So in my hunt for something approaching unique, I have cooked up this thing - it definitely has it's weaknesses, but the overall concept really appeals to me.
Here it is:
Here we have an interesting arrangement of J1, Q1 and Q2. Forming what I'm calling 'the stack'.
J1 makes the magic happen, providing not only constant current to Q1 but also a low impedance drive to the output transistor, connecting Q3's base directly to Q1's collector results in almost 10x distortion!
Q1 is the only voltage gain device and also deals the feedback correction from the base, meaning the whole amplifier is in phase.
Q2 is needed simply to drive Q1 with a low impedance source while presenting a high impedance to the audio source.
Yes, Q4 is biased by a LED, forming probably the simplest current sink possible. This decision was made earlier on as I had trouble with oscillation, and besides, it works and distortion and stability is good enough for my liking 🙂
Lets talk distortion.
Here we have 1k - 8Vpp into 60Ohm's:
And 1k - 16Vpp into 60Ohm's:
10k 8Vpp into 60Ohm's:
10k 16Vpp into 60Ohm's:
....
But here's the thing, the FFT for a Loop back reading (straight wire from output to input) of my kinda crappy sound card.
Loopback 1k - 16Vpp:
So yeah, what I take away from this analysis is I need a better sound card 😕 but.. the distortion from my amp is probably much lower than I first thought 🙂
Stability seems OK, I have tortured it with 100nf capacitance and up on the output without a problem, but it did not like 33nf and oscillated without a large capacitor over the rail next to the output transistors. I may need help or at least conformation this is not a problem.
Here I have the 1k square response:
I'm surprised that this good performance can be had with just the voltage gain and feedback from one transistor (yes even a low gain circuit).
This idea really appeals to me and I don't see why just about any transistor can't be used.
I hope some will find this useful somehow and if it gets the pass here I'm going to design a PCB and make it neat in a nice box!
Maybe someone else can build it to see what they think? maybe someone with better equipment too 😉
I have attached the LTspice files and models. 👍
Here it is:
Here we have an interesting arrangement of J1, Q1 and Q2. Forming what I'm calling 'the stack'.
J1 makes the magic happen, providing not only constant current to Q1 but also a low impedance drive to the output transistor, connecting Q3's base directly to Q1's collector results in almost 10x distortion!
Q1 is the only voltage gain device and also deals the feedback correction from the base, meaning the whole amplifier is in phase.
Q2 is needed simply to drive Q1 with a low impedance source while presenting a high impedance to the audio source.
Yes, Q4 is biased by a LED, forming probably the simplest current sink possible. This decision was made earlier on as I had trouble with oscillation, and besides, it works and distortion and stability is good enough for my liking 🙂
Lets talk distortion.
Here we have 1k - 8Vpp into 60Ohm's:
And 1k - 16Vpp into 60Ohm's:
10k 8Vpp into 60Ohm's:
10k 16Vpp into 60Ohm's:
....
But here's the thing, the FFT for a Loop back reading (straight wire from output to input) of my kinda crappy sound card.
Loopback 1k - 16Vpp:
So yeah, what I take away from this analysis is I need a better sound card 😕 but.. the distortion from my amp is probably much lower than I first thought 🙂
Stability seems OK, I have tortured it with 100nf capacitance and up on the output without a problem, but it did not like 33nf and oscillated without a large capacitor over the rail next to the output transistors. I may need help or at least conformation this is not a problem.
Here I have the 1k square response:
I'm surprised that this good performance can be had with just the voltage gain and feedback from one transistor (yes even a low gain circuit).
This idea really appeals to me and I don't see why just about any transistor can't be used.
I hope some will find this useful somehow and if it gets the pass here I'm going to design a PCB and make it neat in a nice box!
Maybe someone else can build it to see what they think? maybe someone with better equipment too 😉
I have attached the LTspice files and models. 👍
Last edited:
Power supply ripple rejection look very good also, I'm thinking separate board's for L R and start ground between them. I am still unsure if an output relay will be needed. It is probably a good idea if in doubt though.
Started PCB design. The ceramic caps will be the multi-layer types.
Connectors are JST and 'Spade' / Faston types.
Connectors are JST and 'Spade' / Faston types.
This idea can be simplified down and made more minimal, but I think it will need M2 to be carefully picked and matched to acquire the correct bias voltage.
BJT:
Bootstrapped MOSFET Version:
Both have similar distortion profiles.
All just thoughts, these have not been tested, only simulated.
BJT:
Bootstrapped MOSFET Version:
Both have similar distortion profiles.
All just thoughts, these have not been tested, only simulated.
Most of the distortion is from the DAC its self.
It works well, very stable and sounds very transparent and effortless bass (from what I can tell from 1 channel).
It maybe a while to order PCB's as I'll probably do a bulk order with some other projects.
Looking forward to hearing this project in stereo properly 🙂
It works well, very stable and sounds very transparent and effortless bass (from what I can tell from 1 channel).
It maybe a while to order PCB's as I'll probably do a bulk order with some other projects.
Looking forward to hearing this project in stereo properly 🙂
