JLH1969 MOSFET Class A/AB amp

Biglee_163

92#
同时我也建议你尝试下 tpa6020 (or ths6012) 或者类似的电流反馈型的放大器.他们是在世的最杰出的芯片放大器同时非常适宜做(桌面型或移动性)耳机放大器.

Biglee_163

93#
thank you so much!
但是我作品微型fet-JLH是使用1或者2节镍氢电池的,没有了沉重的变压器和大量整流\滤波电容的束缚,其具有相当移动能力.( 难道电池没有重量吗??).
你相信JLH可以用电池来工作吗?事实上,我的耳机是AKG's K271s,低阻抗缺少低音,我记得要增益不是很大而且要喂了大电流才可以推好它.然后我实验了我的双极性的 JLH ,但我不是很满意.但是你的帖子让我认识到 JLHs有很多种类.所以我希望能做一个适合我的需要的....我最爱好的耳机放大器是移动的,特别是电池供电的,我已经成功制作了一些分立组件电池放大器.读了你的帖子,关键字FET 和 CLASSAB 看来意味着 "可移动" "电池".啊哈,我只是个小型放大器爱好者!
(老兄,是中国人还要我帮你翻译 )

Biglee_163

94#
maxyeah: 这个很有挑战性!但你要有所取舍.

1) 移动能力要求放大器必须要效率高,这样就意味着你希望使用 mosfet 并工作在 Class AB.
2) 9v电池意味着你希望管子使用尽量低的开启电压来获得足够的输出.这有明确了你不适宜使用 mosfets. Mosfet's 通常的开启电压要超过 2-3v, 且典型我估算是4-5V.所以只要每个管子 的压降达到 2-3v的话,你的两个管子就已经损失了 4-6v且这样输出摆幅只有 3-5v.

我 google了你的耳机,它有 91db/mw 灵敏度 - 对耳机来说算低的, 55ohm, 你要输出3V峰值或者7V摆幅电压才可以产生大约 100mw (110 db? 很响了).

你可以找到低开启电压的 mosfets  (Zetex's E-line 为例子) 并且他们不是 to92封装就是r d-封装的.所以,他们可以做的很小.这样的话,对你而言,我认为还是使用 IC象 tpa6020 或者类似的产品.他们相对效率较高且输出极的损失电压比较小.

ateast99

向 翻译官 Biglee_163  致敬

maxyeah

嘿嘿~下次我自己翻译吧....不过我们真的还是要学好不同语言，因为外国网站真的有很多很好的设计和设计师

maxyeah

我接力翻译吧

Biglee_163

我们共同努力欢迎国外的高手多来我们论坛坐坐!

maxyeah

不如我们开个版面专门翻译国外优秀文章吧....我在隔壁坛经常干这事了

Biglee_163

可惜我时间不是很多,而且我的水平实在是太菜

millwood

原帖由 Biglee_163 于 2007-11-4 22:11 发表
我们共同努力欢迎国外的高手多来我们论坛坐坐!

I am Chinese too.

I wish I could have a more efficient way to input Chinese.

xr000

原帖由 millwood 于 2007-11-4 22:31 发表


I am Chinese too.

I wish I could have a more efficient way to input Chinese.

ateast99

I am Chinese too.

I wish I could have a more efficient way to input Chinese.

上 http://www.dhghost.cn/ 下个中文XP，安个双系统，中英问你都能对付了

maxyeah

号召管理员：一开耳放区，二开外文区

摇滚铁心

很希望有个区，请些大侠们把外国著名的网站的一些优秀的贴子翻译过来，我们国内的网站的许多制作都是只懂照抄电路图，很少有对些特殊些的电路做深入的研究。技术性很强的贴子太少，许多人的理论水平都不高。

millwood

quite a few readers have asked that if jlh1969 mosfet version (jlh1969M) can work at lower voltages.

mosfets tend to have higher saturation voltages (Vdss = 4v-5) than BJTs (Vces = 1-2v). so between two mosfets, you have lost about 10v - which is the minimum voltage required to run jlh1969m.

practically, you should have about 20v to run jlh1969M with good results.

Here is a quick schematic for such an amp. no material revision from the prior version but changes to the boot strap resistor (R1) and DC operating point adjustment resistor (R10).

enjoy.

millwood

one great benefit of the jlh1969M, in addition to its ability to run in Class AB and transition smoothly between Class AB and Class A, is that it has very soft clipping when overdriven.

Here is the same amp's output when R5 is set to 1k (about 10x gain).

just see those nice, soft and tube-like clippings, . try that with any of your solid state amplifiers.

millwood

the opposite question is often asked: how much power can the JLH1969M produce?

a couple limiting factors:

a) thermal stability: at high supply voltage, a little bit of Vgs decline will induce a large amount of heat on the device. While the circuitry itself is thermally stable, such a Vgs decline can destroy the device quickly: I have blown quite a few fdp3682 because of it.

b) power dissipation: the bootstrape network dissipates considerable amount of power. This is more of an issue when higher idle current through the driver / phase splitter is used to drive larger mosfets. most variable resistors are rated 1/2w and to do so reliably you should really derate it at least 50%.

I think 56v supply rail is the practical limit. here is the schematic to get you roughly there.

it is capable of V46v swing into a 4ohm load, or about (23)^2/(2*4)=66wrms.

you should use irfp140 class devices.

if you use igbts, you can get 5 more volte out of it and get to about 100w.

enjoy.

gogowatch

用面包板及手上现成的元件, 做了一次JLH1969M的实验, 已上实验时的线路图.

当R2的电流若为100mA, 得以下两个有问题的输出波形(at point 'out', .5ms/DIV, 2V/DIV).  C5 改为 30pF 或 100pF, 问题都不能解决.

但当R2的电流调至300mA时, 波形变回正常, 请问millwood, 这是否正常呢?

注: 实验时, 已接上4 Ohm 的喇叭.

millwood

that's interesting. looks like that the amp got into oscillation during a portion of the positive cycle.

I can think of two possible cures.

a) 2n5550 is too fast. I used a medium power bjt (mje15030) which is a 30mhz transistor, vs. 300mhz for 2n5551. However, if you have used a 100pf to slow it down, I am not sure if it would indeed be a reason.

b) use a gate stopper to slow down the mosfets. I would suggest a 110ohm resistor on each of the irf540n.

Also, be careful with irf540n. its SOA is considerably smaller than irf540's SOA. the best part for audio application in that family is irf540Z but they are hard to find.

Please let us know how it goes.

Thanks.

gogowatch

原帖由 millwood 于 2007-11-13 03:29 发表 #117

a) thermal stability: at high supply voltage, a little bit of Vgs decline will induce a large amount of heat on the device. While the circuitry itself is thermally stable, such a Vgs decline can destroy the device quickly: I have blown quite a few fdp3682 because of it.

I've read an interesting article http://www.edn.com/article/CA404515.html.  There exists a point of inflection in MOSFET device that the temperature coefficient is zero.  The temperature coefficient with Vgs lower than that in the point of inflection is negative, whilst the opposite is positive.  Attached is the Typical Transfer Characteristics curve from IRF540N datasheet.  The point of inflection is about Vgs ~= 5.5v.  

MOSFET devices are multi-cell devices, a number of power-FET cells are connected in parellel within the package.  The negative temperature coefficient (at a lower Vgs) will cause thermal runaway of a single cell one by one, especially at a higher supply voltage.  So biasing the MOSFET below the point of inflection is not thermally stable.

I think millwood have learnt this in a hard way.  Thanks millwood for sharing his valuable experience.

Cheers,