Transistor sound, cables, and emitter followers

[Wavebourn]

I'm opening a new thread since this discussion goes deeper than just a modification of a condencer microphone.

Originally Posted by ulysses

Okay, there are a few of your claims that I will dispute:

1. Impedance matching for maximum power transfer. The world has abandoned it, as far as short-run interconnections are concerned. The audio world is a world of voltage transfer, and for good reason. We can afford to have amplifiers at the far ends of our cables, and we want immunity from cable capacitance and other real-world interconnectivity issues.

An ostrich wants immunity from enemies when it hides his head under the sand. I understand your point.

Low impedances driving high impedances is the way we do this. It's true that microphones are still a special case, but not because they benefit in any way from impedance matching. It's only due to "legacy" microphones still in use which require proper termination of their output transformers to give flat frequency response. And in any case, we're never going for impedance matching or maximum power transfer. Typical legacy microphones with a source impedance of 150 ohms feed inputs ten times that. It's still a voltage transfer situation. If this were a discussion of dynamic microphones, you might have a (weak) case for power transfer. But we're talking about active source circuitry, and you haven't given a credible argument for insisting on impedance matching.

Yes, the world goes with Microsoft Windows, that's why we, Unix professionals, are still in demand. And because of the world that goes to impedance mismatch high-end audio boutiquea have own bread and butter, offering an impedance match for a big money.

2. A 2SK170 will be perfectly happy with 1-3mA, there's no need to use up 10mA in the transimpedance stage. That reduces the voltage drop across the phantom build-out resistors, and there no reason we can't put those milliamps through a couple of emitter followers on the way.

Yes, it will be perfectly happy if the current is not below 1 mA. And yes, we can put few miliamps through emitter followers getting a "transistor sound" as a result.

3. "Transistor Sound"? Please. I doubt there has been a sound reproduced in recent years without a transistor someplace in the signal chain, whether it be in the source, recorder, or reproduction equipment. You think adding two to a microphone is going to destroy the purist fidelity of your wax cylinder?

We don't get a transistor sound from an equipment that don't amplify much power. For example, my Yahaha MX-400 contains lots of transistors and opamps, but don't add a transistor sound. Why?
But some microphones and majority of modern power amps add the "transistor sound".

Why?

It is not because of usage of transistors. It is because of inproper usage of them.

Back in 1970'th there was a belief, that transistor sound is caused by crossover distortions in the output stage, so transistors must be biased such a way they eat a power turning themself into heaters.

This my design of amplifier disproves this!
Output transistors ARE NOT BIASED AT ALL!

The sectet is, a powerful driver WITHOUT ANY EMITTER FOLLOWER supplies output by initial power through RELATIVELY LOW RESISTANCE enough to discharge cable capacitances. When voltage between output of a drivea and output of the amp is enough for BJTs that start to amplify. Due to selected values of emitter resistors BJTs work up to 3 amperes only (when they are more linear), the rest is handled by FETs. As you may see, no bias at all that as is believed causes crossover distortions and a "transistor sound phenomena", but no transistor sound at all. It is a 3-step approximation amp.

******//wavebourn.com/images/audio/swinik-III.gif

Why?

Because no PN-junctions work directly on capacitance of speaker cords and on a complex load of speaker drivers that on some frequencies mimic capacitive load, no bias shift because of rectification that causes audible "transistor sound".

The similar approach was taken in English amp "Quad 405-2", but they decided to hide the fact that it is a well known approximation in order to patent it under the fancy name "Current Dumping". Anyway, it is the amp WITHOUT ANY BIAS of output stage that nevertheless DOES NOT HAVE A TRANSISTOR SOUND.

The same with microphone amplifiers that are loaded on capacitances of cables. If bias currents are low, they slower discharge cable capacitances. If input impedance of the mic amp is high it slower discharge cable capacitances. It means, more distortions on higher end of the spector because of rectification of the signal on PN junctions of output emitter followers.

[Tim Farrant]

Originally Posted by Wavebourn

The similar approach was taken in English amp "Quad 405-2", but they decided to hide the fact that it is a well known approximation in order to patent it under the fancy name "Current Dumping". Anyway, it is the amp WITHOUT ANY BIAS of output stage that nevertheless DOES NOT HAVE A TRANSISTOR SOUND.

I'll have to disagree with you on this. The Quad 405 was the most brittle sounded power amp I have ever heard.

A nicely biased Phase Linear Class AB from the 70's eats it for breakfast!!

Tim.

[Geoff_T]

Originally Posted by Wavebourn

An ostrich wants immunity from enemies when it hides his head under the sand. I understand your point.

Hi

Totally irrelevant to the thread but I believe that's a misunderstanding like Kangaroo meaning "I don't understand you!"

I believe that an ostrich can run pretty darn fast and kick aggressors where it hurts.. I don't think that it's a stupid bird.

I think it puts it head near the ground to pick up vibrations of approaching creatures.

If they did it just to hide they would be extinct by now!

[tINY]

I hear you Wavebourn. My day job requires the use of balanced impedances. We find 85 ohms to be more harmonious for our latest designs. We also find that line drivers with two sets of outputs with a delay applied to one of them tends to work better.

How is that Humolt Harvest this year, good?




-tINY

[Wavebourn]

Originally Posted by Geoff_T

Hi

Totally irrelevant to the thread but I believe that's a misunderstanding like Kangaroo meaning "I don't understand you!"

I believe that an ostrich can run pretty darn fast and kick aggressors where it hurts.. I don't think that it's a stupid bird.

I think it puts it head near the ground to pick up vibrations of approaching creatures.

If they did it just to hide they would be extinct by now!

May be you are right, I should use another illustration for the industry, some animals who always go after the leader.

[Wavebourn]

Originally Posted by Tim Farrant

I'll have to disagree with you on this. The Quad 405 was the most brittle sounded power amp I have ever heard.

A nicely biased Phase Linear Class AB from the 70's eats it for breakfast!!

Tim.

I have nothing against nicely biased AB except a power consumption. I'll start to use again the amp with such outputs as soon as weather allows:


******//wavebourn.net/images/audio/tower.gif

Source follower loaded by a current source controlled by input sighal. The best sound I ever heard. Also, a good heater.

[Marik]

Originally Posted by Wavebourn

I'm opening a new thread since this discussion goes deeper than just a modification of a condencer microphone.

An ostrich wants immunity from enemies when it hides his head under the sand. I understand your point.

Yes, the world goes with Microsoft Windows, that's why we, Unix professionals, are still in demand. And because of the world that goes to impedance mismatch high-end audio boutiquea have own bread and butter, offering an impedance match for a big money.

I think it would be helpfull here to recreate the original context for sake of balanced discussion.

So... here was your first statement:

Originally Posted by Wavebourn

But I told about loading microphones by low impedance on AC, because it is desirable to match impedances for better power transfer that means wider dynamic range.

and

Originally Posted by Wavebourn

My point is, we amplify neither just a current nor, voltage, we amplify a power. The less we need to amplify, the less distortions we can add, both of linearity, and of a frequency response. That's why an impedance match is so vital both in microphone and power amplifiers.

followed by Ulysses' passage you quoted above.

I am not sure why you ommited my response, which was showing that power transfer/impedance matching in fact reduces dynamic range and bandwidth.

So here was my response:

Originally Posted by Marik

Everything would be fine if you did not overlook two things--NOISE and BANDWIDTH. It was not such a big of the issue half of the century ago in broadcasting, but it is now in studio recording.
Yes, some 50 years ago they used a 600 ohm standard for the reason of preserving every milliwatt of power because there was only one real power device--transformer. Bulky, expensive, with very high turn ratio (often 600 ohm to 150K grid), they required special techniques to reduce capacitance, leackage, and losses, and were additional sources of distortions and noise as a result of high copper resistance.
Any slight mismatch often caused a significant change in frequency response.
And Ulysses has already mentioned interconnectivity issues driving low impendances.

To make sure we are all on the same page, it is important to notice that for power transfer/impedance matching, the load input impedance is equal to a source output impedance.

In 50s they realized that this technique substantially loads the output of microphones, especially ribbon ones, causing irregular frequency response, limited output, and increased distortions. Ever since, the votage trasfer was prefered.

For further reading I'd suggest AES paper On Electrical Loading of Microphones, from October 1955, volume 3, number 4.

Best regards, M

[Wavebourn]

Originally Posted by Marik

I am not sure why you ommited my response, which was showing that power transfer/impedance matching in fact reduces dynamic range and bandwidth.

Because I am a nice guy and don't argue against nonsenses.

Marik, my post about biasing FET is totally based on elementary physics and Toshiba datasheet for the transistor in question. I don't want, and can't argue against beliefs if you don't want to look at datasheet and perform elementary calculations.

This topic is about "transistor sound" phenomena that is the result of rectifications of a signal on P-N junctions and load capacitances.

[Marik]

Originally Posted by Wavebourn

Marik, my post about biasing FET is totally based on elementary physics and Toshiba datasheet for the transistor in question. I don't want, and can't argue against beliefs if you don't want to look at datasheet and perform elementary calculations.

Well, since you mentioned it... based on elementary physics, and most importantly, based on practical experience in real life, in that particular case, i.e. 2SK170 with Rg=1GOhm with 30V@10ma, or even 30V@3ma, the Vdg will be too high and the noise will be unacceptable for use in a condenser microphone.

Best, M

[Wavebourn]

Originally Posted by Marik

Well, since you mentioned it... based on elementary physics, and most importantly, based on practical experience in real life, in that particular case, i.e. 2SK170 with Rg=1GOhm with 30V@10ma, or even 30V@3ma, the Vds will be too high and the noise will be unacceptable for use in a condenser microphone.

Best, M

This is what you believe in.

Please open the datasheet and look at pictures. You'll be surprised.

Once more, for your convenience:

******//www.ampslab.com/PDF/2sk170.pdf

Practical experience in the reall life corellates well with datasheet Toshiba presents for their production. Input capacitances, especially Miller capacitance, several times less on 30V than on 1V. Forward transfer admittance drops down abruptly below 2 mA current. Equivalent noise voltage is lower between 3-10 mA. Noise figures go up abruptly below 1 MA. Noise figure is flat between 5 and 35V. And, what is significant as well, goes straight up on input impedances below 2 KOhm and above 200 Kohm. RG=1GOhm is not a source impedance. It is a purely leakage resistor. Source impedance is purely capacitive, and 1G is so big not because we "amplify voltage", but because we stilll need a leakage resistor, but don't need a resistive component on input, so it's resistance is too high to take in account speaking of a source impedance.

It is all from the datasheet, if you still don't want to open it...

Let's go back to the topic.

Speaking of "current amplifications", "voltage amplifications", we just simplify the explanation, but in reality we always amplify a power. At least, engineers amlify power, while technicians speak of voltage and current amplifiers. Purely current amplifier is a voltage repeater, but it does not mean that the voltage disapeared from the Earth. And a purely voltage amplifier is a current repeater, but nevertheless it presents on input and output. Typical example of a voltage amplifier is an amplifier with the common grid (base, source). Current amplifier is an amplifier with common anode (collector, drain), called "cathode repeatrer". But an amplifier with common cathode (emitter, source) is a complex amplifier that amplifies both voltage and current.
Yes, sometimes we go on loss of power in order to achieve better linearity between interconnected stages, but it does not mean that it is good for the entire amplification. Loosing power, we increase the need to amplify it again, adding more stages, more noise, more distortions... I.e. decreasing dynamic range.

[Marik]

Originally Posted by Wavebourn

This is what you believe in.

Please open the datasheet and look at pictures. You'll be surprised.

Once more, for your convenience:

******//www.ampslab.com/PDF/2sk170.pdf

Huh... how to say it... yes, I saw the datasheet... many times!
The pictures you are refering to gives the parameters for certain conditions, in that case for Rg=1K. I am talking about R=1GOhm, as used in the modern condenser microphones.
For example input capacitance listed as 30pf can go as high as 600pf in different conditions and varies significantly, depending on gm.

Best, M

[Wavebourn]

Originally Posted by Marik

Huh... how to say it... yes, I saw the datasheet... many times!
The pictures you are refering to gives the parameters for certain conditions, in that case for Rg=1K. I am talking about R=1GOhm, as used in the modern condenser microphones.
For example input capacitance listed as 30pf can go as high as 600pf in different conditions and varies significantly, depending on gm.

Best, M

R-1GOhm is not a Rg, it is a leakage resistor. Is is taken soo bieeeeaaag because we don't need to affect purely capacitive Rg adding frequency distortions, but we still need it to set a bias.

What capacitance has that nice Chinese capsule?

[Marik]

Dear Anatoliy,

I am trying to find where is our disagreement and it seems to be in your use of terminology.

Of course, as long as we deal with voltage and impedance we deal with current and power and it is impossible to have voltage without power, as it would mean we are working with infinite impedances.
Often it is hard to discriminate sharply between "voltage" and "power" amplifiers and depends on how you see things, but traditionally, the voltage amplifier is the one in which voltage gain is a dominating criteria, no matter whether you are an engineer or technician.
The same goes to the current, or power amplifiers.

Of course, you can use it as you wish, but maybe if we stick to that there would be much less mis-understanding and mis-communications, like "ya emu pro Fomu, a on mne pro Eremu".

Originally Posted by Wavebourn

Yes, sometimes we go on loss of power in order to achieve better linearity between interconnected stages, but it does not mean that it is good for the entire amplification. Loosing power, we increase the need to amplify it again, adding more stages, more noise, more distortions... I.e. decreasing dynamic range.

But likewise, the power transfer has its own flaws and is not good, either. It depends where you want to make a compromize. Regardless of your beliefs, it would not work with MOST microphones, adding even more noise and distortions, affecting frequency response, and decreasing dynamic range.

Oh, and sorry for a little correction, but sometimes it is hard to understand if you are typing "poorly", meaning "purely".

Best, M

[Wavebourn]

Originally Posted by Marik

Oh, and sorry for a little correction, but sometimes it is hard to understand if you are typing "poorly", meaning "purely".

Best, M

Unfortunately, they did not teach us in Tomsk how to communicate with the "Potential Enemy" by English words, they taught how to communicate with better quality of equipment. Thank you for the English lesson!

It is an axiom, about power amplification, regardless of what part of it we pay attention on: voltage, or current. And it works everywhere, first of all in microphone and power amplification, where we need the most of power amplification. And this parts add more distortions and noises than any other in the audio track, especially when thinking of voltage and current amplifications we loose power between stages. If we neglect Laws of Physics, regardless how big part of the industry goes this way, we get punishment from the Nature.

Let's return back to the topic. Well known Doug Self spent lot of time carefully designing power amplifiers. I've found very interesting article on his site, very close to our topic:

******//www.dself.dsl.pipex.com/ampins/discrete/ef.htm

If to look at his results we may find that character of distortions may be well obtained by a single diode, without any transistors... Now, add here a capacitance of cables, and you will get exactly "Transistor Phenomena" distortions, that we can hear both from power amplifiers of 1970'th and condencer microphones of 2000'th... Both are results of attempts to "amplify a current" using emitter folloowers loaded by complex loads.

[Marik]

Originally Posted by Wavebourn

Unfortunately, they did not teach us in Tomsk how to communicate with the "Potential Enemy" by English words...

Hahahaha!

Screw the electronics!!!
I went through the same... but only in Kemerovo.

Privet zemliak

[Wavebourn]

Originally Posted by Marik

Hahahaha!

Screw the electronics!!!
I went through the same... but only in Kemerovo.

Privet zemliak

Privet!

Me: TIASUR, 1976-1981.

0705, "Design and Manufacturing of Radio and Electronics Equipment".

You? Who taught you to use well known in AM radio emitter demodulators instead of transformers?

[Marik]

Originally Posted by Wavebourn

You? Who taught you to use well known in AM radio emitter demodulators instead of transformers?

Normally, I would not even consider replying.
Just felt like letting you know--it is harsh, arrogant, and for the record, has no any foundation--I use transformers extensively.

Now, get back to the topic.

Ciao zemliak

[ulysses]

Originally Posted by Wavebourn

May be you are right, I should use another illustration for the industry, some animals who always go after the leader.

Right, because that's me. A lemming, I think you mean. I like to feed low-impedance sources into high-impedance loads not because it's the technically sound thing to do, but because I think Bill Whitlock is a snappy dresser. And quoting my comments on the topic of microphone output interfacing in the middle of your discussion of microphone capsule interfacing makes perfect sense, so thanks for that.

[Wavebourn]

Originally Posted by Marik

Normally, I would not even consider replying.
Just felt like letting you know--it is harsh, arrogant, and for the record, has no any foundation--I use transformers extensively.

Now, get back to the topic.

Ciao zemliak

If you use transformers extensively I don't understand why you was arguing so harsh and arogant against them in favor of emitter followers that cause "transistor phenomena" sound. The whole topic is an attempt to explain you details in technical terms.

[Marik]

Originally Posted by Wavebourn

If you use transformers extensively I don't understand why you was arguing so harsh and arogant against them in favor of emitter followers that cause "transistor phenomena" sound. The whole topic is an attempt to explain you details in technical terms.

Sorry for pulling up this thread, as nobody seem to be interested in the topic. I should say here, I also have been thinking a lot about "transistor sound phenomenon" (was it Mr. Shushurin, who invented the term some 25-30 years ago and posted an article in Russian Radio magazin in late 70s, or early 80s?). However, I never found the answer (at least with microphones).

To put things straight, no, I never was arguing about superiority of EFs compare to transformers. Likewise, I'd never argue about superiority of transformers compare to EFs (despite my personal beliefs). My point (in another thread) was everything has its use, depending on application. I am for one, who believes that topologies, which from technical standpoint would be assosiated with such illusory and subjective things as "transistor sound phenomenon" despite of technical imperfections, often can be used for benefit (whatever it is).

Here is one of the tracks I made a couple years ago for a commercial CD. For the session I had quite a few mic bodies and capsules at my disposal--2 pairs of custom (mine) ribbons (one pair in fig8, and another in "true" omni patterns), three pairs of custom (mine) bodies--EF, CCS SF, and tube (inline transfomer coupled), with choice of pairs of B&K 4133, AKG CK62, Studio Projects C4 (both omnies and cardiods (just for fun)), and Gefell M94 capsules. All of the capsules had adapters to accomodate the bodies.
The polarization voltage for each pair of capsules was coming from the tube pre PSU, and was carefully adjusted with a switch for each capsule optimum voltage.
After a couple hours of miking and trying different combinations, both, me and the performer came to quite a shocking for me conclusion.
The artistic choice for a sweet spot between piano/performer/hall/type of music/mic position relationship, was a pair of Chinese SP4 omni capsules (with Jecklin disk) with emitter follower topology bodies . Go figure.

[Wavebourn]

The term was invented probably everywhere when cheap Japanese transistor amps come to market.

I don't hear unpleasant distortions on highs, the material is a bit different for them to be well heard. Piano playing piano, i.e. quiet, with low energy of high harmonics. Also, highs seems to be rolled off.
I'll post later my record when they are heard well in speech and singing. Despite the artist and his administrator were impressed by quality of sound I was dissatisfied. That time I switched off 10 dB pad on the mic (I really don't know why I did that, mic preamps in Yamaha MX-400 are quiet enough), and switched on 80 Hz roll-of on the mic.