I had always thought that if the input cap is more than 2uF, there's probably an error. . . somewhere else.
My LM1875 projects, 22 watts max, are fine tuned at the voltage amplifier and all run on 36va or less (class ab efficiency of 62%. . . 36VA*0.62=22Watts, so that the chip doesn't break). None have more than 11,000uF per monobloc. Despite the fact that the optimum (for split rail monobloc) is 13,200uF~15,000uF, the main thing is that your transformer get the caps charged INSTANTLY following every bass note. Sometimes the transformer is strong enough to do that with big capacitance, but when it is not, we use smaller capacitance. The caps are clearer when full, no matter how that is arranged. The amplifier is clearer when it can amplify whatever bass is at input.
Starting from 2k7 with 240u for feedback-shunt leg, you can half and double your way to finding a proper RC for good clear bass (and clear everything else) (caps of 680u and larger need a <> bypass diode clipper to prevent excessive charge-up that could otherwise blow out the inverting input). In this case, if the inverting input coupler turns out inconveniently large, you can either deal with it (see my TDA7293 parallel amp thread for this mod) or use a smaller non-inverting input coupling cap. Whatever bass is inputted, needs to be amplified cleanly in order for the amplifier to have useful resolution.
If you were to halve the cap value above, such as 2k7 with 120u, then the bass distortion would increase only very slightly and would have a "warming" effect. In the case of the overture chips, a bit of warming may be desirable, such as 1k with 220u or 2k7 with 100u or 3k3 with 100u (a really very easy way to reduce inverting coupler sizes so that the amplifier might resist a bit of input error).
Last edited:
Hi danialwritesbac, I was thinking in terms that the 0 volt rail isn't behind the feedback network like the +/- rails are. The feedback to a large degree will compensate for voltage and current fluctuations (assuming your well below clipping) but voltage fluctuations in the 0 volt rail could have an audible effect on the sound
Given that electrolytic caps tend to have higher ESR at low frequencies than they do at high frequencies this would effect bass more than treble in terms of amplifier output impedance, it could also be a source of IMD due to the production of low frequency voltage fluctuations in the 0 volt rail that all the other frequencies then ride on
Given that electrolytic caps tend to have higher ESR at low frequencies than they do at high frequencies this would effect bass more than treble in terms of amplifier output impedance, it could also be a source of IMD due to the production of low frequency voltage fluctuations in the 0 volt rail that all the other frequencies then ride on
Last edited:
Wouldn't you need to place the regulators at the chips pins to get the best result?
I started a thread http://www.diyaudio.com/forums/chip-amps/252122-conflicting-theories-please-help.html
I could not get a good explication why capacitor size does change bass output. Maybe yours does explain this.
I agree with Andrew on the 90ms filter, although I find it to also be cap dependent as well... With some slower or lower bandwidth ones ones I tend to like ~200ms more.
I am currently using 20mF per channel and I plan to double it for a test before getting my final Sikorel caps.
As for the error elsewhere, if we had 100K instead of 22K in that input impedance we wouldn't have to use so big and ugly coupling caps...
The datasheets explain how to design such an amplifier, but I am just scratching the surface of how these things work and what happens to their bandwidth etc with changes in gain, feedback etc and how they have to be balanced... it is way out of my league for now...
The difference for me was quite bigger than plain intelligibility of bass because of lower SNR.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.