ART USB Dual Pre
Denon DP-61F Dynaco St-70 Hafler DH-110
Hafler DH-500
Hafler XL-280

  Klipsch KG1
Klipsch Promedia 2.1
Mogami/Canare RCAs
Nakamichi CM-300
Scarlett 2i2
Tannoy 603
VOX Pathfinder 15


  • I recently decided to buy a large-diaphragm condenser mic for my home studio. One option was to build a U87 clone using plans found looked like a great project. But buying a shell and constructing a clone isn't exactly cheap. And finding a quality review of a clone's sound quality was oddly I took a different path...I decided to modify a new microphone. After comparing specs and reviews I decided to use the Behringer B2 Pro. I liked the switchable polar patterns, excellent specs, U87 look...and the price. Its visually identical to the 797 NT2S and MXL 2010 and was born in the early 2000s using a mic circuit created by 797 Audio. For some reason the B2 is rarely mentioned as a candidate for modding though, and I'm puzzled why its treated differently than the other U87 clones. According to their website, 797 Audio Co. Ltd was "originally founded in 1952 with the help of Germany." Behringer was originally a German company in 1989 but now makes all its products in "Behringer City" in Zhongshan. Both companies are considered pretty formidable in the audio world.

    Despite the mystique, a microphone circuit-board is relatively simple...not rocket science. Most are based on the Shoeps circuit and online photos of the original B2 show several Wima capacitors and a handful of resistors and diodes. However when I opened up my new mic I was in for a surprise. It appears Behringer has moved away from its older circuit using traditional parts...the mic is now largely SMD(!) SMD parts may not be as exciting as traditional ones...but benefits are lower noise, shorter signal paths, and less parasitic inductance from long leads. Compared to older photos, this newer version contains twice as many parts as the original B2. I haven't seen any photos of the newest NT2S or MXL 2010 but I suspect they now use this newer, more complex 797 SMD design. NOTE: the boards have plated-through holes and tracing on both sides...meaning care must be taken when de-soldering so as not to pull the trace from the holes...otherwise re-soldering can be tricky.

  • Above is the main board which removes easily by taking out 2 screws and sliding it uses a gold-plated terminal pin connector to affix to the capsule...meaning fewer wires. The four blue electrolytics are Sanwha SE caps of 22uf/63V. Sanwha spec sheets list them as standard models "suitable to replace tantalums at low cost". The little tan SMD capacitors may be Sanwha ceramics. I find it interesting that several Wima capacitors are still used instead of going all SMD. Possibly Behringer determined that these were critical locations that benefited from standard film caps.

  • - The first red Wima cap on the left is a 10% .015 250V which looks like a FKS2 (Polyethylene-terephthalate/PET) film.

  • - The pair of red Wima 10% caps are isolated in a hole in the board with the leads hanging out the bottom. The one connected to the capsule is a "1000/630" and looks like a FKP2 (film/foil polypropylene), the one connected to the FET is a "1000/100" and looks like a FKS2 (Polyethylene-terephthalate/PET) film.

  • - The three black components on the top are no-name 51M ohm resistors. The board has several SMD transistors, but a single black Fairchild BF245A was used for the diaphragm circuit. This transistor was obsolete as of 2012.

  • In the photo above are the pair of Wima caps that couple the capsule with three 1G ohm is soldered to a FET lead that passes through a hole in the board. This isolates the FET/cap/1G resistor from the PCB and solder traces...which reduces noise. A simple solution. The smaller board uses a gold 4-pin connector and pops off by removing two screws. The two green components are inductors of 220mh and 68mh at 5%. The 220uf 25v electrolytic is a Rubycon XYF which is listed as a "105C, long life, low impedance" type.

  • NOTE:
    Hot glue was applied at the factory to secure the wire joints on the board and filled up the inside of the XLR jack (see above). It was sloppy though and I didn't like the idea that it touched all the connectors simultaneously as well as the chassis. The wires are short, light, and securely soldered, there's no reason to use glue here.

  • Also, I noticed something interesting. The leg of the FET passing though the PCB to a 1G ohm resistor was touching the edge of the circuit board (fixed in photo). This is a big no-no as it definitely increases circuit noise, which defeats the point of isolating the connections in mid-air. I used a small screwdriver to carefully pry the leg away to center it in the hole. Behringer should pay more attention to this during production.

  • For capacitor replacements my choices were the Elna Silmic or the Nichicon KZ. Interesting to note that while Silmics have pages of forum comments describing their sound, the KZs have far fewer despite being known to the audiophile community since the mid 1990s. Maybe its because Nichicon's data sheets are so minimalist compared to their other electrolytics. NICHICON: you should release some white papers on how your Muse audio electrolytics are constructed and why they are suited for audio... Elna has. In any case I tried both the Silmics and KZs, comments below.

  • - I changed the pair of 10% 1000pf caps to Wima FKP2 1000pf/100V 2.5% (film/foil polypropylene). I replaced the 10% Wima .015 with a .015 1% Wima FKP2.

  • - The single 220uf/25V cap handles DC and since the factory chose a low impedance cap here I decided against an audio-grade electrolytic. I'm a big fan of Nichicon so I replaced it with a PW (low impedance, high reliability, for switching power supplies). I also added a .1uf Russian K73-11 PETP bypass.

  • - I replaced the no-name 1G ohm resistors with Ohmite MOX-200001007JE 1G ohm 5% versions which are rated for "Avionics, Medical electronics, High gain feedback applications, Current pulse limiters, Vacuum and space application". I replaced the three no-name 51M ohm resistors with 50M Ohmite MOX-200005005FE 1% versions. I installed these resistors on the bottom of the board because of crowding caused by the new .015 Wima on the top.

  • - I removed all the old hot glue and replaced the three wires from the XLR to the circuit board with 21g Canare L-4E6S wire. My mic cables are Canare Star Quad so this will maintain the same wiring characteristics all the way from the preamp to the mic circuit.

  • - The factory had the mic's XLR jack pin 1 and the chassis ground tab wired together. Conventional wisdom when it comes to grounding is that a microphone circuit should always 'float' to prevent ground loops. So I clipped the jumper inside the mic to make the B2 fully floating (using only pins 1,2,3). NOTE: I have a set of cheap XLR cables that also connect pin 1 and the ground tab together. But my Canare Star Quad XLR cables don' there appears to be no industry standard.

    - The upper 22uf/50V Nichicon KZs fit but I had to bend one over to reach the solder joint. The lower pair were too tall for the case so I relocated them to the bottom of the board. Despite my comments against hot-glue, I used some to fix the new electrolytics in place to cut down on microphonics, making sure that it didn't touch any leads or signal traces.

    BELOW: The new 51M ohm and 1G ohm resistors are visible on the far left.

  • When I first powered the B2 up after the mod I was surprised through headphones to hear a very low-level rumble across the noise floor. It wasn't on the pre-mod recordings and sounded like very distant bass notes bleeding through from a radio station, I've seen references to this calling it "ghost rumble". It was unrelated to the mic cable or preamp and was puzzling. So I took a very close look and realized I had not defluxed the board properly, especially around the new 1G resistors. I had used Chemtronics flux-off and Q-tips but noticed several near-invisible fibers had remained stuck to the board, touching leads, etc. By squinting under a strong light I used a small plastic pick and tweezers to clean these fibers off which seemed to be everywhere. From now on I will use foam swabs rather than cotton ones that shed. The rumble was reduced but still present and so I went through the circuit with my Fluke and found I had left a bad solder joint on one of the 22uf electrolytics.

    SUCCESS! The 'new' B2 is super clean and clear with a detailed, natural sound. Super responsive with zero ghost rumble even with my Scarlett 2i2 cranked to max on sensitive headphones (MDR-V6). A pre-mod recording of my voice had been clean but had an audible layer of pink noise across the noise floor. In comparison the post-mod circuit is vanishingly silent and cranking the pre-amp gain to max only produces circuit noise from the pre-amp...the mic is so quiet it may as well be out of the circuit. Whether this was due to the misadjusted FET or the new parts is hard to say, but I believe it was due to both.

    I didn't want to obsess over treating the electrolytics like tone controls, but when dealing with microvolt input signals subtle differences between caps are going to appear. Detailed vs tube-like, etched vs smooth, which is correct? To my ears the 22uf Silmics are dynamically correct with 'body' and realistic bass. On playback over speakers and headphones voices sound more natural and dynamic. The KZs had a clean, natural presentation but I found them to sound leaner tonally and a bit polite. After some comparisons I settled on Silmics for the 'lower' pair of 22uf output caps and KZs on the 'upper' pair.

    On another note my Hafler DH500 amplifier uses a single 1000uf 6.3V Silmic in a critical spot on each channel. As a test I replaced it with a 1000uf 25V KZ. It took on a cleaner taut sound, thinner bass, and transients seemed to lose their 'pop' a bit. It was a more polished sound, with advertised "...qualitative and quantitative comfortableness" but just wasn't what I was looking for. So I switched them back to a 1000uf/6.3V Silmic and the sound took on the open, dynamic sound I like.

    CONCLUSION: Behringer should double-check the FET leg/board clearance at the factory, remove the hot glue from the solder joints to prevent bleed-over, and float the circuit by not using the XLR ground lug...all cost nothing and increase performance. Additionally, instead of the no-name ceramic SMD capacitors they should consider using Nichicon UQ versions rated for audio...this would appease the mod crowd.

    Six years ago it was commented online that Chinese mics needed to EQ their circuits to tone down the slight brightness of the capsules they used. The newest B2 is said to use a custom-made capsule, and taking into account the redesigned SMD circuit boards and gold-pin connectors it appears Behringer has consciously updated the older version. The result is the final clarity and low noise of this sub $200 microphone with $20 of tweaks was beyond what I expected...I am VERY pleased and satisfied.

    The Red links below are 44khz 24-bit wavs with corresponding traces
    (play loud with headphones).

    The noise floor between stock and modified is both audible and visible. NOTE: your browser may play these in Quicktime which truncates them. For best results use another audio program.


    POST- MOD RECORDING with Silmics

    POST- MOD RECORDING with KZs © 1997-2017
All mods are illustrative only, perform at your own risk.
Datsun is a registered trademark of Nissan ®