You should use output transistor with high fT and you can push slew rate higher.
I have several design with this topology. It is very good at mid frequency, warmth, easy listening.
Last edited:
You can also try driver transistor from Toshiba like 2SC4793 or 2SC5171. I think for low power amplifier, those are better than ON Semi MJE15032.
Aren't Toshiba devices out of production, and those the regular DIYer can found, most likely be fakes?
For the same parts count some enhancements to try:
- Move the RC filter formed by R9 & C5 to the -ve rail, between the VAS and driver stage. The current sources arranged at the top will provide ample filtering (provided the Vrefs are stabilised) and filtering the -ve rail will improve PSR by reducing modulation of Cdom. D2 is probably best omitted.
- Personally I would never put a trimmer in the LTP tail - what if the wiper opens through adjustment or old age? Trimming the LTP tail current this way effects amplifier stability, and you will need to pay close attention to Cdom if you do this. If your goal is to achieve excellent DC offset then simply:
a) make R1 and R18 equal value
b) use high beta and reasonably well matched transistors for the input pair. BC560C (or similar) parts for Q1 and Q4.
c) equalise the LTP collector currents by sizing R14 appropriately. If you a "perfectionist" and really want to aim for 0.0mV offset then this is the place to put the trimmer. On the other hand you can get very good results with an R12 series resistor and excellent results with an R24 one, properly dimensioned.
- You can reduce ripple into the LTP by increasing R27. I would probably make R27 and "VR1" (both fixed resistors) equal value with the decoupler at the midpoint.
- Move the RC filter formed by R9 & C5 to the -ve rail, between the VAS and driver stage. The current sources arranged at the top will provide ample filtering (provided the Vrefs are stabilised) and filtering the -ve rail will improve PSR by reducing modulation of Cdom. D2 is probably best omitted.
- Personally I would never put a trimmer in the LTP tail - what if the wiper opens through adjustment or old age? Trimming the LTP tail current this way effects amplifier stability, and you will need to pay close attention to Cdom if you do this. If your goal is to achieve excellent DC offset then simply:
a) make R1 and R18 equal value
b) use high beta and reasonably well matched transistors for the input pair. BC560C (or similar) parts for Q1 and Q4.
c) equalise the LTP collector currents by sizing R14 appropriately. If you a "perfectionist" and really want to aim for 0.0mV offset then this is the place to put the trimmer. On the other hand you can get very good results with an R12 series resistor and excellent results with an R24 one, properly dimensioned.
- You can reduce ripple into the LTP by increasing R27. I would probably make R27 and "VR1" (both fixed resistors) equal value with the decoupler at the midpoint.
Interesting, will try all these, against the O/P original.
Aren't R9, C5, and D2 only in the +VE rail, part of a chef recipe mutuated from the AKSA 55 ?
Hi, I was playing with this circuit in LTSpice, and it seems to work pretty well, only annoyance is the assimetrical clipping behaviour (negative part of the waveform clip by first).What can cause this? it's related to the classA single transistor VAS?
Attachments
The LTP does not need much +ve voltage to operate with very high signal levels.
The feedback through C9 also allows the +ve side signals to go to very high levels.
It's the -ve side that is sensitive to maximum level and determined by the supply rail voltage.
The asymmetical clipping is inherent to the circuit operation and exaggerated by the C9 feedback.
Check that the amplifier exits from clipping cleanly. This is where differences in amplifier sound often come from. How they misbehave has a profound effect on the apparent performance.
Ensure you have sufficient unclipped output such that you never clip any of your audio signals.
20dB of overhead from average level to clipping level ensures that most music is not clipped, but some music/audio types can require 30dB of overhead from average to clipping. And some can manage with only 10dB of average to clipping overhead.
eg 500mW of average signal level into a 86dB/W @ 1m speaker gives you an average SPL of 83dB @ 1m and ~75dB @ 2.5m
20dB of overhead means specifying a 50W amplifier and 30dB of overhead means specifying a 500W amplifier.
Avoid clipping if you can afford it.
The feedback through C9 also allows the +ve side signals to go to very high levels.
It's the -ve side that is sensitive to maximum level and determined by the supply rail voltage.
The asymmetical clipping is inherent to the circuit operation and exaggerated by the C9 feedback.
Check that the amplifier exits from clipping cleanly. This is where differences in amplifier sound often come from. How they misbehave has a profound effect on the apparent performance.
Ensure you have sufficient unclipped output such that you never clip any of your audio signals.
20dB of overhead from average level to clipping level ensures that most music is not clipped, but some music/audio types can require 30dB of overhead from average to clipping. And some can manage with only 10dB of average to clipping overhead.
eg 500mW of average signal level into a 86dB/W @ 1m speaker gives you an average SPL of 83dB @ 1m and ~75dB @ 2.5m
20dB of overhead means specifying a 50W amplifier and 30dB of overhead means specifying a 500W amplifier.
Avoid clipping if you can afford it.
Last edited:
Thanks Andrew
I observed that raising the -ve a couple of volt over the +ve solve the problem , but it seems rather unortodox to do that in a real world amplifier...
I observed that raising the -ve a couple of volt over the +ve solve the problem , but it seems rather unortodox to do that in a real world amplifier...
Last edited:
Using battery or asymmetrical winding could be unorthodox, but when the front end has its own supply may be it is not.
But your asymmetrical supply includes the power output stage hehe. The original idea has "simple" word in it BTW (Blame it on Andrew 😀)
Do not go to asymmetrical supplies.
It's normal for your amplifier to have different clipping levels on +ve and -ve excursions.
It is far more important how it comes out of clipping.
Reduce R19 and increase C1
Remove D3
Change Q10 to low Cob (Try 2sc3423Y, or even a low Cob To92)
Decrease R8 & R11
Decrease R13 & R22
Decrease R15 & R23
Decrease R4
Decrease C8, massively.
Increase L1, massively.
Add resistor load between Q10 collector and ground. (Try 1k to 2k)
It's normal for your amplifier to have different clipping levels on +ve and -ve excursions.
It is far more important how it comes out of clipping.
Here are a few more "tweaks" to look at.
Reduce R19 and increase C1
Remove D3
Change Q10 to low Cob (Try 2sc3423Y, or even a low Cob To92)
Decrease R8 & R11
Decrease R13 & R22
Decrease R15 & R23
Decrease R4
Decrease C8, massively.
Increase L1, massively.
Add resistor load between Q10 collector and ground. (Try 1k to 2k)
Last edited:
Andrew, I don't agree with all your suggestions: -
- Q10's collector is grounded, therefore it is not modulated with signal and the Cob is a nonissue. The main VAS transistor should be a low Cob video transistor type.
- Why do you say the driver base stoppers need to be reduced? How much voltage is dropped at full power?
- If the purpose of putting a resistor into Q10's collector leg is for current limiting how big does this resistor need to be to stop it burning up? At a value it is going to be useful don't you think it will adversely affect slew rate?
- Status
- Not open for further replies.