Perhaps you can give my another little help:
1.
I want to use a single big toroid for 4, maybe 5 lm3886.
I'm going to start with 2200µF (+&- each) soldered directly at the pins.
If I understood these snubber theories, the transformer and the capacitors form an oszillating circuit. All people recommend not to use more capacitance because of that.
But in my case, the transformer will "see" 22.000µF of all ics.
What will happen?
Better use 5 small transformers?
2.
If I decide to big capacitors with snubbers later, wich disign is the best?
Mayby a stupid question, but there has been written so much that it's hard to find the optimun for a newbie...
I do not want to use more than 10.000µF-15.000µF (+&- each) per IC.
3.
Many chipamp disigns use 8 diodes for rectifying. Has that any
advantages compared to the classic rectifyer disign with 4 diodes and the low ends of the transformers secondary windings directly to the circuits ground?
Thanks,
Stephan
1.
I want to use a single big toroid for 4, maybe 5 lm3886.
I'm going to start with 2200µF (+&- each) soldered directly at the pins.
If I understood these snubber theories, the transformer and the capacitors form an oszillating circuit. All people recommend not to use more capacitance because of that.
But in my case, the transformer will "see" 22.000µF of all ics.
What will happen?
Better use 5 small transformers?
2.
If I decide to big capacitors with snubbers later, wich disign is the best?
Mayby a stupid question, but there has been written so much that it's hard to find the optimun for a newbie...
I do not want to use more than 10.000µF-15.000µF (+&- each) per IC.
3.
Many chipamp disigns use 8 diodes for rectifying. Has that any
advantages compared to the classic rectifyer disign with 4 diodes and the low ends of the transformers secondary windings directly to the circuits ground?
Thanks,
Stephan
1. The cap's are (should be) in paralell so 1 circut cant "see" all cap's. It can only "see" the one in front of it. The current doesnt go in the cap and back out again so to say.
I have read about the snubber's and personaly i dont feel the need to use it. I have a setup of one toroid, then three (3) 22000uF cap's/voltage channel and on the pcb i have one 2200uF/voltage channel. And that is more then enough. The bass is tight and fast and it stores enough current for my 2 pcb's with LM3886 on so they can continue playing for about 20 secconds after i have taken the plug out of the walljack.
You shouldnt be needing 5 transformer, just one big with alot of VA, maybe 500 or 600?
2. Well... i started with only one toroid and after that one 10 000uF onboard/voltage channel. It's really a matter of taste. If you use to small cap's you want get enough bass, and if you use to big you get bass, but it can be slow.
3. Dont know about this one, i use 2 classic rectifyer's.
I have read about the snubber's and personaly i dont feel the need to use it. I have a setup of one toroid, then three (3) 22000uF cap's/voltage channel and on the pcb i have one 2200uF/voltage channel. And that is more then enough. The bass is tight and fast and it stores enough current for my 2 pcb's with LM3886 on so they can continue playing for about 20 secconds after i have taken the plug out of the walljack.
You shouldnt be needing 5 transformer, just one big with alot of VA, maybe 500 or 600?
2. Well... i started with only one toroid and after that one 10 000uF onboard/voltage channel. It's really a matter of taste. If you use to small cap's you want get enough bass, and if you use to big you get bass, but it can be slow.
3. Dont know about this one, i use 2 classic rectifyer's.
Bassterror said:
Maybe? It would be good to nail-down the needed current drive for your speakers and THEN, process with what it'll take to get there.
Not quite, the higher inductance but it's additive, you can parallel caps to improve this, but it "might" cause oscillations. Remember that if one single person out of 1000 has oscillations, then doing something that "might" have caused it, becomes the voodoo child for everyone. (Gotta luv Jimi). You can use more capacitance but don't use ONLY very high mfd, large caps. Put some reasonably small caps close to the chipamp, but larger than the domain of film or ceramic ~ < 5uF (in cost effective sizes) realm.
What'll happen is that with loud volume and high bass, there is the possibilty that current will backflow from the caps local to another channel, to those on the channel playing that loud low bass note. However, 22,000 uF is nothing to sneeze at, I have a chipamp with about 50% more capacitance that plays for about 4 seconds after the AC power cord is pulled from the wall, at fairly high volume. In other words, I suspect with 22,000 mfd you are in a safe region so long as all those caps have a low impedance path to each other. IE- thick traces or heavy gauge wiring.
You could, and it might be slightly better channel separation, but that's pretty messy for a layout, if you were considering an unregulated PSU, I'd just go regulated before I'd do that... not to knock the idea, it'd work, but that's a ton of real-estate and cost with questionable benefit.
There are as many designs as builders, better to mention what you are contrasting. With 5 chipamps, 10K mds per rail is enough, probably even overkill with low impedance between each. It's not that I'm trying to discourage you from 10K mfd per, just providing a point of reference that it's enough given a good layout.
You will find people that swear they hear a difference, but I suspect it's only making up for other deficiencies in the amp. Even so, 5 chipamps is a non-trivial amount of current if you use fewer trannies, do the math and figure out if the target configuration is within the peak current capabilities of the diodes you choose (keeping in mind that for most higher-current diodoes, you need a 'sink on them to reach the rated current for a sustained period of time to remote that heat which you can also calculate.
Bassterror said:
Yes that would prevent it from happening but it's not necessarily a problem, is a trade-off but possibly a positive thing if you want to limit the amount of capacitance, to have the entire load buffered by more than one pair. This is a big project though, are you sure you want to only stick with some 2200 caps you already had? Might want to get 4 x 10K mfd or so for a centralized reserve right after the rectifiers then the 2200 mfd pairs closer to each chipamp. It depends on how loud you want it though, 5 x chipamps is a lot of volume for some situations if building for max power. If it'll never be at max power (for certain) you dont' have to design for that.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.