实做和仿真高性能低成本耳放 JLH (John Linsley Hood) 8V低压供电 补图

e3po

Many thanks to millwood for walking me thru such a long road!!

-----------------------------------------------------------------------------
原贴地址:
http://www.crystalradio.cn/bbs/thread-134183-1-1.html
***  感谢原贴中诸位的回复和关注 !   

最开始的原理图:


经过几天的琢磨,  第一次把理论 THD 降到 0.001% 以下时的原理图:

实物照片



波型:
下面是波型,  黄色波型是输出, 蓝色波型是输入.



2am BJT Aug 17, 2010    增加了耳机插座,  剪掉了散热片, 测了一下 100KHz 的表现 (用 555 产生的 100KHz)
黄色是信号,  蓝色是输出.



附录:   失真度测试 看看您的耳朵对失真是否真的敏感
不妨用下面的工具检验一下加了某种设备/元件/改进以后,  对您本身听音到底有没有帮助
点击第一个连接是测试开始,
http://www.klippel-listeningtest.de/lt/startformular.php

第二个连接是英文说明
http://www.klippel-listeningtest.de/lt/instructions.html

测到最后就会生成一份报告
例如最下方的条型图,  
"红色" 就是您的测试结果.

吾画的简单中文说明




自己 2am BJT, 2010AUG09 的测试结果贴上来供大家参考


2am BJT, 2010AUG11 的另一个测试结果


###########################################################################################
附录2:    关于耳放(耳机放大器) , 运放, 您的健康以及听力损害 etc.
http://www.crystalradio.cn/bbs/thread-133834-1-1.html

[ 本帖最后由 e3po 于 2010-8-17 02:10 编辑 ]

millwood

that 100Khz square waveform is telling:

1) no instability - no ringing at all. If anything, the amp appears to be slew limited.
2) slew rte of about 6v/2ns=3v/ns. pretty good and more than sufficient for a headphone amp.

I would be interested in how the amp reacts to overdrive / clipping.

also, how about a pair of small signal mosfets (irf510 or 2n7000) as the output device?

e3po

原帖由 millwood 于 2010-8-17 06:46 发表
that 100Khz square waveform is telling:

1) no instability - no ringing at all. If anything, the amp appears to be slew limited.
2) slew rte of about 6v/2ns=3v/ns. pretty good and more than suffici ...

I dont have any 2n7000 or IRF510 on hand.  2n7000 is $?  EA

Yeah, I limited the slew rate by adding a 10pF cap parallel to the 900 ohm resistor.
Without the 10pF cap, there is ringing.

millwood

if there is ringing, you can put a small cap, like 22pf - 110pf on the VAS's collector / base.

high speed amps are prone to ringing, due in part to their high speed and driving reactive loads - like long speaker cables, etc.

putting a small resistor on the output (in serial with the load) helps a lot.

millwood

to give you a sense as to how "stable" this amp is driving capacitive loads: most power amps will have a hard time driving a capacitor as the load.

when I built my first JLH1969M (with IRF540 as output devices), I tried a 2uf capacitor on the output (in parallel with the speaker). You could see sparks coming out of the capacitor when it was put on line but the sound quality just suffered slightly.

JLH1969 really is an amp too ahead of its time.

e3po

原帖由 millwood 于 2010-8-17 21:36 发表
to give you a sense as to how "stable" this amp is driving capacitive loads: most power amps will have a hard time driving a capacitor as the load.

when I built my first JLH1969M (with IRF540 as ou ...

          could you share the sch of the MOSFET version ?

millwood

2n7000 is about a few cents each.

the mosfet version is identical to yours, except that the output devices are replaced with two mosfets, with gate stoppers (resistors < 110ohm on their gates). you will need to play with biasing a little bit.

If you are building a small headphone amp, you can use irf510 (my favorite medium power mosfets), without heatsink.

if you are building a large power amp (> 20w RMS), try IRFP140 or FQA25Nxx class mosfets in TO3P packaging.

millwood

the good thing with JLH1969M is that it can run in Class A, Class AB or Class B. You can bias it in class A for smaller power (around 1W RMS for example) and if a large signal comes, it will smoothly transition into Class B.

No more expensive heat sinks.

e3po

It seems for the MOSFET version,  8 Volt supply (Li-BATT) is not sufficient ,  12 Volt (Car Charger Plug) is ok.

[ 本帖最后由 e3po 于 2010-8-17 23:35 编辑 ]

e3po

原帖由 millwood 于 2010-8-17 22:24 发表
2n7000 is about a few cents each.

the mosfet version is identical to yours, except that the output devices are replaced with two mosfets, with gate stoppers (resistors < 110ohm on their gates). you ...

In mainland China,  2N7000 is around a quarter EA (1 RMB),  not cheap at all.   on the other hand, IRF610 is cheap, 2 RMB EA ( $0.35).

millwood

you are correct that the mosfet version is less efficient, because of their larger Vds drop for linear applications. So consequently, you need higher rail voltage to achieve the same output swing.

millwood

I just checked, 2n7000 is < 5 cents each at avnet.

e3po

原帖由 millwood 于 2010-8-17 23:37 发表
I just checked, 2n7000 is < 5 cents each at avnet.

Avent seems to be a good source.  Do they ship to Canada at same rate?


I just took a look and found out that the 4-cent price is for 20K, gee~~~   ....  and they are out of stock :p


http://avnetexpress.avnet.com/store/em/EMController/Discrete/Transistor/MOSFET/_/N-100099?action=products&cat=1&catalogId=500201&cutTape=&hrf=www.avnet.com&inStock=&langId=-1&proto=&regionalStock=&rohs=&storeId=500201&term=2N7000&topSellers=

Mouser is selling at $0.24~$41 EA.
http://ca.mouser.com/Search/Refine.aspx?Keyword=2n7000

[ 本帖最后由 e3po 于 2010-8-18 00:00 编辑 ]

millwood

didn't realize that.

another n-channel to try is bs170. less beefier but roughly the same.

you can also do a more generic search for mosfets with Ids in the 1 - 2amp range.

e3po

原帖由 millwood 于 2010-8-18 01:11 发表
didn't realize that.

another n-channel to try is bs170. less beefier but roughly the same.

you can also do a more generic search for mosfets with Ids in the 1 - 2amp range.

Thanks for your input!

I'll play with 2sc1815/2sa1015 or 2n5551/2n5401 for a while, since they are everywhere, and dirt cheap.
I have 10 pairs of IRFP250/9250, 8 pairs of IRF610/9610 and  40 pairs of 2sk246/2sj103,  hopefully they won't stay in my drawer for ever.
One day, I'll build a roof-blowing amp (I've been conceiving this idea for 12 years....

I'm wondering whether my toddler son will be interested in my junk 12 years later?  or they'll be  donated to 1-800-got-junk ....

[ 本帖最后由 e3po 于 2010-8-18 01:49 编辑 ]