NAP-140 Clones

millwood

I have always been a fan of quasi complementary output stages: they are inexpensive to make, they are very rugged, and you don't have to match them.


so I was quite happy when I saw NAP140 PCBs available here.

However, I don't particularly like the circuit. In particular, it uses too many active load and I want to use N-channel mosfets.

so I simulated a few of the circuits and here is what I learned.

here is the schematic. I typically design my amps to run on 4-ohm load, and on 1Vp input. Here, the amp is driven by a 1Vp 20khz sine wave.

millwood

here is the output waveform.

you see clearly massive cross-over distortion on the negative side. Not good.

millwood

how about its frequency response?

here it is.

Something is seriously wrong in the 7-800Khz range. the amp is obviously self-oscillating there.

FUMAC

原帖由 millwood 于 2008-6-1 04:28 发表
how about its frequency response?

here it is.

Something is seriously wrong in the 7-800Khz range. the amp is obviously self-oscillating there.

you can use it at self-oscillating  class-d

millwood

does that mean it is hopeless?

not exactly.

some easy fixes, as proposed in this version of NAP140.

1) input resistor: 100K is simply too much.
2) I added R43/R44 to just balance out the effect of unmatched input transistors. 11 - 47ohm is typical. Higher numbers also reduce open loop gain, something I like.
3) R31: that was an obvious mistake in the original schematic. corrected here.
4) I also added R29, again to reduce open  loop gain and to make the amp more linear open loop.
5) R33/47/51 changed to a much lower number to increase the idle current on the drivers. They now idle about 30ma - just about right.

other changes are added to address stability.

As you can see, there is no more cross-over distortion, even at 100ma idle current.

millwood

this is the frequency response. Not perfect but much better.

millwood

here is the same amp, with IRF540 as output devices and Vgs multiplier.

I added two gate stoppers and change some values, all highlighted.

millwood

here is the waveform, at 100ma idle current. Nice.

here is the FFT for the waveform. mostly 2nd order harmonics (0.6v/27v=1.8%?) and higher order harmonics go disappear very fast. Nice too.

millwood

here is the FFT profile for the original NAP against MOSFET.

As you can see, while the MOSFET's FFT (red curve) disappear very fast, the higher order harmonics in the original NAP (green curve) continues into multi-meghertz, thanks in part to its high open loop gain.

millwood

Now, comes the sad  part: the slew rate.

as you can see, it takes the amp about 15us to get to 30v, or about 2v/us. Not a great number for a high performance power amp.

However, that is at an output power of (30)^2/(2*4)=100wrms.

BTW, the amp has superb control of the output, with no ringing on either upper edge or lower edge.

millwood

the amp presented in post #7 can be assembled with the existing nap140 pcb. you just need to omit a few parts and use some different parts. It should be an excellent amp for anyone, especially if you use those inexpensive power mosfets as output devices.

you can make more analysis to it, or add some parts (like r/c decoupling for the front stages).

but this is going to be a good amp for you to try, especially if you already have the PCB.

millwood

the advantage of mosfets comes in at its high current capabilities.

here are two waveforms.

1) the improved BJT driving into a 1ohm load.

2) the mosfet version driving into the same 1ohm load.

I would rather listen to the mosfet version.

millwood

for those who are adventurous, here is another version of the same amp, running off a bootstrap network instead.

major changes are highlighted:

a) R105/R109/C29: the bootstrap network.
b) if you use an LED (forward biased) in place of the 1n4148s, you can change the value of R108 to get the amp to idle at roughly the same current (1ma).
c) R102 downsized to 22ohm, to provide more output swing. You can set R102 back to 110 ohm but you will lose a volte or so of output swing.

BTW, the amp clips quite with grace and no harsh harmonics.

millwood

in summary:

we presented three Nap140 clones that will be a better amp than the original nap140 and that can be built on the same pcb. Of course, I perfer the mosfet versions for their lower costs, higher performance and better current capabilities.

enjoy.

aeon

非常好，我的一些疑问你给我解析了。非常支持你的设计。nap140毕竟是生产于那个年代的东西，电路设计是要顾及当时零件的性能的。在naim现代新型的功放里面，已经不见到以前的设计。

shimu

good,good,我把末级改成了ON的MJL21193，21194的互补，不知道有啥差别。。

qjdj

这个电路坛子上很多同学都有基本相同的电路板，包括抄板的nap140，论坛的H-140，新板子mx50也相似。谢谢楼主提供改进建议。
有个疑问：R31在原电路图中取22k，楼主说这是个错误。能详细说明吗？有没有可能这是个特殊的设计。
另外，楼主这样改动比较大了，干脆改得更大些，比如电压放大级加入共射共基，电压放大级的输出再加个射极缓冲。天，这还是140吗

红双喜

这么好的精华贴看不懂

jason2511

1楼。I have always been a fan of quasi complementary output stages: they are inexpensive to make, they are very rugged, and you don't have to match them.
so I was quite happy when I saw NAP140 PCBs available here.
However, I don't particularly like the circuit. In particular, it uses too many active load and I want to use N-channel mosfets.
so I simulated a few of the circuits and here is what I learned.
here is the schematic. I typically design my amps to run on 4-ohm load, and on 1Vp input. Here, the amp is driven by a 1Vp 20khz sine wave.
翻译：我一直是范准互补输出阶段：他们是廉价作出，他们是非常崎岖的，你不用，以配合他们。
所以我很乐意当我看到nap140多氯联苯可在这里。
不过，我并不特别喜欢的电路。特别是，它使用了太多的积极负荷，我要使用的N沟道MOSFET 。
因此，我模拟的几个电路和这里是我的经验教训。
这里是示意图。 i通常设计我的安培上运行4 -欧姆的负载，以及1vp投入。在这里，安培是驱动一1vp 20 kHz正弦波。

2楼。here is the output waveform.
you see clearly massive cross-over distortion on the negative side. Not good.
这里是输出波形。
你清楚地看到，大规模的跨超过失真在消极的方面。也不是好事。

3楼。how about its frequency response?
here it is.
Something is seriously wrong in the 7-800Khz range. the amp is obviously self-oscillating there.
又如何，其频率响应？
在这里，这是。
一些是严重错误，在7 800khz范围。该安培，显然是自振荡。

5楼。does that mean it is hopeless?
not exactly.
some easy fixes, as proposed in this version of NAP140.
1) input resistor: 100K is simply too much.
2) I added R43/R44 to just balance out the effect of unmatched input transistors. 11 - 47ohm is typical. Higher numbers also reduce open loop gain, something I like.
3) R31: that was an obvious mistake in the original schematic. corrected here.
4) I also added R29, again to reduce open  loop gain and to make the amp more linear open loop.
5) R33/47/51 changed to a much lower number to increase the idle current on the drivers. They now idle about 30ma - just about right.
other changes are added to address stability.
As you can see, there is no more cross-over distortion, even at 100ma idle current.
这是否意味着它是无可救药的？
不完全。
一些容易修复，建议在这个版本的nap140 。
1 ）输入电阻： 100K的简直是太多。
2 ）我补充说： r43/r44公正平衡的影响无与伦比的投入晶体管。 11 -4 7ohm，是典型的。较高的数字也可以减少开环增益，一些我喜欢。
3 ） r31 ：这是一个明显的错误，在原来的示意图。纠正在这里。
4 ）我也补充说： r29 ，再次以减少开环增益，并作出安培更多的线性开环。
5 ） r33/47/51改变，以低得多的数目增加闲置目前对司机。他们现在闲置约三〇毫安-只需约的权利。
其他的变化，加上地址稳定。
你可以看到，有没有更多的跨超过失真，甚至在一百毫安闲置电流。

6楼。this is the frequency response. Not perfect but much better.
这是频率响应。并非十全十美，但好得多。

7楼。here is the same amp, with IRF540 as output devices and Vgs multiplier.
I added two gate stoppers and change some values, all highlighted.
这里是相同的安培， irf540作为输出设备和vgs乘数。
我已将二门塞子和改变一些价值观，都突出。

红双喜

谢谢楼上的 虽然翻译的不是很通俗，但也勉强看懂一点点了