Real Mic from a Chinese Toy

[Marik]

[quote author="Wavebourn"]
Cello and double bass have different formants. All singers have the same. 2 of them are well known since antiquity "Belkanto" formants.[/quote]

I was quite interested in that topic and asked a friend of mine--professional singer to comment. He sent me an interesting article:

The Journal of the Acoustical Society of America -- June 1977 -- Volume 61,
Issue 6, pp. 1622-1629

Acoustic properties of voice timbre types and their influence on voice
classification
Thomas F. Cleveland
University of Southern California, School of Medicine, Los Angeles, California
90033
University of Southern California, School of Music, Los Angeles, California
90033


"An investigation to determine the significance of formant frequencies, pitch,
and source-spectrum slope on voice classification is reported. Eight
professional singers sang five common vowels on four common pitches, and, in a
forced-choice test, vocal pedagogues classified the vocalization as tenor,
baritone, or bass. Formant frequency and pitch differences accounted for most of
the jury- evaluation results. To evaluate more accurately the significance of formant frequencies, pitch, and voicing and source-spectral
slope on voice classification, vowel synthesis was employed. Measurements of
spoken vowel vocalizations revealed higher formant frequencies in tenor timbre
type and lower formant frequencies in bass timbre type. Long-time-average
spectra seem to afford a good estimate of voice classification.
Formant-frequency percentage differences between basses and tenors were similar
to those found between males and females, suggesting that the mouth and pharynx
lengths, known to differentiate females and males, probably also differentiate
tenors and basses."

His response to the last soundfile (unfortunately, the previous aria clip does not work anymore) was that it might be sung by bass-baritone.
A quick googling reveals that indeed, Sergey Sadvorny is a bass-baritone.

 

[bcarso]

[quote author="Marik"] <from the cited paper:> ... Long-time-average
spectra seem to afford a good estimate of voice classification. ...
[/quote]

This is quite interesting and in significant contrast to some other instrumental identifications, where the attack transients are so important.

 

[Guest]

[quote author="bcarso"][quote author="Marik"] <from the cited paper:> ... Long-time-average
spectra seem to afford a good estimate of voice classification. ...
[/quote]

This is quite interesting and in significant contrast to some other instrumental identifications, where the attack transients are so important.[/quote]

Attack transients are important for acoustical instruments, say to distinguish a guitar from a violin.

Mouth and pharynx sizes determine frequencies of speech articulation formants, and indeed they are different for different people. What is interesting, people can't understand foreign speech if they can't pronounce what they heard correctly, i.e. training of muscles is necessary to hear and understand! Microphones or EQs can't change such formants because they are dynamic in nature. In my Bi-El-Countour processor I play with another formants... I can't show records made with Bi-El-Contour, because I'm going to "labelpoop" it before.

And indeed, Sergey Zadvorny can sing both bass and baritone parties.

 

[Rossi]

To come back to the original topic. Well, kind of. I think there are a few points that have not been put in perspective regarding the Schoeps circuit.

I don't think the DC converter is such a bad idea. With a fixed 60V DC bias for the capsule you are no longer dependent on low current draw. By contrast, if you want to bias the capsule directly from P48, you have not only the 48V limit but also a small window regarding total power consumption. The more current you use, the less voltage you'll have left for your capsule bias. Take also into consideration that many preamps don't really conform to the "modern" P48 specification. I've talked to a developer from a pretty well known manufacturer who was dead sure P48 allowed for no more than 2 mA current draw. As a result, the PSU was undersized and there was noticeable hum with just *one* mic drawing a totally acceptable amount of current. It was an 8-preamps-in-a-box unit... :roll: The later revision supplied enough current for eight TLM103 type mics, but still wasn't fully compliant to P48 standards.

Anyway, that's the kind of stuff that mic manufacturers face. There is a chapter on P48 crap in the Schoeps microphone book, btw. For all I know, modern Schoeps mics draw a little more current than the Schoeps circuit you'll find on the internet, so I'm not sure this one's up to date. The old circuit sure had to face insufficient power supplies. Early P48 was only specified for 2 mA, while the more modern version allows for 10 mA. Well, in theory. In practice, you have to make a compromise between best performance and least disturbance under suboptimal conditions. The DC converter provides a good basis for that.

Of course you also get the benefit of optimal S/N from the capsule. A schoeps-like circuit without DC converter will often draw enough current to drop P48 to say 35V. Maybe even lower on a crappy supply. The 60V from the DC converter will thus give you about 5 dB extra signal level without increasing the noise, so your S/N improves about as much. Given the small size of a Schoeps capsule, the difference is in a critical area. Something like 21 dBA vs 16 dBA self noise. That's a very appreciable difference, and an improvement almost without drawbacks. A DC converter for the capsule bias will hardly draw any extra current, if properly designed. And I don't think the circuit used in Schoeps mics affects the sound quality in a negative way. At least I haven't heard any complaints. You're likely to run into problems with bad board layout, though. Which is what you may find in cheap microphones.

To sum it up, the DC converter gives you better S/N, a hotter signal (both particularly relevant for small diaphragm capsules) and allows you to draw more current in the EF output stage without running into problems with improperly designed P48 supplies.

 

[Guest]

Rossi; I agree that connected to the appropriate custom input (enough DC current and higher AC impedance) and well biased emitter followers will give much less distortions. And I agree that properly constructed DC/DC converter will not leak too much of HF energy, but the reality was, this particular design could be greatly improved using less elements than originally (however, a transformer cost more than all of them, but we pay for usability of the mic itself).

 

[clintrubber]

[quote author="Rossi"]A DC converter for the capsule bias will hardly draw any extra current, if properly designed.[/quote]
I'm not sure about the practical lower obtainable limit w.r.t. current consumption of the DC/DC-converter, but apart from the 'zero current consumption' of the capsule itself, the oscillator of the DC/DC-converter has to spend some current to operate of course. You'll understand, this is not meant as nitpicking, just an observation.

I ran a circuit-simulation of that oscillator a while ago and that typ. circuit (just threw in a 'normal' NPN) had an average current consumption of near 2.6mA (without any added zener-current consumption or such).

In the MCA-SP1-version of the Schoeps circuit, the added DC/DC section (including a zener which otherwise wouldn't be required*) even seems to use 3.5mA (based on the posted measurements in the pdf from Zapnspark @ micbuilders), but that figure needn't be the lowest obtainable current.

Regards,

Peter

*: here the circuit topology differs from the Schoeps because the JFET-stage is connected to a higher voltage than in the Schoeps which feeds both the oscillator & the JFET-stage from the 6.2 V Zener-node

 

[Rossi]

Well, I said "hardly" any extra current, not none at all. :wink:

To be honest, I don't know how much current this DC converter draws, but I built a cmos converter which drew almost nothing. The cmos stuff is very prone to become a HF crap radiator, though, and definitely requires clever placement and good design. One of these days I'm going to experiment with the Schoeps oscillator approach.

 

[clintrubber]

[quote author="Rossi"]Well, I said "hardly" any extra current, not none at all. :wink: [/quote]
Sure :thumb: As said, I wasn't after nitpicking but in this context of phantom-powered stuff, just a few mA's is not totally insignificant. But hey, your :wink: adressed all this already

To be honest, I don't know how much current this DC converter draws, but I built a cmos converter which drew almost nothing. The cmos stuff is very prone to become a HF crap radiator, though, and definitely requires clever placement and good design. One of these days I'm going to experiment with the Schoeps oscillator approach.

I didn't do any further checking now, but both the lower current consumption and higher level of generated crap might be explained by assuming that the CMOS-version does square-waves (most likely topology), while the Schoeps is a class-A sinewave-oscillator.
So in the former it's just dynamic current consumption (charging of the caps) but the current drawn by the dr.Schupp oscillator is a sinewave around a DC-term (IIRIC that typ. sim gave 2.6mA DC with a few hundred uA pp AC).

Bye (& enjoy tonight all),

Peter

 

[Rossi]

I'm quite confident that the Schoeps oscillator could be worked at very low current, because there is almost no output load. I don't think 2.6 mA is a correct real world figure; that's about as much as most mics use in total. I'd guess that the oscillator uses less than 1 mA. The best way to find out would be to compare a mic that uses that circuit to the same mic with the oscillator bypassed. The MXL603 comes to mind as a possible lab rat. I think some people around here bypassed the oscillator, because they liked the sound better at lower capsule voltage.

I actually have a t.bone mic here, that follows the schoeps circuit but without the oscillator. I'll see what it draws as soon as I can find the time. I think I once measured the P48 voltage drop at about 35V, IIRC. I don't think I have a mic that uses the oscillator, though.

 

[clintrubber]

[quote author="Rossi"]I'm quite confident that the Schoeps oscillator could be worked at very low current, because there is almost no output load.[/quote]
That simulation was just with 'a circuit' of that topology, so indeed it can already be (or made to be) less.

But the fact that the capsule draws about zero current doesn't necessarily mean that the DC/DC-converter can run at about nothing as well.

In general that's what can be frustrating about DC/DC-converters in case the 'load' isn't really drawing much current: you need to spend some current to get stuff going and/or to get low-ripple etc, but then the 'load' hardly consumes anything :evil:
So your resulting total efficiency-figure is very poor.

Maybe Schoeps opted for a (more consuming) sinewave i.s.o. a currentwise cheaper 2-BJT squarewave-generator because of the less worse disturbance pattern ?

The posted info from Zapnspark of a Schoeps-kinda circuit (MCA SP1) indicates such currents though (assuming his data was indeed measurements). Which again doesn't mean that it really needs to be this high.

Hope to find the time later today to do a simple current measurement in the Schoeps-derived MCA SP1. Which doesn't need to mean that all such oscillators are running at these currents, but at least it can give some indication.

Cheers,

Peter

 

[Rossi]

I just did a quick sim, but in a different way. I have the Dorsey circuit on my harddrive, which I quickly modified to Schoeps values (6.2V Zener etc) but without the oscillator. According to my simulation, P48 drops to 34.5V. Now, if I look at the Schoeps circuit (the original document), the measurement point value given is 34.4V - and that of course includes the oscillator. So the oscillator ought to draw very little current.

 

[zapnspark]

The polarizing oscillator is a variation on a Hartley oscillator.
It produces a very clean ~2 mHz sinewave.
The typical current draw of this polarizing oscillator is .5 mA
The zener regulator needs approx. 1 mA or more for decent regulation.
The FET phase splitter draws about 1 mA.
So -- we have a current draw of .5 + 1 + 1 = 2.5 mA minimum.
Around 3 mA total current draw is typical of most of these Schoeps-like microphone circuits.

ZAP

 

[Guest]

Here is the record I made yesterday using this mics made an year ago:

http://wavebourn.com/rain_helicopter.mp3

Happy New Year! :grin:

 

[clintrubber]

I measured the voltage drop across the 1k5 resistor in two MCA SP-1 mics and found 4.81 & 4.87V, which means around 3.2mA. This is the current for the zener & oscillator together.

It is ~ in agreement with both that circuit simulation I did and the posted 'GenericChinaMic' voltages from ZAP @ micbuilders.

Phantom drops from 47.3V (unloaded) to 35.2 V (at mic-pins), indicating a total mic current draw of 3.6 mA.

So at least for these mics the oscillator & associated zener take about all the current.

I guess the conclusions are as simple as:
#1. the 'original Schoeps-circuit' draws considerable less current in its oscillator (see data from ZAP here above) than the 'GenericChinaMic' incarnations, with this SP-01 as an example
#2. despite being more wasteful with its oscillator-current, 'GenericChinaMic' seems to get away with it.

The Schoeps-circuit has a trimmer for polarizing-voltage adjustment, but the clones not. The clones were probably put at higher juice to economize on this trimmer and still be safe over component tolerances etc.

Regards & happy new year !

Peter

 

[Guest]

[quote author="clintrubber"]
#2. despite being more wasteful with its oscillator-current, 'GenericChinaMic' seems to get away with it.
[/quote]
Real and laboratory conditions are very different. Think what is going on when SPL changes forth and back.

 

[clintrubber]

[quote author="Wavebourn"][quote author="clintrubber"]
#2. despite being more wasteful with its oscillator-current, 'GenericChinaMic' seems to get away with it.
[/quote]
Real and laboratory conditions are very different. Think what is going on when SPL changes forth and back.[/quote]
I've looked at capacitor-mics with oscillators only for a short while now, so might not be aware of quite a few more or less obvious 'implications'. Could you elaborate a bit ? It might be quite obvious, but first some sleep now, maybe tomorrow morning it's all clear to me :wink:

You mean perhaps that higher SPL's and the 'hot' oscillator might conflict w.r.t. the available amount of juice as provided by P48 ?
For now I'm assume that high SPL doesn't directly lower the capsule polarization voltage.

Regards,

Peter

 

[Guest]

[quote author="clintrubber"][quote author="Wavebourn"][quote author="clintrubber"]
#2. despite being more wasteful with its oscillator-current, 'GenericChinaMic' seems to get away with it.
[/quote]
Real and laboratory conditions are very different. Think what is going on when SPL changes forth and back.[/quote]
I've looked at capacitor-mics with oscillators only for a short while now, so might not be aware of quite a few more or less obvious 'implications'. Could you elaborate a bit ? It might be quite obvious, but first some sleep now, maybe tomorrow morning it's all clear to me :wink:

You mean perhaps that higher SPL's and the 'hot' oscillator might conflict w.r.t. the available amount of juice as provided by P48 ?
For now I'm assume that high SPL doesn't directly lower the capsule polarization voltage.
[/quote]

I mean something like percussions, with fast attack then decay. Power consumption of DC/DC converter leaves not too much for emitter followers. 3.4K resistors that power the mic and 200 Ohm input impedance is not good for them anyway. Such conditions were defined for transformer output mics. However, some short cables and a phantom power box with 10 kOhm or more impedance on stage would be good for such mics, it is the option to consider instead of modifying them, like tube power supplies that stay on stage I had in mind such a solution an year ago...

...but I decided to modify mics for myself instead of making external boxes with pre-amp, power, and an equalizer for sale.

 

[Rossi]

@ clintrubber: Are you sure your P48 supply is as powerful as it should be? The 47.3 V unloaded could suggest that it may not be fully compliant to P48. Dead short, each of the legs should supply about 7 mA.

The original Schoeps circuit is specified for 4 mA, which is pretty much what I get in simulation - without the oscillator.

@ wavebourn - Most real world preamps have input impedances of about 2k (not 200 ohms), the absolute maximum would be 13.6k due to phantom resistors. If you lowered the value to increase the max current, you'd also lower the input impedance limit.

Also note that mics are usually specified for a minimum input impedance of 1k. Below 1k, THD is likely to increase, and maximum SPL will be lower than specified, sometimes considerably so.

Another way to get more current for the output EF stage would be to use a single ended, impedance balanced topology. For all I know that's what Neumann use in the KM18x series. The downside is that this approach requires large output capacitors, i.e. lytics.

 

[clintrubber]

[quote author="Rossi"]@ clintrubber: Are you sure your P48 supply is as powerful as it should be? The 47.3 V unloaded could suggest that it may not be fully compliant to P48. Dead short, each of the legs should supply about 7 mA.[/quote]
Hi Rossi,

For that measurement I used a MX8000, so a desk which should be able to supply a lot more than just one channel. So it surprised me that the unloaded voltage wasn't more accurate... you have/had one as well, you'll have measured the P48 as well one day, what did you get ?

Regards,

Peter

 

[jdbakker]

[quote author="Rossi"]@ clintrubber: Are you sure your P48 supply is as powerful as it should be? The 47.3 V unloaded could suggest that it may not be fully compliant to P48. Dead short, each of the legs should supply about 7 mA.[/quote]
Last I checked P48 was defined as 48V +/- 4V , so 47.3V looks perfectly legal.

JDB.
[couldn't easily find an online copy of IEC/DIN EN 61938 (the Phantom spec), so no link, sorry]