I solved my problem today, but I thought some people here might be interested in the process. I have active speakers and performance is quite acceptable, but I hear a small buzz from my speakers when it the music is off and the area is quiet.
Potential culprits:
-active crossover board situated immediately above power supply due to limited space (with aluminum mounting plate and copper ground plane providing electrical shielding)
-on crossover filtering is limited to small poly bypass capacitors
-an edit was made which involves a small extra board wired to with unshielded wires
-psu is straightforward datasheet implementation of standard regulators without any extreme filtering or snubbing
I was avoiding testing because I didn't want to make a measurement jig to safely measure my psu ripple with my sound card based analyzer. I checked with my multimeter and it showed < 3mV. I got smart and realized I don't care about psu ripple, I care about the noise floor on the output. So I hooked up the output to ARTA's spectrum analyzer.
Here is the baseline noise floor, it's quite low relative to full scale but not acceptable:
So I tested: massively increased psu capacitance on crossover board, crossover board without edit, and board location. Board location was the only thing that made any difference. So I thought to myself, uh-oh, do I need to buy some expensive mu-metal or make a full sub-enclosure. Then I rummaged around and found some giant steel washers in my bin. I placed those on top of the aluminum mounting plate and the problem is mostly gone, and now inaudible. I'm hoping replacing the aluminum plate with a full steel plate will kill the rest, but its quite inaudible.
The reason why I posted this is that a lot of people in this hobby don't really measure and look for root cause, and they may just resort to glomming on more junk until the percieved problem is solved. If I hadn't measured, I might have kept the extra capacitance as a safety blanket. Turns out my solution was quite cheap, I will just replace the aluminum mounting plate with a steel one. No need for mu-metal or massive increase in bypass capacitance to address this problem.
Thanks for listening,
Lee
Potential culprits:
-active crossover board situated immediately above power supply due to limited space (with aluminum mounting plate and copper ground plane providing electrical shielding)
-on crossover filtering is limited to small poly bypass capacitors
-an edit was made which involves a small extra board wired to with unshielded wires
-psu is straightforward datasheet implementation of standard regulators without any extreme filtering or snubbing
I was avoiding testing because I didn't want to make a measurement jig to safely measure my psu ripple with my sound card based analyzer. I checked with my multimeter and it showed < 3mV. I got smart and realized I don't care about psu ripple, I care about the noise floor on the output. So I hooked up the output to ARTA's spectrum analyzer.
Here is the baseline noise floor, it's quite low relative to full scale but not acceptable:
So I tested: massively increased psu capacitance on crossover board, crossover board without edit, and board location. Board location was the only thing that made any difference. So I thought to myself, uh-oh, do I need to buy some expensive mu-metal or make a full sub-enclosure. Then I rummaged around and found some giant steel washers in my bin. I placed those on top of the aluminum mounting plate and the problem is mostly gone, and now inaudible. I'm hoping replacing the aluminum plate with a full steel plate will kill the rest, but its quite inaudible.
The reason why I posted this is that a lot of people in this hobby don't really measure and look for root cause, and they may just resort to glomming on more junk until the percieved problem is solved. If I hadn't measured, I might have kept the extra capacitance as a safety blanket. Turns out my solution was quite cheap, I will just replace the aluminum mounting plate with a steel one. No need for mu-metal or massive increase in bypass capacitance to address this problem.
Thanks for listening,
Lee
Comment