John Curl's Blowtorch preamplifier part III

[scott wurcer]

Scott, here is a zip with 2 recorded files, 300Hz+10kHz and 200Hz+10kHz. Recorded from 2 sources, microphone placed about 10cm from the tweeter membrane. Sorry for some background noise modulation by room noise.

http://pmacura.cz/imtest.zip

P.S.: attached room background noise, same dB scale.

OK, I'll see what it shows.

 

[1audio]

Isn´t it mainly calibrator distortion? Iirc the GR "omnical" was specified for THD with not so low " <1% " .

In that case it is. I'm working through my collection of calibrators and they are all pretty high distortion. I have some headphones to try as well.

 

[john curl]

Wow, doesn't anybody use a measurement mike that you can rely on?

 

[PMA]

Studio mike has lower distortion, for me 😉

 

[john curl]

PMA, I don't understand this problem. I have been around measuring mike capsules, and even measurement mike systems for the last 45+ years. I built my own capsule preamp from AES apps, back in 1972, and it still works well today, certainly better than most studio mikes. Since it is a 1" element (B&K), it does only measure accurately to about 18KHz, but for distortion, it seems to run rings around quality speaker drivers. I used it to evaluate speakers for the 'Wall of Sound" (mostly JBL) but also Electrovoice, Klipsch, etc. and discern just about everything that was useful at the time.
I also worked with and modified Nagra microphone electronics that mated with the B&K 4133 1/2" mike for the GD, with no real problem, and finally John Meyer and I purchased and used 2619 electronics for our speaker testing while we worked together in Switzerland in 1974-75, with no problem.
However, most studio microphones, either have output transformers and/or are 48V powered, and this is potentially limiting due to the compromises of current and transformer size, so I would be careful with those.

 

[elektroj]

Wow, doesn't anybody use a measurement mike that you can rely on?

That surprises me as well. Why use Behringer or some other "toys" when there are proper mics available. Ok, they aren't exactly a 30 Euro a piece, but for a little more there are plenty of former East German (Gefell) measurement mics available on 'worlds largest marketplace'. If you're based in Europe, IMHO, it's a sin not to buy those for proper acoustic measurements. They are as good as B&K, but could be had for less money.

http://iesservice.de/MIKROappt.pdf
REISS-Service ROBOTRON Akustische Messtechnik

PMA, here's info about them in your own language 😉 :

ELEKTROAKUSTIKA® RFT MV102 - measuring microphone
ELEKTROAKUSTIKA® RFT MV201 - measuring microphone

P.S. The company is still around and still producing them
Sound / Noise

 

[1audio]

The Gefell mikes are very good. Earthworks uses a MK301 1/4" as a reference for transient response. Really rare in US and eBay sellers seem to be Russian so caution is indicated.
However none of this is as simple as it seems. First measurement microphones are about level and stability. Distortion is a separate and less important issue. As evidenced in the current efforts there are no simple ways to test for HD. IM is not a direct equivalent but easier to test for. I tried the electrostatic actuator and got the predicted 1.3%. (Single side drive, 800V bias and 30V audio). I'm not sure if a predistortion can be calculated that would translate into "0" distortion. I'm game to try if possible.

 

[scott wurcer]

First measurement microphones are about level and stability. Distortion is a separate and less important issue.

Folks are losing sight of this, distortion from a weakly non-linear transfer function suppresses the amplitude of the fundamental and this is almost the only concern in the microphone literature.

 

[31697B]

This doesnt look like a promising way to learn what the speaker/driver distortion is ... even those very stable sensitivity B&K microphones have too high distortion.

??

https://pdfs.semanticscholar.org/e9f8/5980d82b1d8f9bcc2574e38f0c1ef162d07d.pdf



THx-RNMarsh

 

[Jakob2]

Not sure why the B&K distortion should be considered as to high, as it depends on the acoustic pressure level. Below 95 dB (SPL) HD from the mic/preamplifier combo should be below 0.01% .
So far i only have seen large electrostatic panel speakers getting such low distortion numbers.

Wrt to the Dessein thesis there was a predecessor from Pastillé:

Pastillé,Holger. Ueber die Nichtlinearitaeten am Kondensatormikrofon unter besonderer Berucksichtigung der Membran

https://depositonce.tu-berlin.de/bitstream/11303/700/1/Dokument_10.pdf

Unfortunately only available in german; translated title would be something like
"On condensator microfon nonlinearity especially considering the membran contribution"

As it is a PHD thesis it covers the theory quite extensively and the contribution of the amplifiers (impedance converters) is considered/measured as well. Some measurement solutions were developed including laser equipment for membran mode monitoring.

Due to collaboration with different microphone manufacturers some mechanical modifications could be examined.

 

[PMA]

However, most studio microphones, either have output transformers and/or are 48V powered, and this is potentially limiting due to the compromises of current and transformer size, so I would be careful with those.

I do not intend to generalize, I can only say that in my case the studio 1" condenser microphone has much lower distortion than the ECM8000 (which is a cheap and entry level measuring microphone) and this 1" microphone has very low distortion in absolute terms. It is a transformerless microphone with Schoeps-like el. circuit (2SK170, balanced output). It seems that any microphone should be tested if used for measuring distortion and that the 2-source IM test might be valuable.

 

[scott wurcer]

2SK170

Unlikely, the lowest noise 1" mic readily available (Rode NT1a) uses a (selected?) ordinary SOT23 FET, the substitution with a 2SK170 would increase the noise and distortion due to the input capacitance.

I have some results. It seems that using the two separate speakers nicely gives the distortion of the mic alone and as expected is purely AM but around -73dB ,is this good enough? I do't think the fancy maths brings anything to the party, you could have culled this from a simple high resolution FFT. They might help when there is FM from both tones on one speaker, the displacement on the mic's diaphragm is just too small.

Another comment on distortion, all mics based on electret capsules with built in FET's are biased by the balance of leakages on the input junctions i.e. there are back to back diodes on the input. The Linkwitz mod helps but they are not something you would use for this. I published a circuit in Linear Audio to bootstrap out the input capacitance, but these capsules are still probably too noisy for this application.

 

[PMA]

I have some results. It seems that using the two separate speakers nicely gives the distortion of the mic alone and as expected is purely AM but around -73dB ,is this good enough? I do't think the fancy maths brings anything to the party, you could have culled this from a simple high resolution FFT. They might help when there is FM from both tones on one speaker, the displacement on the mic's diaphragm is just too small.

Thank you, do you read the AM components as sidebands 10kHz-F1, 10kHz+F2 related to F1 amplitude? This I am able to do, of course, it can be seen from the posted spectrum.

 

[scott wurcer]

Thank you, do you read the AM components as sidebands 10kHz-F1, 10kHz+F2 related to F1 amplitude? This I am able to do, of course, it can be seen from the posted spectrum.

Yes, I remain a little curious about the difference in height of the sidebands (maybe it just falls out of the math I can't remember right now) because it is still there when I use exact 1Hz bins. I grab 20sec of your data and take a 882000 point FFT where the 10kHz tone is better than 1ppm on exact frequency.

 

[PMA]

Yes, I remain a little curious about the difference in height of the sidebands

I am not sure why is that. Of course I made a FR measurement in the same, 2-source setup (which is not optimal, there is a dip between 1kHz-3kHz due to mike position and phase cancellation), but it did not give an answer. According the FR, their height effect should be opposite 🙂.

 

[Jakob2]

Afair Sennheise used the two speaker distortion measurement with a fixed frequency difference between the two sources. "Sweeps" (apparently stepped frequencies are a bit easier) from 200 Hz - 20 kHz (spkr 1) and 270 Hz - 20070 Hz (spkr 2), measured signal filtered for the 70 Hz component.

To get better results at higher frequencies they developed a push-pull condenser capsule design.

If -73 dBr is good enough? Depends on the level uses for the measurements.
At around 90 dB (SPL) it wouldn´t be good enough for really good electrostats for example.

Edit: Sennheiser used the RF principle in this so-called MKH microphones.

 

[PMA]

Let me have a look from another perspective.

We get some 0.02 - 0.03% IM distortion or HD distortion at 200Hz and even 10kHz from the whole chain, DAC - preamp - power amp - speaker - microphone - microphone preamp - ADC. 1 ppm at 10kHz frequency. This is not bad.

Now, this is an amplifier, better say a preamplifier thread. We may find many (even highly prized) power amplifiers, and even preamplifiers, that will certainly worsen the distortion found by acoustical path measurement (especially the 10kHz distortion) . So? Any thoughts? Famous "subtle sound differences"? 😉

 

[PMA]

At around 90 dB (SPL) it wouldn´t be good enough for really good electrostats for example.

It was more (SPL) because of the near-field measurement technique.

 

[31697B]

Near field and coupler distortion data is unrealistic for speaker distortion/spl measurments in a listening room.

I think =/< .1% should be fine as it is lower than driver being tested for our experiemnts on improving the driver.

There is so little publish distortion data though... Just the Max spl level at 1% to 3%... some at 10%



THx-RNMarsh

 

[scott wurcer]

If -73 dBr is good enough? Depends on the level uses for the measurements.
At around 90 dB (SPL) it wouldn´t be good enough for really good electrostats for example.

The push-pull idea is good, I wonder if there is a way to use a dual diaphragm capsule of this. BTW I looked more closely and the distortion was closer to around -77dB in PMA's test. It should be OK for the woofer low frequency distortion measurement.