Quick question

One extra benefit not mentioned enough - or is the common knowledge? - is that the cascoded transistor will be less prone to "funny" hot-spot modulations in high-voltage and/or high power/insufficiently cooled/hot-running transistors, because the hot-spots is moved to the the other transistor.

Glenn

The Vbe or Vgs drift will be lot less of a problem in a DC-coupled amplifier even when thermally coupling the cascoded transistor to a later opposite polarity stage.

Glenn

Meant:

The Vbe or Vgs drift will be lot less of a problem in a DC-coupled amplifier _especially_ when thermally coupling the cascoded transistor to a later opposite polarity stage.

Glenn

There is the oft-rediscovered thermal-hook supressor circuit, used in older oscilloscopes. It's a cascode with a parallel RC between the lower transistor's collector and the upper guy's emitter. The resistor is chosen to plop the lower transistor at the flat spot on its parabolic power-vs-Ic curve.

--

John Larkin         Highland Technology, Inc 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

Integrated circuit processes have different constraints than processes being used to produce discrete devices. I'm perfectly prepared to believe that manufacturers could produce tighter-tolerance devices, but what they'll actually do is produce cheaper ones, and they might be exploiting the better control to put in the same quantity of dopant faster, rather than at a better controlled concentration.

If you can wander down to the discrete production corner of your local fab I'd be interested to hear what they told you. Always assuming that they don't import all their discrete parts from an obsolete fab somewhere in Russia.

--
Bill Sloman, Sydney

Cascoding can make a significant improvement in audio power amplifier specifications. A cascode allows the use of a low voltage, fast transistor to drive the inherently slower high voltage gain transistor. This reduces the effect of non dominant poles at that stage, which allows a larger unity gain bandwidth, which allows lower distortion. The cascade also gets you higher LF gain, which reduces distortion in the loop and improves LF PSRR low frequency. Cascode current sources also improve PSRR. Additionally, eliminating the highly non-linear Cob of the gain stage also reduces distortion, more so than any Early effect non-linearity.

Whether the resulting 0.0001%, 20khz THD/IMD is actually audible, is another matter :-)

Kevin Aylward B.Sc.

- SuperSpice

Harmonic distortion seems well-tolerated by humans... it's transient intermodulation distortion (TIM) that's intensely painful in something like a Mozart wood-wind ensemble.

And most audio-phool amplifier manufacturers carefully avoid mentioning TIM. ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    | 
| San Tan Valley, AZ 85142     Skype: skypeanalog  |             | 
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  | 
| E-mail Icon at http://www.analog-innovations.com |    1962     | 
              
I love to cook with wine.     Sometimes I even put it in the food.

That is correct.

According to some article, our sensor ears make up 10% harmonic distortion.

Why electron tubes and single stage power MOSFET power amplifiers sounds so good, according to the audiophiles themselves, it is because of domination even harmonics, and it generally has lower power for higher frequencies.

Glenn

I don't believe in TIM. It was just a word used when someone designed an amp that measured poorly at HF due to slew limiting. This was because amp manufactures like to use Orwellian double speak. They quote a small signal bandwidth of 40KHz, and hope no one notices that at 20KHz it slews to only

1% of its rated output.

If an amp has

Probably the definition of TIM is at issue. Sure, x-over distortion can well be a problem at high frequencies as no loop gain, but I don't consider that (TIM) distortion caused by large amplitude high frequencies high corrupting lower frequencies. Say a square wave from a synth, intermodulating with other instruments. x-over distortion is, well... x-over distortion.

However, it trivial to get low distortion, full power, HF today, and has been for 35 years, where I did a 500W 200KHz, full power bandwidth amp :-)

Kevin Aylward B.Sc.

- SuperSpice

My first "solid state" amplifier, built in 1962, used Motorola RF transistors and had over 1MHz bandwidth. ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    | 
| San Tan Valley, AZ 85142     Skype: skypeanalog  |             | 
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  | 
| E-mail Icon at http://www.analog-innovations.com |    1962     | 
              
I love to cook with wine.     Sometimes I even put it in the food.

My recollection is that the amps which were measured out as having significant TIM problems, were ones which used high-bandwidth input sections, slow output transistors, and lots of global negative feedback. Their loop stability was marginal, and they'd behave very badly if the input section tried to make the outputs slew faster than the output devices would allow. Lots of overcorrection took place :-(

I believe I recall one writer stating that much of the blame for these products reaching the market, could be placed on audio designers who were ignorant of history and simply didn't know how to do a Bode plot and analyze loop stability and phase margin properly.

I agree, having been in the "Pro-Audio" around 30 years ago, I can state that I knew of one commercial 500 W power amplifier that every single resister and operating current, was determined by trial and error. Additionally, it originally had no main (Miller)compensation capacitor at all.

Kevin Aylward B.Sc.

- SuperSpice

That's very impressive.

Kevin Aylward B.Sc.

- SuperSpice

Some manufacturers wanted to prove how good their amplifiers are by feeding a 20 kHz square wave and showing an oscilloscope display of the output, demonstrating the 100+ kHz bandwidth at unspecified output voltage swing.

According to one explanation of TIM, if you had too much gain in the input differential pair and use slow output transistors like the

2N3055, the feedback voltage arrives so late that the differential pair is driven to the other edge, cutting off any high frequency signals, until the differential pair is in the linear region again.

A simple solution is to reduce the input differential stage gain by adding small degradation resistors to each emitter in the differential pair.

Also it helps using a simple low pass filter in front of the amplifier input to cut off frequencies above the audio range, reducing input slew rate. Unfortunately, the marketing division can no longer produce pictures of 20 kHz square waves :-)

A well-designed audio power amplifier is architecturally just an OpAmp with classic pole-splitting compensation. ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    | 
| San Tan Valley, AZ 85142     Skype: skypeanalog  |             | 
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  | 
| E-mail Icon at http://www.analog-innovations.com |    1962     | 
              
I love to cook with wine.     Sometimes I even put it in the food.

Nonsense explanation. If the system was linear at HF, it doesn't matter that the output gives delayed feedback. A large linear gain will get the feed back in time irrespective of any frequency response of the output stage. Even if the output stage BW was 1khz, it wouldn't cause TIM, assuming it can actually swing the full output at HF. What matters is non-linear behaviour.

The essential issue is the amp not having a full power bandwidth. Insufficient current to drive the output devices is possible in principle, but more usually, slewing at the main gain stage was the problem. To fix that requires both gm reduction, e.g. by the aforementioned emitter resistors *and* lowering of the compensation capacitor to get a faster slew rate, but still having the same unity gain stability BW. Resisters on their own are unlikely to do much.

Yes.

Kevin Aylward B.Sc.

- SuperSpice

Maybe, but maybe not. Speakers (especially at crossover and resonance frequencies) have impedance fluctuations that will hurt the sound, but hurt it less if the amplifier output Z is the same as speaker impedance (i.e. maximum power transfer). So, you generally want to have a higher output impedance for an audio amplifier than a fully voltage-feedback'ed op amp. Six ohms is typical.