I can second this amplifier, I tested on breadboard and clip leads using IRFP140's. good little design and had no problems with it.If you get 0.04% distortion in the simulator on ideal components, then in the real world you will get 0.4%. This is a lot for such a complex circuit that works in class A. A few pages earlier, I gave a link to my project of the Hyperbolic amplifier - it contains half as many parts and at the same time produces distortions of 0.01-0.02% when operating in class A. These are measured distortions of the assembled amplifier, and not computer simulations.
Oh for sure there will be differences but I would hesitate to put a blanket figure on it and say its going to be ten times worse. So much depends (in the real build) on layout and track routing and grounding and of course component matching.If you get 0.04% distortion in the simulator on ideal components, then in the real world you will get 0.4%.
Somewhat 🙂somewhat fixed
You found the collector of Q6 and the resistor to Q4 collector were not even connected electrically. I'm not on my main PC where I have an updated version which I'll post later.
You still have a massive imbalance in input currents, R1 is way to low for a headphone amp where you need only a few milliamps in Class A to supply the load. R2 is far to low and you still need that resistor to Q4 collector.
ive set my eyes on the hifiman sundara closed or any other planar headphone and these things need some power the guy at the store said to look for something with like 1to 2 watts of power, i also demo-d them with an fiio k19 but it costs kinda a lot
replaced it with an op amp and i think a better front end + a current buffer should fix the issue, also i realised that c2 could be replaced with a supercapacitor
As a first port of call, I'd suggest lateral MOSFETs for the output stage. Some vertical MOSFETs are probably fine, as long as the bias circuit settles to a reasonable current, or has some active control which is harder to do, and the "safe operating area" graph in the datasheet includes a DC trace.
Maybe I'm wrong, but BJTs seem like more trouble than they are worth, at least for the output stage. One nice thing about bipolars is that they make nice cascodes / impedance converters. Say, for instance, you have a long tail pair, and you are trying to drive an output MOSFET, configured as common source. The gate will probably swing around ±1V, and the LTP will distort due to the Early effect (among others) if you attempt to drive it directly. But if you connect a cascode or folded cascode inbetween, the output of your LTP will only swing a few mV.
For low power, ie headphones, you could even use a sacrificial cascode above or below the output device, so the voltage modulation is in a separate device rather than across the output device.
Maybe I'm wrong, but BJTs seem like more trouble than they are worth, at least for the output stage. One nice thing about bipolars is that they make nice cascodes / impedance converters. Say, for instance, you have a long tail pair, and you are trying to drive an output MOSFET, configured as common source. The gate will probably swing around ±1V, and the LTP will distort due to the Early effect (among others) if you attempt to drive it directly. But if you connect a cascode or folded cascode inbetween, the output of your LTP will only swing a few mV.
For low power, ie headphones, you could even use a sacrificial cascode above or below the output device, so the voltage modulation is in a separate device rather than across the output device.
mosfets are more complicated devices and have all kinds off gate capacitances, weird voltage drops that vary from fet to fet and so on.
In post #126, Q4 is a common-emitter stage and the feedback goes to the negative op-amp input, leading to two inversions in the feedback path. Hence, the feedback around the loop is positive and the loop gain is huge. Should be a nice class A Schmitt trigger.
As well as the problem @MarcelvdG pointed out, it also looks like you might be using a generic opamp model rather than picking a suitable one from the library files.
how good would did be in practice as a speaker amp? im also thinking about introducing a double feedback loopTry this and try and understand how it works. The preset is a 100 ohm to trim offset.
View attachment 1335964
View attachment 1335965
View attachment 1335966
It would be pretty limited in its capabilities tbh without some rework, mainly because of the limited current capability of the output stage. Single output devices haven't enough current gain to be driven by Q4 and also the current sink would need more current. Darlington's or FET's might be an answer.
I'm not quite sure what you are thinking here. The output impedance of a common emitter stage is high and could not drive a speaker to useful levels. You could make a simple two transistor voltage amplifier with added output stage that could sound good. A single rail AC coupled design would be best and would avoid offset problems that would make a DC design much more complicated.
Hi All,
I'll apologize up front because I know my question does not belong here.
How do I convert a .asc file to a .jpeg file so that I can post it on the stack exchange?
Thank you,
Ralf
I'll apologize up front because I know my question does not belong here.
How do I convert a .asc file to a .jpeg file so that I can post it on the stack exchange?
Thank you,
Ralf
How do I convert a .asc file to a .jpeg file so that I can post it on the stack exchange?
You need to open it in LTspice first.
.asc is the file extension for 'schematic capture'.
- Home
- Amplifiers
- Solid State
- how do i improve this mess of an Class A amplifier?