Buzz in gainclone, looking for source.

[traw]

oh yeah... along similar lines. what happens if only one input plugged in? I'd had buzzes where dead silent with nothing plugged in. hook up, buzz. remove on jack, buzz subsides. once used hack ignorantly placing link between 2 grounds of rcas directly, uh, creating a loop, and squashed it. but don't like that as not right.

[AndrewT]

Hi Traw,
your last post is kind off alluding to two faulty sources. I think this is close to what Minion has.
I am beginning to agree the fault is not with the Power Amp.

[traw]

yeah... it's more being more sensitive to grounding, although it should be black and white. yeah, computer driven sources, especially stock, can be extremely noisy.

[MikeBettinger]

Here's a ground layout to experiment with. It eliminates a questionable loop area between the source and the star point and defines the return paths for the electrolytics.

Regards, Mike.

[AndrewT]

Hi Mike,
I would make two changes to your layout.
1. The on board decoupling (often used in discrete) taken straight to main system ground.
2. The reference to chassis removed from the input ground.

To clarify the need for safety, I would add a note about wiring in a safety ground directly to and permanantly attached to the chassis. This removes any ambiguity the builder may have about the safety ground location. Then a note on the various ways to interconnect the safety ground to the main system ground using the optional components to form a disconnecting network.

[MikeBettinger]

Originally posted by AndrewT
Hi Mike,
I would make two changes to your layout.
1. The on board decoupling (often used in discrete) taken straight to main system ground.
2. The reference to chassis removed from the input ground.

To clarify the need for safety, I would add a note about wiring in a safety ground directly to and permanantly attached to the chassis. This removes any ambiguity the builder may have about the safety ground location. Then a note on the various ways to interconnect the safety ground to the main system ground using the optional components to form a disconnecting network.

Hi Andrew,

First one note: the input jacks need to be next to each other. In this case drill two more holes 3/4" to 1" apart in the wood and run the ground wire as shown; using twisted pair wiring back to the amp circuits and single lo-Z back to the central ground.

2.) The input grounds need to be implemented as shown. There is a good reason for this but, until it's been tried, common knowledge has most designs (that I see published) doing it differently and the proof is in the manual simulation.


1.) The board decoupling caps are still part of the power supply filter loop and should be returned back to the filter caps. I actually haven't used/needed these but most applications guides call for them. If there is a chassis, then connect them to the chassis and use that as a low-z return (prefered).

My thought here is that there is no chassis ground; most of the chip amps I've seen here appear to use wood or have most of the circuitry floating (wood chassis' seem to be popular).

The circuit ground to the chassis (if there is one) should be connected wherever the two power supply grounds (because of the two transformers and filter pairs) come together, which is, as I noted, as close to the tranformers as possible. By this I intend that the center ground from the secondaries be as short as possible. (I have experimented with this approach extensively)

As for the safety ground (if there is a chassis) it should connect directly to the chassis where it enters. With no chassis, take it directly to the central ground (I've tried this and unless there is a short there is no noise circulating through any of the functional loops (it's quiet).

Thanks for your comments, but experimenting with the layout as shown is what my intention is. There's a noise voltage being developed in the grounds, this approach controls this.

Regards, Mike.

[MikeBettinger]

Originally posted by AndrewT
Hi Mike,

1. The on board decoupling (often used in discrete) taken straight to main system ground.

Hi again Andrew,

I just realized I mis-read the above to think you were refering to the decoupling caps at the output stage, not that you were making reference to any decoupling located on the drive board. Sorry bout that. I agree, actually either the main ground or filter grounds will work on drive circuit bypassing. Most of what's being returned through that path is pretty weak.

I should probably slow down when I read, but then the speed reading course I took would be a waste on money; although, it gets me into more trouble than it was worth...


The input grounding is correct from my experience.

Regards, Mike.

[AndrewT]

Hi Mike,
bypassing is a different issue again.

By bypassing I mean the tiny fast caps used to attenuate voltage glitches due to current changes in high current devices.
These bypass caps should be inserted into the circuit to bypass the current spike back into the device and do not need to be returned to the ground.

[mod_evil]

Hi,

Do you have a Central Ground?

How many capacitance you have on the filters? For Chip-Amps, you like to use 10.000uf PER VOLTAGE LINE!!! If you use 1000uf, 2200uf, 3300uf, 470uf, you will have HMMMMM and BUZZZZZZZZZ

Can you put your POWER SUPPLY on a PCB? This is very dangerus use a Power Supply, how you are using in this moment.

Do you have coupling on the input? You REALY NEED a high pass filter on inputs. You can test, a Non-Inverted operation and using 2.2uf and 100k resistor. To make a high-pass.

Can you post in here your schematic.

Best Regards.

[kmj]

Hi MOD!

I currently have the supplied caps from Peter Daniels KITs (1500uF, 50V) so 3000uF per channel.

No PCB, it might be more dangerous this way but he whos going to use it will never even open it.

I don't have a filter on the input but I could add one on the Pre-amp if it is needed. The pre is the Freebird designed by Russ White here at the forum.

The shematic is the one used in the chipamp.com LM3875-manual and the groundingschematis is the same as MikeF used here