JLH1969 MOSFET Class A/AB amp

aili

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.

我是中国人。

我希望我能够有一个更加有效的途径来输入中文。

millwood

原帖由 xiaolu36 于 2007-12-5 10:15 发表
末级的图来了，大家研究一下。

this may not work and this should most definitely not work.

the front end idles at about 25ma each. the maximum current going through the 1k resistor.

when the input goes to its negative, the maximum current going throught he 1k resistor will be about 50ma. that means the current swing through the base of the output devices is 25ma (=50ma - 25ma).

typical hfe for highpower output devices is <50. that means your current ability is about 1amp (=50*25ma). or about 4w rms into an 8-ohm load.

you can try to use mosfets as your output devices but they tend to have low gm as well.

locky_z

可以借鉴一下TDA2003内部输入结构电路，改造一下JLH，既可以保留射极耦合电流反馈特征，又能适应不同供电电压时中点的稳定。


Q1的基极由Vref(1.4V)提供偏置电流，其发射极电压约0.7V，这样流过Rf上的电流为If=(Vo-0.7)/Rf,；
  Q2、Q3组成镜像恒流源，流过Q3的电流为(Vcc-0.7)/R2，因为Q2镜像Q3，所以电流为（Vcc-0.7）/R2,
  这样Q1发射极节点电流方程式为恒流源I1+Rf上的电流等于Q2上面的电流，即I1+If=Iq2
I1+(Vo-0.7)/Rf=(Vcc-0.7)/R2
  整理得出V0=Vcc*Rf/R2-(0.7-0.7*Rf/R2-I1*Rf)
  如果Rf/R2=1/2,
Vo=Vcc/2 -(0.35-I1*Rf)，
  如果I1*Rf=0.35或者0.35-I1*Rf远远小于1/2Vcc，那么Vo就自动等于1/2Vcc了，和电源电压大小无关了。

  这种输入级用在JLH或者Zen4就可以无调整自适应各种不同电源电压了。

millwood

原帖由 locky_z 于 2007-12-6 09:13 发表
可以借鉴一下TDA2003内部输入结构电路，改造一下JLH，既可以保留射极耦合电流反馈特征，又能适应不同供电电压时中点的稳定。

Q1的基 ...

it is actually a lot simpler than that. see the attached chart.

take a look at note A.

from the left of it, the current going into it is about I1 (ignoring the base current of Q1).

the current flowing down the node (collector current of Q2) is I2, because Q2/Q3 forms a current mirror.

the current flowing into the node from the right (OTL  output) is If.

at any given time, I1+If=I2.

if I1 is sufficiently small, If=I2.

well, I2 ~= Vcc/R2, If ~ = (Vo / Rf) - because Q2/Q3 will be sufficiently saturated.

so Vo =Vcc * (Rf/R2).

In other words, if Rf = 0.5*R2, Vo = 0.5*Vcc.

xiaolu36

this may not work and this should most definitely not work.

the front end idles at about 25ma each. the maximum current going through the 1k resistor.

when the input goes to its negative, the maximum current going throught he 1k resistor will be about 50ma. that means the current swing through the base of the output devices is 25ma (=50ma - 25ma).

typical hfe for highpower output devices is <50. that means your current ability is about 1amp (=50*25ma). or about 4w rms into an 8-ohm load.

you can try to use mosfets as your output devices but they tend to have low gm as well.

请高手翻译。

Biglee_163

这个不能正常工作而且是明确无误的。
前面一级的静态电流是每管 25ma.指流过1K电阻的最大电流。
当输入信号转到负极性的时候，流过1K电阻的最大电流将是大约50ma。这意味着通过输出管基极的电流摆幅是 25ma (=50ma - 25ma).
高功率输出管的典型的 hfe  <50.这意味着你的电流输出能力只有大约 1amp (=50*25ma).或者相当于 4w rms功率输出在 8-ohm 负载上。
你可以尝试用mosfets来充当输出管但他们的跨导gm 同样不高 l

xiaolu36

谢谢楼上的，还有millwood 。
另外还有个问题，能否减小1K的电阻，达到加大电流，提高输出功率的目的？
还有输出功率如何计算？

locky_z

原帖由 millwood 于 2007-12-6 09:35 发表
the current flowing into the node from the right (OTL  output) is If.

at any given time, I1+If=I2.

if I1 is sufficiently small, If=I2.

well, I2 ~= Vcc/R2, If ~ = (Vo / Rf) - because Q2/Q3 will be sufficiently saturated.

so Vo =Vcc * (Rf/R2).

In other words, if Rf = 0.5*R2, Vo = 0.5*Vcc.

I1是Q1工作电流，不可能太小的，I2是白白消耗的电流，也不可能设置的太大，所以I2不会>>I1,
所以仍旧要根据I1的值设定好Rf的值。否则达不到自适应电源电压的效果。

millwood

in a typical jlh setup, I1 is about 1ma.

in a Vcc=40v, Rf=1k set-up (fairly standard JLH), If=20v/1k=20ma.

so in that case, If is >> I1.

locky_z

一般jlh电路输入管基极电位是接近1/2*Vcc,这样Rf上的电压不大，电流也不大。

另外在20mA这么大电流下，下不一定能保持电流镜像。所以If不适宜大。

300B

我想问一下大家,小甲最高DC电压可以上到好多?我现在手里有双ＡＣ32V的牛,不知道安全吗?

zhengfaming

请问 millwood 兄你的MOSFET版 JLH1969 输出多少W? 如果想提高功率，有何办法？？？

millwood

you should expect about 5v of voltage loss per mofet. so the maximum positive voltage is Vcc-5, and the maximum negative voltage is 5v, with a Vpp of about Vcc-10v.

mine runs on 40v Vcc, with a output swing of 15v (=30vpp, confirmed on a scope). so maximum power output is 15^2/(2*8)~=15w into 8 ohm, and 30w into 4ohm.

if you run on higher voltage, you can get more output. At 50v Vcc, you should get a swing of 20v, and the maximum power is 20^2/(2*8)=25w @ 8ohm, or 50w @ 4ohm.

I used just one pair of output devices and it is stable into 2ohm (didn't try anything lower), AND with a 2uf capacitor in parallel with the speaker.

for voltage much higher than that, you will need to have special voltage compensation to avoid thermal run-away. I think 60v is the practical limit for this circuit. At that level, you should get 25^2/(2*8)=40w into 8ohm and 80w into 4ohm.

millwood

原帖由 300B 于 2007-12-7 13:00 发表
我想问一下大家,小甲最高DC电压可以上到好多?我现在手里有双ＡＣ32V的牛,不知道安全吗?

32vac rectified will give you 32*1.2=38.5v. that is almost identical to what I have (40v rail). you can either use just one winding or parallel them - a tricky proposition.

this kind of voltage can be tough for the BJT version. the bjt version should idle at about 2amp, and each output device dissipates 2*19=40w of heat, while outputing just (2)^2*8/2=16w into 8ohm.

millwood

someone asked earlier the output power difference between BJT and mosfet versions.

a bipolar output device losses about 2-3v of Vce when fully saturated. a mosfet losses about 5v Vds when fully saturated.

so Vpp on the output for the BJT version is about Vcc-5v, and Vcc-10v for the mosfet.

so the output from a BJT is ((Vcc-5)/2)^2/(2*Rl), and ((Vcc-10)/2)^2/(2*Rl) for the mosfet.

Obiviously, under the same Vcc, the BJT will have slightly higher power.

However, the BJT version can only work in Class A, and the Mosfet version works in Class A or Class AB seamlessly. That is, if you underbias the BJT version, you get clipping distortion.

if you underbias the mosfet version, the amplifier simply slides into Class AB when it is overdriven.

a side effect of this is that the mosfet version works more efficiently, runs cooler and doesn't require as big of a heatsink.

Mines runs on an old Pentium heatsink, without a fan while idling at about 50 - 60ma.

zhengfaming

谢谢millwood 兄解答，还是有点看不懂！

audiokey

Dear Millwood,

That’s to excite to read all the  article. Actually I want to build a power amplifier that have 40-50W. Do you have any advise? Would you please post the diagram again?

millwood

season's greetings, audiokey.

you may want to check  out post 158 on page 8.

if you need to tailor it for your transistors or want to use different parts, please let me know and I can redraw it for you.

millwood

zhainm asked, in a different thread, if we can use mj11033 (PNP, 300w, 50a darlington power BJT) as output device in a JLH1969. Because of its high current capability and darlington construction, it is ideally suited for a JLH. However, it is a (relatively) high speed transistor so we will need current compensation to make sure that the amp is overall stable.

as requested, here is how it can look like.

it is basically a regular NPN JLH1969 upside down.

Optional components marked with a red trace and optional but recommended components marked with a green trace.

[B]PLEASE PAY ATTENTION TO C10. WITHOUT IT, THE AMP WILL OSCILLATE[/B]. so use a larger one (maybe 220p or 470p) and reduce its size. 110p is the smallest per simulation but your may change.

R22/R24 changes the gain for the amp (about 10x now). Setting R24 at 47 gives a maximum gain of 20x.

R19 is a variable resistor setting the mid point.  if you don't care about maximum output voltage swing, use 10K for R19. As is, the mid point is about 16v, the amp idles at 2amp, and maximum output power is about 14^2/(2*6)=16wrms into a 6ohm speaker.

If you adjust R19 to 7k, the mid point is at about 20v, the amp idles at 2.3amp, and the maximum output power is about 18^2/(2*6)=25wrms into a 6ohm speaker.

the power devices each dissipates about 45w, so you must use large heatsink. or your transistor will die in a split second.

the 2nd chart is the output waveform produced with R24 = 110ohm, R19=10k.

the 3rd chart is the THD content. Notice how low the 2nd and 3rd harmonics are (2.5mv and 5mv), vs. the output (10v).

the 4th chart is the frequency response. no issues of instability so I didn't draw the phase shift chart.

millwood

the same approach applies to P-channel mosfet. You can reconfigure a N-channel jlh1969 into a P-channel jlh1969 as well.