Microphone boosters?

[Newmarket]

Hi all!
I'm a new member of this forum. I was searching for a schematic and I found it here

I play with electronics since I was 14 and since 1988 it have also been my work. I played seriously with audio, but also other linear electronics, digital and RF. In 2015 I discovered Arduino and now the ATmega328p is very often in my projects. My work has been lab servicing on broadcast cameras and video mixers, then HDTV studios design and now A/V infrastructure and main switching systems.

If I would design a mic booster, the specifications would be:
- a very simple device (less components, less noise);

Not a real basis for design. 'More components can mean more noise or less or the same.

- unbalanced input: it is directly connected to the microphone, then a balanced input is not required;

What ???

- low input impedance (2~3kohms);

Doesn't seem to take account of mics requiring a higher input Z for best oerformance, And that is sort of the topic here - apart from the noise issues.

- balanced output;
- phantom power;
- gain: about 8dB.

I think the general requirement seems to be for upto 20dB gain ?

 

[ccaudle]

Attached is a screen shot from one of his videos that compares EIN of several interfaces

Comparing EIN or noise figure at different gains is useful. Those EIN figures would indicate to me that they were all at maximum gain. Depending on the amp topology there may be a fixed noise floor that is not dominant at high gains, where the boosted noise of the source dominates, but becomes dominant at lower gains. That would show up as EIN which is higher than expected at low gains compared to high gains. Possibly not relevant in a case where the complaint is not enough gain available.

 

[JohnRoberts]

Comparing EIN or noise figure at different gains is useful. Those EIN figures would indicate to me that they were all at maximum gain. Depending on the amp topology there may be a fixed noise floor that is not dominant at high gains, where the boosted noise of the source dominates, but becomes dominant at lower gains. That would show up as EIN which is higher than expected at low gains compared to high gains. Possibly not relevant in a case where the complaint is not enough gain available.

That output noise floor is an important consideration for the later poster asking about a mic preamp with only 8 dB gain. Not enough gain to drive other constant noise sources to insignificance.

For fun and games when comparing preamps at different gains, also look at LF frequency response. An often dominant HPF is formed by gain resistor and DC blocking capacitor in series. When comparing mic preamp specs look at frequency response at max gain. LF response can be compromised by size of that blocking cap, HF response can be compromised by marginal open loop gain. A preamp with less LF and HF content in WFO (wide ___ open) listening tests can sound quieter when listened to compared to flat preamps.

JR

 

[ruffrecords]

Comparing EIN or noise figure at different gains is useful. Those EIN figures would indicate to me that they were all at maximum gain. Depending on the amp topology there may be a fixed noise floor that is not dominant at high gains, where the boosted noise of the source dominates, but becomes dominant at lower gains. That would show up as EIN which is higher than expected at low gains compared to high gains. Possibly not relevant in a case where the complaint is not enough gain available.

I completely agree. Every mic pre ever built is nominally specified for EIN at maximum gain because this always gives the best figure. A very small number of manufacturers will provide a graph of EIN versus gain (Neve did this back in the 70s) and as soon as the gain is reduced enough for output noise to dominate, the EIN figure worsens. The problem with specifying noise performance as EIN is that what the end user wants to know is the S/N ratio and the EIN does not give him that directly. For example, 1970s style Neve mic pres typically achieved and EIN of -126dBu and had a maximum gain of 80dB. When set to 80dB gain, this meant the noise at the output would be -46dBu so if the nominal output was 0dBu then the S/N was a mere 46dB which is useless for music recording. But if you have a fairly insensitive boom mic in a TV studio recording fairly quiet dialogue you just might need that amount of gain. You might choose to roll of the top and a little to reduce his but the probability is that acoustic background noise in the studio would mask it anyway.

If the EIN remains the same, the S/N increases to 66dB at 60dB gain and to 86dB at 40dB gain and these are much more typical gains for music recording. With sensitive condenser mics and loud sources gains of only 20dB or even 0dB might be required. AT 20dB gain, if the EIN stayed the same the S/N would be106dB but in 1970s Neve mic pres, output noise starts to dominate around -95dBu so the S/N ratio would bottom out at about -95dB.

The problem for the designer of a mic pre is that it might be used at gains of anything from 0dB to 60dB or more and maintaining optimum noise performance over such a wide range is quite difficult especially if you want equally good distortion and frequency response figures over the same range.

Cheers

Ian

 

[JohnRoberts]

The problem for the designer of a mic pre is that it might be used at gains of anything from 0dB to 60dB or more and maintaining optimum noise performance over such a wide range is quite difficult especially if you want equally good distortion and frequency response figures over the same range.

Cheers

Ian

This is a bit of a veer but I have killed many brain cells soaking this in beer... The optimal preamp design for a front end into a modern A/D running from 5V rail. Suggests we don't need +20dBu output. Likewise with sensitive mics we don't need 60+ dB of preamp gain.

The right way to do this is probably a dedicated design optimized for a specific mic and specific A/D.

I have never seriously considered this because I don't have preference for either end of that custom chain but I have taken notice that some of the modern uber-low noise op amps, also have 600 ohm drive capability. When we start trying to milk the last dB of NF (or S/N) from a given combo we may find a need to drop down the feedback network resistor values into the 600 ohm ballpark or less.

Not saying anybody should do this, just sayin...

JR

 

[ruffrecords]

While we are just sayin, I thought the trend these days was for external mic pres to give people tonal options. This also allows choosing one with a higher Z input for those pesky ribbons. They will mostly all output +20dBu, which I think is essential to take advantage of the dynamic range of 24 bit convertors and I have also noticed that many recent interfaces will accept up to 20dBu on the line input.

Just my twopennyworth

Cheers

IAn

 

[swpaskett]

Thanks, Abbey for posting your mic booster schematic. I may give it a spin; something I could have used about 50 years ago.

 

[abbey road d enfer]

Thanks, Abbey for posting your mic booster schematic. I may give it a spin; something I could have used about 50 years ago.

Suitable FET's didn't exist that long ago...
In some domains, our life is easier than our elders'.

 

[abbey road d enfer]

Thanks, Abbey for posting your mic booster schematic. I may give it a spin; something I could have used about 50 years ago.

Note that for correct operation you need the "A" version of the LSK389.

 

[Ricardus]

Note that for correct operation you need the "A" version of the LSK389.

I don't think I used "A"'s.

 

[abbey road d enfer]

I don't think I used "A"'s.

I had "A"s on the first few, then I had non-identified ones, and their Idss was too high, resulting in very low Vds in operation. Installing one instead of two resulted in correct operation, at the detriment of the noise performance.

 

[Tubetec]

Any spare boards left over from the previous run ?

 

[moamps]

Suitable FET's didn't exist that long ago...

From the information I know and my experience with vintage devices, jFET has been used since the late 60s. Examples are SONY 2SK18 (dual), 2SK23, 2SK56 (dual), 2SK97 ...
I don’t know why they weren’t used for microphone boosters, but probably at the time audio engineers had more to worry about tape machine noise. High output condenser microphones were mostly used, and gains in well-designed preamplifiers like the V76 were quite sufficient, IHMO. From my experience, ribbon microphones (and consequently the need and hype to use boosters) have had their comeback in the last ten to twenty years but only when neodymium magnets have proven themselves in hard drives. (rant) At the same time, each average external sound card happily got a XLRF connector and a button called + 48V, all powered together from a USB connection. The formula for success. (/rant)

 

[JohnRoberts]

From the information I know and my experience with vintage devices, jFET has been used since the late 60s. Examples are SONY 2SK18 (dual), 2SK23, 2SK56 (dual), 2SK97 ...
I don’t know why they weren’t used for microphone boosters, but probably at the time audio engineers had more to worry about tape machine noise. High output condenser microphones were mostly used, and gains in well-designed preamplifiers like the V76 were quite sufficient, IHMO. From my experience, ribbon microphones (and consequently the need and hype to use boosters) have had their comeback in the last ten to twenty years but only when neodymium magnets have proven themselves in hard drives. (rant) At the same time, each average external sound card happily got a XLRF connector and a button called + 48V, all powered together from a USB connection. The formula for success. (/rant)

I will echo what Abbey said, we didn't have low noise audio JFETs back in the day... especially not in the 60s. I vaguely recall some exotic mosfets in the early 70s (for high impedance applications) but they would fry if you looked at them crosseyed,,,

By the mid/late 70s I recall some low ohms JFET switches (like 231s, 5957, etc... but not very low noise. ) In one of my middle phono preamp designs (1980) I used a low noise (cough) JFET with a whopping 4.5nV rt/HZ (about the same noise voltage as a 5532 op amp, but less noise current). By the mid 80s we started to see some quite nice 1nV rt/HZ Japanese (2sk117, etc) JFETS...

IIRC back in the 70s there were some crazy expensive low noise JFETs mainly for military use (Teledyne Crystallonics?), I don't think I ever sourced one of them even to play with (way too expensive).

JR

 

[moamps]

Here is the 1976 DOA schematic



and here is one from 1979, DOA with +/- 47V power supply



both have low noise JFETS for LTP.

 

[abbey road d enfer]

both have low noise JFETS for LTP.

Most often, DOA's are considered low noise when their EIN density is about 3-5 nV/sqrtHz and their OSI about 5-10kohms. It seems very much the case here (although the OSI might well be much higher).
For a mic booster, the targetted noise voltage density is about 1nV/sqrtHz.
I believe the first commonly available FET's capable of such performance were the 2SK170 and its complementary (2SJ74?).
Only relatively recently Linear Systems introduced their take on 2SK170 and 2SK89.

 

[abbey road d enfer]

Any spare boards left over from the previous run ?

PM addressed.

 

[JohnRoberts]

Most often, DOA's are considered low noise when their EIN density is about 3-5 nV/sqrtHz and their OSI about 5-10kohms. It seems very much the case here (although the OSI might well be much higher).
For a mic booster, the targetted noise voltage density is about 1nV/sqrtHz.
I believe the first commonly available FET's capable of such performance were the 2SK170 and its complementary (2SJ74?).
Only relatively recently Linear Systems introduced their take on 2SK170 and 2SK89.

The only DOA I ever designed used a LM394 for the input LTP to realize 1nV rt/HZ (technically a LM394 is more like an IC than discrete component).

I recall reading about the very low noise bipolar transistors (2sb737, and 2sd786) in an IEEE journal back in the 70s. They were less than 1nV rt/Hz

The first very low noise JFETs that I could get my hands on, appeared in the 80s, but I am repeating myself.

"Low noise" is relative and there were many discrete devices called low noise that weren't really very quiet. Using the more generous definition of low noise, your schematics document earlier use. My sense about low noise component availability is in agreement with Abbey's.

JR

 

[swpaskett]

Memory is such a sketchy thing... I remember that back when I was a teenager, RCA introduced something called COSMOS, soon changed to CMOS. In the early 70s it was thought of as a slow, lower power replacement for TTL with some exotic properties like insensitivity to power supply voltage and some of the stuff, while technically "logic", could be used to switch audio, which we routinely did at the PBS station where I volunteered. Our homebrew on-air audio switcher was full of the stuff, as well as discreet JFETs (I know -- I drew up the schematics). Noise? Nobody thought about noise, and we cared even less about headroom. Everything hissed and hummed back then and TV was the lowest of Lo-Fi anyway. Our studio output was bandpassed to 300-5kHz anyway. (several paragraphs deleted, you probably haven't read this far, anyway; even I acknowledge limits)

 

[JohnRoberts]

CMOS logic was huge back in the 70s and arguably still is (CD4000 series). I used CMOS transfer gates (CD4016/4066) for all kinds of audio switching back last century.

RCA cosmos, how about RCA BiMOS..? mosfet input opamps... not very Hifi. As I recall didn't gain much traction.

CMOS opamps didn't really come into their own until relatively recently. I suspect thanks to process technology improvements developed for big digital processors.

JR

PS: Search engine finds an RCA records band COSMOS, and an old RCA acronym COS/MAC.... There was a link to RCA COSMOS IC manual but the link did not work.