[RnB180]

From ground zero noob status to designing youre own headphone amplifier,

is it difficult?

 

[Garbz]

Depends on what you class as your own headphone amplifier. It's not too difficult to learn to use opamps and strap together some pretty creative and great sounding amps, however if your own amp means giant mess of discrete circuitry then there's quite a steep learning curve involving lots of maths to do it properly.

 

[skyskraper]

having recently started playing with a pcb design (albeit with the schematic design done by someone far more talented then i

) my advice to you would be to go and read as much literature as you can, and also try and build as many different amps as you can so you can get an understanding of the circuits and circuit boards.

 

[SnoopyRocks]

Quote:

Depends on what you class as your own headphone amplifier. It's not too difficult to learn to use opamps and strap together some pretty creative and great sounding amps, however if your own amp means giant mess of discrete circuitry then there's quite a steep learning curve involving lots of maths to do it properly.

It's a misconception that a lot of math is required. Math is only really meaningful in this context when it directly leads to insight into design trade-offs or helps with understanding functionality. With proper modeling, relatively simple analyses can reveal a lot. Unfortunately, by the time you can do this yourself, you've long since passed noob status and can (probably) work through the brute force solution that a timely observation completely obviates.

Knowledge of a handful of basic structures is all that's necessary to get started with your own discrete designs. The advanced stuff is merely the simple structures creatively arranged. Putting it all together is the tricky part.

 

[tangent]

Quote:

Originally Posted by SnoopyRocks It's a misconception that a lot of math is required.

True. Frequently, you can use a free SPICE program to act as an overgrown calculator to avoid having to do any math. (But beware, SPICE sometimes tells lies.)

The main reason I'm responding, though, is that a lot of people use the word "math" to mean everything from arithmetic through postgraduate mathematics. But those who know advanced mathematics tend to reserve it for only the more advanced topics, and use specific words -- arithmetic, algebra, etc -- when talking about the more basic things. So when someone knowledgeable about such things says "you don't need a lot of math", that doesn't mean you should ignore the opportunity to exercise a little trivial algebra (say, Ohm's Law) now and then. It means that if the finding an accurate answer to a problem requires calculus and circuit analysis, there may be a way to bull your way past the problem some other way.

 

[RnB180]

interesting, I think I will stay with the typical DIY amps for now

 

[SnoopyRocks]

Quote:

Originally Posted by tangent True. Frequently, you can use a free SPICE program to act as an overgrown calculator to avoid having to do any math. (But beware, SPICE sometimes tells lies.)

This is a slippery slope because dependence on the simulator tends to be substituted for actually thinking about the circuit. Rarely will the software actually tell a lie: i.e. incorrectly perform a computation. Garbage in - gargabe out.

Quote:

Originally Posted by tangent The main reason I'm responding, though, is that a lot of people use the word "math" to mean everything from arithmetic through postgraduate mathematics. But those who know advanced mathematics tend to reserve it for only the more advanced topics, and use specific words -- arithmetic, algebra, etc -- when talking about the more basic things. So when someone knowledgeable about such things says "you don't need a lot of math", that doesn't mean you should ignore the opportunity to exercise a little trivial algebra (say, Ohm's Law) now and then. It means that if the finding an accurate answer to a problem requires calculus and circuit analysis, there may be a way to bull your way past the problem some other way.

+1

 

[jefemeister]

It's easy to make an amplifier that will pass sound through it. To make it of the highest quality is another issue and requires years of experience. Some things just can't be taught and need to be learned.

 

[pabbi1]

Absolutely correctly spoken. Simply look at the sheer number of designs that Kevin Gilmore and Pete Millett have between them (posted - more than 20), and who knows of the number they never bothered to post due to whatever unacceptable parameters. While they have learned something from the (successful OR failed) design, there was no need to share with the public at large.

That said, I agree with the OP to work with / tweak / destroy existing designs first as a flatter learning curve, then share what you know, or learn what you don't.

More often in DIY, we share what we don't know, and simply re-learn what we do.

 

[tangent]

Quote:

Originally Posted by SnoopyRocks Rarely will the software actually tell a lie

That's strictly true, but there are a whole lot of underspecified macromodels out there. It's common for these models to approximate reality quite well as long as you're doing normal things, but diverge wildly when you do something a little strange.

Also, your average simulator won't tell you, "this is the correct answer, but you might beware that it requires putting 45W through R3...."

 

[drewd]

Regarding math, I'm with Tangent 100%. I took more math classes than I want to think about when I studied for my EE degree, but the end result of having to know the math wasn't to be able to sit down and calculate currents through a transistor or to figure out the impulse response of a digital circuit - the reason for it was to have a feel for what the solution should be, at least within an order of magnitude. That way if the simulator (or the actual circuit) kicks out a result that is outside the bounds of reason, it's recognizable as such, or at least you'll notice that it doesn't match up with what you expected.

But if rough approximations aren't your thing, there are all sorts of reasonable approximations using arithmetic or algebra that get fairly close to calculus-based solutions. And any physicist worth his salt will tell you that engineering is all about approximations

The last time I calculated an integral or solved a differential equation was in college. I'm pretty sure that I couldn't do it now. But I can intuitively recognize if the solution given is way off. And, broadly speaking, that's the value of math in amplifier design.

Oh, and the simulators will simulate whatever you tell them. Even if you tell them wrong.

-Drew

 

[jbloudg20]

Quote:

Originally Posted by drewd The last time I calculated an integral or solved a differential equation was in college. I'm pretty sure that I couldn't do it now. But I can intuitively recognize if the solution given is way off. And, broadly speaking, that's the value of math in amplifier design. -Drew

Thats really not what I wnat to hear, considering im in Uni for engineering right now. However I have all my math slasses out of the way.

 

[gsferrari]

Take my advice - talk to the current designers and request a 3 month internship with them (not full time...just on-off).

Examples : kevin Gilmore, Pete Millet, Phil. AMB/Morsel/Tangent...

I'd rather eat my gym socks than intern with Kevin Gilmore and he would rather eat a transformer than teach me

But I cant think of anyone more colorful and opinionated and talented - perfect if you ask me

The idea is to follow their progress and learn as they move along. Nothing better than to learn by experience and hear everything straight from the designer's mouth. Sign an NDA...whatever is required...

Books can only teach you so much. I have wanted to do this for a long time but once you get sucked into a job - finito!

If you have the time now - go for it!

gs

 

[pabbi1]

For instance...

I spent a couple of hours picking Pete Millett AND Ray Samuels brains Saturday, and was so pumped I could hardly stand it. It certainly helped that many of their designs were there in tangible form, and rather than starting from scratch, what could be done to several to make them better.

My dream is that someone would revisit Pete's HA-4 (on his web site) - while it needs some updating, as Ray said, "With this amp, what else could anyone need?". Now, certainly that was a compliment to Pete, but I verified it with a couple hours of listening - UFB.

Millett HA-4

While Pete only made 5, there should be more, but updated with some newer parts, and biased into class A. For a 6 year old design, this thing rocks, AND has built in EQ and cross feed (with a defeat switch). THAT is what an experienced (Master?) designer brings to the table.

GS and KG? Well, start with a radioactive power source from an ion implanter, and go from there.

 

[chesebert]

took me 4 years to learn, 3 weeks to design/build and 2 years to forget everything.

now I am back being a noob with a EE degree...