How do you know how good your diy amp is?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Jump to Latest
east electronics

east electronics

Member
Joined 2006
eastelectronics.gr
akis said:
A listening test may perhaps pick up gross problems, like huge distortion, hum, other noise, but if it sounds "OK" you will never know what "OK" is supposed to be. So you may listen to a music track on 3 different systems, and they all could be "OK" but at the same time all different. Which one is closer to the original sound? And how do we define "original sound". There are so many variables it is impossible to quantify.

I have been building a guitar pre-amp for about 6 months (on version 3 now). First version worked, with slight bugs that were fixed on the PCB but overall design could be heavily improved so am re-designing it.

Here's what I do for pre-amps and low wattage ( < 1 W ) amps:

I do not use op-amps.
I do not use feedback.
I do not use class AB or long tail paired transistors (except final stage of power amp which is AB).
Use a twenty-times rule for biasing circuits (to help with temperature changes, tolerances and soldering changes).
I do not use more than one active device on the signal path at any stage (eg no current mirrors, cascade etc).
I *may* just use a transistor to regulate the supply rail but I am not too keen on that idea either.


I use an Excel spreadsheet to calculate bias points and then I build a bread board to do some measurements with oscilloscope and function generator and bench power supply. The problems here are: (a) the breadboard has stray capacitances and resistances which will have you scratching your head and waste hours chasing shadows (b) the bench power supply is perfect and in no way simulates a basic PSU that you would build yourself, so then you get hit with hum when you do not expect it and (c) it takes a very long time to make slight adjustments and retest.

So now I use a SPICE simulator (Multisim 10) to set out the parameters across the circuit. It takes a very long time on Multisim, it would take 10x time on breadboard.

I aim for low distortion (less than 0.1% typically 0.020%-0.050%) and also eye inspection of the curves on simulated oscilloscope to make sure they look OK (eg symmetrical).

Clip every stage on purpose to make sure it clips symmetrically (as much as possible).

Try to keep currents as low as possible to minimise noise, and at the same time over a threshold to maintain a linear beta. Beta and low Ic threshold measured experimentally (not depending on SPICE models).

I hope the end result will sound better than version 1, which does not sound bad at all anyway. It is already the better amp between the Fender and the Vox.
Click to expand...

this particular approach to my opinion is wrong ... this approach akis might be enough for some type of amplifiers only ...

example what is your approach for an amplifier rated 100 w ??? if you work like that you stand no chance ...

as a conclusion of this quoted and the other that doesnt like measurments i would say that designing is chapter A and has to be done under specific targets and then specific rules ....

and after that, constructing is chapter B that also requiers expirience and focus to the primary target ....

under that rule evaluating is relativelly easy ...the only thing you have to do is a check list to your primary traget .....

example : all the above is pointless if you design /construct/evaluate a 400w amplifier that is done with bugjet limitations ....or power supply limitations or poor soar limitations ...and goes on and on and on ....

there is got to be NO GENERALL RULE .... everything has to be tested and evaluated versus all the above
 
C

Conrad Hoffman

Member
Joined 2007
conradhoffman.com
Some say measurements don't tell you how an amp will sound. I say bee s. If your measurements don't correlate with sound, then you're measuring the wrong stuff, or measuring it poorly. There's another assumption people take on faith as correct and it probably isn't- that an electrically perfect amp, i.e., a straight wire with gain and zero impedance, will be the best sounding amp. I suggest it won't be. You can almost always fiddle around with response and impedance to get a more pleasing sound with a given set of speakers, or even a bunch of different speakers. Only you can decide what your design target is. Ain't no magic here. Amplifiers that produce a deep null in a null test will sound the same. Amplifiers that don't produce a deep null have errors that can also be measured on the test bench. There's no evidence I know of that suggests the human ear can detect error levels below what modern equipment can measure in the electrical chain of the system. Most arguments against this will require you to believe that identical time/voltage signals at the speaker terminals can produce different audible results- a claim that would require extraordinary proof to even consider. :deerman:

CH
 
N

nikwal

Member
Joined 2005
dsl.mine.nu
First obviously, the signal has to look ok on my digital scope during development,on different loads and so on.. Then listening and if I still sounds ok after listening a long time(several cd's) then it's time for a brake a few days at least. If it sounds Awesome(should be like, "DAMN, did I build that?") after that and so on I pretty much consider it ok.
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Top Bottom