What is techniques exist to preventing interferences between selfoscillating amps if two channels single chip used? Total and accurate DC decoupling, ok, i think it's done, gaped RM8 inductors with 60mm distance, whole GND layer (i've tried cut this layer to the half between the midline.. and yeah, not better) -done, external comparator -i've added diff. input stage (two BJT, 3 res). So, i achieve interference product from 520-524=4Khz about -95db in the output, maybe it's not so much level, but due to the live carrier behavior, it is VERY appreciable. Synchronisation..it's reduce loop gain, and do bad work around clipping too (hysteresis type modulator in use). Difference carrier (40-50khz) gave good result, but in the high output levels, carriers can be too closer, however interferences well masked, i can't hear it at least. To keep carriers difference about 50khz up to clipping, possible by the modulate the hysteresis..yesterday i have tried it too- nice clipping behavior, but the loop gain is reduced again. Now i'm looking for the "new" ideas about it, though simplest approach - slightly difference carriers, sounds the best so far (for >2 channels the troubles will back again, maybe).
I will try to add my little contribution...
Non synchronous designs (self oscillating) have that kind of problems. I have heard someone in this forum who proposed adding a capacitor between speaker terminals. That made the modules sync and run at the same frequency up to a given output level, but at the expense of increased distortion and possibly noise.
With synchronous (clock based) amps, the solution is to run all of them from the same clock. This increases transient supply requirements a little bit, but that can be alleviated by generating several phases of the same original clock, thus spreading the transitions and avoiding that they occur at the same time.
However, in both cases, proper wiring is the key in my opinion. From my experience (with clock based ones), using shielded twisted pair for the input connections is vital, as well as running it far from speaker and supply wires.
Separating the modules physically is also helpful, although I still haven't found a definitive solution other than synchronising them.
Grounding has also something to do, but I prefer that someone with more experience (i.e, Bruno and/or Lars) poste their advices here.
Experiences in this area are welcome!
Non synchronous designs (self oscillating) have that kind of problems. I have heard someone in this forum who proposed adding a capacitor between speaker terminals. That made the modules sync and run at the same frequency up to a given output level, but at the expense of increased distortion and possibly noise.
With synchronous (clock based) amps, the solution is to run all of them from the same clock. This increases transient supply requirements a little bit, but that can be alleviated by generating several phases of the same original clock, thus spreading the transitions and avoiding that they occur at the same time.
However, in both cases, proper wiring is the key in my opinion. From my experience (with clock based ones), using shielded twisted pair for the input connections is vital, as well as running it far from speaker and supply wires.
Separating the modules physically is also helpful, although I still haven't found a definitive solution other than synchronising them.
Grounding has also something to do, but I prefer that someone with more experience (i.e, Bruno and/or Lars) poste their advices here.
Experiences in this area are welcome!
- Status
- Not open for further replies.