I tried paralleling with one feedback resistor, but it didn't work. So now, I made a 2 channel amp. I'm using 2 lm3886s, one for each channel. The psu is +-25v with a 2 amp fuse. I'm not sure about the ratings. In the feedback network, I'm just using 470k resistors for each chip. Turned up about half-way (before I can hear the cracking), I can run a 3 ohm sub off each chip for about 45 secs before they get really hot. At lower volumes, I can run them for a long time, and they just get a little warm. Also, my heatsink isn't very big, but I'll upgrade to a bigger one once I get it working right.
ok, the chips have two issues -- gain difference and ossilitions.
in parallel, a 0.1 ohm resistance (copper wire, of x length), on each output, the feedback resitor attaches as close to the chip as normal to reduce common inductance.
the gain setting resistors must be chosen to be close on each IC, the idea is to have a very similar gain.
these two things are to keep current from one chip from flowing to the other chip. basically if you have a gain of 10 on IC 1 and a gain of 11 on IC, and a 1V input signal, you will hit the 5A limit on the ICs as you will have 1V difference on the outputs and about a dead short... the copper wire will make this 0.2ohm minimum, and close 0.1% tolerance resistors will make the gains closer.
ossilations can be helped by placeing capacitors near the PSU pins and bypassing them with 100nF ceramic (or other good RF type) caps. the datasheet has a lot of mods to get more stability.
in parallel, a 0.1 ohm resistance (copper wire, of x length), on each output, the feedback resitor attaches as close to the chip as normal to reduce common inductance.
the gain setting resistors must be chosen to be close on each IC, the idea is to have a very similar gain.
these two things are to keep current from one chip from flowing to the other chip. basically if you have a gain of 10 on IC 1 and a gain of 11 on IC, and a 1V input signal, you will hit the 5A limit on the ICs as you will have 1V difference on the outputs and about a dead short... the copper wire will make this 0.2ohm minimum, and close 0.1% tolerance resistors will make the gains closer.
ossilations can be helped by placeing capacitors near the PSU pins and bypassing them with 100nF ceramic (or other good RF type) caps. the datasheet has a lot of mods to get more stability.
The heatsink has to be very large. My first 3886 amp (a single chip) had a large but small heatsink, and when it got too hot, the chips would make a popping sound in the speaker.
You need a VERY large heatsink for these chips.
Also, you are running these in parallel, not bridged, right? If you were running them in bridged, you already solved your problem when you said you were running a 3ohm speaker. Thats only 1.5ohm a chip, and those chips aren't designed for much lower than 4ohm.
If not, are you sure they are paralleled, where the outputs tie together with a .1ohm resistor and not bridged, where the output of each amp drives one side of the speaker?
-Mike
You need a VERY large heatsink for these chips.
Also, you are running these in parallel, not bridged, right? If you were running them in bridged, you already solved your problem when you said you were running a 3ohm speaker. Thats only 1.5ohm a chip, and those chips aren't designed for much lower than 4ohm.
If not, are you sure they are paralleled, where the outputs tie together with a .1ohm resistor and not bridged, where the output of each amp drives one side of the speaker?
-Mike
"You need a VERY large heatsink for these chips."
-really, you need to quantify this. a radiator from a car is a heatsink. is it large? very large? i have a 6x8x0.5 aluminum heatsink. big? a CPU heatsink. big?
as to xplod. a good idea would be to read the highpower app notes linked to in this thread. then if you have any specific questions, ask first then build.
-really, you need to quantify this. a radiator from a car is a heatsink. is it large? very large? i have a 6x8x0.5 aluminum heatsink. big? a CPU heatsink. big?
as to xplod. a good idea would be to read the highpower app notes linked to in this thread. then if you have any specific questions, ask first then build.
As I said before, I tried paralleling with one feedback resistor, but that didn't work. Now, I'm using one chip for each channel. Each chip drives a 3 ohm sub. I tried it with a 4 ohm 6x9...same result. No matter whether it's hot or not, when I turn up the volume to more than half, it sounds really bad. Does anybody know what the problem is?
I knew somebody would laugh at my description 
Let me explain more. The heatsink was large, about 3 1/2 inches wide by 1 1/2 inches tall, with fins about 1 1/2 inches deep. There were only 4 of those fins, though. That made the heatsink look large, but in operation, compared with a small heatsink. When I put a huge heatsink on it, which about 5 inches wide and 3 inches tall, with at least 100 little fins, the chip still got warm, but never overheated.
About the popping, when the chip overheated, an explosion sound in a movie or a boom in a song would make a loud pop, repeating for the duration of the sound. I think it was getting too hot and kept switching into thermal mode, or just the transistors were too hot to be operating at the volumes they were.
-Mike
Let me explain more. The heatsink was large, about 3 1/2 inches wide by 1 1/2 inches tall, with fins about 1 1/2 inches deep. There were only 4 of those fins, though. That made the heatsink look large, but in operation, compared with a small heatsink. When I put a huge heatsink on it, which about 5 inches wide and 3 inches tall, with at least 100 little fins, the chip still got warm, but never overheated.
About the popping, when the chip overheated, an explosion sound in a movie or a boom in a song would make a loud pop, repeating for the duration of the sound. I think it was getting too hot and kept switching into thermal mode, or just the transistors were too hot to be operating at the volumes they were.
-Mike
I have built a amp with two lm3886's.. used a 2x28V/120VA transformer..
at first i had a too small heatsink and the lm3886 got so hot the solding melted at it legs. but, the sound was superb all the time.. i've never experienced any popping or something.
But i never tried to use the mute fuction on the lm3886, i built a seperate loudspeaker-delay unit instead.. worked really fine.
i used to caps at 10000uF/50v in the power supply unit..
xplod, if you want my schematics i got them in OrCad format..
at first i had a too small heatsink and the lm3886 got so hot the solding melted at it legs. but, the sound was superb all the time.. i've never experienced any popping or something.
But i never tried to use the mute fuction on the lm3886, i built a seperate loudspeaker-delay unit instead.. worked really fine.
i used to caps at 10000uF/50v in the power supply unit..
xplod, if you want my schematics i got them in OrCad format..
So your saying that no matter how hard you pushed it's volume, you couldn't get the amp to clip before bringing up the voltage?
If so, could you please post those schematics in a standard image file?
and when you said there were 2 3886s, did you mean stereo, bridged or parallel?
Thanks!
Mike
If so, could you please post those schematics in a standard image file?
and when you said there were 2 3886s, did you mean stereo, bridged or parallel?
Thanks!
Mike
soundNERD said:
Entirely possible if the gain is not too high and the input signal is not to large either. The moment you apply higher PS voltage you increase the power disipated by the amp and the protection circuit will kick in and it'll start distort. This is one of the reasons when having 4 Ohms speackers you need lower PS voltage to minimaze the loss over the IC and nothing else.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.