Capacitance multiplier: which Darlington to choose?

[JohnRoberts]

Yet they all started out subjectively, or as marketing gimmick (eg. THD&N as quality metric, that is on Harold J. Leak who just changed the way he qualified rather generic Mullard circuits as "point one" series claiming his Amplifiers had .1% THD while others claimed 3% or 10% THD for the exactly identical circuits).

aka marketing... I have worn that hat before too...

I spoke very generally, it applies as much to hardcore subjectivists as to objectivists and applying also outside audio. I did NOT target any specific individual, but more like...

If we are honest, every microphone, amplifier and speaker is to some degree an "effect".

being honest (or trying to be) , microphones and loudspeakers are not very precise....

amplifiers OTOH operated within their linear range of operation (not clipped, or otherwise limiting), can be very transparent, especially compared to microphones and loudspeakers. 2 out of 3 ain't bad.

What is the point of making a 200W/8R Amplifier with 0.00001% THD&N at half power when it drives a speaker with 0.1% THD&N at 1 Watt input?

Thor

Selling amps is the goal, the customer is always right, even when horribly misguided and chasing phoolish specs...

JR

 

[abbey road d enfer]

Therein lies the rub. Mind you, compared to 40 Years ago we have a much better understanding of how human hearing works

Do I know that. When I was the Neutrik distributor, I was invired to a seminar given when Neutrik absorbed a promising startup that dealt with psychoacoustic measurement systems, which are now part of NTI. This startup was founded by students of Zwicker and Fastl. For about 10 years I was liaising between NTI and the CCFA (french automobile manufacturers comittee).
It never occured that the evaluation provided by these pseudo-subjective measurements gave an indication on how to fix any issue.
I don't think it's a case of focusing too much on objective mesurements, it's just that there is no direct bridge between subjective measurements and engineering.

 

[abbey road d enfer]

What is the point of making a 200W/8R Amplifier with 0.00001% THD&N at half power when it drives a speaker with 0.1% THD&N at 1 Watt input?

That is an old chestnut. Quite often, a person with a half-decent audition training detects the difference between loudspeaker distortion and electronics distortion.

 

[ruffrecords]

Yet they all started out subjectively, or as marketing gimmick (eg. THD&N as quality metric, that is on Harold J. Leak who just changed the way he qualified rather generic Mullard circuits as "point one" series claiming his Amplifiers had .1% THD while others claimed 3% or 10% THD for the exactly identical circuits).



Thor

That sounds like urban myth to me. Can you provide evidence?

Cheers

Ian

 

[thor.zmt]

It never occured that the evaluation provided by these pseudo-subjective measurements gave an indication on how to fix any issue.

No measurement gives an indication how to fix an issue, it merely shows that there is an issue.

I don't think it's a case of focusing too much on objective mesurements, it's just that there is no direct bridge between subjective measurements and engineering.

YET the equipment being engineered will ultimately be judged subjectively.

amplifiers OTOH operated within their linear range of operation (not clipped, or otherwise limiting), can be very transparent, especially compared to microphones and loudspeakers.

I am not sure what "transparent" means in your use. Does it mean "low THD&N" or "audibly transparent". As Earl Geddes (among others) showed, relatively low measured distortion (0.01% THD with a full scale signal) does not guarantee audible transparency (in fact the distortion was highly audible and objectionable) while relatively high measured distortion (nearly 10% THD with a full scale signal) does not reliable cause a lack of audible transparency.

http://www.gedlee.com/Papers/The Perception of Distortion.pdf

Personally, for example, I found that a system using non-feedback single ended tube circuitry with ~ 0.3% THD @ 1W/8R and 3% THD @ 10W/8R driving a 97dB/2.83V sensitive 16 Ohm speaker appears subjectively transparent with a huge range of music, more-so than many highly regarded solid state amplifiers with very low measured distortion.

Of course, if we listen to music with decent dynamic range and to a stereo pair of speakers in a "normal" setup with ~105dB Peak SPL (THX playback level setting) most of the music will be happening 18dB below 10 Watt with 3% THD at around 87dB average SPL (which is quite loud really) at ~ 0.16 Watt or lower, where the THD of the Amplifier will be ~ .1% or lower and thus pretty much reliably inaudible, because of auditory masking, as HD of this amplifier is pretty much only H2 & H3.

So in this system context, a horribly distorted and broken by design Amplifier (objectivist view) nevertheless passes the psychoacoustic requirements for audible transparency (truly objective view accounting for context and human hearing) and "sounds realistic, dynamic and engaging, more-so than many solid state amplifiers" (subjectivist view).

Surely it would seem that to allow such a dichotomy to exist and persist is highly undesirable and that it should be resolved?

Thor

 

[thor.zmt]

That sounds like urban myth to me. Can you provide evidence?

The schematics of the .1 series are standard Mullard, the same as used other british marques. They have the same distortion as other amplifiers using the same design. There is nothing in the .1 series that makes it lower distortion than other products on the market using the same Mullard designs.


What is different is that nominal output levels / power are lower than Mullard Spec and Distortion is also lower. The Mullard 5-10 actually was rated at 14 Watt & 1% THD, Leak TL12 which is essentially identical was rated at 10 Watt & 0.1% THD.

And this low distortion (which is the same as any other product using the same design) is advertised.



I am unaware of any formal advertising of Amplifiers with "low distortion" prior to Leak. If you have an example I'd be interested to see it.

HJ Leak was quite a piece of work actually, you might remember how he stole Dinsdale's solid state amplifier design and his reaction on being called out on it...

http://ukhhsoc.torrens.org/makers/Leak/Dinsdale_HFNLetter.txt

"HJL then looked me straight in the eye and added 'I have far more money than you, and if you rock my boat I will destroy you.'"

HJL was a great marketeer, not much of an engineer. I see people still fall for it.

Thor

 

[thor.zmt]

That is an old chestnut. Quite often, a person with a half-decent audition training detects the difference between loudspeaker distortion and electronics distortion.

That surely depends entirely on actual distortion spectrum...

With both speakers and amplifiers below 1% of H2 and 0.1% H3 and little else at SPL's in the upper 80dB range, can anyone distinguish if the distortion they are hearing are the 30% H2 generated in the ear, or the speaker or the Amp?

Thor

 

[JohnRoberts]

I am not sure what "transparent" means in your use. Does it mean "low THD&N" or "audibly transparent".

audibly transparent or neutral... not contributing a sound coloration of it's own.

As Earl Geddes (among others) showed, relatively low measured distortion (0.01% THD with a full scale signal) does not guarantee audible transparency (in fact the distortion was highly audible and objectionable) while relatively high measured distortion (nearly 10% THD with a full scale signal) does not reliable cause a lack of audible transparency.

a tired old straw man... the type of distortion matters to its relative audibility. Small numerical amounts of crossover distortion can be very audible while low order harmonic distortion not so much, duh... amplifier circuit design 101.

Personally, for example, I found that a system using non-feedback single ended tube circuitry with ~ 0.3% THD @ 1W/8R and 3% THD @ 10W/8R driving a 97dB/2.83V sensitive 16 Ohm speaker appears subjectively transparent with a huge range of music, more-so than many highly regarded solid state amplifiers with very low measured distortion.

Of course, if we listen to music with decent dynamic range and to a stereo pair of speakers in a "normal" setup with ~105dB Peak SPL (THX playback level setting) most of the music will be happening 18dB below 10 Watt with 3% THD at around 87dB average SPL (which is quite loud really) at ~ 0.16 Watt or lower, where the THD of the Amplifier will be ~ .1% or lower and thus pretty much reliably inaudible, because of auditory masking, as HD of this amplifier is pretty much only H2 & H3.

So in this system context, a horribly distorted and broken by design Amplifier (objectivist view) nevertheless passes the psychoacoustic requirements for audible transparency (truly objective view accounting for context and human hearing) and"sounds realistic, dynamic and engaging, more-so than many solid state amplifiers (subjectivist view).

Surely it would seem that to allow such a dichotomy to exist and persist is highly undesirable and that it should be resolved?

Thor

Sounds like you are hard at work slaying these imaginary dragons.

JR

 

[thor.zmt]

audibly transparent or neutral... not contributing a sound coloration of it's own.

Ok, we pretty much know what that takes, at ~ 90dB SPL 200Hz -4kHz appx. > .3% H2 and > .1% H3, > .03% H4 etc.... & > -90dB (< 0.003% ) H10 and up for HD, higher levels are permitted at higher and lower frequencies.

Noise that is lower than the background noise in a quiet room (say -80dB SNR @ ~ 90dB.

Freedom from dynamic distortion such as thermal memory, TIM/SIM including anything potentially not formally qualified yet, including being presented with relatively large amounts of ultrasonic broadband noise.

a tired old straw man... the type of distortion matters to its relative audibility.

Not as tired as THD. You just confirmed that THD is completely useless. Yet it continues to be used as figure of merit, never mind the Kult of SINAD at Audiocargocultsciencereview.

Sounds like you are hard at work slaying these imaginary dragons.

The "drag on" (s).are hardly imaginary. Further up Abbey Road bemons the lack of actionable test results to provide "good sound" and notes that he instead selected to focus on objective performance.

Since Olson and Shorter of the BBC in the 50's we have seen well researched, well founded attempts to replace THD with a metric that actually has a meaningful correlation with how audible distortion is and the reaction is "tired oldcstrawman, let's just keep using THD instead".

I mean for real? Seriously?

Well, it's spring. I'm off poisoning pigeons in the park.



Thor

 

[thor.zmt]

Single ended and Push pull Tube Power and preamps. The Design of class A SET or SEP is very old indeed, but still offers lowest distortion Sound. Modern Shindo hasnt much else in Common with those old dinosaurs.

Walter, it seems you removed a post where you contrasted modern tube sound which broadly correlates with objective spec driven tube studio sound to "dampfradio" sound.

Dampfradio = Steamradio, a colloquial German term for those lovely large wooden Boxes that contained many drivers and more formally are known "GrossSuper" (Big Super [Radio]) in German.
The one we had had a 10" woofer forward and a pair of 8" full range drivers mounted in the sides with a bevy of 4" cone tweeters pointing pretty much everywhere and and a PCL86 based PP Amp for the woofer and PCL86 SE Amplifiers for the rest.

It was "Allstrom", meaning it operated without mains transformer and my first ever DIY electronics project at age 7 or 8 was to find a way to connect a modern stereo turntable with build in transistor amp and dinky terrible speakers, so I could listen to the LP's I borrowed from the local library (like Beatles, Jethro Tull, Vangelis, Tangerine Dream, Tomita and so on) in glorious "DampfRadio Klang".

The whole job was quite involved, an EE uncle helped me, the unit was recapped, converted to stereo (it already had two ECL86 Amp's for the Full Range) a stereo volume control with 3 Taps and physiological Volume control and a "wide" stereo function was added. We even added a FM stereo decoder (already IC based at the time, east german clone of Motorola MC1310A) and naturally the isolation capacitors and "Diodenbuchse" for crystal Pickup in the record player.

All that made that old DampfRadio sitting in the corner fit for the 80's and I ever so much discovered music on this setup.

Now what made "studio gear" different from my old DampfRadio? Not the basic circuitry. That all remained the same with minimal changes. The difference was the Iron. The low cost commodity transformers in old Radio's used thick lams, low grade steel and undersized cores with extremely simple (often just 3 section) winding schemes. Other parts were also quite "commodity".

Studio gear (Power Amp) at the same time used ultrathin high grade nickel steel wound into a "schnittbandkern" (appx. a double-c-core) and used a hugely oversized core with airgap even for PP and very com plex sectioning for windings and gen erally much higher grade passive parts.

Dedicated "HiFi" tube gear was generally in-between but cost you. Mind you, studio gear was xing desk system insanely expensive. Even the 1980's transistors + IC's cassettes for the east german modular mixing desk systems I helped design towards the late 80's.

So the studio gear usually simply had better iron, passive parts and power supplies than consumer radios, but little else in actual schematic differences.

Thor

 

[JohnRoberts]

Ok, we pretty much know what that takes, at ~ 90dB SPL 200Hz -4kHz appx. > .3% H2 and > .1% H3, > .03% H4 etc.... & > -90dB (< 0.003% ) H10 and up for HD, higher levels are permitted at higher and lower frequencies.

Noise that is lower than the background noise in a quiet room (say -80dB SNR @ ~ 90dB.

Freedom from dynamic distortion such as thermal memory, TIM/SIM including anything potentially not formally qualified yet, including being presented with relatively large amounts of ultrasonic broadband noise.

We agree far more than disagree

Not as tired as THD. You just confirmed that THD is completely useless. Yet it continues to be used as figure of merit, never mind the Kult of SINAD at Audiocargocultsciencereview.

THD was adequate for characterizing magnetic tape path linearity. SMPTE IMD was adequate for back when 7kHz was considered a HF (long ago).

I discovered the limitations of simple THD+n back in the 70s designing phono preamps. The RIAA EQ is effectively a low pass filter that attenuates HF harmonic distortion overtones helping preamps measure cleaner than they sound. SMPTE IMD with only 7kHz was not fast enough to stress phono preamp designs to reveal their flaws. Since I could not afford SOTA bench test equipment, I rolled my own. I modified my cheap Heathkit SMPTE IMD analyzer to instead operate two tone IMD using 19kHz and 20kHz 1:1. This was very revealing for RIAA phono preamp design because the distortion product was no longer getting rolled off by the RIAA EQ LPF.

The "drag on" (s).are hardly imaginary. Further up Abbey Road bemons the lack of actionable test results to provide "good sound" and notes that he instead selected to focus on objective performance.

In my judgement this "good sound" metric is still subjective so different listeners will be focussed on different sonic characteristics. Some may even prefer more euphonic distortions.

Since Olson and Shorter of the BBC in the 50's we have seen well researched, well founded attempts to replace THD with a metric that actually has a meaningful correlation with how audible distortion is and the reaction is "tired oldcstrawman, let's just keep using THD instead". I mean for real? Seriously?

I have seen plenty of these "new" distortion metrics promising to zero in on some intangible sonic flaw. They came and left because at the end of the day it comes down to simple linearity (output is same as input just amplified). Of course with your experience designing power amps you know there are plenty of other ways to drop the ball. Several are involved with recovery from overload conditions (clipping, current limiting, slew limiting, etc).

Well, it's spring. I'm off poisoning pigeons in the park.



Thor

Almost time for me to start shooting squirrels up in my pecan trees, coming soon.

JR

 

[fred dieckmann]

A Mosfet has a much higher transconductance than a BJT

????????????

 

[JohnRoberts]

????????????

GM= ID/Vgs

JR

 

[fred dieckmann]

I know what transconductance is (BSEE and analog designer) and am curious why someone would say that mosfets have greater transconductance than BJTs under similar operating currents. Just the opposite is the case in my experience. Mosfets require a high drain current to reach there highest Gm point which puts them at further disadvantage for a lightly loaded follower applications. Good PSRR ratio is an advantage many fet followers have over BJTs and ease in finding devices with Vds of even 1200 V dc. I have watched so many BJTs pop in high voltage regulators for tube high end audio preamps and amps that I wouldn't use them for that reason alone. Always use gate stoppers on Mosfets and mount them as close to the gate as possible. I will avoid glib responses in the future but was mystified by this statement.

 

[Matador]

I know what transconductance is (BSEE and analog designer) and am curious why someone would say that mosfets have greater transconductance than BJTs under similar operating currents.

I think maybe it applies in this particular instance, in the case where high voltage BJT's are considered (and their relatively low current gains)? The TIP50 for example (400V), running a few hundred milliamps has a gm of about 5-10. This is comparable with an IRF830 (500V), so long as the VGS voltage >> VT (and other conditions apply). I don't think this would apply when considering only small-signal operation.

 

[JohnRoberts]

I know what transconductance is (BSEE and analog designer) and am curious why someone would say that mosfets have greater transconductance than BJTs under similar operating currents. Just the opposite is the case in my experience. Mosfets require a high drain current to reach there highest Gm point which puts them at further disadvantage for a lightly loaded follower applications. Good PSRR ratio is an advantage many fet followers have over BJTs and ease in finding devices with Vds of even 1200 V dc. I have watched so many BJTs pop in high voltage regulators for tube high end audio preamps and amps that I wouldn't use them for that reason alone. Always use gate stoppers on Mosfets and mount them as close to the gate as possible. I will avoid glib responses in the future but was mystified by this statement.

Could be a mistake (brain fart?), I am pretty sure he knows that (maybe we can let him sleep).

I recall a metric from back in the day called Vth referring to FET op amp inputs that benefited from lower transconductance in their LTPs to enjoy higher slew rate. Higher Vth correlates with lower transconductance. Vth characterized the input voltage before the LTP saturated (likely slew limit). BiFets delivered several volt Vth numbers, while bipolar input op amps Vth were tens of mV. Some bipolar op amps added degeneration resistors in series with the emitters of the LTP to reduce transconductance, but those resistors introduced new problems.

Sorry for the veer but that life on the internets...

JR

 

[thor.zmt]

Always use gate stoppers on Mosfets and mount them as close to the gate as possible.

I said that. And the reason is?

I will avoid glib responses in the future but was mystified by this statement.

A simple BJT commonly has lower in circuit Gm than vertical Mosfets, IRF830 is Vertical Mos. Lateral Mosfet's are a different story.

Together with the parasitic capacitances and inductances it makes a better oscillator than a BJT.

Thor

 

[Walter66]

In

Walter, it seems you removed a post where you contrasted modern tube sound which broadly correlates with objective spec driven tube studio sound to "dampfradio" sound.

Dampfradio = Steamradio, a colloquial German term for those lovely large wooden Boxes that contained many drivers and more formally are known "GrossSuper" (Big Super [Radio]) in German.
The one we had had a 10" woofer forward and a pair of 8" full range drivers mounted in the sides with a bevy of 4" cone tweeters pointing pretty much everywhere and and a PCL86 based PP Amp for the woofer and PCL86 SE Amplifiers for the rest.

It was "Allstrom", meaning it operated without mains transformer and my first ever DIY electronics project at age 7 or 8 was to find a way to connect a modern stereo turntable with build in transistor amp and dinky terrible speakers, so I could listen to the LP's I borrowed from the local library (like Beatles, Jethro Tull, Vangelis, Tangerine Dream, Tomita and so on) in glorious "DampfRadio Klang".

The whole job was quite involved, an EE uncle helped me, the unit was recapped, converted to stereo (it already had two ECL86 Amp's for the Full Range) a stereo volume control with 3 Taps and physiological Volume control and a "wide" stereo function was added. We even added a FM stereo decoder (already IC based at the time, east german clone of Motorola MC1310A) and naturally the isolation capacitors and "Diodenbuchse" for crystal Pickup in the record player.

All that made that old DampfRadio sitting in the corner fit for the 80's and I ever so much discovered music on this setup.

Now what made "studio gear" different from my old DampfRadio? Not the basic circuitry. That all remained the same with minimal changes. The difference was the Iron. The low cost commodity transformers in old Radio's used thick lams, low grade steel and undersized cores with extremely simple (often just 3 section) winding schemes. Other parts were also quite "commodity".

Studio gear (Power Amp) at the same time used ultrathin high grade nickel steel wound into a "schnittbandkern" (appx. a double-c-core) and used a hugely oversized core with airgap even for PP and very com plex sectioning for windings and gen erally much higher grade passive parts.

Dedicated "HiFi" tube gear was generally in-between but cost you. Mind you, studio gear was xing desk system insanely expensive. Even the 1980's transistors + IC's cassettes for the east german modular mixing desk systems I helped design towards the late 80's.

So the studio gear usually simply had better iron, passive parts and power supplies than consumer radios, but little else in actual schematic differences.

Thor

I've studied those pro- tube electronics after retirement and came to the conclusion, that often they cut the technological grade which was achivable in sound quality at the time of those designs. Just look at tab V76, the ultimate of studio tube preamp. Very low noise figures due to big anode choke loaded output tube designs, the whole unit uses many of those choke transformers and signal transformers, the PSU rather simple but good enough for those fantastic measurements. This real sounds much different compared to any steam powered tube radio we found in the past for nothing on the streets. And the iron, just tasty. My Klangfilm mono cinema amp system, just the best iron available and mostly Bosch MKP PSU caps combined with other ultra long life parts. And this all leads to a sound quality, clear as a blue sky. No euphonic sound, no weak bass response, ultra fast, ultra sharp audio reproduction. Studer C37 the same, ECC88 cascoded designs, tube regulated PSU, output trans, the best of that time. The best ever?
But there is a proceeding in tube circuit design that can be seen with gear like Shindo today. Modern PSU combined with old tube technology can lead to phantastic results, too. Have to experiment with the suggested circuits in this thread. Thanks for your help, much appreciated.

 

[JohnRoberts]

A simple BJT commonly has lower in circuit Gm than vertical Mosfets, IRF830 is Vertical Mos. Lateral Mosfet's are a different story.

Vertical mosfets AFAIK were designed to switch on/off crisply for use in switching PS (or class D). Lateral MOSFETs like the Hitachi devices were briefly popular in linear audio amplifiers. (Hitachi published a white paper promoting their class H multi rail approach to realize higher efficiency).

Together with the parasitic capacitances and inductances it makes a better oscillator than a BJT.

Thor

I wasted some time trying to make vertical mosfets behave in an audio amplifier and gave up (it really did like to oscillate). It was possible to trick a vertical mosfet combined with a BJT to behave like a compound BJT but I did not pursue that.

JR

 

[thor.zmt]

Vertical mosfets AFAIK were designed to switch on/off crisply for use in switching PS

Yup, Steilheit/Transconductance is very high as result. They can be used in linear amplifiers, but need different circuits and approaches than lateral FET's.

Lateral MOSFETs like the Hitachi devices were briefly popular in linear audio amplifiers.

I still like them. But they are relatively expensive. My personal favourite is a "Bimos" output that uses the 10A Mosfet's as drivers for banks of modern bipolar outputs.

Bias the MOSFET drivers just a smidgen above Zero tempco current (~150mA) and keep them on the main heatsink. Add a 2Vbe multiplier and use a resistor to set quiescent current.

Overall the tempco is mildly negative (minimal slope).

No need for more drivers, very VAS friendly load.

I wasted some time trying to make vertical mosfets behave in an audio amplifier and gave up (it really did like to oscillate). It was possible to trick a vertical mosfet combined with a BJT to behave like a compound BJT but I did not pursue that.

There are many interesting options, but for audio lateral Mosfet's are preferred on my experience.

Thor