Active ribbon-mic

[trubac]

Thanks for the link !

 

[trubac]

You should definitely check this for yourself and let us know.

Like I said, I'm not yet convinced in either direction, but i have only used my ears, and haven't done a strict signal-to-noise test. I built three slightly different boosters with different transistors, and also tried the commercially available FETHEAD booster, which has a bit more gain.

What I found - or at least what I perceived - using my API A2D preamp was that for many ribbon mics there is very little difference between using a booster and just cranking the API. In both cases the noise from the ribbon mic is amplified, especially with an old oxidised ribbon and a non-optimised transformer.

But if you have a nice quiet mic then you'll be happy!

Maybe then an alternative will be to design a simple preamp, regular one, not phamtom powered, with no bells and whisles, that could deliver a fixed 40db ultra clean gain ...

 

[ihscoutlvr74]

Would the circuit near the top of page 10 posted eons ago by PRR work with a supermatch transistor pair like a JAN 2N2920?  I have seen several DOA circuits that utilize a similar transistor and curious if this would be applicable to a balanced ribbon preamp  post trafo using phantom power.  thoughts??

 

[Rossi]

As has been said before (I think) the usefulness of this circuit depends on the base spreading resistance (Rbb) of the transistors. There are not that many transistors that are optimized for super low Rbb.

The best ones are transistors specifically made for low Z input stages such as Toshiba 2SC3329 (hard to find), Hitachi 2SC2546 (easier to find).

Some switching transistors are usable such as the 2N4401 (differnt brands and different batches will perform better or worse), you may be able to improve noise performance slightly by paralleling two for each leg, but you'll still have a bit more noise than with the above ultra low Rbb transistors (for which paralleling is unnecessary and won't improve noise performance noticeably).

Matched transistors may enable you to get rid of the input caps, but if their noise performance (i.e. Rbb spec) isn't excellent, what's the use?

One advantage of the ribbon booster approach is that you don't have to crank up your mic pre to its absolute limit. Most preamps get somewhat "cloudy" when you use them with 60+ dB gain. The PRR ribbon booster + 50 dB gain will give more clarity. Of course that may be different with super high class preamps, but standard or middle class preamps usually lose some of their punch and clarity in their highes gain settings.

Another advantage is the PRR ribbon booster's higher input impedance compared to most mic preamps, which results in slightly better treble and bass respons as well as lower gain loss. That's especially true for ribbons with somewhat higher output Z. Some ribbon (e.g. most Chinese ones) mics' actual output Z is higher than specified in the data sheet.

[edited for correct transistor designations]

 

[ihscoutlvr74]

Do you mean the Toshiba 2SC3329 and Hitachi 2SC2546 instead of 2SA?  2SC3329 says it has Rdd of 2 ohms and is used as input on MC headphone  amps. 

http://www.datasheetcatalog.org/datasheet/toshiba/991.pdf

If the 2SC3329 is applicable they are available for $0.25/ea from futurlec.com

 

[trubac]

As has been said before (I think) the usefulness of this circuit depends on the base spreading resistance (Rbb) of the transistors. There are not that many transistors that are optimized for super low Rbb.

The best ones are transistors specifically made for low Z input stages such as Toshiba 2SA3329 (hard to find), Hitachi 2SA2546 (easier to find).

Some switching transistors are usable such as the 2N4401 (differnt brands and different batches will perform better or worse), you may be able to improve noise performance slightly by paralleling two for each leg, but you'll still have a bit more noise than with the above ultra low Rbb transistors (for which paralleling is unnecessary and won't improve noise performance noticeably).

Matched transistors may enable you to get rid of the input caps, but if their noise performance (i.e. Rbb spec) isn't excellent, what's the use?

One advantage of the ribbon booster approach is that you don't have to crank up your mic pre to its absolute limit. Most preamps get somewhat "cloudy" when you use them with 60+ dB gain. The PRR ribbon booster + 50 dB gain will give more clarity. Of course that may be different with super high class preamps, but standard or middle class preamps usually lose some of their punch and clarity in their highes gain settings.

Another advantage is the PRR ribbon booster's higher input impedance compared to most mic preamps, which results in slightly better treble and bass respons as well as lower gain loss. That's especially true for ribbons with somewhat higher output Z. Some ribbon (e.g. most Chinese ones) mics' actual output Z is higher than specified in the data sheet.

Thanks Rossi that makes sense ...

Anybody has tried the Cloudlifter ? http://www.youtube.com/watch?v=pyaSgqQ6lAg

 

[Rossi]

Do you mean the Toshiba 2SC3329 and Hitachi 2SC2546 instead of 2SA?  2SC3329 says it has Rdd of 2 ohms and is used as input on MC headphone  amps. 

http://www.datasheetcatalog.org/datasheet/toshiba/991.pdf

If the 2SC3329 is applicable they are available for $0.25/ea from futurlec.com

My bad! Yes, 2SC not 2SA. I'll correct the designations in my above post so as not to confuse future readers.

25 cents is very cheap.

 

[greenmanhumming]

Performance with 2SK170BL:

Biased at 2mA per device, with 300Ω||2KΩ||2*6.7KΩ load, I get a gain of only 6dB with the Sources shorted together. Hardly worth it. 4.6mA (about the limit of Phantom) isn't a lot better. Slide-rule from the datasheet numbers gives slightly better gain, so my model may be skewed, but not enough to save it. We can't really short the Sources together because we can't match two devices so exactly that we won't get disturbing current in a subsequent input transformer. (If a Dual becomes available, that may help.)

what about adding a 10R trimpot between the source resistors and ground to balance the current through each fet?

I had built something like this independently, then discovered this thread. Mine performs poorly into transformer inputs, possibly because of DC offset between the drains of the 2 fets although they are matched as close as I could from the parts I had. I'm about to try adding a trimpot, but i will have to build it in a box rather than in a plug.

has anyone come to a conclusion about the lowest noise transistor to use? and if a BJT version is actually quieter in real life? So far I have found little or no noise advantage using a fet booster relative to just cranking my preamps harder.

what about lowering the input resistor on the fet version to 5k or so? why raise it to 22k on the fet version? surely the load that is optimum for the ribbon transformer output is the relevant thing for that?

 

[greenmanhumming]

I'm not surprised this has low gain, it has a 300R between 2 out of phase amps with 6.8k output Z. effectively loading 6.8k output with 150R.

what about replacing each half with a cascode? or some other DC coupled second transistor to lower output Z?

or is it better just to use BJTs to start with?

 

[PRR]

> if a BJT version is actually quieter in real life? So far I have found little or no noise advantage using a fet booster relative to just cranking my preamps harder.

> what about lowering the input resistor on the fet version to 5k or so? why raise it to 22k on the fet version? surely the load that is optimum for the ribbon transformer output is the relevant thing for that?

> this has low gain, it has a 300R between 2 out of phase amps with 6.8k output Z. effectively loading 6.8k output with 150R.

> what about replacing each half with a cascode? or some other DC coupled second transistor to lower output Z?

> or is it better just to use BJTs to start with?

What application are we talking about today? A 100-200 ohm microphone?

The preamp designer has the same options that you do. If the source is roughly mike-impedance, and the preamp isn't junk, you are unlikely to do much better than the preamp.

If the source is well below 30 ohms, transformer it. There are no well-specified devices for such low noise impedance.

Lowering the input impedance with shunt resistors makes S/N _worse_. Don't do that unless you "must" match the source for some other reason, and can accept the loss of available signal power.

This pre-preamp "must" show ~~300 ohms to the main preamp because some preamps get unhappy with high source impedances. Yes, this limits gain. No free lunch.

If Phantom Powered there is a limit to the current we may reasonably pull. Gain and noise resistance are a function of Gm which is a function of device current. There is a fundamental limit on Gm/mA, the Ktq formula, about 28mV. BJTs usually come very-very close to theory. FETs (Silicon or vacuum) approach 28mV at VERY teeny currents but deviate high at practical currents. The 2SK170 does not suck at a few mA; most JFETs don't come close.

Is preamp hiss really limiting your work? All mikes hiss, and many have hiss very nearly as low as their resistance implies. Most studio ambient noise is higher than mike self-hiss.

 

[emrr]

I did a classical guitar recital remote recording recently, and took a pair of Beyerdynamic M130/M160, that were maybe 8 feet out from the performers.  This pair was picked by the artist after a rehearsal run.

I also had the house stereo condenser (AKG I think), which was flown roughly 12 feet high and 6 feet back.  This was the room, which is of fairly recent design and I understand exceeds the NC-15 noise criterion. 

http://cpf.uncsa.edu/watson.html
http://www.engineeringtoolbox.com/nc-noise-criterion-d_725.html

The ribbons revealed more hiss, and had fairly low output.  The condenser had less hiss, but the sound of air movement in the room was louder than the hiss in the ribbon chain, and more distracting due to the non-constant nature.  In the end the differences came out in the wash, even with a room as quiet as this. 

 

[greenmanhumming]

Thanks PRR, that confirms pretty much what I was thinking.

My current interest stems from getting my hands on a Beyer M160N re-ribboned by Stephen Sank. Output impedance nominally 200ohms, not sure if changing the ribbon changes that.

It sounds really nice, but it is by far the quietest mic I have. It is the only mic I have where preamp hiss is (subjectively) louder than room noise.

When recording acoustic guitar with it I can hear hiss throughout the recording if listening on headphones, although not in a mix.

I have several preamps that have enough gain, but they all hiss at that gain level by a roughly similar amount. For reference I've mostly been trying it with a preamp I made using the THAT1512/1646 chips, but i have a valve pre as well that is very nearly as quiet. (based on the NYD mila design)

I guess I was just hoping I'd missed something - or that putting the first preamp stage right at the mic would help.

I'll try the BJT schem posted here as well, but I'll be surprised if it makes any noticeable difference.

Why does lowering the input impedance make S/N worse? I had thought the opposite, as long as the input impedance is not so low as to drag down the level of the signal. What is optimum? why not make the input 100k or more then if using a fet?

 

[trubac]

I had a M260 by Sank ... I sold it after i've found an original M260 (grey model) ... the original was way better, more output AND better sounding.
Sank mod on the Beyers are IMHO not good : he doesn't use the right ribbon, and even the corrugate RCA ribbon he puts in are not as gap adjusted as the original one ---> very low output.

 

[greenmanhumming]

I had a M260 by Sank ... I sold it after i've found an original M260 (grey model) ... the original was way better, more output AND better sounding.
Sank mod on the Beyers are IMHO not good : he doesn't use the right ribbon, and even the corrugate RCA ribbon he puts in are not as gap adjusted as the original one ---> very low output.

interesting, what exactly was the difference in the sound? was the Sank one the one with the hipass transformer, or also an original?

have you compared to the current stock model (hipass) too?

 

[emrr]

If I'm within 3 feet of a classical guitar with an M160/130 pair and a decent preamp, there is no audible hiss. 

 

[trubac]

I had a M260 by Sank ... I sold it after i've found an original M260 (grey model) ... the original was way better, more output AND better sounding.
Sank mod on the Beyers are IMHO not good : he doesn't use the right ribbon, and even the corrugate RCA ribbon he puts in are not as gap adjusted as the original one ---> very low output.

interesting, what exactly was the difference in the sound? was the Sank one the one with the hipass transformer, or also an original?

have you compared to the current stock model (hipass) too?

The Sank one was also a grey one, not the more recent M280.80.
The Sank M260 had a lower input than my B&O BM6 ! an already very low output ribbon.
As for the sound, the original Beyer was just better, better overall sound : I've plug them into a Phoenix Audio preamp (they can go to 80 db) side by side, and just by speaking into them one after the other, the original was better, plus the Sank one had to be pushed really hard (added hiss) to matched the gain.
I didn't compared the old 260 aith the newer ones, but the older grey ones are easy to find and Beyer is still able to remplace the ribbon if needed.
Hope that helps 

 

[PRR]

> ...only mic  .... preamp hiss is (subjectively) louder than room noise.

That, if true, says the mike SENSITIVITY is low.

> as long as the input impedance is not so low as to drag down the level of the signal.

10X impedance is 1dB worse NF. Even the ~~2K input of most preamps is significant.

 

[mjk]

>Why does lowering the input impedance make S/N worse?

So as I understand it, the source resistance and load resistance in parallel results in signal ratio of Rl/(Rs+Rl) which is a voltage divider and noise ratio of sqrt((Rs||Rl)/Rs) relative to the unloaded source resistance. The decrease in noise is smaller than the attenuation of the signal, resulting in the worse signal to noise ratio.

>why not make the input 100k or more then if using a fet?

Indeed. Unless the mic likes to see lower impedance for some reason and the small increase in noise is a tradeoff.

 

[Rossi]

I never built the FET version, so I don't know how it actually performs. Back in the thread it was suggested that the gain may be pretty low. The BJT version works very well with the right transistors (see above). As I said before, the main advantage is not in noise performance, but in keeping your mic pre from working at its limits. So you do get better sound with low end or mid priced pres but not with high-end extra-clean super-low-noise pres. That said, the BJT version is fairly low noise at about -129 dB A.

I recently saw a commercial product from Cloud Microphones that works with 4 FETs and provides 20 dB gain. Its noise performance was about -127 dB A. Which is pretty good for a transformerless FET input at those low impedances. About as good as a INA163 preamp chip, acutally.

 

[Samuel Groner]

Even the ~2k input of most preamps is significant.

It is possible to synthesize specific input impedances using active elements, with resulting lower noise contribution compared to passive implementation. Hardly worth the trouble for a standard mic pre, but worth keeping in mind for other things...

Samuel