Forums

Class D commercial audio amp

Started by N_Cook November 29, 2016
On Wed, 30 Nov 2016 08:51:05 -0500, bitrex
<bitrex@de.lete.earthlink.net> wrote:

>On 11/30/2016 07:00 AM, N_Cook wrote: >> >> Should 1.5v remnant 200KHz stay constant with whatever frequency or >> amplitude of audio going in? Even at zero audio input there is this 1.5V >> 200KHz. > >You know how PWM works, yeah? > >AFAIK unless the amp has some fashion of automatic shutdown after the >audio signal is removed for a time, the modulator and output stage will >always be active. > >With no input signal the output waveform will just be a 50% duty cycle >square wave at 200kHz, so yes I imagine you'd still be seeing the >residual at the load terminal.
There are topologies in some commercial designs that do not present a continuous conversion frequency ripple amplitude. Ternary modulator topologies generally don't - one type being employed in Crown/Amcron SMPAs is an example. This is invisible as far as audio performance evaluation is concerned, but has advantages in efficiency, weight and emc. Some implementations require more parts. RL
>"With no input signal the output waveform will just be a 50% duty cycle
square wave at 200kHz, so yes I imagine you'd still be seeing the residual at the load terminal. " I read somewhere about another type of class D that did not have the constant 50 % duty cycle. This scheme reduces the duty cycle at low levels and runs like a low bias. Seems to me this is not easy to implement ad I don't know where they are implementing it, but it does seem like it would be more efficient. In other words, a quiescence the duty cycle might only be 1 % positive and negative. When a positive signal comes it increases the positive duty cycle and the same for negative. Seems like it would even be more efficient than Crown's circuit which uses two output chokes per channel and the damping diodes dump back into the other leg of the power rails. I am not so sure that patent is any good though because I have seen other class D amps with two coils per channel. However those could have been two stage filters, I do not know. I rarely work on class D amps because of the possibility of wasting silicon. I do not have the right equipment to cold test it. If I did, I would but the employer is not going to buy the stuff and I am not buying it for him. Actually, because of the nature of the beast I think many problems in class D amps are caused by bad solder connections. high frequencies and currents destroy the best of solder, even that silver bearing Tektronix stuff. When you got the lead free stuff it is much much worse. And then it is almost impossible to tell when a connection is bad by looking at it, I find it better to actually troubleshoot it down. Resoldering everything is a bad option. But bad connections to the base or gate of switchers will kill them. To the other terminals the heat generated might do them in. I almost want to say to resolder them periodically but most stuff today is built in a pan, meaning there is no access to the bottom without removing the whole board. Then frequently testing is impossible. You might see it work a little bit but with no fan you can't do a high power check or any of that. Eventually it will all be like toasters. We used to tell people about VCRs that it is not a toaster, but the price made it so. Right now B&O is charging good money for those Icepower modules but in time as other people start building them the price will come down. They want like $100 for the fifty watt version, hundred watts into four ohms. The thing is smaller than a 1960s transistor radio, and I mean AM only, littlest thing out there. Two chips, two coils, one transformer, three transistors and a few twelve cent components. Someone is going to take them out of the game sooner or later.
On 11/30/2016 09:20 AM, jurb6006@gmail.com wrote:
>> "With no input signal the output waveform will just be a 50% duty cycle > square wave at 200kHz, so yes I imagine you'd still be seeing the > residual at the load terminal. " > > I read somewhere about another type of class D that did not have the constant 50 % duty cycle. This scheme reduces the duty cycle at low levels and runs like a low bias. > > Seems to me this is not easy to implement ad I don't know where they are implementing it, but it does seem like it would be more efficient.
Interesting, yeah it seems like they'd have thrown some clever techniques at the problem by now, particularly in the higher-end stuff. Once you add a DSP to the mix, rather than doing the modulation stage in analog, that probably opens up a lot of possibilities. Sample the input waveform into memory, and then you basically have a DAC driving two switches. At that point I imagine if one were clever (more clever than I am at least) you could do all sorts of DSP hacks to create a "botique" switching waveform which was still perfectly reconstructible, but was otherwise optimized to minimize stuff like no-signal output level, harmonic rejection, spread spectrum, etc.
On 11/30/2016 09:38 AM, bitrex wrote:
> On 11/30/2016 09:20 AM, jurb6006@gmail.com wrote: >>> "With no input signal the output waveform will just be a 50% duty cycle >> square wave at 200kHz, so yes I imagine you'd still be seeing the >> residual at the load terminal. " >> >> I read somewhere about another type of class D that did not have the >> constant 50 % duty cycle. This scheme reduces the duty cycle at low >> levels and runs like a low bias. >> >> Seems to me this is not easy to implement ad I don't know where they >> are implementing it, but it does seem like it would be more efficient. > > Interesting, yeah it seems like they'd have thrown some clever > techniques at the problem by now, particularly in the higher-end stuff. > > Once you add a DSP to the mix, rather than doing the modulation stage in > analog, that probably opens up a lot of possibilities. Sample the input > waveform into memory, and then you basically have a DAC driving two > switches. At that point I imagine if one were clever (more clever than I > am at least) you could do all sorts of DSP hacks to create a "botique" > switching waveform which was still perfectly reconstructible, but was > otherwise optimized to minimize stuff like no-signal output level, > harmonic rejection, spread spectrum, etc. >
Don Lancaster (who used to post here) had a series of articles on "magic sinewaves", where through some rather challenging mathematical optimizations you could create a PWM representation of a sine wave where you could force many of the non-relevant harmonics to zero through careful selection of the switch event timing.
On 11/30/2016 07:00 AM, N_Cook wrote:
> > Should 1.5v remnant 200KHz stay constant with whatever frequency or > amplitude of audio going in? Even at zero audio input there is this 1.5V > 200KHz.
It'll probably be at its worst near zero output, because the switcher is producing a symmetrical square wave. At nonzero outputs, the amount of fundamental in the output goes down, and the second harmonic goes up. Cheers Phil Hobbs -- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics 160 North State Road #203 Briarcliff Manor NY 10510 hobbs at electrooptical dot net http://electrooptical.net
On Wed, 30 Nov 2016 10:41:59 -0500, Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:

>On 11/30/2016 07:00 AM, N_Cook wrote: >> >> Should 1.5v remnant 200KHz stay constant with whatever frequency or >> amplitude of audio going in? Even at zero audio input there is this 1.5V >> 200KHz. > >It'll probably be at its worst near zero output, because the switcher is >producing a symmetrical square wave. At nonzero outputs, the amount of >fundamental in the output goes down, and the second harmonic goes up. > >Cheers > >Phil Hobbs
What does that non-zero remnant do to the speaker... produce some heat? ...Jim Thompson -- | James E.Thompson | mens | | Analog Innovations | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | STV, Queen Creek, AZ 85142 Skype: skypeanalog | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at http://www.analog-innovations.com | 1962 | There is no fool quite like an audiphool.
On Wed, 30 Nov 2016 04:13:42 -0800, Phil Allison wrote:

> Rob wrote: > > >> That does not have anything to do with the achievable quality, doesn't >> it? >> >> Today, Class-D is not only used for small amps in a consumer device, >> it is used in high-power stage amplfiers as well, and the high-end >> audio people love them. >> >> > ** That has been true for over 20 years now. > > The really ground breaking class D amps were the K1 and K2 models from > Crown - released in the mid 1990s. When bench tested as a "black box" it > was genuinely hard to tell you were dealing with anything unusual. The > only give away was the low heat production so the amp needed no fan. > > http://adn.harmanpro.com/site_elements/resources/995_1425481007/K1-K2-
Data-Sheet-136713_original.pdf
> > THD measures about 0.03% at rated power, no switching noise is evident > at the output terminals and the DF is over 3000 at low and mid > frequencies. Plus the case is hermetically sealed so no dust ever gets > inside. > > Widely used and still used for pro-audio and home theatre applications. > > > > .... Phil
Sound companies in the US seem to favor the L-Acoustics V-DOSC Line Array system, which is all digital and cannot be matched for sound quality at large outdoor concerts by any analog system ever made. They have recently introduced a 12 kW digital amp, but mostly I have seen 2 and 4 kW in the field (you need a lot of them) with on the order of 100 kW total amplifier capacity but only about 10 kW generator load during a loud concert for ~ 10k audience. (Sound having a separate generator is left over from the era of incandescent lighting with big SCR dimmer packs dumping HF noise into power lines, which would go right through the old analog amplifiers. Totally pointless today.) Setup of these systems at outdoor concerts starts with entering a topo map of the concert area with speaker tower locations and height. The setup program then calculates optimal angles for each element in the speaker line arrays. After the speakers are hoisted at the prescribed angles and connected a pink noise source and a wireless microphone carried out to designated spots on the topo map are used by the setup program to automatically adjusts levels for flat response across the full concert area. Monitor setup is a bit harder :-).
On 30/11/2016 15:45, Jim Thompson wrote:
> On Wed, 30 Nov 2016 10:41:59 -0500, Phil Hobbs > <pcdhSpamMeSenseless@electrooptical.net> wrote: > >> On 11/30/2016 07:00 AM, N_Cook wrote: >>> >>> Should 1.5v remnant 200KHz stay constant with whatever frequency or >>> amplitude of audio going in? Even at zero audio input there is this 1.5V >>> 200KHz. >> >> It'll probably be at its worst near zero output, because the switcher is >> producing a symmetrical square wave. At nonzero outputs, the amount of >> fundamental in the output goes down, and the second harmonic goes up. >> >> Cheers >> >> Phil Hobbs > > What does that non-zero remnant do to the speaker... produce some > heat? > > ...Jim Thompson >
There is what I took to be a Zobel,between speaker line and ground, 2.2R 10W in series with 430nF but may be there to absorb some of this remnant I suppose
On Wed, 30 Nov 2016 17:04:24 +0000, N_Cook <diverse@tcp.co.uk> wrote:

>On 30/11/2016 15:45, Jim Thompson wrote: >> On Wed, 30 Nov 2016 10:41:59 -0500, Phil Hobbs >> <pcdhSpamMeSenseless@electrooptical.net> wrote: >> >>> On 11/30/2016 07:00 AM, N_Cook wrote: >>>> >>>> Should 1.5v remnant 200KHz stay constant with whatever frequency or >>>> amplitude of audio going in? Even at zero audio input there is this 1.5V >>>> 200KHz. >>> >>> It'll probably be at its worst near zero output, because the switcher is >>> producing a symmetrical square wave. At nonzero outputs, the amount of >>> fundamental in the output goes down, and the second harmonic goes up. >>> >>> Cheers >>> >>> Phil Hobbs >> >> What does that non-zero remnant do to the speaker... produce some >> heat? >> >> ...Jim Thompson >> > >There is what I took to be a Zobel,between speaker line and ground, 2.2R >10W in series with 430nF but may be there to absorb some of this remnant >I suppose
What is the impedance of a speaker at 200kHz? ...Jim Thompson -- | James E.Thompson | mens | | Analog Innovations | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | STV, Queen Creek, 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'm looking for work... see my website.
On 11/30/2016 10:45 AM, Jim Thompson wrote:
> On Wed, 30 Nov 2016 10:41:59 -0500, Phil Hobbs > <pcdhSpamMeSenseless@electrooptical.net> wrote: > >> On 11/30/2016 07:00 AM, N_Cook wrote: >>> >>> Should 1.5v remnant 200KHz stay constant with whatever frequency or >>> amplitude of audio going in? Even at zero audio input there is this 1.5V >>> 200KHz. >> >> It'll probably be at its worst near zero output, because the switcher is >> producing a symmetrical square wave. At nonzero outputs, the amount of >> fundamental in the output goes down, and the second harmonic goes up. >> >> Cheers >> >> Phil Hobbs > > What does that non-zero remnant do to the speaker... produce some > heat? > > ...Jim Thompson >
Yes, and if you're unlucky it might excite higher resonances of the tweeters. That's really bad in woofers, dunno if it matters at the high frequency end. We seem to have a bunch of audio guys coming out of the woodwork who will probably know that. I have an unused matched set of electret omni mics (Earthworks) left over from a project 15 or so years ago. Their claim to fame was a super clean and fast impulse response (10 us FWHM iirc, good to 40-50 kHz). I was going to try to use the Helmholtz resonance (*) of the box of a delivery truck to measure how full it was, but the project never got funded. (It only works in certain circumstances anyway. We've all seen small but valuable boxes mounted on their own pallet in swathes of plastic wrap. That takes up a lot of floor space but not much volume, so it wouldn't move the resonance much. Cheers Phil Hobbs (*) The Helmholtz resonance is of the mass-spring type, the sort you get by blowing across the mouth of a beer bottle. The mass is the air in the neck, and the spring is the compression of air in the bottle proper, which goes as 1/(free air volume). This resonance is always lower than the organ-pipe resonances. so it's easy to find. -- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics 160 North State Road #203 Briarcliff Manor NY 10510 hobbs at electrooptical dot net http://electrooptical.net