"Audio Grade" Transistors

[DanyBoy55]

Hello guys,
I've got an ok understanding of the theory and some basic practical knowledge of analog amplifier circuits (good old Sedra and Sergio Franco) And while these books dive deep into frequency response, in/out impedance and some other basic parameters, they discuss next to nothing about the selection process for actual electronic components. Which leads me to the question, what makes a component "good" for being in the signal path of an audio amplifier? Why is a TIP31/32 pair not ok but a 2SA1859/4883 pair is? Is it something about the Ic/Vce curves? Is it the unity bandwidth frequency? I hope you see what I'm trying to ask. On the op-amp side it's a bit clearer to me: low noise, fairly high slew rate, high CMRR, unity gain stable (I think). I hope you guys can help me a little bit to understand why a certain transistor is better than another. Or if its the case, why it doesn't matter. I'm genuinely curious.

Since we are in lockdown, I've decided to build and measure a variety of headphone amplifier circuits. Everything from the crappy class A to AB with and without feedback and various biasing techniques. Ultimately I would like to build an AB amplifier with feedback and potentially feedforward (once I finish with Ogata) I'm doing this to learn not to compete with current audio products. Attached is a ultra realistic simulation of one of the circuits I'll eventually build

Thank you!

 

[amirm]

The first order of business is selecting a transistor with current (Ic) and Voltage (Vce) capability and margin for the design. These are definitely critical aspects for the final drive transistors in your push-pull configuration. Packaging, heat dissipation and cost round up the minimum requirements.

 

[solderdude]

My first remark would be why you made a decision to build an inverting amp with a 500 Ohm input resistance (at least it is at 1GHz, at 20Hz it is 140 MOhm) and why the high-pass cutoff point (-3dB) is at 6MHz ?
Do you intend to only drive 600 Ohm headphones with it ?

Just now saw the input cap is 56uF and not 56pF so -3dB point is higher than 20Hz, with a 32 Ohm headphone -3dB = 150Hz.

When you actually would build this device I don't expect you will hear much of anything.

Why not build a design that is already out there that would actually work ?

 

[DanyBoy55]

My first remark would be why you made a decision to build an inverting amp with a 500 Ohm input resistance (at least it is at 1GHz, at 20Hz it is 140 MOhm) and why the high-pass cutoff point (-3dB) is at 6MHz ?
Do you intend to only drive 600 Ohm headphones with it ?

When you actually would build this device I don't expect you will hear much of anything.

Why not build a design that is already out there that would actually work ?

Simply because I really don't know what are real world requirements for audio equipment. I have not actually built anything yet, but I'm still reading and widening my understanding of theory. What's a better input impedance? I've read that about 20k is good for line level? Any good references so that I don't have to invent the wheel twice? (which would probably result in a square wheel)
Thanks!

 

[solderdude]

The wheel has already been invented many times over.

The input impedance can be anything you like it to be.
There is something to be said for every value between 5k and 100k.

You will need to worry about other aspects. You need to decide on output power, output resistance, feedback, feedforward, topologies, load impedances, frequency and phase response, distortion products and profiles, costs, balanced in/out or not, stabilty, capacitive load capabilities.
Then there is component choices to be made and power supplies as well as heat management.

Using someone's design and taking it to the next level or re-inventing the wheel ?
Top of the line or budget ? Specific components in mind... and why.
Just realize that every design is a compromise, at least when you want it to be affordable.

When you made up your mind on specs and design you can do simulations and ultimately the PCB design. The higher the spec the more important the PCB design plays a role.
You need to be quite savvy with layouts and understand groundplanes, decoupling and limits of parts to get this going right.

Here you can find info about designs and some practical pointers/explanations

 

[Bob-23]

Simply because I really don't know what are real world requirements for audio equipment. I have not actually built anything yet, but I'm still reading and widening my understanding of theory. What's a better input impedance? I've read that about 20k is good for line level? Any good references so that I don't have to invent the wheel twice? (which would probably result in a square wheel)
Thanks!

Here's very practical advice with regard to some of your questions:
"Designing an Opamp Amplifier"
https://web.archive.org/web/20150315014729/http://headwize.com/?page_id=147
Have a good built!

 

[JohnYang1997]

I see a few steps.
1, Objective 2 is a good starting point. It allows you to understand a whole lot without worrying about actual operation of transistors. Once you are able to modify to achieve better performance you can proceed to the next one.
2, Take a look at the circuit of Lehmann, Beyerdynamic A1, LISA3. These are opamp + discrete output stage. You are touching more about operation of transistors and negative feedback.
You can build and or mod similar products to pursue similar objectives as first step. You need to be comfortable with simulation of loop gain, stability, square wave etc. Once you are comfortable with these you can go to the next step.
3, Read power amplifier books by Dougals Self and Bob Cordell. Learn everything in the books, build the circuits, test the circuits. Try to modify the circuits. Understand why it's better or why it doesn't work. Have a few circuits built and working then go to the next step.
4, Now you should know what you are looking for and how to do it. Go ahead and create something amazing!

 

[DanyBoy55]

Thanks to @JohnYang1997, @Bob-23 and @solderdude . From what you tell me, I've been looking at the design process in the wrong way. I'll set myself some reasonable goals and start with that in mind. Thanks for your advice and knowledge!

 

[Bob-23]

In addition to NwAvGuy's 'Objective 2' site, which JohnYang1997 pointed to, I'd recommend the world's best diy-audio-site: Rod Elliott's 'ESP' - what this man has done for the world wide diy-audio-community is simply great!. His Project 113, btw, ressembles yours.
https://sound-au.com/index.html
http://nwavguy.blogspot.com/2011/07/o2-design-process.html
O2-pcbs, and even whole kits are available, for instance, here:
http://www.headnhifi.com/
But it might be a good idea to develop your own board...

 

[LTig]

Before diving into unknown territories I recommend to read a good book which is understandable for most laymen: The Art of Electronics by Horowitz/Hill.

 

[amirm]

Before diving into unknown territories I recommend to read a good book which is understandable for most laymen: The Art of Electronics by Horowitz/Hill.

That's a bible everyone needs to have. But it is more of a reference than a tutorial for a novice.

 

[amirm]

I am going to against the grain of suggestions and say to build discrete circuits and learn from them. Building op-amp amps is just an exercise in assembling parts. If your goal is to build functional stuff, then yes, that is a good path. But if you want to learn, the path you are on is a good one.

 

[DanyBoy55]

Thanks again for the suggestions! As soon as I have some half decent results (better than my scope noise floor and dynamic range) Ill post some results.
Cheers!

 

[Bob-23]

Thanks again for the suggestions! As soon as I have some half decent results (better than my scope noise floor and dynamic range) Ill post some results.
Cheers!

Yes, go ahead! Your building from scratch was very 'courageous', and look at Rod Elliott's Project 113: the similarity shows you're on the right path.

 

[solderdude]

Just remember to decouple power pins and for God's sake use higher value or no coupling caps at the output.