Solid State Amplifiers > Amplifier Discussion

Grounding techniques

(1/7) > >>

teemuk:
This thread was started so that people struggling with humming amps could have an information resource that helps them to understand what’s going on. The internet is full of information but it is quite scattered and I hope this thread will develop to collect this information together. There are few resources I can recommend such as…

Epanorama: Tomi Engdahl, Ground loop problems and how to get rid of them (http://www.epanorama.net/documents/groundloop/) and Ground loops and equipment design (http://www.epanorama.net/documents/groundloop/in_circuit.html)

Randall Aiken, Star Grounding (http://www.aikenamps.com/StarGround.html)

R.G. Keen, Star Grounding in Tube Amplifiers (http://www.geofex.com/Article_Folders/stargnd/stargnd.htm)

I’m sure a more thorough search would reveal even more. From top of my head I have summed up few rules that I regard as most important. Feel free to participate to this thread by comments or presentation of methods that you have found important.


Update 30.01.2007
Some corrections to the terminology. Corrected some spelling issues and added a small description of "loop breaker" circuits.


1) The essential part is to start treating the grounding as a part of the circuit itself: Each time a part of your circuit makes a connection to common it will not have a reference of ideal zero volts. Instead, this connection is just the other end of conductor connected to the only real common point that can exist in the system - a conductor with resistance. Since the circuit stage’s current goes through this resistor, then according to U=IR the "common reference" point of the stage must have a voltage potential. It would be better to refer to these points as “common returns” not as commons.


2)  Different common spots and safety ground
Even a typical amplifier circuit can have at least three "grounds": Mains "ground", safety ground and secondary circuit ground, which really is not a ground at all. The first two should be ignored: Mains "ground" or neutral is NOT a ground either and should never be mistaken as one. It has zero volt potential but one should treat it as a conductor carrying mains return currents. Safety ground (of a device) is just an alternative for neutral in the case of faults. Safety ground has a sole purpose of saving you and from the viewpoint of removing hum and noise that should not make any difference whatsoever.

In practically all circuits, the reference used is not ground, it's common. A term common is better also because it indicates that the point might be "floating" and does not necessarily have a zero potential in reference to earth. Any voltage chosen from the secondary side can be used as the common reference: This is also the reason why you can get positive, positive and negative and negative voltage potentials from center-tapped transformers.

In some point of the house/mains wiring, the safety ground is connected to neutral. Usual location to do this is either inside the breaker box (right way) or inside the mains receptacle (works but is a bit unsafe). Having a permanent connection somewhere in housewiring is a safer than trying to make the connection by yourself inside the device since you have 50/50 change of succeeding if you can plug the power cord into the wall outlet in any direction. Since the chassis (safety ground return) and the common are usually tied together they will appear in nearly same potential. If the chassis is insulated from the common the common will float in whatever potential it sees best.


3) Choosing the right common spot
Only one true common point where the returns connect can exist in the circuit and therefore it should be selected with great care. Also, remember that current flows in loops. Consider these points:

a) Diode rectifiers make switching noise that appears as transients. These transients have the worst effect in close proximity to the diodes. You do not want switching noises to affect any other common returns. Ever wondered why commercial devices have those capacitors in parallel with diodes? Well, now you know.
b) Filter capacitors draw current as pulses. A filter capacitor charged by rectified signal never becomes “full” and never stops being charged, this is especially notable when the circuit’s current draw is high. Currents created by these charge pulses are high and form a loopfrom the rectifier to the filters. The loop - as always - includes common wiring as well. You want to keep these charge pulses away from any other common returns.
c) Using the center tap as common point and connecting it to chassis most likely prevents you from making any other high current common connections to chassis (ie. speaker jacks).

I would not recommend using either one of the first two as the common point. So far I have found the following arrangements ok:
a) Single supply: Transformer connects the rectifier, rectifier connects the main filter capacitors and capacitors connect the common point. All other common returns connect only the common point. Safety ground may, or may not connect this point directly.
b) Dual-supply: Center tap connects the main filter capacitors and capacitors connect the common point. All other common returns connect only the common point. Safety ground may, or may not connect this point directly.


4) Correct ways to route the common returns
A ground loop is formed if the return currents of one stage inflict the return currents of another stage. Most severe situation is when high current stage (ie. speaker load) return currents share the same route with the return currents of the input stage. Most common example is when a ripple containing supply return shares the return of the input. Since nearly all gain stages use only a common return as a reference, any voltage appearing at this point will be heard in the output. Differential input amplifiers (opamps etc.) are a little more tolerant against this than class-A discrete stages.

The common returns should be separated by their type: audio (signal) / power (supply) and high current / low current returns should never mix up, neither should analog/digital returns. All these will eventually connect each other in the best and correct place, which usually is the common point. If nothing else, you want to at least separate high current and low current returns as well as signal returns from other "dirty" returns. Make it a principle for yourself that connecting these returns together anywhere else than in common point is always a conscious risk of adding a source of hum or noise to the circuit.

a) Star grounding
Ideally, all common returns should be routed separately and connected to common point in a form of “star” in order to prevent the returns from capacitively coupling each other. This method is ultimately effective but very hard to construct in real circuits. Very often the result is just a massive amount of long wires that may catch interference. Many people question the effectiveness of star ground but the reason for this is that they have most likely done something wrong: Either they have very long wires that are highly inductive or run near hum sources, they have bundled a few returns or made a chassis connection. (A true star ground can have only one chassis connection. This connection is made from the common point itself). In latter two arrangements one no longer even has a star ground. I see no reason why a correctly made star ground should hum – ever.

b) Various bundled common returns
Bundling up common returns can make the "grounding" scheme more compact. This is advisable, and properly done will not increase the hum or noise substantially. Usual procedure is to bundle the returns of each stage together. However, this is a pitfall: Many people bundle the supply to the signal return and wonder why their “bundled star ground” hums. The reason is the topology is no longer a star ground and the supply returns are infested by – sometimes – high current ripple or other noise that comes from switching devices etc. Also, bundling up the input stage and chassis (where several other common return connections will be made) together is asking for trouble. A better way is presented next….

c) Ground bus
The signal common returns should be bundled together this way:

Lowest current stage -> higher current stage -> etc -> highest current stage -> common.

The benefit of this arrangement is that the common return currents of the higher current stages will not return to the common point through the lower current stages. Then, the supply wiring returns should be bundled together in the same manner. If, besides these, the returns of power devices and speaker load are routed separately there is a fair chance that one can bundle all preamp/low current stages of power amp returns together with no hum whatsoever. In the end you have only few return wires running to the common point. Even better chance to success is to use bundled returns from multiple stages. Most likely the "grounding" scheme still stays fairly compact. Also, all passive stages returns can be fairly safely bundled to the return of the preceding active stage.

d) Large ground plane
This was a very often used technique in older amplifiers and usually involves connecting a vast amount of common returns to chassis thus forming a large and low resistance path for all return currents. Sometimes this works, sometimes it doesn’t and I would not recommend this technique for anyone building their own amplifier unless you are willing to experiment with several prototype boards and layouts. The problem in ground planes is that the return currents choose the least resistive path - not the path we wish - and thus may interfere with each other. The use of either bundled ground or ground bus topologies is far more effective, less confusing and logical. At least with it one can control what way the currents return. For people fixing or tweaking amps ground planes, however, are a necessary evil.

A more common approach nowadays is to use the technique in PCBs, where – sometimes even a complete layer is dedicated to "grounding" purposes only. Properly done this is even more effective than using ground buses. But then again, the whole topology can be ruined by a simple mistake that happens to form a ground loop. (One can, for example, originate from connecting inputs to both PCB and chassis and connecting chassis to common). I have also seen circuits utilizing massive ground planes connected together by a single wire - thus the whole idea of the topology is ruined. In my opinion, a careful following of ground bus topology leaves less room for mistakes than reliying on massive ground planes.


A little more about safety ground...
As said, this is tied into the Neutral wire - usually in the breaker box of the house wiring. The safety ground connection should tie together all conductive parts that the user can touch. Make the connection near the place where the mains wiring connects the chassis and leave the safety ground wire longer than any other mains wiring so that it will be the last one to break loose if tension is applied to the mains cord. Be sure the chassis (or any part of it) will not corrode and break this connection by making it highly resistive. A use of teeth lock washers and lock nuts is advisable. If you ground heat sinks make sure a proper connection is made regardless of the non-conductive anodisized layer they might have. Bear in mind that the safety ground is effective only when it is connected in house wiring as well. All attempts to locate a hum problem should be started by checking the continuity of the safety ground.

Common potential does not neccessarily need a connection to safety ground. However, in guitar amplifiers, where the user can access common potential via the strings of the guitar, this is highly recommended since reference to earth prevents the common point from floating up to high potential. The connection may either be direct or indirect. Since a direct connection to safety ground may introduce ground loops when two devices are connected together the connection is sometimes made through a "loop breaker" circuit: This is usually a 10 ohm resistor in parallel with a 100 nF capacitor. The resistor introduces enough resistance to limit ground loop currents and the capacitor bypasses radio frequencies. Since in fault conditions the circuit can burn "open" it should have parallel high power diodes that pass fault currents through. Diodes usually burn "short" so the protection would remain. Note that the loop breaker circuits are illegal in some countries.

idle_chatterbox:
nice tutorial  8)

willy1usa:
Also in the grounding / hum department is the word interference. I lived with various interference hums until I discovered shielded wire. I now use shielded wire for my inputs/outputs. Clip one end of the shield off, and run the other end to the star ground. This has really cleaned up a lot of my work. Also, on a budget, I keep an eye out at the flea mkts. for old computer speakers, usually for a couple bucks. Most of these have shielded wire. Some of the wire is too small for larger amps, but work very well for opamp/preamp circuits.

idle_chatterbox:
agreed.

Old RCA cables that came with VCRs can be used as shielded wire too.

I haven't tried shielding the output wires, like you say. Big difference? I assume you mean wires that terminate at the speakers?

willy1usa:
Yes, tight budget encourages inginuity. Some of these old computer speaker cables have 2 wires + a shield. Especially in opamp circuits, I'll use the 2 wires for +/-, then ground one end of the shield. Some cables have foil for shielding, and some stranded wire. I've cleared up quite a few hums with this method.

Great article on grounding. Everyone, at some time, have had grounding problems.

Navigation

[0] Message Index

[#] Next page

Go to full version