400W Amplifier (Ebay/Aliexpress)

Hello everyone,

I ordered two PCB's from Aliexpress some time ago for an active crossover project where i just needed a decent bass amplifier.


The design looked reasonable and at a price of like 2 bucks each they fit the bill :)


So i just finished retracing the board and it looks like a simplified version of the leach amp.


The transistors used are a little too expensive for my project, thinking about replacing the input transistors with a pair of 2SA1015/2SC1815, 1145/2705 to 2SB647/2SD667 and drivers and output transistors to BD911/912.

Any thoughts or idea's are welcome.
Last edited:
You could also try KSA992/KSC1845 in the LTP, and KSA1381/KSC3503 in the VAS.

Stick a ".four 1khz V(out)" statement into your .asc, run the sim, open the Spice Error Log and scroll down a bit to see the THD.

(You may also need to modify your .transient statement. Maximum timestep should be on the order of 0.5u or 1u for 20kTHD, and 10u for 1kTHD.)
Are Q7 and Q8 correct in the first schematic? EF? The second schematic looks better as far as Q7 and Q8.


It's only Q7 in the first schematic that's wrong.

There are loads of transistors in wrong way around on the schematic if you follow power supply polarities.
Q5, 6 and 8 and wrong way around.
What a mess !

Typical half cocked Chinese attempt.......
Amazed it works at all.


2008-06-12 3:43 am
Yep, I just noticed Q9 is bassackwards too! I wonder if this is the manufacturer's schematic or the OP's, if the OP's it usually takes a few attempts to get it all correct. Sometimes you get lost staring at the board going from one side to the other. Been there, done that. My first attempt at an Ampzilla IIa schematic was a real challenge.

Last edited:
What a mess !
Typical half cocked Chinese attempt.......
Amazed it works at all.

I blame it on my reverse engineering 'skills' :rolleyes:

Thanks a bunch anyways for all the pointers! Jeff, i was able to get the simulation running. Really happy with that. Now with what i think is the correct mess i get:

Simulated distortion at 1khz, 2.83v into 8 Ohms gives me 0.008. Not bad at all.

And here's the latest schematic. I only had to add a bypass cap on the feedback resistor.

Spice file: View attachment 400W_AMP-mod_1.2.asc

Again any comments or pointers are very welcome!
Last edited:
The transistor working condition is different in pre-amplifier and power amplifier application, so that we can't change resistor value or supply voltage to bad condition.

According to your power amplifier application, I suggest values for you in your circuit diagram.

Very good recommendations, your feedback network works really well. Using the suggested feedback network with R7/R8 back to 22k makes distortion drop from 0.008 to 0.002.

However increasing R17/R18 increases distortion a lot from 0.002 to 0.016. Could anyone explain this?
I found more info on this VAS (or DIS-PP-VAS to be exact) in Self's 'Audio Power Amplifier Design', but i fail to see why distortion increases in the simulation. Can i test for DC offset with spice?

Audio Power Amplifier Design - Douglas Self - Google Boeken
This looks like a pretty good power amp for general use. We have made 1000's of similar amps over the decades. It can be improved of course. A DC servo would eliminate 2 electrolytic caps, and complementary, matched jfets instead of transistors would be a little better, but a big hassle with amateurs. That is what we use at virtually all price and power levels. It is good circuit and should be left alone, mostly.
However increasing R17/R18 increases distortion a lot from 0.002 to 0.016. Could anyone explain this?

R17/R18 value decided Q7,Q8 collector current which normal design suggest in 2.5mA to 3.5mA range for it's application, so that R17/R18 value range can be 220R(2.5mA) to 150R(3.5mA).

But I think Q7,Q8 use higher collector current 120R(4.3mA), the amplifier performance will sound more live with good bass. You can built the amplifier and try different values for your actual listening and make final value.
But why specify only TO220 power transistors? If you expect more than about 60W, you'll have to think again. If you want cheap, the Chinese Toshiba TTC5200/TTC1943 complements are are hard to beat, though they don't have all that much more to offer.

B aware that the h-fe of the BD911-912 differ from 40 to 250 !!
You have to select them by h fe . Buy 50 each at €0,40 each and
use the rest for winter faithful........
I remember an ELEKTOR amp with big problems caused by above reason

The Elektor amp was actually the inspiration to configure it this way. Long time ago on a DIY meeting there was someone who had built that amp, and i was really surprised of the deep bass and clarity. I've also used those BD911/12's on plenty of repairs, they are quite unique spec wise.

For sure a single pair of 1943/5200 could outperform 4 pairs of 911/912, but the cost would be the same. I like the high HFE and it helps to lower distortion, but like ingo says you have to match them. I've used CDIL's and ST's and usually out of every 10 transistors there is one that has a low HFE.
Power wise i'm not interested in high wattage, it will be used in an active setup with two amplifiers per speaker.

Now that i've looked up that Elektor amp again it looks awfully familiar: