Low open loop gain from a mic preamp-why?

>"Phil Allison"  wrote in message 
>news:837f7e3f-ffd6-4583-82ba-26b8caf08cf0n@googlegroups.com...

Kevin Aylward wrote:
==================
>
> Fundamentally, its a poor design. Its inherently NOT low noise, and loads
> the mic way to much, effecting the frequency response.

>** Nonsense.

It will, I didn't say how much..... :-)

>  A 1200 ohm load will not affect dynamic mics and only drop the level from 
> an electret.

Its on the low side. its a fair drop of 33% on a 600 ohm mic, its a 10% drop 
on a 200 ohm mic.

Its not a good idea for some powered types of electret mic amps. If its a 
simple emitter follower buffer, running at low current, with say its 
internal emitter resistor at say 10k, or even higher. This is ac coupled to 
the output. This means the negative going voltage will clip at 1k/10k x 
emitter resistor voltage. The emitter resistor voltage might only be ~ 0.7V 
if the system is only using a 1v5 battery. 70 mV signal handling is pushing 
it a bit, even for a voice signal.


> The 1k2 input resistor dominates the noise.

>** Maybe, but only just.  BC549s are non ideal for low source impedances.

The data sheet for the BC549 has a noise figure of 1.2 dB with a 2k source. 
This calculates to  rbb' = ~ 52 ohms

This is around what I expect. Typically rbb' is going to be 10 to 100 ohms.

Thus the 1k2 totally  dominates the noise.

Actually, I knew immediately on inspection about this problem from 
experience gained the early 80s when I actually used to design pro mixers. 
:-)

The Studiomaster mixers I was designing had around -127 db ein. A competitor 
had -120 dB, specifically because it used a single op amp differential 
configuration with 1k source resistors.


>> For a standard mic of 200/150
>> ohms it destroys the inherent noise performance of the source resistance 
>> due
>> to thermal noise of the resister, and because of the input current noise
>> dropped across the input source resistor.

>**  Still low enough for a great many real life uses and quite OK with an 
>internally  FET buffered electret capsule.current noise.

Sure, a buffered mic will kill the current noise.

Its about the definition of "low noise" mic amp.

In standard pro mixers, "low noise" means -127 dBv to -130 dBv ein.

>> The issue is that the design is of an inverting feedback amplifier. Low
>> noise amplifiers need to be non inverting if a feedback topology is used.
>
>** Correct.

>> This allows having a standard input bias resistor setting the mic load of
>> say 6k8.

>**  Not needed at all  - 1.2K is fine and standard practice.

Well.... see above....

"Standard" practice on mic input stages would range from 1k to 10k or so.

My view is that one should really keep it above 5k. There is no reason not 
to do so, and it minimises the issues noted above.

>>  Note: the input capacitor should be increased to 100uf so that the
>> 1/f current noise of the input transistor is also shorted through the mic
> >resistance.

>**  Has no effect,  all the white noise energy is concentrated above 1kHz.

It most certainly does have a significant audible effect...... as noted, 
some of us actually have pro mixers out in the field. This is a standard 
well known issue,  and these large capacitors are indeed used on 10s 
millions of pro mixers. The point being that the cap is chosen for noise 
reasons, not the 3 dB LF response.

1/f noise of transistors is a fundamental problem. It's clearly audible. 
Indeed a "low noise" transistor, *IS* a transistor with low 1/f noise, 
because, by and large, most "modern" transistor have around the *same* 
flatband noise noting that that typically their rbbs are *all* around 10-100 
ohms mark. Although I do note that historically a BC109 was around 400 ohms.

The early Studiomasters used 2N4403, chosen because as they were switching 
transistors. They had an rbb' of only 12 ohms to keep the Ccb.rbb' time 
constant low.. Unfortunately, their 1/f noise was quite variable. Actually, 
the main problem with these devices was RTS noise. Thus there was a noise 
room test setup to measure every transistor before being used in the mic 
amp. There were bags of rejected devices.

The large input capacitance was indeed a requirement to reduce the audible 
LF noise.

> https://www.kevinaylward.co.uk/ee/micampdesign/MicAmpDesign.html

** Hmmm -  my Project 66 is simpler and measured performance better.

Well... the basic topology of the compound transistor input stage is used in 
10s of millions of real pro mixes. Its a fair reference to the design of 
actual products such as Soundcraft, and many others.

The reason the topology is used is to reduce the distortion from that of a 
simple diff pair. Its local feedback.


-- Kevin Aylward

http://www.anasoft.co.uk/ SuperSpice
http://www.kevinaylward.co.uk/ee/index.html 

On a sunny day (Wed, 22 Dec 2021 10:35:10 +0100) it happened Arie de Muijnck
<noreply@ademu.com> wrote in <61c2f14d$0$9613$e4fe514c@usenet.xs4all.nl>:

>On 2021-12-22 09:16, Jan Panteltje wrote:
>> On a sunny day (Tue, 21 Dec 2021 15:08:59 -0600) it happened amdx
>> <amdx@knology.net> wrote in <sptfpd$6ek$1@dont-email.me>:
>> 
>>> I would like to do the best I can with an electret mic driving a low
>>> noise high gain preamp.
>> 
>> I think the electret mikes have just a JFET source follower.
>> And quite a large signal output.
>> good quality opamp is likely all you need.
>
>
>Generally not a follower but an amplifier, hence the high output:
>     <https://en.wikipedia.org/wiki/Electret_microphone>
>
>However, some do rewire it to source follower, like:
>      <https://www.firstpr.com.au/rwi/mics/2009-09-b/>

Very nice article, much detail!
Thank you

 Kevin Aylward wrote:
=================
> 
> > Fundamentally, its a poor design. Its inherently NOT low noise, and loads 
> > the mic way to much, effecting the frequency response. 
> 
> >** Nonsense.
> It will, I didn't say how much..... :-)

**  Very cute   .....   
  Now do a small dance for me. 


> > A 1200 ohm load will not affect dynamic mics and only drop the level from 
> > an electret.

> Its on the low side. its a fair drop of 33% on a 600 ohm mic,

** Such mic are rare and not used in noise critical situations. 


> > The 1k2 input resistor dominates the noise. 
> 
> >** Maybe, but only just. BC549s are non ideal for low source impedances.

> The data sheet for the BC549 has a noise figure of 1.2 dB with a 2k source. 
> This calculates to rbb' = ~ 52 ohms 

**  Please cite.  My info is the ideal is 50k. 

> 
> The Studiomaster mixers I was designing had around -127 db ein. A competitor 
> had -120 dB, specifically because it used a single op amp differential 
> configuration with 1k source resistors.

**  That would be the old H-H  12 and 16 ch desks -  right ?? 
    Studiomasters of that era were super quiet by comparison. 

    Studio v live sound, I suspect. 

 
> >> This allows having a standard input bias resistor setting the mic load of 
> >> say 6k8. 
> 
> >** Not needed at all - 1.2K is fine and standard practice.
> Well.... see above.... 

  ** Been there,  done that.

> "Standard" practice on mic input stages would range from 1k to 10k or so.

**   The mean is about 2k, balanced.
      Euro mike makers ( AKG, Beyer, Sennheiser)  say 1kohms or greater. 
      Yanks go for even lower - Shure say 300 ohms is the go. 


> My view is that one should really keep it above 5k. There is no reason not 
> to do so, 

** Yes there is,  mics with output transformers need to see a resistive load to damp HF ringing. 
     
  Shure  SM57 /58s,  all ribbon mics, older condenser models and a host of others.


> >> Note: the input capacitor should be increased to 100uf so that the 
> >> 1/f current noise of the input transistor is also shorted through the mic 
> > >resistance. 
> 
> >** Has no effect, all the white noise energy is concentrated above 1kHz.

> It most certainly does have a significant audible effect...

**  Bollocks.  Try really answering my obvious point. 
    The audible noise from an input loaded mic pre is all above 1kHz. 
 
> The early Studiomasters used 2N4403,

** Used to see quite a few of them.
   Very familiar with the mic pre fitted

> chosen because as they were switching 
> transistors. They had an rbb' of only 12 ohms to keep the Ccb.rbb' time 
> constant low.. Unfortunately, their 1/f noise was quite variable. Actually, 
> the main problem with these devices was RTS noise. Thus there was a noise 
> room test setup to measure every transistor before being used in the mic 
> amp. There were bags of rejected devices. 

**   Really?   Never seen a bad new one in my life. 

 
> The large input capacitance was indeed a requirement to reduce the audible 
> LF noise.

**  Low frequent noise is not audible at low levels  - at all.  


> ** Hmmm - my Project 66 is simpler and measured performance better.

> Well... the basic topology of the compound transistor input stage is used in 
> 10s of millions of real pro mixes. Its a fair reference to the design of 
> actual products such as Soundcraft, and many others. 

**  FFS take look at P66.  

   It should look VERY familiar   ....   ;-) 
 

.....   Phil 

"Phil Allison"  wrote in message 
news:72caa4e3-f202-4338-a465-5f0c2f11eefan@googlegroups.com...

Kevin Aylward wrote:



> > The 1k2 input resistor dominates the noise.
>
> >** Maybe, but only just. BC549s are non ideal for low source impedances.

>> The data sheet for the BC549 has a noise figure of 1.2 dB with a 2k 
>> source.
>> This calculates to rbb' = ~ 52 ohms

>**  Please cite.  My info is the ideal is 50k.

What do you mean by 50k ?

I assume that you mean the optimum value of source resistance for a given 
operating point..

Its calculated from

re = RS/sqrt(hfe)

re=1/(40.IC)

where this value of IC makes the ein due to base current noise equal to the 
ein due to collector current noise.

its from:

Vibn = RS.sqrt(2q.ib)
Vicn = re.sqrt(2q.Ic)

re goes down linearly with ic, but the noise goes as sqrt()

>
> The Studiomaster mixers I was designing had around -127 db ein. A 
> competitor
> had -120 dB, specifically because it used a single op amp differential
> configuration with 1k source resistors.

>**  That would be the old H-H  12 and 16 ch desks -  right ??
>    Studiomasters of that era were super quiet by comparison.

>    Studio v live sound, I suspect.

Yes. The mixers are general purpose, but inherently designed for recording.

> >> This allows having a standard input bias resistor setting the mic load 
> >> of
> >> say 6k8.
>
> >** Not needed at all - 1.2K is fine and standard practice.
> Well.... see above....

  ** Been there,  done that.

> "Standard" practice on mic input stages would range from 1k to 10k or so.

>**   The mean is about 2k, balanced.
>      Euro mike makers ( AKG, Beyer, Sennheiser)  say 1kohms or greater.
>      Yanks go for even lower - Shure say 300 ohms is the go.

The problem with manufactures, is that they don't actually know what they 
are doing. Secind, there is often errrors in the docs.

Someone  at some point has a design, and it gets blindly copied.


> My view is that one should really keep it above 5k. There is no reason not
> to do so,

** Yes there is,  mics with output transformers need to see a resistive load 
to damp HF ringing.

There is no sensible reason to use a transformer....its one of those 
historical audiophool things.....

> Shure  SM57 /58s,  all ribbon mics, older condenser models and a host of 
> others.

Ribbon mics with their low output voltage, highlight the issues of requiring 
very low noise.

> >> Note: the input capacitor should be increased to 100uf so that the
> >> 1/f current noise of the input transistor is also shorted through the 
> >> mic
> > >resistance.
>
> >** Has no effect, all the white noise energy is concentrated above 1kHz.

> It most certainly does have a significant audible effect...

>**  Bollocks.  Try really answering my obvious point.
>    The audible noise from an input loaded mic pre is all above 1kHz.

As I explained, it is a real problem, known to all in the industry.

I guess you haven't sat in a test room actually listening to LF noise.

Hint:   Put on 15 dB of bass boost at 40 Hz.


>> The early Studiomasters used 2N4403,

>** Used to see quite a few of them.
>   Very familiar with the mic pre fitted

>> chosen because as they were switching
>> transistors. They had an rbb' of only 12 ohms to keep the Ccb.rbb' time
>> constant low.. Unfortunately, their 1/f noise was quite variable. 
>> Actually,
>> the main problem with these devices was RTS noise. Thus there was a noise
>> room test setup to measure every transistor before being used in the mic
>> amp. There were bags of rejected devices.

>**   Really?   Never seen a bad new one in my life.

Well... it was certainly a fact in 1980.

RTS noise is typically a processing issue. Its likely that newer, modern, 
transistor versions are much better.

RTS noise is such a problem in modern IC process principally designed for 
cmos, that I (my current employer) had to run test chips on different 
processes to locate a BiCMOS process with no RTS.

We got burnt on the IBM7WL. Horrendus RTS. To date we have only located one 
commercial BiCMOS process with both isolated npn and pnp with no RTS noise. 
Its the XFAB XT018

We use gilbert cell multipliers to generate chebychev functions to generate 
an inverse temperature v voltge profile to correct for the xtal oscillator 
temperature response. 1/f and RTS noise destroy the close in phase noise of 
the oscillator.

CMOS have ~100 time larger 1/f noise for the same size device, so 
constructing CMOS multipliers is really problematic.



>> The large input capacitance was indeed a requirement to reduce the 
>> audible
>> LF noise.

>**  Low frequent noise is not audible at low levels  - at all.

It most certainly is audible. Its a fact. Note, it depends on the 
sensitivity of the mic and associated required gain.

I guess you haven't sat in a test room actually listening to LF noise.

Hint:   Put on 15 dB of bass boost at 40 Hz.

Such a level of boost is quite typical in mixing band sound, especially for 
recording.


> ** Hmmm - my Project 66 is simpler and measured performance better.

> Well... the basic topology of the compound transistor input stage is used 
> in
> 10s of millions of real pro mixes. Its a fair reference to the design of
> actual products such as Soundcraft, and many others.

>**  FFS take look at P66.

>  It should look VERY familiar   ....   ;-)

I checked out https://sound-au.com/project66.htm

To the best of my knowledge, this compound pair was first used in a mic amp 
by Studiomaster. A guy named Alan Pound introduced it, prior to my arrival.

I note the 3k input resistors :-)

Its always a good idea to put reverse diodes across the base emitter of the 
input transistor. Overvoltage on power up can destroy the noise 
characteristics.

-- Kevin Aylward

http://www.anasoft.co.uk/ SuperSpice
http://www.kevinaylward.co.uk/ee/index.html 

On Wed, 22 Dec 2021 01:27:08 -0800 (PST), Phil Allison
<pallison49@gmail.com> wrote:

> Jan Panteltje wrote:
>================
>>>
>> I think the electret mikes have just a JFET source follower. 
>> And quite a large signal output. 
>
>**  Most are a JFET with drain as output needing a +DC source and a series resistor of about 4000 ohms 

I'd guess that the best load on the jfet would be constant-voltage,
namely a cascode NPN or the inverting input of an opamp. That
minimizes d-g feedback and likely distortion.

D-g leakage current increases radically at higher drain voltages,
which would forward-bias the g-s diode and do all sorts of bad stuff.



-- 

I yam what I yam - Popeye

On Wed, 22 Dec 2021 11:24:06 GMT, Jan Panteltje
<pNaOnStPeAlMtje@yahoo.com> wrote:

>On a sunny day (Wed, 22 Dec 2021 10:35:10 +0100) it happened Arie de Muijnck
><noreply@ademu.com> wrote in <61c2f14d$0$9613$e4fe514c@usenet.xs4all.nl>:
>
>>On 2021-12-22 09:16, Jan Panteltje wrote:
>>> On a sunny day (Tue, 21 Dec 2021 15:08:59 -0600) it happened amdx
>>> <amdx@knology.net> wrote in <sptfpd$6ek$1@dont-email.me>:
>>> 
>>>> I would like to do the best I can with an electret mic driving a low
>>>> noise high gain preamp.
>>> 
>>> I think the electret mikes have just a JFET source follower.
>>> And quite a large signal output.
>>> good quality opamp is likely all you need.
>>
>>
>>Generally not a follower but an amplifier, hence the high output:
>>     <https://en.wikipedia.org/wiki/Electret_microphone>
>>
>>However, some do rewire it to source follower, like:
>>      <https://www.firstpr.com.au/rwi/mics/2009-09-b/>
>
>Very nice article, much detail!
>Thank you

Yikes. What is the source voltage?

http://www.linkwitzlab.com/images/photos/mic-amp.jpg

-- 

If a man will begin with certainties, he shall end with doubts, 
but if he will be content to begin with doubts he shall end in certainties.
Francis Bacon

On Wednesday, 22 December 2021 at 17:23:15 UTC, jla...@highlandsniptechnology.com wrote:
 
> >** Most are a JFET with drain as output needing a +DC source
>  and a series resistor of about 4000 ohms
> I'd guess that the best load on the jfet would be constant-voltage, 
> namely a cascode NPN or the inverting input of an opamp. That 
> minimizes d-g feedback and likely distortion. 

Yes, connecting the mic output to the inverting input of a low-noise
op-amp via a capacitor works very well.  No series resistor is needed
or even desirable.  The electret mics being used
by the OP are intended to operate at 0.5mA drain current.  The pullup
resistor can be chosen according to the supply voltage to give that
current when the output dc level is about 1.9V.  These are the conditions
under which the mics are characterised by the manufacturer.
Running at much lower current would result in the drain voltage falling
which reduces the dynamic range.  There are advantages in using
a fairly high bias voltage of up to a maximum of 10V with a
correspondingly higher pullup resistor to keep the the output at
about 1.9V and the current at 0.5mA.  A 9V battery makes an
excellent low-noise bias source.  The higher pullup resistor which,
for a 9V bias  supply would be 14.2k, reduces the noise gain of the
op-amp.
As the acoustic input modulates the drain current of the FET in
the mic, all the audio current ends up in the feedback resistor of
the op-amp and the drain voltage stays almost constant.
With a feedback resistor of 17.4k and using the PUI Audio
AOM-5024L-HD-R microphone the output of the preamp will be
about 0.5Vrms/Pa.

John

On Wed, 22 Dec 2021 10:37:41 -0800 (PST), John Walliker
<jrwalliker@gmail.com> wrote:

>On Wednesday, 22 December 2021 at 17:23:15 UTC, jla...@highlandsniptechnology.com wrote:
> 
>> >** Most are a JFET with drain as output needing a +DC source
>>  and a series resistor of about 4000 ohms
>> I'd guess that the best load on the jfet would be constant-voltage, 
>> namely a cascode NPN or the inverting input of an opamp. That 
>> minimizes d-g feedback and likely distortion. 
>
>Yes, connecting the mic output to the inverting input of a low-noise
>op-amp via a capacitor works very well. 

I meant a direct connection to the inverting input. Bias up the ni
input of the opamp to a suitable electret operating voltage. The only
resistor is the opamp fb.

That hard forces the jfet drain voltage.

-- 

If a man will begin with certainties, he shall end with doubts, 
but if he will be content to begin with doubts he shall end in certainties.
Francis Bacon

 Kevin Aylward wrote:
=================
> 
> > > The 1k2 input resistor dominates the noise. 
> > 
> > >** Maybe, but only just. BC549s are non ideal for low source impedances. 
> 
> >> The data sheet for the BC549 has a noise figure of 1.2 dB with a 2k 
> >> source. 
> >> This calculates to rbb' = ~ 52 ohms 
> 
> >** Please cite. My info is the ideal is 50k.
> What do you mean by 50k ? 

** I see no cite of the noise data. 

 > > >> This allows having a standard input bias resistor setting the mic load 
> > >> of  say 6k8. 
> > 
> > >** Not needed at all - 1.2K is fine and standard practice. 
>
> > Well.... see above.... 
> 
> ** Been there, done that. 
> 
> > "Standard" practice on mic input stages would range from 1k to 10k or so. 
> 
> >** The mean is about 2k, balanced. 
>
> > Euro mike makers ( AKG, Beyer, Sennheiser) say 1kohms or greater. 
> > Yanks go for even lower - Shure say 300 ohms is the go.
>
> The problem with manufactures, is that they don't actually know what they 
> are doing.

** But you are ( or were) one ? 

> Someone at some point has a design, and it gets blindly copied.

** What a desperate red herring. 

> 
> ** Yes there is, mics with output transformers need to see a resistive load 
> to damp HF ringing.
> There is no sensible reason to use a transformer...

**  Wow -  now Kev goes for " bury head in the sand"  method of debate.

> > Shure SM57 /58s, all ribbon mics, older condenser models and a host of 
> > others.
>
> Ribbon mics with their low output voltage, highlight the issues of requiring 
> very low noise.

**  Another redirect and red herring. 

> > >> Note: the input capacitor should be increased to 100uf so that the 
> > >> 1/f current noise of the input transistor is also shorted through the 
> > >> mic resistance. 
> > 
> > >** Has no effect, all the white noise energy is concentrated above 1kHz. 
> 
> > It most certainly does have a significant audible effect... 
> 
> >** Bollocks. Try really answering my obvious point. 
> > The audible noise from an input loaded mic pre is all above 1kHz.

> As I explained, it is a real problem, known to all in the industry. 

**   What is ?? 

> I guess you haven't sat in a test room actually listening to LF noise. 

**   Another desperate red herring,  plus absurd. 

> >> The early Studiomasters used 2N4403, 
> 
> >** Really? Never seen a bad new one in my life.

> Well... it was certainly a fact in 1980. 

** Nope, it was not.


> >** Low frequent noise is not audible at low levels - at all.

> It most certainly is audible. Its a fact.

** See Flectcher Munsen curves. 

White noise is predominately high frequency energy,  masking any low frequency component. 

 I guess you haven't sat in a test room actually listening to LF noise. 

** I guess you cannot stop obfuscating. 
> 
> >** FFS take look at P66. 
> 
> > It should look VERY familiar .... ;-)
>
> I checked out https://sound-au.com/project66.htm 
> 
> To the best of my knowledge, this compound pair was first used in a mic amp 
> by Studiomaster. A guy named Alan Pound introduced it, prior to my arrival. 
> 
**  The same or closely similar topolgy is used in many desks.  
     But you don't get to see them. 
  
> I note the 3k input resistors :-) 

 ** Yawnnnnnnnn....

> Its always a good idea to put reverse diodes across the base emitter of the 
> input transistor. Overvoltage on power up can destroy the noise 
> characteristics.

** With 48V phantom that is an issue.


......  Phil 

On 12/22/2021 12:37 PM, John Walliker wrote:
> On Wednesday, 22 December 2021 at 17:23:15 UTC, jla...@highlandsniptechnology.com wrote:
>   
>>> ** Most are a JFET with drain as output needing a +DC source
>>   and a series resistor of about 4000 ohms
>> I'd guess that the best load on the jfet would be constant-voltage,
>> namely a cascode NPN or the inverting input of an opamp. That
>> minimizes d-g feedback and likely distortion.
> Yes, connecting the mic output to the inverting input of a low-noise
> op-amp via a capacitor works very well.  No series resistor is needed
> or even desirable.  The electret mics being used
> by the OP are intended to operate at 0.5mA drain current.  The pullup
> resistor can be chosen according to the supply voltage to give that
> current when the output dc level is about 1.9V.  These are the conditions
> under which the mics are characterised by the manufacturer.
> Running at much lower current would result in the drain voltage falling
> which reduces the dynamic range.  There are advantages in using
> a fairly high bias voltage of up to a maximum of 10V with a
> correspondingly higher pullup resistor to keep the the output at
> about 1.9V and the current at 0.5mA.  A 9V battery makes an
> excellent low-noise bias source.  The higher pullup resistor which,
> for a 9V bias  supply would be 14.2k, reduces the noise gain of the
> op-amp.
> As the acoustic input modulates the drain current of the FET in
> the mic, all the audio current ends up in the feedback resistor of
> the op-amp and the drain voltage stays almost constant.
> With a feedback resistor of 17.4k and using the PUI Audio
> AOM-5024L-HD-R microphone the output of the preamp will be
> about 0.5Vrms/Pa.
>
> John
>
  Isn't there a trade off? Does the higher 14.2k pullup resistor have 
it's own noise vs the reduced noise off the opamp.

  I experimented the as much as 20V on the mic FET from a 25v supply. It 
only gave me about 30% more signal over 1.5V

on the mic FET. What other ± effects would a high pullup voltage and 
resistor have?

                                               Mikek


-- 
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus