Over on danielwritesbac's thread about the TDA8932 there's some interest in winding one's own input trafo for these chips so here's my quick-and-dirty guide to saving yourself a fair wad of cash (at the expense of your time) in building your own transformer. Their audible advantages are already taken as read so here I'm just going to deal with the steps to be taken, with the absolute minimum of math.
Step 1 - what frequencies do you need your transformer to pass?
You only need to be concerned with the low-frequency end as the top will look after itself. Transformers get bigger (and generally need more turns) the lower their working frequency so if you want to end up with the smallest possible size, don't ask your creation to deal with more LF than you absolutely need.
For the purposes of this example I'll assume you want to go down to 20Hz at full level, but no lower. The trafo can handle lower frequencies than that but only at reduced amplitude.
Once you've answered this initial question you can move on to calculate the Volt-seconds for your design. This is a crucial parameter in the design, the more V-s the more turns of wire you have to put on, given a particular physical size of trafo.
I've made a sketch of how to calculate the V-s parameter given your answer to the first question. The top squiggly line is my attempt at a sinewave with a geometrical horizontal axis, the lower one is what you might see on a scope with time as the X-axis. I'm showing the first one because we typically use calculus to determine the area under a curve, the area calculation is what's need to determine V-s, shown as the Greek letter 'lambda'. The lower wave shows a time axis - with 20Hz the whole cycle takes 50mS, for an amplitude of 1V peak, the V-s in the hatched area comes out to be 0.0159.
I got this result by using the area calculation from calculus which says the area under the sine is 2/Pi times the horizontal dimension (25mS in this case).
Step 1 - what frequencies do you need your transformer to pass?
You only need to be concerned with the low-frequency end as the top will look after itself. Transformers get bigger (and generally need more turns) the lower their working frequency so if you want to end up with the smallest possible size, don't ask your creation to deal with more LF than you absolutely need.
For the purposes of this example I'll assume you want to go down to 20Hz at full level, but no lower. The trafo can handle lower frequencies than that but only at reduced amplitude.
Once you've answered this initial question you can move on to calculate the Volt-seconds for your design. This is a crucial parameter in the design, the more V-s the more turns of wire you have to put on, given a particular physical size of trafo.
I've made a sketch of how to calculate the V-s parameter given your answer to the first question. The top squiggly line is my attempt at a sinewave with a geometrical horizontal axis, the lower one is what you might see on a scope with time as the X-axis. I'm showing the first one because we typically use calculus to determine the area under a curve, the area calculation is what's need to determine V-s, shown as the Greek letter 'lambda'. The lower wave shows a time axis - with 20Hz the whole cycle takes 50mS, for an amplitude of 1V peak, the V-s in the hatched area comes out to be 0.0159.
I got this result by using the area calculation from calculus which says the area under the sine is 2/Pi times the horizontal dimension (25mS in this case).
Attachments
Last edited: