No, please read it: there are 4 amplifiers working from a single psu in a single box (and a box for left and a box for right). I have two "bass amplifiers", a "mid amplifier" and a high amplifier" in a chassis. Photo here: https://www.diyaudio.com/community/threads/chip-amp-photo-gallery.79303/post-7808209
Fair enough. But isn't there still a risk that I get DC feeding through if there's a significant offset between signal ground and power ground? The hum breaking resistor should try to keep them close in potential, but if they drift too far apart, that will definitely translate into DC on the speakers. Here's an example. V4 simulates a 1V offset between signal ground and power ground. As you can see, that offset is present on the output.
I believe there is no path for DC offset between grounds into the PA. This is in contrast to hum, since hum can pass through C2 into the amp--- but DC cannot.
To elaborate a bit, any DC imposed across V4 will be amplified by the amp's DC gain, which is 1.0. So the voltage at the amp's output will be V4. So there would be some DC imposed across the speaker (R2) but the current flowing will be V4/(R5+R2). There would also hum but with gain of only 1, rather than normal AC gain. So any DC current is limited by R5. DC between ground networks should be rare and small if present. Any AC would see unity gain and be further attenuated by R5--- a few dB.
Dissenting members, please weigh in. This assessment is a new analysis, at least to me. Thanks.
Edit: revised opening sentence.
Dissenting members, please weigh in. This assessment is a new analysis, at least to me. Thanks.
Edit: revised opening sentence.
Last edited:
That is not close to reality. In your amp, signal ground and the return from your speakers will both be going to the star point of your psu. That star point of the psu is what is "ground" in your simulation. But the "safety earth" (what you call power ground) is completely separated from that. The question is "where do you connect 'the ground of your electronics" (*) to the "safety ground" which is external to your electronics. And there is only a single connection between both.
Now that 10 Ohm shouldn't be there at all. It is an easy remedy to a bad design. Any accredited laboratory will refuse this.
(*) all your electronics should refer to a single "ground". Now inside your amplifier you can divide it up with the intention to keep interaction between parts low. Like separating the ground your audio signals use from those that your protection relays use. Or digital and analog circuits. But somewhere they will meet. In a power amp this is mostly at the "star ground", there where your psu makes the voltages available and where the (very) large currents going into and out of the smoothing capacitors after the rectifiers do not flow anymore.
Safety ground or "earth" is a different concept. This is the reference for the power distribution system that almost always run a a high dangerous voltage. In most cases this somewhere is literally "the earth", a copper rod driven into the soil. Both at the generator side and at the home side.
Somewhere both have to meet to keep you safe from a mains voltage coming on the housing of the equipment. But unless you are making telegraph or telephone equipment, you do not want any of your (audio) currents to flow through the mains related earth. So in most cases the best it to connect them together at a single point and put that point as close to where the "signal ground from the external source" comes into your equipment. In that case any current that is related to mains and coming from outside your equipment will be diverted to mains earth before it can flow through your circuitry.
If it was 120Hz hum then I'd be looking after rectification eg, a tired/failing voltage regulator, perhaps it's a load issue and said regulator can handle one channel but falls over when 2 channels connected?
I think you're missing my point. I'm not talking about protective earth. I'm talking about signal ground vs. power ground (which is the same as speaker return from what I understand).
Now, my question was whether the feedback loop should be grounded to signal ground or power ground. You can make the argument that the 0V reference in the feedback loop should be the power ground. Why? Because any idle voltage on the amplifier output that's different from power ground (a.k.a. speaker return) causes DC offset on the output. We don't want that. Removing the 10ohm resistor between signal ground and power ground would make matters even worse. If the signal ground completely floating relative to the power ground, you could (at least in theory) have any arbitrary potential difference between the two. And since the amplifier probably has a relatively high input impedance, it will gladly pass that voltage through and boost the current as needed. That could be very bad.
Maybe I'm chasing ghosts here. Maybe the 10ohm (or whatever value you pick) resistor is enough to equalize the voltages to the point where any offset would be harmless. Here's an example. The input sits at 0V, but there's a 1V offset between power ground and signal ground. As you can see, that offset is present on the output.
Now, my question was whether the feedback loop should be grounded to signal ground or power ground. You can make the argument that the 0V reference in the feedback loop should be the power ground. Why? Because any idle voltage on the amplifier output that's different from power ground (a.k.a. speaker return) causes DC offset on the output. We don't want that. Removing the 10ohm resistor between signal ground and power ground would make matters even worse. If the signal ground completely floating relative to the power ground, you could (at least in theory) have any arbitrary potential difference between the two. And since the amplifier probably has a relatively high input impedance, it will gladly pass that voltage through and boost the current as needed. That could be very bad.
Maybe I'm chasing ghosts here. Maybe the 10ohm (or whatever value you pick) resistor is enough to equalize the voltages to the point where any offset would be harmless. Here's an example. The input sits at 0V, but there's a 1V offset between power ground and signal ground. As you can see, that offset is present on the output.
Hello, I've been busy with work these past few days, so I haven't been able to test much. I'm trying to finish a different wiring setup. I'll have to post a diagram of the amplifier's ground plane over the weekend.
As for the voltage regulators, they work well. Oscilloscope tests are good up to the amplifier clipping, including on a capacitive load.
As for the voltage regulators, they work well. Oscilloscope tests are good up to the amplifier clipping, including on a capacitive load.
I work a little on the amp and simply by reduce the surface on the cross channel loop the hum are very low now. I think I'll leave it at that.