I'm sorry to reanimate such an old thread but it is the subject that I am addressing and why not?
I know there are those who just abandon all efforts to do any matching and count on the designs to be tolerant of mismatches, but in any case, it's always better to do the matching than not. The designs may be somewhat tolerant, but they're not totally impervious to the mismatches, so if it can be done, why not make the effort and get a better result.
For one thing, one side effect to be expected from the matching, especially for the power amps input diff transistors, is a possible reduction of a DC offset on the output, which is a good thing, not to mention the improved linearity.
Obviously, in the hope of finding decent matches, especially on more than just pairs, the batches to draw from must be significantly large enough, or there is likely little chance of finding close enough matches.
It's not just small signal input stage transistors, and in fully symetric designs, the vas, the drivers and pre-drivers, and the output power transistors, that can benefit from the matching, but even zeners, when used, and certain resistors and sometimes other diodes.
I am thinking that since they are ubiquitous and dirt cheap, with plenty of versatility, plus the built-in computer link to more data, the arduino can be used as the base to make the testers.
I've seen some examples of such setups, that were rather single minded and aimed at only specific types of parts, but that can be expanded upon to tackle a larger variety of parts. And with the available freeware that can be used on the computers to process the data, there would not be very much more needed to get a really efficient matching system.
With an arduino and a little more on a breadboard, the testing can be set up, and the data sent back to the computer, then that data can be plotted by other software for display on screen. And I would go even further, with a little more software that would have to be written, to take data sets for parts and do comparison by program instead of doing it visually.
This way matches could be detected more easily and precisely with little effort and be less prone to errors. Data sets from parts could even be plotted together on a single graph, superimposed, which quickly allows to visualize differences and matches.
I have a few arduino boards and I will be looking for what else could be needed to make something like this to work.
I am thinking about a set up that would allow matching of power transistors, in their area of expected function.
I know there are those who just abandon all efforts to do any matching and count on the designs to be tolerant of mismatches, but in any case, it's always better to do the matching than not. The designs may be somewhat tolerant, but they're not totally impervious to the mismatches, so if it can be done, why not make the effort and get a better result.
For one thing, one side effect to be expected from the matching, especially for the power amps input diff transistors, is a possible reduction of a DC offset on the output, which is a good thing, not to mention the improved linearity.
Obviously, in the hope of finding decent matches, especially on more than just pairs, the batches to draw from must be significantly large enough, or there is likely little chance of finding close enough matches.
It's not just small signal input stage transistors, and in fully symetric designs, the vas, the drivers and pre-drivers, and the output power transistors, that can benefit from the matching, but even zeners, when used, and certain resistors and sometimes other diodes.
I am thinking that since they are ubiquitous and dirt cheap, with plenty of versatility, plus the built-in computer link to more data, the arduino can be used as the base to make the testers.
I've seen some examples of such setups, that were rather single minded and aimed at only specific types of parts, but that can be expanded upon to tackle a larger variety of parts. And with the available freeware that can be used on the computers to process the data, there would not be very much more needed to get a really efficient matching system.
With an arduino and a little more on a breadboard, the testing can be set up, and the data sent back to the computer, then that data can be plotted by other software for display on screen. And I would go even further, with a little more software that would have to be written, to take data sets for parts and do comparison by program instead of doing it visually.
This way matches could be detected more easily and precisely with little effort and be less prone to errors. Data sets from parts could even be plotted together on a single graph, superimposed, which quickly allows to visualize differences and matches.
I have a few arduino boards and I will be looking for what else could be needed to make something like this to work.
I am thinking about a set up that would allow matching of power transistors, in their area of expected function.