Hi. Besides from the package, voltage, current, heat dissipation specifications etc, is there a difference between bjts when it comes to the sound? I've recently made a class AB headphone amplifiers using MJE2935 and MJE3055 transistors, but I have no idea of whether these are any good in an audio application? This amplifier is biased well above the cutoff-point and is mostly operated in class A, but I've several times thought to myself that the amp sound s better after it has heated up. I just now made another amplifier just like it, but thought it was a little strange that the sound didn't seem as good. I later noticed that the amplifier wasn't getting as hot as the first one and later noticed that I had chosen wrong resistors thus making the bias current much lower (still way above the cutoff-point though). I won't put any faith into my mega biased ears though, but I still think it's odd. So the question is, can bjts alone cause audible distortions if lets say the amplifier is biased a little bit above the cutoff point (so it operates mostly in B)? Let's put the line at 1% THD, I believe anything below this is inaudible.
MJE2955 and MJE3055 are designed for medium power audio amplifiers, so yes they are fit for purpose. Like all amplifiers, class A to B, one must consider the characteristics of the transistors, gain wise. At low current the gain will be different to medium current etc. So the signal will differ at the crossover point.
There's almost Mega-threads here relating to Bias current and sound quality, I believe the consensus is; it depends!
But mostly on the type of transistors you are using.
So - yes!
But mostly on the type of transistors you are using.
So - yes!
These work OK as audio transistors and in TO3, TO3P or even TO220 form, the've been used in countless, highly regarded amplifiers. Whether you need something with such a large die depends a lot on what peak currents the headphones draw. I notice a lot of builders assume their headphones are more like toasters than miniature speakers in a tiny room and design them for crazy high output levels. Sure, using them in class A requires more than 100mA bias current continuously, so whatever rail supply voltages are specified, this means using parts with the dissipation rating to handle it.
I'd also consider TIP31/32, TIP41/42 at the budget end of costs. If you want premium audio parts, the MJE1503X family look to be just right for low power amplifiers.
Last edited:
The quality of amplifier is not depend only from THD figure. Many variables involve here, such S/N ratio, frequency response, intermodulation distortion, PSRR, slew rate, etc.
If you satisfied 1% of THD then I think it is easy to design an amplifier using transistor which available at the market. Even using MJE2935 and MJE3055, you can make an amplifier with THD below 1%.
If you satisfied 1% of THD then I think it is easy to design an amplifier using transistor which available at the market. Even using MJE2935 and MJE3055, you can make an amplifier with THD below 1%.
Last edited:
Thanks for the responses. I would really like a way to measure distortion, but those analyzers are quite expensive and DIY solutions require great precision.
I see that the Richard Clark amplifier challenge sets a limit at 2% THD so it's clear that anything below this shouldn't be audible. He has also said several times that it's impossible to hear a THD value of less than 1%. This makes me wonder why people are so keen on measuring measuring distortion figures below this, seeing it as a victory to get the distortion from 0.2% to 0.02%.
My design is very simple, it has a +-12V transformer, NE5532 as an input stage with a class AB output stage. The output stage is biased using four resistors. I know that thermal runaway is a potential problem with this design, but I figure that it's not worth worrying about if the temperature settles at a reasonable level (which it does). I'm still struggling to understand where the distortion in this design will come from though. I'm using global feedback so with my limited knowledge I'm thinking that the output distortion would be the same as the inherent distortion of the op-amp as long as the op-amp manages to get the output it wants. As the output stage is biased well about the cutoff point, one transistor will begin to turn on before reaching the other half of the sine wave thus removing the need for the op-amp to overcome the threshold voltage of the bjts. Seen from the perspective of the op-amp, could someone please try and explain to me how the op-amp even knows that it has an output stage connected to it? Can the crossover distortion be educed by adding some component to this design: https://www.circuitlab.com/circuit/ynfes6/slew-rate-test/
I think I've handled the other aspects. The frequency response is good and the noise is only slightly audible when turning the pot all the way up (this is a hissing noise, not humming). I don't really know what intermodulation distortion is, how is this different from THD?
I see that the Richard Clark amplifier challenge sets a limit at 2% THD so it's clear that anything below this shouldn't be audible. He has also said several times that it's impossible to hear a THD value of less than 1%. This makes me wonder why people are so keen on measuring measuring distortion figures below this, seeing it as a victory to get the distortion from 0.2% to 0.02%.
My design is very simple, it has a +-12V transformer, NE5532 as an input stage with a class AB output stage. The output stage is biased using four resistors. I know that thermal runaway is a potential problem with this design, but I figure that it's not worth worrying about if the temperature settles at a reasonable level (which it does). I'm still struggling to understand where the distortion in this design will come from though. I'm using global feedback so with my limited knowledge I'm thinking that the output distortion would be the same as the inherent distortion of the op-amp as long as the op-amp manages to get the output it wants. As the output stage is biased well about the cutoff point, one transistor will begin to turn on before reaching the other half of the sine wave thus removing the need for the op-amp to overcome the threshold voltage of the bjts. Seen from the perspective of the op-amp, could someone please try and explain to me how the op-amp even knows that it has an output stage connected to it? Can the crossover distortion be educed by adding some component to this design: https://www.circuitlab.com/circuit/ynfes6/slew-rate-test/
I think I've handled the other aspects. The frequency response is good and the noise is only slightly audible when turning the pot all the way up (this is a hissing noise, not humming). I don't really know what intermodulation distortion is, how is this different from THD?
Heat and Hot have little to do with the sound quality.
Excessive Heat will impact on long term performance.
Generally amplifiers like to run Cool.
Bias makes a big difference to sound output.
Whether single ended, or push-Pull output stage, the bias determines the maximum ClassA output current.
If the output is always inside the ClassA region, then you cannot have any crossover distortion, because there is no crossover.
Most music and other audio have a big range from peak transient to average output levels.
This results in us listening at levels that can be inside the ClassA region for much of what we hear.
A small increase in bias could make a big increase in the ClassA region and this could make a big difference to the sound quality.
Excessive Heat will impact on long term performance.
Generally amplifiers like to run Cool.
Bias makes a big difference to sound output.
Whether single ended, or push-Pull output stage, the bias determines the maximum ClassA output current.
If the output is always inside the ClassA region, then you cannot have any crossover distortion, because there is no crossover.
Most music and other audio have a big range from peak transient to average output levels.
This results in us listening at levels that can be inside the ClassA region for much of what we hear.
A small increase in bias could make a big increase in the ClassA region and this could make a big difference to the sound quality.
How loud would that source be if the vol pot is turned that far up?
It could be that you simply have too much gain.
Have you read the article about system gain?
http://www.diyaudio.com/forums/diyaudio-com-articles/186018-what-gain-structure.html
The consensus is that it depends on the design and implementation of the circuit. One can build a great amp with cheaper inferior transistors but the designer must allow for in the design to mitigate the weaknesses these devices have. Gain and SOA is limited? Use more components to share the duty. Now Cob is relatively high? A more robust driver stage may be a solution. Perhaps some sort of distortion correction circuitry to linearize the transfer of a not so linear component is in order. Certainly it is easier and simpler to use better quality components to arrive at a set goal. It may become more complex and use more components to achieve the same goal to use inferior parts, but to say that amplifier quality depends on particular transistor type is arguably not true.
- Status
- Not open for further replies.