[ProAudio] Microphones question

Bill Whitlock engineer_bill at verizon.net
Mon Jun 14 02:25:22 PDT 2021

Noise contribution of feedback (gain setting( resistances dominates most "cookbook" designs at low gains.  The Jensen Twin-Servo(R) design (as available from the John Hardy Company) avoids this by cascading two gain stages, gain adjusted in tandem by very low-value networks - made practical by using the 990 amplifiers which can easily drive 150 Ω loads ... and has lots of headroom due to bi-polar 24 V rails.  It has exemplary noise and dynamic range performance that beats any other "fule" design I've ever heard of ... and it's won quite a number of listening tests as well.  In my humble opinion, most makers of ribbon mics have really messed up by going for such high ratio transformers, with unreasonably high secondary impedances (due in part to cheap transformer designs, i.e. high leakage inductance) in their quest for higher and higher output - not a market I'd care to cater to ...
Just sayin'
Bill WhitlockAES Life Fellow(and owner of Jensen Transformers 1988-2014)Ventura, CA

-----Original Message-----
From: Crispin HT via ProAudio <proaudio at bach.pgm.com>
To: 'Dan Lavry' <dan at lavryengineering.com>; proaudio at bach.pgm.com
Sent: Mon, Jun 14, 2021 1:39 am
Subject: Re: [ProAudio] Microphones question

#yiv3011626908 #yiv3011626908 -- _filtered {} _filtered {} _filtered {} _filtered {}#yiv3011626908 #yiv3011626908 p.yiv3011626908MsoNormal, #yiv3011626908 li.yiv3011626908MsoNormal, #yiv3011626908 div.yiv3011626908MsoNormal {margin:0cm;font-size:11.0pt;font-family:sans-serif;}#yiv3011626908 a:link, #yiv3011626908 span.yiv3011626908MsoHyperlink {color:blue;text-decoration:underline;}#yiv3011626908 pre {margin:0cm;font-size:10.0pt;}#yiv3011626908 span.yiv3011626908HTMLPreformattedChar {font-family:Consolas;}#yiv3011626908 span.yiv3011626908EmailStyle21 {font-family:sans-serif;color:#1F497D;font-weight:normal;font-style:normal;text-decoration:none none;}#yiv3011626908 .yiv3011626908MsoChpDefault {font-size:10.0pt;} _filtered {}#yiv3011626908 div.yiv3011626908WordSection1 {}#yiv3011626908 In my experience, it all comes back to the EIN figures.  Any fule can make a mic amp with EIN> 128dB at 60dB gain, because the mic pre output noise is defined by the effective feedback resistance, not the noise voltage of the amp.  The more interesting case is where in the field, you are actually amplifying those pesky mics with low gains of 20-40dB (close micing etc).  Here the feedback resistor values of typical mic amp circuits start to dominate the output noise, and you can hear it.We spent a lot of time making our pre’s give a lower EIN than a mic across all gains, and you can hear this effect in recordings made with our gear.  Our EIN at 60dB gain is actually worse than most, but you can’t hear it.  Then there are then the issues of mic loading die to the input impedance of the pre ( valve vs semi vs passive) which subtly affect the frequency response of the mic, and finally high Z mics like ribbons, where the bias current noise of your amp, starts to matter.  Lots of fun, but all quite easy maths, the engineering to achieve is inevitably harder than reading the spec sheets.Kind RegardsCrispin Herrod-Taylor
Managing Director, Crookwood
www.crookwood.com  Tel: +44 (0)1672 811 649
Mobile:+44(0)7910 637 634 Sign up for our great newsletter here! and keep up to date with the audio world  From: ProAudio <proaudio-bounces at bach.pgm.com> On Behalf Of Dan Lavry via ProAudio
Sent: 14 June 2021 03:37
To: proaudio at bach.pgm.com
Subject: Re: [ProAudio] Microphones question  I am surprise that you say 2N4403 has 1nV/sqrtHz. Are you  sure? The data sheet says nothing like that. I looked up the 2N4124, it has some plots of noise figure vs freq for 200Ohm, and also noise figure vs resistance. The noise figure is in dB, and the 2N4124 is better, but the noise figure with a given source resistance combines the impact of noise current and noise voltage. Not easy to separate it.I don't view the LM394 as today's transistor. Match pair is better then singles, but I cant find it at major distributors. It seems like history to me. I guess there is no point to claim that it can be done much better. I would have to actually do it. I know it can be improved a lot, but I will stick to my area - conversion. I think that experience is very valuable.  I have some. But there is new stuff all the time...RegardsDan Lavry  On 6/13/2021 6:41 PM, Bill Whitlock via ProAudio wrote:
At the risk of belaboring the point, extremely low-noise mic preamps were built using quality PNP bipolar transistors such as the 2N4403 or 2N4124, both having e-noise of 1 nV per root Hz across the audio band and corresponding i-noise making their optimum source impedance around 200 Ω to 800Ω.  Today's transistors, even including the LM394 (50 NPN transistors connected as one) won't do any better.  Granted, if your design world is ICs or FETs, they've gotten way better.  But mic preamps, built with today's "state-of-the-art" transistors and "gee-whiz" circuit topologies are insignificantly quieter than a good one built 50 years ago.  I submit that a single low-noise PNP transistor is the "state-of-the-art" input device (PNP devices are inherently quieter).  While differential pairs and op-amps make circuit design simpler (you don't have to deal with those pesky h-parameters for one), simply using a differential pair increases noise over that of a single device.  I'm often dismayed by today's convenience-driven "cut and paste" circuit designers.   But I do agree that, if one is to accurately predict noise performance of any given mic and preamp, both the reactance and resistance of the mic are necessary.  Bill Whitlock

-----Original Message-----
From: Dan Lavry via ProAudio <proaudio at bach.pgm.com>
To: proaudio at bach.pgm.com
Sent: Sun, Jun 13, 2021 5:33 pm
Subject: Re: [ProAudio] Microphones questionI agree that "basic principles" stay the same. But I think that a lot has changed. In 1971, I met sales reps from semiconductor, resistor, name it companies, and they left behind tons of data books and specs. Today, I can get a data sheet (internet) in seconds or a minute... The materials are better, the process is tighter, the test equipment is better and more affordable. I keep reading everyday, articles and about product introductions. One can wait for technology to advance, and it has since 1971. We can get better results as technology moves forward.But yes, I would agree that the basic principles of networks, components, theorems and circuits are the same. I know that, I am a circuit designer... RegardsDan LavryThis is all well-known and much written about - and well explored territory - and was when I got into pro audio in 1971 at Quad-Eight, and at Jensen dug even deeper to understand how input transformers can greatly benefit noise performance of vacuum tubes, for example. Nothing has really changed ...   Bill WhitlockAES Life Fellow

-----Original Message-----
From: Dan Lavry via ProAudio <proaudio at bach.pgm.com>
To: proaudio at bach.pgm.com
Sent: Sun, Jun 13, 2021 2:25 pm
Subject: Re: [ProAudio] Microphones questionI guess I should be more precise:

1. Measure noise voltage with a short (with gain).

2. Measure with 100K (with gain), remove (compute) the resistor noise 
and noise voltage, leaves i*R where i is noise current.

Clearly the sums and differences are not add linearly, you do sqrt of 
the "sum or difference" square...

That would yield noise voltage and noise current of the micpre, and 
leaves the mics out of it.


Dan Lavry
On 6/13/2021 1:00 PM, Dan Lavry via ProAudio wrote:
> Hi again,
> Let me get away from sales and what people say, and back to the 
> technical stuff:
> Some here suggested to look at the mic output impedance to tell us 
> about noise. That is only true for the mic pre noise current 
> component. And in most real world cases, that is the smallest 
> component. I think there is some confusion regarding the noise 
> generated by the mic pre, it does not directly relates to output 
> impedance. Say some noise is due to a PNP transistor, how does that 
> relates to some output inductor? The combined causes of noise in a 
> analog circuit is complex, circuit dependent and component dependent. 
> The question is how to model it.
> I would start by using 100KOhm (not 150 Ohm), with gain it is enough 
> noise to be measured well. That noise (divide by gain) is due to 
> current noise (the voltage component is negligable). Now I will go 
> back to a short and measure the noise voltage of the mic pre (of 
> course gain is needed). Now you have in and en (noise current and 
> noise voltage).
> That is what I want to know. I agree, it is difficult to translate to 
> the consumer. I am not going to insist on much. Just turned 76, got my 
> own problems. fs=24KHz would work fine. Youtube can have a narrow 
> bandwidth for the old, you add lossless compression and a 4KHz 
> wireless connection to the hearing aid...
> Dan Lavry
> On 6/13/2021 5:06 AM, Scott Dorsey via ProAudio wrote:
>>> Scott Dorsey writes=20
>>>> In a perfect world the lowest noise would be when the input 
>>>> impedance =
>>> of the
>>>> preamp matches the output impedance of the microphone, ....
>>> Um, no. Your perfect world would need to redefine a lot of other =
>>> parameters for that to be so. A typical emitter-follower microphone =
>>> output may have an output impedance of 100 ohms but if you load it 
>>> with =
>>> that value the distortion will be significant, the output level 
>>> severely =
>>> limited. Noise performance depends as Bill mentioned on the operating =
>>> point where current and voltage noise of the input stage is optimal 
>>> for =
>>> the source impedance of the microphone. This is why IEC 60268-4 
>>> requires =
>>> specification of both output impedance and minimum load impedance, 
>>> which =
>>> is typically 10-20x the output impedance.=20
>> This is all true, unfortunately.  It is far from a perfect world.
>> I'm waiting for a noiseless transformer that gives me free voltage 
>> gain and
>> I'm not expecting to see one any time soon.  But once I get one I'll be
>> able to get the largest possible signal into an input in order to 
>> swamp the
>> noise.
>> I'll point out that specifying minimum load impedance is sufficient for
>> a condenser microphone but that a dyanmic microphone should also be 
>> specified
>> for maximum load impedance.  In some cases with weak coupling where the
>> electrical load does not make a large proportion of the total damping 
>> that
>> maximum might be infinite, but not all microphones are like that.
>>> Dan is pointing out that one needs to know the output impedance of 
>>> the =
>>> microphone to design an optimal preamp input stage, and this 
>>> information =
>>> is seldom supplied. There are preamp designs that don=E2=80=99t play =
>>> well with very low output impedance mics, for instance, because of 
>>> their =
>>> negative feedback structure (applying NFB to the input stage in 
>>> parallel =
>>> with the input signal.) There are many mics that don=E2=80=99t do 
>>> well =
>>> when loaded with anything less than about 1500 ohms, and if you 
>>> parallel =
>>> a few consoles at a venue without considering this, performance may 
>>> be =
>>> impacted. Performance may also be different depending on impedance 
>>> from =
>>> each side of the input to ground, which may be related to the =
>>> differential output impedance, or not.=20
>> This is all true, and it's why some preamps perform better with some
>> microphones while other preamps perform better with other microphones.
>> But nobody sells an "optimized for condenser microphones" or "optimized
>> for moving coil dynamic microphones" preamp although there are a 
>> couple of
>> "optimized for ribbon microphones" preamps out there.
>> --scott
>> _______________________________________________
>> ProAudio mailing list
>> ProAudio at bach.pgm.com
>> http://bach.pgm.com/mailman/listinfo/proaudio

This email has been checked for viruses by Avast antivirus software.

ProAudio mailing list
ProAudio at bach.pgm.com

_______________________________________________ProAudio mailing listProAudio at bach.pgm.comhttp://bach.pgm.com/mailman/listinfo/proaudio  
|  | This email has been checked for viruses by Avast antivirus software. 
www.avast.com  |

ProAudio mailing list
ProAudio at bach.pgm.com

_______________________________________________ProAudio mailing listProAudio at bach.pgm.comhttp://bach.pgm.com/mailman/listinfo/proaudio
ProAudio mailing list
ProAudio at bach.pgm.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://bach.pgm.com/pipermail/proaudio/attachments/20210614/c740a568/attachment.html>

More information about the ProAudio mailing list