I'm seeking advice on how to reduce noise (hiss) and hum (mains pickup?) for my chip amp. I'm using an LM3886TF board that has only the amp on it (no PS). This has neither RF blocking on the input nor the usual parallel inductor and resistor on the output to prevent RF ingress there. I am using a small toroidal transformer and a rectifier + cap board. These are build into a wooden loudspeaker box. The transformer, cap board, and chip amp are all within about 6" of each other. Power wiring is done using 16ga wires, routed through some Euro terminal blocks. Speaker wire is also 16ga, with one run from amp board to driver, and the return (cold) speaker lead connecting directly back to the cap ground. No attention was paid to twisting any wires together. There is little to no metal - only the aluminum heatsink - so no "chassis" that would provide EMF shielding. The amp ground connection includes a 4.7R resistor in between the PS ground point and the amp board, inserted at the amp end of the wire. The PS ground is connected to earth ground.
As described above, the amp is connected to a 100+dB/W compression driver tweeter, 8 ohms impedance.
When I switch on the amp, with the input unterminated, there is a small but noticeable amount of hiss. If I connect a 12' RCA-terminated coaxial cable with the source equipment on but muted I also get some mains pickup, with a bit of a grainy tone to it. This is the same whether or not the source end of the cable is plugged in or not. At the amplifier, the shield is connected directly to the amp ground plane.
If I instead power the compression driver using a consumer amp, and use the same interconnect, the hiss (but no hum) is present but at least 10dB less (e.g. much less) and is barely audible.
What plan of attack should I follow to try and clean up the noise floor of the chip amp?
-Charlie
As described above, the amp is connected to a 100+dB/W compression driver tweeter, 8 ohms impedance.
When I switch on the amp, with the input unterminated, there is a small but noticeable amount of hiss. If I connect a 12' RCA-terminated coaxial cable with the source equipment on but muted I also get some mains pickup, with a bit of a grainy tone to it. This is the same whether or not the source end of the cable is plugged in or not. At the amplifier, the shield is connected directly to the amp ground plane.
If I instead power the compression driver using a consumer amp, and use the same interconnect, the hiss (but no hum) is present but at least 10dB less (e.g. much less) and is barely audible.
What plan of attack should I follow to try and clean up the noise floor of the chip amp?
-Charlie
Here's another piece of info - I added in another 4.7R resistor where the input coax coming from the RCA panel jack shield is connected to the amp board. With the 12' RCA coax line disconnected there is almost no noise and no discernible hum. If I connect the 12' coax to the source, there is now even more hum than before. The source equipment is not earthed and the outputs are transformer coupled.
Why is the input coax picking up so much noise and hum?
Should I try using a ground loop breaker?
Why is the input coax picking up so much noise and hum?
Should I try using a ground loop breaker?
Here is one more possibility - I traced the PCB and the ground for the local rail decoupling caps is connected to the ground for the input components. I wonder if the rail caps are coupling some PS ripple into the ground, which is separate from the PS ground by the 4.7R resistor?
Also, the source component outputs aren't transformer coupled as I had first thought - they are capacitor coupled, with a series cap, a series resistor, and then a grounded cap (AC coupling plus LP filter). The source is powered using a linear regulated wall wart PS. Could connecting the (-) lead of the PS to earth help to reduce noise?
I'm still a little baffled that when I connect the input cabling the noise is so much greater...
Also, the source component outputs aren't transformer coupled as I had first thought - they are capacitor coupled, with a series cap, a series resistor, and then a grounded cap (AC coupling plus LP filter). The source is powered using a linear regulated wall wart PS. Could connecting the (-) lead of the PS to earth help to reduce noise?
I'm still a little baffled that when I connect the input cabling the noise is so much greater...
Last edited:
SOLVED!
After some sleuthing I discovered that the problem was actually upstream with a component, something that I have used before and not had a problem with, and not with the amp itself.
This just ends one mystery and starts another...
After some sleuthing I discovered that the problem was actually upstream with a component, something that I have used before and not had a problem with, and not with the amp itself.
This just ends one mystery and starts another...
The coax cable you are using, is it 1+shield.
Use 2+shield cable and connect shield only to the source end.
Gajanan Phadte
Use 2+shield cable and connect shield only to the source end.
Gajanan Phadte
I think I have narrowed this down even further. The "problem" component in question in an outboard computer soundcard. With the computer and soundcard off, I turned on the rest of the chain and the amp and the noise and hum were audible. Turning the problem component on or off did not change anything. Unplugging the interconnect from the problem component caused the hum to go away and noise to be reduced.
The problem component has combo TRS/XLR inputs and outputs. I use RCA, so I have a set of RAC to TS adapters that I plug into the jacks. These seem to be the source of the problem and seem to be corrupting the ground. I have some home made XLR to RCA adapters that I can try instead. I know that these are wired correctly.
The problem component has combo TRS/XLR inputs and outputs. I use RCA, so I have a set of RAC to TS adapters that I plug into the jacks. These seem to be the source of the problem and seem to be corrupting the ground. I have some home made XLR to RCA adapters that I can try instead. I know that these are wired correctly.
- Status
- Not open for further replies.