Class D amp discussion

millwood

I would like to start a thread on Class D, knowing that we have a resident expert (fumac) on it.

Class D amps (or PWM amps) have its output stage work like a switch. It was invented in the 1950s and never took off due to poor sound quality and lack of highspeed switching devices.

a few things happened in the last few years.

1. power electronics got a lot better, and mosfets can switch in mhz range.
2. the use of DSP allows noise shaping (pre-distortion) to improve sound quality.

The first successful Class A amp is the Panasonic SA-XR20 which uses the TI TAS PurePath chips. It was called Poorman's Krell when it was introduced. That series goes all the way to XR-70 (XR20, 25, 40, 50, 55, 57 and 70). you could get them for less than $200 in the US, which includes AV functions, 125wx6 or x7 power. pretty good for the money.

then you have T-amp, using Tripath's class D chips.

after that, you have a rebirth of the Class D era.

cglvisa

看了半天，只认识了一个　YOU   真是晕啊!!!

millwood

There are basically three types of Class D amplifiers.

a) open loop analog: this topology takes an analog input, run it through a PWM modulator to get a PWM signal, and have a driver stage that runs either in half bridge or full bridge.

it usually runs on fixed frequency and uses onboard oscillator / clock.

millwood

this is a typical open loop analog class D amplifier (LM4673):

syhkiss

好象是一个D类放大器的问题

opqgopqg

lz是外国人`?

syhkiss

millwood ~~~Did you make D enlarger?

问他做过 D 类放大器没有

syhkiss

$200 在美国,可以做个比较好的 7声道 125w的 AV 功放

好象挺便宜

millwood

the second type of Class D is open loop digital class D amplifiers.

Rather than taking an analog input, they take a digital input (PDM, or PCM or S/PDIF, depending on what the situation may be), and use it to directly drive a PWM modulator.

earlier ones do not pre-distort the input signal, and don't sound quite good due to distortion introduced in the PWM modulation. TI, through an acquisition, found that you can pre-distort the signal in a way so that after it has been distorted by the PWM modulator, it forms great sounding signal after the low-pass filter.

That becomes the TI PurePath family of DSP-based PWM modulators. they typically consists of a PWM controller / processor (TAS5518) driving a few power stages (TAS5182).

millwood

all of the two class d amps are open loop: the output signal is never fedback to the input.

so they suffer from poor PSRR, among other things. TI actually had an application note that uses supply voltage for volume control for its PurePath chips.

the 3rd type of Class D uses feedback. they are called self oscillating class D. They use positive feedback to great oscillation. the feedback signal can be picked up from before the filters (pre-filter feedback) or after the filters (post-filter feedback).

The first that came to the market is IRF's IRAudAmp 1 (https://ec.irf.com/v6/en/US/adir ... QueryName=iraudamp1).

the front-end is a LT1220 that takes the input signal, mixed it with pre-filter feedback. the output is a IR2011S half bridge driver driving two N-channel mosfets.

I have built this one and it sounds quite good.

millwood

IRF made a few versions of its self-oscillating class D amp. the latest is IRAudAmp5, using IRS2092 single chip that integrates the front end (now a transconductance amplifier for speed, comparators and half bridge driver).

Here is its block diagram. Unfortunately, it is not avalable yet. the price is $1 - 2 per chip.
You can build a whole amp for probably less than $10.

syhkiss

通过楼主的改造声音非常的棒~

syhkiss

Use English expression, we understand

But thank very much

Julien

谢谢millwood的资料，希望多一些这类的讨论。

millwood

at the same time that IRF was developing its class D amp, Phillips was also working on its class D amp. However, the Phillips design uses post filter feedback for better sound quality and better PSRR.

Here is the original article in AES article (http://www.ciaudio.com/ucd_aes.pdf).

Bruno soon left Phillps and Hyperx started to market a similar design called Universal Class D (UcD) at their website. the first product is UcD180 and now they have UcD700. I have a UcD180 and it sounds very good.

Here is the block diagram for UcD. As you can see, it takes post-filter signal, and feeds it into a conventional amplifier to start a positive feedback loop.

Later in the article, Bruno put forth a schematic that doesn't have device selection nor values.

However, in case you want to do it, Phillips put forther a full application note on almost the same thing (http://www.semiconductors.philips.c...s/UM10155_1.pdf).

I also built it and found it difficult to get started. I have also tried to build one with a half bridge driver but that had problem oscillating during the negative half of the signal. But this is in fact the Hypex module so it should work. Other people have reported success with the Philipse application note.

农民工！

看不懂！

millwood

the IRF amps are nothing but sigma delta modulators. a lot of small power class D chips are of such a design. the AD1990 family for example is a typical sigma delta modulator and you can take the output and drive a set of half bridge for high power applications.

the most promising, however, is the Phillips TDA8931 (http://www.nxp.com/acrobat_download/datasheets/TDA8931_1.pdf). If you look at the blockdiagm, it is identical to IRS2092: a front end preamp that functions like a comparator, driving a half bridge with integrated mosfet power devices.

so you can use it to form either a UcD type amplifier or IRAudamp class D amp.

and it costs less than $2.5 each from digikey.

ideal-v

恩,希望来点比较深入点的理论知识.以上不够

ideal-v

这个是IR方案的,小功率还可以.但对布局的要求高

日王贝才

26个英文字母偶都认识