Class D amp discussion

lm1875a · 发表于 2007-10-25 22:57

Hi Millwood,
welcome to this bbs!!! Can you tel me, are you Chinaese?

Sorry! I just know a little English

FUMAC · 发表于 2007-10-25 23:01

he is chinese, but he cant type chinese

靠,搞错了
他是中国人,但是他的电脑无法打中国字

lm1875a · 发表于 2007-10-25 23:14

原帖由 FUMAC 于 2007-10-25 23:01 发表
he is chinese, but he cant type chinese

靠,搞错了
他是中国人,但是他的电脑无法打中国字

收到！

xdgw · 发表于 2007-10-25 23:28

我发个输出入法给你试试能不能输入中文，yes OR no

xdgw · 发表于 2007-10-25 23:35

智能五笔应能在英文系统下出中文的

xdgw · 发表于 2007-10-25 23:40

http://www.sofriend.cn/bbs/viewthread.php?tid=246

FUMAC · 发表于 2007-10-26 00:45

有些时候,在国外的中国人的电脑并不方便安装中文输入法,
不要难为他了

millwood · 发表于 2007-10-26 01:42

原帖由 xdgw 于 2007-10-25 23:28 发表
我发个输出入法给你试试能不能输入中文，yes OR no

sure.

杰哥哥 · 发表于 2007-10-26 02:55

金山快译勉强能行吧？哈哈。。

millwood · 发表于 2007-10-26 08:35

here is a simplified UcD class D amplifier.

the front end is a typical comparator. the driver stage is theoretical so you cannot build it yet. You can, however, use a half bridge driver and two switchers to build the 2nd stage to make it functional.

the switchcad file is updated in .rar format. you can play with it and see how it works.

enjoy.

younglee · 发表于 2007-10-26 08:40

millwood,在哪？Class D好玩吧？

millwood · 发表于 2007-10-26 08:40

here is the simulation results for the previous amp, driven by a 20khz sine signal.

the green line is pre filter pwm output, and the blue line is the post filter audio signal.

millwood · 发表于 2007-10-26 08:50

you can use a typical half bridge driver, like ir2011 family (from IRF), or lm5100 family (national), or ucc27200/ucc27201 (TI) or any other mosfet drivers - On Semi has a few as well.

millwood · 发表于 2007-10-26 08:51

I have to say that I have not been able to get UcD with mosfet driver working. Not sure why but it wouldn't work on the high side. Need more experiment, I guess.

音响DSP · 发表于 2007-10-26 09:39

在某些仿真软件，如EWB8就无法启动UCD的电路，也许有些收敛条件要改下。
至于你提到的TDA8931是自激振荡（NXP叫它为“SODA”）的结构，那双声道的TFA9810T倒可以让我去试下UCD的结构，其实应该是普通D类滤波器前反馈+滤波器后相移修正反馈网络。这样不用拿NXP的UCD分立电路来搞了，直接拿块TFA9810T的DEMO板来试。

音响DSP · 发表于 2007-10-26 10:00

原帖由 millwood 于 2007-10-25 22:15 发表

yes, you will need to block the DC.

the tda8931 datasheet has a nice way of doing it.

TI and IRF

NXP

gogowatch · 发表于 2007-10-26 12:36

回复 #54 millwood 的帖子

I did make UCD working good on a breadboard with IRF540N a few months ago. Attached are the pictures of the breadboard setup and some oscilloscope traces. I have made some a PCB off from the same circuit giving even cleaner output.

Attached also my LTSpice simulation files. For those who interested in the simulation, you can download the software from http://www.linear.com/designtools/software/#Spice.

Cheers

millwood · 发表于 2007-10-26 18:41

gogowatch: have you tried to use an integrated driver rather than a discrete driver there? my discrete version of the ucd worked as well but i have never been able to get the version with an integrated driver (I tried three of them) to work.

here is my version based on the phillips application note.

millwood · 发表于 2007-10-26 18:43

here is the schematic:

followed by the switchcad simulation file.

gogowatch · 发表于 2007-10-26 20:29

Hello millwood,

Thanks for starting this thread.

I'm also trying to make a UCD by using IC gate drivers but still not yet get it done.  Those drivers are very easy to blow and I did several times just because of careless mistake.

I took an approach to develop the circuit stage by stage.  Following is my sequence,

1. Get a triangular wave from a Signal generator or generate it from a simple opamp circuit.
2. Feed the triangular wave and the ground signal to a comparator.  I'm using LT1016.
3. Level shift the comparator output down to the negative rail.
4. Implement a dead time generator for the pair of PWM signals.  I used a XOR gate, a cap and a resistor to do the job.
5. Feed the final PWM signals with dead time to the IC gate driver.  I tried UCC27201, HIP20101, LM5101.
6. Feed the output of the gate driver to a half bridge mosfet network.
7. Feed the output of the mosfet to the low pass LC filter.
8. Measure the signal delay from comparator to the output of the low pass filter, calculate the value of RC network for self oscillation and finally remove the siggen input and complete the feed back loop

I'm still at step 6 today.  I found the fast rise and fall of the IC driver output make transient and ringing of the mosfet output very difficult to control.  I plan to make a PCB to do the experiment.

Hope this helps.

BTW, have you checked this thread in diyaudio.com?

http://www.diyaudio.com/forums/showthread.php?postid=778406

Cheers

[分析测量] Class D amp discussion

UCD on breadboard with IRF540N

回复 #58 millwood 的帖子

新人进步