[k brown]

Even bypassing the filter, the 10ks, and doing the supply noise reduction, I still feel the circuit is let down by the quality of the FET and BJTs used.

 

[Khron]

"Quality" meaning...?

 

[kingkorg]

I replace q5 with a fet (2sk208 usually), but i was just wondering why they went this route to begin with.

Also i find these too convenient to ignore. I stumble upon these occasionally in diverse cheap donor bodies.

 

[k brown]

"Quality" meaning...?

Sound quality.

Bypassing the board entirely and using a 2SK660 FET with the guts of an Audio Technica power module always ends up sounding better to my ears (especially if the AT trafo is replaced by impedance-balanced out - Oktava MK-012 style).

 

[tbone8]

A simple 470 ohm resistor between the zener-diode and base of the regulator transistor (together with the filter cap) will work wonders.

Wow, what a HUGE difference that mod makes!!! A little history on this project if I may...

So, bought a BM800 a couple weeks ago to maybe do some videos with. Was $25 on the Amazon and I knew it was cheap but on the upside, would learn a ton building a pre-amp and modding the microphone. I built a little amp based on the TL072 and the setup sounded horrible. Wasn't sure if it was the mic or the amp so built another amp based on the SSM2019 (https://www.instructables.com/Build-the-Four-Channel-SSM2019-Phantom-Powered-Mic/) and man did that sound good. I modified the schematic for just a single mic:
There is not a OPA2134 to be found but I did find some OPA134's at Antique Electronic supply. Had issues with the first amp with 120Hz harmonics so investigated capacitor multiplier circuits and came up with this mod for a power supply:
It ran real quiet in LtSpice and seems to work nice with this amp.

However, there was pretty noticeable noise on my test recordings. I could filter them out with Audacity but figgered I would try to reduce it. This is my Zingyou BM800 schematic:
I laid out the schematic to correspond with the board components so that is why it looks like it does. Anyway, added that 470 ohm resistor and a 220uF cap and the noise is completely gone! This is how it did it on the PCB:
Just took off the SMD Zener and slipped in these, mod schematic:
So, wanted to thank the board and share my project. Am open to any comments or suggestions, particularly on the power supply I came up with. Thanks!

My setup, amp is done, PS still in breadboard pending possible revisions...

​

 

[Pariah Zero]

Very interesting reading, given I just bought a BM-700 and BM-800 for $15 each during Amazon's Prime day -- really just to be mic bodies; but maybe I'll find a use for the original electronics after all.

Was $25 on the Amazon and I knew it was cheap but on the upside, would learn a ton building a pre-amp and modding the microphone.

I'm going to have to check out how the circuit differs vs the $15 ones I got -- mine are Pyle and Neewer, and no doubt came from the same factory, but I'm interested to see if they have the same PCB/circuit.

Given my goal was a donor body, I'm just happy they were metal...

 

[Pariah Zero]

The 10 K's will only increase the self noise of the microphone.
(We discussed this before and for the people who don't believe it: try it yourself and you will hear the difference!

I sim'd it in LTSpice, and tried it out on the board. I can see the difference in LTspice, but I'm not seeing anything in the real world, sadly. Granted, my testing methodology is crude at best.

Wow, what a HUGE difference that mod makes!!!

I'm glad it worked for you. While I can see an effect in LTspice (assuming I'm using it correctly -- it's been 20 years since I've used pSpice), when I try to measure the noise level of the mic, there was no measurable change in the 'before' vs 'after' noise level.

Recordings in the quietest room of my house, wrapped in a weighted blanket have a RMS sound level @ -42 dB before modification, and -42 dB after modification. Swapping the capsule to a JLI-2555, however, dropped the noise level by 6 dB.

Just to make sure it wasn't my audio interface/preamp, I plugged in my Røde NT1 (but didn't bother moving it to different rooms, wrapping it up, etc.) Even in a room with fans, and A/C still going, the NT1 clocked in 10dB quieter than the modified BM-700.

However, there was pretty noticeable noise on my test recordings. I could filter them out with Audacity but figgered I would try to reduce it. This is my Zingyou BM800 schematic:

Interesting to note that my two mics (An NW-800 and a Pyle BM-700 rebadge) have a PCB which differs from yours only in the lack of the "700/800-1" silkscreen; otherwise the PCB looks identical. Mine also have two 47µF caps on the backside, rather than a 47 & 22 µF. I soldered mine slightly differently - I have an abrasive fiber pencil, so I was able to wear away the solder mask and expose the trace:



That gave me a lot more real estate to solder down the resistor to. You can also see the wires I replaced the 10k resistors with. Sadly, my flux pen barfed on the board, so I had a bit more cleanup than I bargained for, but when it was all over, it cleaned up really nicely.

After that, I used the thru-hole for the 47k cap as a peg to the ground plane on the back to solder the zener diode and electrolytic capacitor to; it's just a more solid connection physically and electrically. (Looking at the PCB, I don't see any chances of a ground loop, but my career took me down the software route, so I could easily be wrong).

While fun to do, I just didn't see any reduction in noise by dropping in the RC filter around the zener diode.

 

[Khron]

You might have been lucky to catch a zener from a less noisy batch, perhaps? Not all zeners are created equal, despite similar voltage specs...

I recall Doug Ford mentioning about this in one of the EEVblog videos about mics, that they had noose issues from some "hard knee" zeners, but "soft knee" ones were much quieter. Probably not a spec you'll find in many datasheets though...

 

[Pariah Zero]

You might have been lucky to catch a zener from a less noisy batch, perhaps? Not all zeners are created equal, despite similar voltage specs...

That's possible: For my particular noise mod, I wired a 4.3 and 4.7 V zener diode in series, rather than a single 9V zener as - My idea being the diodes operate in zener mode rather than avalanche mode -- zener mode being much quieter between than avalanche mode, IIRC.

Even then, one thing I neglected to say was that the BM-700 was itself about 8 dB noisier than the NW-800 (ie. Neewer's BM-800), with an identical circuit board, and identical part values.

I have swapped the capsules on both, so they both have identical JLI capsules - the NW-800's noise actually increased by ~ 1.5 dB.

So I'd expect applying the same mod would give me about 2.5 dB of noise reduction - not a lot. It might also be interesting to see what swapping the head basket does, since everything else is looks identical (same make/model for parts).

I may also post my LTSpice model into the drawing board forum, just to ask if I'm using the tool properly. (I have an Electrical Engineering degree; my career took a different path and I haven't used the electronics part).

 

[Pariah Zero]

Interesting: Henry Spragens has a couple of more recent posts about alternate designs he has tried, including a charge amplifier (already discussed), and a source follower design - then he followed on with a comparison of the measured THD of the mods.

Mr. Spragens's mods touches on noise, but are far more interested in addressing the limitations of the K596: If you upgrade to a larger electret capsule, you'll find at higher SPL levels the capsule produces voltages that cause the K596's internal diodes to start conducting - introducing distortion.

So his mods are fixing two issues, not just noise. I'll have to try out the source follower design on my BM-800, when I get a chance.

Either way, I'm pretty sure I'll end up replacing both PCB's with something more robust anyway - I intend my cheap mics to be beaters, to be carried while hiking to record nature.

Having a modded PCB with a 1000 µF capacitor hanging off a 1 mm^2 SMD pad ain't gonna end well.