Biamp 3875 problem

Here's the deal.....running a biamp.......active crossover works fine. The amps were fine individually. When pumping music through them they both distorted badly. Also if you let that mess go for a couple of minutes the HF amp shorts from V+ to V-. Aby ideas? Need more info? I have a 600VA toroid feeding a full bridege then to 2200uF caps for PS.
 
schematics

LF side is same just switch the HPF for a LPF
 

Attachments

  • hfbrd.jpg
    hfbrd.jpg
    49.3 KB · Views: 358
Re: schematics

sauuuuuce said:
LF side is same just switch the HPF for a LPF


There are some observations:

First. Do you need such low gain (2x)? I have not tested this type of monolithic amplifiers at gains much lower than recommended (I've got down to 10x with no problems, not below). There exists the possibility of instability, and I suggest to check changing R14 from 1K to at least 10K, or 18K - 22K as recommended and see what happens.

Related to the previous point and with respect to the HF amplifier which is the one depicted, you have a 1 nF DC decoupling capacitor (C5). Not going into details, this converts the amplifier to an (unintended) voltage follower at low frequencies exacerbating the potential instabilities.

C5 (and the corresponding one for the LF side) should be fairly large (about 100 uF), and since they must be electrolytic to achieve this value, should be polarized. A handy alternative is to return the (-) side to the -V rail, but that will couple and amplify any ripple there, so it is better to devise an auxiliary simple zener regulator for biasing. (unless you have the -V rail already regulated and clean).

Lastly, don't forget to bypass the + and - rails to ground with 100nF and at least 10 uF on each side as close to the chips as practical. You may also want to try Carlos's snubberizing.

Hope some of this helps.

Rodolfo

Edit:

..The feed back resistor is supposed to be 20k..

Sorry I posted before your correction. Except for the gain issue, other points hold.

Rodolfo
 

Fossil

Member
2001-09-30 5:18 pm
SG
sauuuuuce said:
Tested without active xover. so just the LM3875 circuit. I am figuring the problem, or at least part of it, roots in joining the circuits. just can't pin point it....I could be wrong

Looking at it as 2 separate sections, there is a selector at the xover input and a pot at the 3875 input.
Personally, I prefer to have fixed resistors to ground at the inputs. But I'm not sure how to apply my preference in your cct.


:D