Zibb

Paul RakoTechnical Editor Paul Rako looks at analog technology in power supplies, interface, the signal path, and life in general.



   Advertisement

Profile

RSS Feed

  • Add this blog to your RSS newsreader!

Recent Posts

Recent Comments

Most Commented On

Archives

By Category

Analog Design Articles

Blog

Monday, June 15, 2009

Audio amplifiers, class-T, class-W, class-I, class-TD and class-BS

Jun 15 2009 12:40PM | Permalink |Comments (17) |


The last few years several companies have been inventing amplifier classes, not as a legitimate architecture class, but as a marketing trick. The first I heard about was TriPath who called their class-D amplifier a class-T. Interesting to note they are out of business now. Last year a nice marketing guy from Wolfson told me about their new “class-W” amplifier in their new codec. He kept asking me what I thought of the class-W moniker that I assumed he dreamed up. He was a nice guy and Wolfson is a really excellent company, so I didn’t have the heart to tell him that I thought inventing amplifier classes as a marketing ploy only degrades the message. At least they put scare quotes around the name.  I was at Fairchild last week and they talked about a class-F amplifier. I sure hope they are referring to the real class-F amp as described in the literature and not some marketing gimmick that uses first letter in their company name. The excellent amplifier class overview in Wikipedia says that Crown dreamed up a Class-I and Lab.gruppen has fabricated class-TD. Well, I consider all these to be class-BS. It does no service to engineers to invent a plethora of amplifier classes when pretty much every possible method already has a class. You can imagine my relief when Maxim called me up last year about their new audio chips and they properly called it a class-G rather then saying it was class-M.

Marketing people love to throw around adverbs and adjectives. Part of an editors job here at EDN is to strip out all the hyperbole and just give you the bare facts so you can make up your own mind. We all remember those op-amp datasheets with “low-power” in the title and they use 12 mA. That may have been low power in 1985, but we do a lot better today. If you want your company name to be in the initials of the amplifier class go buy Allen Bradley and introduce a new class-AB amplifier. ‘Til then, the engineering community is best served by the standard classes described in the Wikipedia article. Those are:

  • Class-A            Used by audio nuts to waste power.
  • Class-B            Used by the LM324 op amp, but you can pull up the output to force it into class-A. Cheap but it has crossover distortion.
  • Class-AB            Used by most op-amps, it is a class-B with both transistors turned on a little to eliminate the crossover distortion.
  • Class-C            For RF amps, a pulse drives a resonant output stage. A church bell is a class-C amplifier, the clapper supplies a pulse excitation and the resonant system makes a sine wave come out.
  • Class-D            A switching power supply modulated at audio frequencies. Seeing a lot of use in audio amps. Texas Instruments, Maxim, Analog Devices and National Semi do great in this as well as fellow biker Skip Taylor over at D2Audio/Intersil. Getting a lot of use in LCD televisions to reduce power supply cost and conform with green power regulations. Inverters and motor drivers are essentially a class-D amp. Class-D has really crappy power supply rejection ratio (PSRR) unless it has feedback like Freescale did with its Symphony series meant for cars, where the battery voltage goes all over the place.
  • Glass-G         This is what Maxim came out with last year. To save power you feed a class-AB output stage with stepped power rail voltages. For small outputs you feed the output transistors a lower voltage. For more power and larger output waveforms you switch in one or more higher power supply voltages. This is great to save battery power in cell phones, also to drive piezoelectric speakers along with a charge-pump, one of Maxim’s cool technologies.

These are the main classes that are in use; there is also a class-I that sounds like a power converter architecture (not sure if this is the Crown invention) and class-J that combines a class-D with a class-B. The class-S is now called a class-D, see the reference website in the Wiki article. Others are:

  • Glass-H            Takes the class-G one step further and supplies a continuously varying power supply to the output transistors so the transistors never clip, but the peak voltages are just a volt or two under the instantaneous power supply. This saves power but has no harmonics and switching noise like a class-D. Class-D amps have gotten so good this class-H may never see much use in audio.
  • Class-E            A single transistor resonant circuit that is fed by an inductor like a class-C and has an output filter like a class-D. I never hear much about this class.
  • Class-F            Like class E, a resonant driver, else it would be a class C otherwise.

These classes might be used more in RF, I described the Doherty amplifier in my article about RF power amps. Now if Jim Williams or Bob Pease or Dave Freeman or Barry Harvey or Barry Gilbert calls me up all excited at four in the morning and tells me that they have invented a new amplifier class, I will be the first to tell you about it, but if it is a marketing person, well I have a very big grain of salt on my nightstand to take along with any claim of a new amplifier architecture class.


Related entries in: Analog | 


Reader Comments



at 6/15/2009 3:40:52 PM, ken said:
Paul,
It sounds like you're crediting Maxim with inventing class G.
They didn't.
It's donkey's years old.
Second generation DSL line drivers used it, and I'm sure it was in use before that but I'm too lazy to search the web.

kc



at 6/15/2009 4:02:26 PM, Paul Rako said:
Sorry for the confusion, yup, the wiki article and the references talks about class-G ages ago. Texas Instruments makes the same kind of parts. Sorry, I just meant to credit Maxim with calling a spade a spade instead of a "sub-soil retrenchment implement" like marketing people like to do. You are right that class-G goes back a long way.



at 6/16/2009 2:26:26 PM, Brad Wood said:
This reminded me of the dialogue in Back to School, where Thornton Mellon (Rodney Dangerfield) is trying to date his English teacher, Diane (Sally Kellerman). After being told she has class each of the nights he suggests, he says, "I'll tell you what, then. Why don't you call me some time when you have no class? "




at 6/16/2009 4:30:58 PM, JDB said:
Hi Paul

"class-H may never see much use in audio" ..... I would add the word "again". Class H became quite popular during the mid 80s with amps like the Crest 8001, leading to 5kW per channel amps like the 10,001. It continues to be used in high power audio to date, but you are correct, Class D will eventually replace it.

I believe Class G originated around 1976 compliments of Hitachi.



at 6/16/2009 5:42:28 PM, Old Geezer said:
Bob Carver used a form of class "G" many decades ago in his high power but incredibly small audio power amps. I think they were marketed under the Carver Audio brand. They modulated the power supply rails through a magnetic amplifier topology.



at 6/16/2009 8:27:55 PM, STSB said:
There are many new classification of audio power amp.
However, Class-AB is still the most practicle amp for quality + energy efficient balanced circuit.

Ever heard of Class-SB (by ST micro) and Class-KB (by Toshiba) ?? They are the improved energy efficient version of class-AB. Any comments about these amps??



at 6/17/2009 6:56:59 AM, arclight said:
Paul: Class A is also used in receiver front ends to maximize IP3.

I commend you on getting into the spirit of "change" advocated by the current occupant of the White House: You are "declassifying" things (snicker groan eye-roll).





at 6/17/2009 8:18:59 AM, JDB said:
At least all of us posting can produce valid birth certificates.



at 6/17/2009 11:26:28 AM, Brad Wood said:
About Crown: although the blurb on their website is drastically in need of editing (did the writer never hear of apostrophes?), Gerald Stanley's BCA invention is not simply a marketing ploy (even if dubbing it Class I may have been). The topology uses opposed polarity buck converters that "T" into the load and a modulation scheme that is the same as the "filterless" class D parts. The result is a doubling of the ripple frequency (as well as easy migration to multiphase operation with more output satgaes and inductors) and an avoidance of deadtime issues and their attendant shoot-through current issues endemic to conventional class D. I'd love to use it for someone besides Harman, but I suspect they would vigorously enforce their patents!



at 6/17/2009 12:05:12 PM, M. Simon said:
If you did much high power RF work you would be familiar with class E. High power short wave broadcasters insist on it.



at 6/17/2009 2:49:22 PM, Mike Fry said:
Paul, Instruments, Inc. makes two kinds of switchmode power amplifiers, and we call them Class D and Class S. The S is for square wave or Stairstep. We use Class S when the customer's frequency is too high for Class D. Then we control the amplitude and duration of squarewaves to meet their power requirement.

You can read the tutorial on our website: instrumentsinc.com



at 6/18/2009 8:16:20 AM, Gerald Stanley said:
The point is well taken that there is much confusion in the assigning of classes to audio amplifiers. There is not and never has been an official body assigning classes to power converters. What there has been is a body of literature setting precedents (IRE and IEEE) and a tendency to use letters consecutively to highlight substantive differences in approaches.

Classes A, B, AB and C are rarely confused. Class D is often assigned blindly to anything with switches. It is most appropriately assigned to uses of the common half bridge with requisite time alternating switching. Classes E & F are resonant forms of switching power that are well documented and in widespread use in the RF design community.

Class G is a voltage-segmented quasi-linear dissipative technology where the dissipative member in use is a function of the signal level. Class H has articulated rails that are switched between discrete levels to maintain operating voltage across a typically class AB output stage. Class H has recently been confused with rail tracking amplifiers that use PWM methods to continuously vary their rails. Unfortunately the latter form of amplifier was not initially given a class designation by Sony (Hamada 1976) and this has been a frustration to others such as LabGruppen that have reasonably sought a class name for a technology that did not name its child until it was of adult age.

Class-I is the next class in alphabetical sequence and is short for class Interleave. Being the most interleaved half bridge theoretically possible and fundamentally different in switch activation than class-D it needs a name not given by some capricious thing such as the name of a company or inventer.

It seems that class-D has morphed into class-"you fill in the blank" or class-UFiB. This is unfortunate as no one needs this form of class warfare.



at 6/19/2009 5:04:34 AM, Brad Wood said:
Thanks for that clarification on Class I Gerald. It had not occurred to me that it was in alphabetical historic succession, and at that, had the happy coincidentally semantic association with "interleave".



at 6/23/2009 2:40:39 PM, Mike said:
Regarding "Class D" amplifiers you should check out this "White Paper" at: www.extron.com/download/files/whitepaper/class_d_ampl_wp.pdf and see what Extron is doing with their Class D Ripple suppression technology.



at 6/24/2009 10:22:19 AM, Frank Whiteside said:
I think there is a more basic issue at work here:
Amplifier class originally referred to just the output stage (dating back well into vacuum tube days) -- whethere switched or linear, and whether the stage switched at voltage nodes, current nodes, etc.

Now with integrated amplifiers it is including more and more of the overall system, including modulation scheme, feedback tricks, etc.
Seems to me these are more worthy of subclass designations, like the old AB1-vs-AB2 designation (which was specific to tubes and is no longer relevant).



at 6/24/2009 4:48:14 PM, WKetel2 said:
To Frank Whiteside: Perhaps you consider the distinctions between tube operating conditions as no longer relevant, but you have not looked very far. It turns out that in the HF amateur radio service it is indeed quite relevant. A class AB1 amplifier requires a fairly constant drive power throughout the cycle, since it is not driven into grid current. Class AB2 amplifiers wind up having a very non-linear input impedance, and so require much more power for part of the input cycle waveform than for the other part. The result is that they have the least distortion when the driver circuit includes a HI-Q resonant circuit, which is able to provide the required drive power for the portion of the cycle where grid current flows. So class of operation is still important to those who use tubes.



at 6/25/2009 9:05:38 AM, Frank Whiteside said:
@WKetel2 -- I stand corrected. I should've stopped at "specific to tubes" :)

Post a comment



Display Name

Change Image
Before submitting this form, please type the characters displayed above.
Note the letters are NOT case sensitive.


ADVERTISEMENT

©1997-2009 Reed Business Information, a division of Reed Elsevier Inc. All rights reserved.
Use of this Web site is subject to its Terms of Use | Privacy Policy

Please visit these other Reed Business sites