Active ribbon-mic

[greenmanhumming]

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.

I have seen the cloud product advertised. I'm curious what the extra fets are doing? do you know?

the fet circuit from this thread has a useful amount of gain if the 300R is increased to 560R. At least one commercial product of this type omits that resistor altogether I am pretty sure.

 

[greenmanhumming]

The BJT version works very well with the right transistors (see above).

I didn't see a definitive answer on what the "right" transistors are for the BJT version, what do you recommend?

EDIT: I see you recommended Toshiba 2SC3329 or Hitachi 2SC2546 - haven't found either yet.... suggestions about where to get them (in europe) or alternatives would be great. I have some 2N4401 I will try it with those to begin with.

 

[mjk]

>It is possible to synthesize specific input impedances using active elements, with resulting lower noise contribution compared to passive implementation.

Is it possible to achieve noise performance better than the ENR of the total input resistance, including the source resistance?

That sounds interesting, though the ~-130 dBu achievable with a 30 ohm source seems "good enough" relative to the self noise of microphones.

Michael

 

[mjk]

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.

Hi,

Do you mean by -129 dB A that you made an A-weighted noise measurement?

Thanks,

Michael

 

[Rossi]

EDIT: I see you recommended Toshiba 2SC3329 or Hitachi 2SC2546 - haven't found either yet.... suggestions about where to get them (in europe) or alternatives would be great. I have some 2N4401 I will try it with those to begin with.

In Germany the Hitachi one is available from www.reichelt.de.
4401s seem to vary between manufacturers. The lowest noise 2N4401 in my bin were the Motorola ones. You can improve noise performance a bit by paralleling two 2N4401 transistors per leg. Single 2SC2546 or 2SC3329 are still a bit better. Unfortunately there's nothing similar that is better available. I tried quite a number of transistors.

 

[Samuel Groner]

Is it possible to achieve noise performance better than the ENR of the total input resistance, including the source resistance?

It is possible to reduce the noise contribution of the input resistance to theoretically zero. The source resistance contribution is of course unaltered, as is the attenuation.

For resistive sources and small attenuation factors the gain is modest. There's more profit for reactive sources, e.g. MM cartridges. Some dynamic mics are in fact quite inductive (e.g. a SM57 rises close to 1k Ohm at higher frequencies).

Samuel

 

[Rossi]

Hi,

Do you mean by -129 dB A that you made an A-weighted noise measurement?

Thanks,

Michael

While I have a measuring rig now, I didn't back then. I simply compared to a number of preamps whose noise figures I know. It's actually not that hard to hear the difference between -127 db A and -129 dB A. Those figures are kind of weird. What they acutally mean is that preamp A is 2 dB noisier than preamp B (at high gain, same amplification, of course).

 

[PRR]

> why not make the input 100k

Other Reasons.

When the mike comes un-plugged, what happens?

The long line picks up all the crap in the room. 2K-10K catches less crap than 100K.

The input device's noise Current raises a noise Voltage in the 100K. This may be small for FETs, true; for a BJT optimized for 200 ohm source the hiss will be large.

Some fools actually judge inputs with nothing connected. Fools with money must be made happy. For a design for low-Z source, a lower impedance will usually be more benign in such mis-use.

There's little point in going to 100K if Phantom is connected: 13K across the mike.

The impedance of a long audio cable CAN'T be 100K at the top of the band. Say 100 feet at 30pFd/foot, 3,000pFd. That's 26K at 20KHz. Is it better to be 100K 20Hz-5KHz dropping to 25K, or to be say 5K 20Hz-25KHz? What about 10' cable? 1,000' cable? (There's no correct answer.)

Transformerless inputs lack RF rejection. Most will not handle RF gracefully: they rectify large RF which usually puts crap in the audio band. A giant capacitor will kill RF but also all the highs from the mike. Adding a coil helps but adds a resonance which could be worse. The usual compromises require L-C damping resistors in the 100-1K range.

As Samuel says, you can do many clever things to shift impedances without actual loss. That's after you get the BASIC action right (as much signal with as little loading as is reasonably possible). And trying to shift RF impedance may demand good RF performance from an amplifier which is already burdened with audio tasks.

 

[greenmanhumming]

In Germany the Hitachi one is available from www.reichelt.de.
4401s seem to vary between manufacturers. The lowest noise 2N4401 in my bin were the Motorola ones. You can improve noise performance a bit by paralleling two 2N4401 transistors per leg. Single 2SC2546 or 2SC3329 are still a bit better. Unfortunately there's nothing similar that is better available. I tried quite a number of transistors.

what about transistor arrays?

the LM394 was mentioned earlier, but it is expensive, what about the THAT300 or similar? (obv I'm not really sure what to look for)

 

[zebra50]

THAT 300 works fine - I've built the circuit above with these paralleled. Think it got mentioned above.
But my experience was that the 2N4401 was just as good, and a little cheaper.

Stewart

 

[greenmanhumming]

well i built it with parallel 2N4401, it is the first one I've built that actually makes a definite (though small) improvement over just cranking the pre.

Curious that you didn't find any improvement with the THAT300 looking at the datasheet it sounds like its purpose built for this sort of application.

 

[Rossi]

Everyone wondering about various transistors, please go back and read the whole thread. PRR and others gave valuable insights. In short, Rbb is the dominating factor in the noise department. If you want optimum performance, get those 2SC3329 or 2SC2546 super-low-Rbb transistors. They're not that expensive, either, just a little hard to find.

 

[greenmanhumming]

I think we are speculating about alternative transistors because we can't get the recommended ones without disproportionate expense and fuss.

I can't find them at any of my usual suppliers, (cpc, farnell, mouser, rapid, RS.)

reichelt have them and also some other parts I want- but they want 19 euro postage for a handful of transistors to the UK, I had decided to just pay it and save time but they don't seem to take payment by credit card! What a waste of time!

 

[zebra50]

I can't find them at any of my usual suppliers, (cpc, farnell, mouser, rapid, RS.)

I had a similar experience.

Hey, I just noticed you're in York. We should compare results sometime. 

Stewart

 

[mjk]

Try futurlec.com 

 

[greenmanhumming]

thanks!! I've ordered from futurelec, very good prices and cheap shipping. easy straightforward site too.

shame they didn't have the SA1085 I am also looking for.

 

[trubac]

I got some 2SC3329 ... do you match them with the hfe figure ?
The ones I've got are very different, hfe speaking, sometimes hfe doesn't stop decreasing ...  ???

 

[greenmanhumming]

well I have been matching the 2N4401s I've been using, haven't got my SC3329s yet

I don't really understand BJTs so I can't say in detail, but seems to me its very important that there is no significant DC difference between the in and out of phase outputs, you don't want DC through yr preamp input transformer!

so I guess the devices need to be matched current or they will draw down the 6.8k resistors by different amounts. This is a big problem with the fet version of the circuit, and if i persevere with that I will use a trimpot to DC balance the output. that could maybe work on this too?

 

[ben_allison]

So, any schems for something like the Cloudlifter?

20db of noiseless gain powered by phantom is what I need! I was planning on buying the single channel version, but I'd rather diy it... way more fun!

 

[zebra50]

Well, the schematic early in this thread will get you about 12dB, and is a good place to start.

If you need more, then a transformer to convert to Hi-Z, followed by the FET buffer is one way to do it.

[Of course, if you have a noisy ribbon then you'll just end up boosting the noise. A lot of the old ribbon mics are ready for a service by now.  ]