仿力士电路的后级功放，达林顿并管输出的。

lsq1463

30对铜壳的11032/33
入了那么多啊！装什么机器那么厉害！
我也很喜欢用这管的，我现在用的仿音乐传真A120.2就用了4对这管，电流一对管1.35A！

流泪的背影

学习中..................

ngzxm

老郑可以试试把推动管省掉，也许又是另一种感觉呢。

millwood

the basic topology bears zero resemblence to Luxman's typical topology. However, if you take a look at Hafler's trans-nova amps and you will see a much better designed amp than this one.

a few mistakes:

a) the designer obviously doesn't understand how the transconductance portion of the VAS works. D14 is actually another A970 - they form a current mirror. As such, R47 = R49.

b) once you figure that out, you will understand why R22=R45.

c) the cascoding makes a lot of sense for Hafler because the VAS is driving mosfets (high Ciss). since you are driving a medium power bjt in a 40v rail application, you don't get much benefit out of the cascoding.

开机十分钟以后，电流穿升到600MA！最后稳定在650MA左右不再上升，看来可以了。为了试验一下温补管子的工作效果，搬过来一个电风扇，用最强档对着散热器吹，同时用表量输出静态电流，随着散热器的温度的降低，电流也在逐步降低，等到散热器冰凉，电流也回到了冷态的150MA左右。

that should have convinced you that the Vbe multiplier is NOT working right - it is under-compensating the thermal impact.

hxx886

试听怎样？

HIFI1231

用669/649也可以,把管子反面固定散热器,电流在10MA左右反面固定没问题的,管子不会过热的

LiangSheng

又见好板了

hl8999696

hao好 ！

millwood

here is the basic Hafler trans-nova topology. I have highlighted those parts that you didn't get right in the attached schematic.

hopefully that will help you correct those design mistakes.

ciao.

HIFI1231

我的也要上电了,有点怕啊,快过年了,不想放烟花啊

goldwayzhang

半铜半铁交织出的声音不知如何。

zheng1

原帖由 millwood 于 2009-1-19 06:35 发表
the basic topology bears zero resemblence to Luxman's typical topology. However, if you take a look at Hafler's trans-nova amps and you will see a much better designed amp than this one.

a few ...

你是说，R22和R45还有R47和R49是上下臂应该是对称的，我直接改成680欧的话，会造成上下臂电流不对称？
第一级R46为68K，R48为10K，比值为6倍，电流太小，靠反馈60倍放大倍数来提高增益？不是靠大电流来驱动？
但是我这样安装下来，末级电流调不起来，声音还很烂。

还有，我的E文实在很差，您说的我只能看一个大概，80年代中旬我就参加工作了，以后就一直没有系统的学习过E文。只是大概的看着多了，李逵李鬼差不多。

zheng1

最后定型了，就按照这个改动，推动管子用669/649就很不错，比15030/31声音个细腻，还原更真实一些，虽然韵味少了一些，用667/647也可以，不过负载电阻就需要增加到330欧左右。
声音吗，我的这种做法和原机的设计原则不一样，结果也有些不同。不过总的来说，这个电路的声音不错，厚实细腻，密度极佳，分析力也很高，低频力度和量感也足够称呼为（力士）。声音收放控制的很好。听交响乐也不错。

zheng1

原帖由 goldwayzhang 于 2009-1-20 21:14 发表
半铜半铁交织出的声音不知如何。

哪个是炮灰管子，调试完成了他们也就该下岗了！主要是我做电路烧管太多，呵呵！

zheng1

35楼图纸里面已经改正了一些错误，以这个最后的为准。末级用达林顿，推动管可以用647/667之类的小管子同时不用上散热，如果用669/649的话，管子需要反过来安装，背后金属面向上，有字的正面涂硅脂上散热器。用669的好处就是，末级就算用21194-15003-15025等管子，也能推得动，同时在小电流下效果也不错。性能最接近小功率管。

石城

刚入了30对铜壳的11032/33。。。。。真有钱啊。我用了几年才舍得花钱慢慢淘了4对。

millwood

some suggestions for you to think about:

1) the current going through R15 is (6.1-0.7)/1k=5.4ma. it is a little bit high but not a big problem.
2) R46/R48 determines the clamping on the jfets. As designed, the jfet's sources are sitting at 22k/(47k+22k)*50v=15v. That is about right for a typical jfet.
3) R49 and the resistor in serial with the diode are too high. those Re resistors exist to just balance out the unmatched parameters in a current mirror. They are typically 11 - 110ohm, and rarely above 220ohm: for a current mirror to work, you have to have high gains here. I would set them at 22ohm - 47ohm.
4) R35 / R27 determines the open loop gain only when they are sufficiently small. At 100k, those two resistors are essentally shorted by the Vbe multiplier and the output stage.
4) R35/R27 determines the idle current for the voltage amplification stage. you want the VAS to idle at about 10ma. that means R35/R27 = 1.4/10ma=140ohm.
5) that value yields a voltage drop of 3*0.7+10ma*60=2.8v on R45, or idle current of 2.8ma on R45. Which is roughly 50% of the current going through R15 -> your differential pair is balanced.

...

you are far better off developing a good understanding of how an amp works BEFORE attempting to modify it.

owl

学习，学习。

zheng1

原帖由 millwood 于 2009-1-24 00:29 发表
some suggestions for you to think about:

1) the current going through R15 is (6.1-0.7)/1k=5.4ma. it is a little bit high but not a big problem.
2) R46/R48 determines the clamping on the jfets ...

明白了！再次感谢一下老兄的热心指点。