Star grounding

Thread Starter

nikolopikolo

Joined Sep 17, 2020
33
Hi, I made a audio amplifier and came to a problem which someone here might be able to help me with. I want to implement star grounding, but on my PCBs I dont have such a place where I could do that. Is it functional to just connect and solder some ground wires together or is there another solution?

Thank you in advance, Niko
 

Ian0

Joined Aug 7, 2020
2,259
Actually you do. Just decide on a point on the board that is going to be the "star ground", a good choice being half way between the 0V connections to the two power supply electrolytics, then make all other 0V tracks join at that point and nowhere else, the most important ones being the loudspeaker return, the input ground and the 0V connection to the feedback ciruit.
You can probably cheat on some low current supplies, such as bias supplies to constant current sources and the like.

Whilst you are doing it, also pay careful attention to where you take the track for the output end of your feedback circuit, because that matters just as much. If it connects to one output transistor so that there is a section of track down which negative current flows but not positive current then you will introduce distortion.
 

Thread Starter

nikolopikolo

Joined Sep 17, 2020
33
Alright that was pretty insightful, but just one thing i still dont understand is the actual star ground point. Should it always be on a board or is it possible to just connect wires between each other at one point. See the thing is everything was already designed and built the way it is, without star grounding and I need a quick solution.

Here is a photo of my two seperate grounds, could the top one which has only three components connected to it as a reference point for star grounding maybe?

It would be really useful if you could give me some guidance on how to execute this, sience its the first time I'm doing such a type of grounding technique.
 

Attachments

Ian0

Joined Aug 7, 2020
2,259
Could you also post your circuit diagram as where they go in the circuit is important.
The concept of grounding/earthing is more complicated that it might seem at first sight - NOWHERE is absolutely "ground", perhaps the best you could get is the case of the power company's generator.
Only SPICE has an absolute ground!
So, the best you can manage is a relative ground. You decided where that is, and you can choose arbitrarily. Once you have decided where it is, then nowhere else is ground. Any other component, pcb track, wire is separated from your ground by a small but finite resistance and inductance, so it isn't ground.
You might choose the point your mains earth connects to the case as your ground, but that would mean long wires to your circuit board, and each long wire has resistance (across which current would cause a voltage to develop) and inductance which (which can produce a voltage from magnetic fields); so the other important feature of star grounds is that the connections are SHORT.
Your SIGNAL earth isn't necessarily exactly the same place as your SAFETY earth.
 

andrewmm

Joined Feb 25, 2011
1,476
the reason for the "star point" philosophy could be of use.

Devices have a +ve and a -ve terminal, the negative can be at "gnd"
ok , that obvious,
but what is often forgotten is the current that goes into the device, tends to come out the -ve side.

Low noise devices tend to want a low noise power input, and that includes a low noise gnd,.

As we said, the current tends to end up gong through the gnd terminal,
larger / more power hungry devices such as loud speakers, have a big current,

Current through a wire makes a voltage,
so if the current form say the speaker shares a wire / track with the low noise parts, the gnd of the low noise parts is going to bounce around as the loudspeaker current changes.

The answer, is to conceptually , connect every device to its own power connections back to the power supply,
I have seen designs where literally there are pairs of power wires from teach device back to the PSU,

in reality, we tend to group parts together,
and try to route the high currents back through a different connection to gnd.
 

Marley

Joined Apr 4, 2016
411
Despite the complexity of the circuit you will notice that there are not that many points that actually connect to ground. So fairly simple to bring all these to one point on the PCB.
 

Ian0

Joined Aug 7, 2020
2,259
Connect each of your two ground connections separately to the junction of the two electrolytic capacitors in the power supply. That will be pretty good, but make sure that the T11/T12 junction doesn't connect to your input ground, . Take that separately to the electrolytics, even if it means cutting a track - quite a bit of current flows there.
EDIT - the R18/R19 junction is less important as the current is a small amount of DC, and any disturbance at that point isn't going to bother anything as it is decoupled by C13 and C14.

Just as a aside - Cyril Bateman did a whole load of distortion tests on capacitors back in the 1990s and concluded that bipolar electrolytics were significantly better than two polarised electrolytics back-to-back. Something that Self isn't aware of in his books. Use a bipolar for C6 as well.
 

Thread Starter

nikolopikolo

Joined Sep 17, 2020
33
Ive connected it together, but when i turn on the power supply the speaker starts making noise and moves up and then down.
Did I mess up anything with grounding here or is it something else. Also I am not feeding it any input. Maybe the wires are too long or something I really dont have a clue. Any advice works.
 

Ian0

Joined Aug 7, 2020
2,259
The only differences between star-grounded and grounded-any-old-way will be reduced hum and noise and lower distortion.
Everything else will be the same. If you connected the grounds, then the problem is elsewhere.
Most people would test it on a dummy load before connecting an (expensive) speaker!
If the DC offset is <50mV after the initial up/down movement as the power supplies stabilise, then it is probably working.
 

Ian0

Joined Aug 7, 2020
2,259
Presumably, this is not your first audio amplifier. I could be many things - wrong component, faulty component, short circuit, missing solder joint etc.
 

Thread Starter

nikolopikolo

Joined Sep 17, 2020
33
I do know it could be many things, also not my first audio amplifier, but I had never made a proffesional one like this. I will check the components and short circuits and try again.
 

Ian0

Joined Aug 7, 2020
2,259
As it's all inside a feedback loop it often gets tricky to find the problem.
My usual technique is to measure the voltages around each transistor, and pay special attention to any where Vbe isn't 0.6V-ish in the right direction.
(Of course, T2 will be saturated and T4 cut-off as it stands, if the feedback is correct)

There is something very peculiar about T20/T24. It is CFP amplifier, so the source of the output transistors should be connected to the supply. On a lateral MOSFET source is in the middle - presumably pin 2, but that seems to go to the output.
So you have the reverse-diode in the output transistor pulling the output high.
 

Ian0

Joined Aug 7, 2020
2,259
SPICE (and other programs) just love to give you the MOSFET symbol with the drain at the top and the source at the bottom regardless of whether it is N-channel or P-channel, and before you know it, the P-channel is in the circuit with drain and source reversed.
 

Thread Starter

nikolopikolo

Joined Sep 17, 2020
33
Source is connected correctly i triple checked, the diodes and transistors seem fine aswell, I havebt found any shorts yet, but am continuing with the troubleshoot
 
Top