Star Grounding in Tube Amplifiers
Copyright 2000 R.G. Keen. All rights reserved. No permission for local copies or serving from sites other than http://www.geofex.com.

[Causes][Star Rules][DIY vs Commercial Amps][How to][How much is enough][Q&A][Example]

"I've built this amp, and it hums - why!!!??? What can I do to keep amps from humming, or make them never hum in the first place?"

The power line hum demon lurks literally everywhere, inserting that low "mmmmmmmmmmmmmm" into your speakers. Sometimes just a little in the background, sometimes a maddening "MMMMMMMMMMMMM". How do we exorcise it?

One certain answer is star grounding. It's not the only way, but it's probably the only slam-dunk, cookbook way to ensure that if you start doing this, you will eventually have as low a hum as you can.

You have to do the other stuff right, of course. Proper grounding  will not prevent floating AC filament supplies from causing hum, or  magnetic field pickup from a power transformer, or capacitively coupled AC hum from signal wires next to filament wiring. But once you've done the other basics right and still have hum, think about star grounding. Better yet, think about it before you start!

What causes shared-ground hum anyway?

Except for certain special materials that are cooled to very cold temperatures - at least liquid-nitrogen cold - all conductors have some resistance, even if it is very low. Copper is a good conductor, and it is the most commonly used wiring material because of this and its affinity for soldering. But it is not perfect. Steel, as commonly used in amplifier chassis is also a conductor, but a much worse one than copper. Worse yet, steel is ferromagnetic - it also "conducts" magnetic fields. A steel chassis actually sucks in the random power line magnetic fields that permeate the air around us. The magnetic field "conducted" in the chassis causes eddy currents to flow in the metal, and the resistance of the material causes a slight power line frequency voltage to exist between two different points on the chassis. This is easily observable with a good oscilloscope. This is eddy current induced hum.

The copper wires carrying "used" current back from the signal circuits also carries a tiny replica of that current in the form of a voltage across its small, but present resistance. This is shared resistance feedback; it does not cause hum, but it can cause your amp to be unstable and oscillate madly.

Further, any conductive loop will have a voltage impressed across it by a changing magnetic field. So if we have literally a ground loop - that is, a full circuit made of conductors that we think should all be at the same voltage - the changing magnetic field in the air around us will cause a voltage to appear in that loop that will mean that every point on the loop is NOT at the same voltage like we think it should be. This is magnetic field ground loop hum.

We attack these hum sources (and coincidentally the shared resistance feedback) two different ways. First, if there are no signal grounds that go through the chassis, then the signal grounds can't pick up the induced magnetic field hum voltage.  Second, if there are no complete loops, there is no induced voltage from ground loops. We make this true by star grounding. Star grounding that is connected to the chassis at one and only one point solves both problems, is always sufficient, and is something that requires no deep understanding of magnetic fields, shielding, or other specialties - it can be a cookbook. The side effect of this practice is that star grounding also breaks the shared resistance runs and removed the possibility of shared resistance feedback.

There is one other instance of induced hum that is also cured by star grounding - rectifier current spike hum. Rectifiers of all types, tube or solid state, conduct currents into the first filter cap of a power supply in short pulses near the very peak of the AC power line waveform. Since the time that they flow is short, the current that flows is big, as the diode has only this short time to load up the filter capacitor with all the electricity that the amplifier will need until the next time a diode can conduct to reload it. These spikes of current are so large that they cause an easily measurable voltage on the wires into and out of the first filter capacitor. If the signal ground connects to anywhere except right at the terminal of the capacitor, then it picks up the rectifier spike noise as well. To eliminate this special form of hum, we take a single wire from the return side of the first filter cap and connect that to the system star ground point so that the resistance of the return wires does not cause a hum voltage from the rectifier spikes.

OK, so what's "star grounding" really mean? How do you do it?

Simply put, star grounding means that you designate some special terminal as the "Star Ground" for the system. All other "grounds" will be referred to this one point. Then *every single place* that is connected to ground in the whole box has its own separate wire run to the Star Ground point. The ground wires radiate away from the Star Ground in all directions, hence the name. This has the unfortunate result that you may have a HUGE number of wires coming into the Star Ground point. Very hard to wire, just because of the large number of wires.

While this is certainly sufficient, it's almost never necessary. It turns out that most circuits have subsections of lumps of circuitry that can share a ground wire safely and hum-free. For instance, let's take a vacuum tube triode gain circuit. This circuit will often have a grid leak resistor, a cathode resistor, and a cathode bypass cap, each of which should have its own ground wire in pure Star Grounding. Instead, it turns out that it works just as well to gather all three of these points into a local ground node, and then connect that point to the system Star Ground by a single wire. This just cut the number of ground wires by three!

Most forms of star grounding use the modified version with local nodes that do not interfere with one another connected to the remote Star Ground point. In either case, it's a lot of work.

Why should we have to use star grounding - very pertinent question from a DIY amplifier builder - 

my commercial Fender amp does not hum, and there's no star grounding in there at all. So why does a DIY amp have to take such drastic measures as star grounding to reduce hum? Even after all the best standard practices, good parts, and good layout techniques, I still get hum in my amps all the time. What's going on? Does the Hum Demon work for Fender and make our homebuilt jobs hum?

What you can't see is all the test amplifier prototypes that Fender and other amp makers did before they ever sold the first one of that kind of amplifier. They can afford to have a junkpile, we can't. 

Think of it this way. There are an infinite number of possible ways to wire an amp. Some of those hum horribly, some just a little, and some small number have no perceptible hum. All we can say is that a good star grounding scheme is guaranteed to be one of that small, non-humming group. Since hum depends on where every wire connects in the amp, we can't in general know the other members of the group without experimentation to find them - or know if there even *is* another member of the non-humming group at all. Star grounding may be the ONLY way, may not be. Can't tell.

Star grounding is a sufficient but maybe not a necessary condition. That is - using star grounding is one possible way to guarantee that you won't have hum. Another is to spend enough time messing with an amp to find where every wire and ground goes so that the hum-causing currents do not happen to flow where they cause a problem, or in another variation, there are canceling currents in the non-star grounding scheme so that the grounds are actively quiet. Experience and understanding of hum, grounding and interference may shorten the prototyping time from twenty humming prototypes down to five with years of experience, but there's still a bone pile.

Unless you're lucky, of course. It could happen that with no knowledge or experience to guide you, you might happen on a non-humming layout first time. No amount of planning will ever take the place of dumb luck. But I personally never plan on being lucky several times in a row 8-)

How do I star ground an amplifier?

1. Establish a star ground point; for practical reasons in soldering, it should have room for many wires to be connected into it using good soldering practices.
2. Isolate the input jack(s) from the chassis and run the(ir) ground wire(s) to the local node of the gain stage that they feed.
3. Isolate the output jack(s) from the chassis and run their ground wire to the star ground point.
4. Connect the ground wire for attenuator style controls such as tone/volume stacks to the local ground node for the amplifying stage that amplifies the output of the control. As an alternate, a separate ground wire for the control is also acceptable, but less easy to wire.
5. Differential, phase inverter, and push-pull stages should have their own ground wire to the star ground point.
6. The loop(s) from the power transformer through the rectifiers and filter caps and then back to the transformer are special. They MUST be wired as follows:
   • wires from the rectifiers to the +  (or  -) side of the first filter capacitor in the power supply must connect to the capacitor
   • wires carrying power TO the circuitry and out of the first filter capacitor must be physically different wires than the wires carrying power into the capacitor from the rectifiers, and connect directly to the capacitor terminal
   • the grounded side of the first filter capacitor connects to the star ground through a wire that is physically separate from the return wires in the rectifier loop; this wire connects to the capacitor terminal, not somewhere along the rectifier return wire.
7. The CT of the filament winding ties to the star ground point. 

How much is enough? Can I only star ground part of the amp and make it better? If so, which parts? 

The point of star grounding is to keep signal circuits with NO common mode noise rejection (that is, all the normally used tube gain stages) from picking up current-induced noise from other currents running through a shared ground. Obviously, the most critical stage is an input stage, and the closer you go to the output stage, the bigger the signal level and the less the milli- and micro-volt signals in the shared ground will be.

So you can achieve some reduction by removing the ground path of either end out of the shared-ground chassis. Remove the ground return paths for the small signal stages from the chassis onto separate ground wires and they don't pick up the current induced noise from the high current stages that you've left in the chassis; remove the ground path for the high current stages from the chassis onto separate wires to the star ground and the currents remaining in the chassis get a lot smaller. Either way, you don't have the small signal stages picking up the ground noise caused by the high current stages.. Is it enough to wire all the signal stages on separate ground wires and leave the output stage currents in the chassis? For many if not most situations, yeah, probably. It's certainly better than having them all in the chassis. There will be some situations where it won't be OK, though. Maybe you'll be lucky.

The ultimate ground noise reduction is achieved by removing ALL the ground returns from the chassis and returning them to a star, but let's face it, there are other ways to get amps to work, as Fender shows us all the time. The star ground method is not the only way to achieve lower noise and crosstalk, it's just guaranteed to do it - that is, it's sufficient, but it may not be necessary, depending on the circumstances.

It's entirely possible that moving the input stages and reverb returns off the chassis might quiet things down just fine, depending on where everything else is inside the amp.

Some detailed questions about grounding:

With cathode biased power tubes, should you run the cathode resistor and bypass cap to the star ground point, or is it okay to just bolt it to the chassis?

I personally would isolate it from chassis and run a wire to the star ground, as I'm sure that would work. However, if all the other things are chassis-isolated, it might be just fine. 

Do I need to isolate the grounds on the speaker jacks from the chassis?

I tried the speaker jack connected/not connected to the chassis in an amp I was re-grounding, and for the life of me could not hear any difference. I don't know whether this was because I already had everything else off the chassis or whether it truly didn't matter at all. I guess I would take a somewhat rough and ready approach - if it can be isolated and star grounded with a reasonable amount of wiring, I'd do that. If not, I'd leave it to the end and see if the results were acceptable from a listening test.

If you notice in the example, I did that with the speed control ground on the tremolo. It was in a position that was truly awkward to run a separate ground wire to, so I decided I was done and just listened to the result. It was OK. Not perfect star grounding, but it works fine, so I left it there.

What about the bias circuit for a fixed bias amp - does that have to be star grounded too, not connected to chassis?

Bias circuits are things that I consider sacred. Make a mistake in a bias circuit and you get big problems. I'd recommend lavishing great care on a bias circuit, and so while it may be perfectly fine grounded to the chassis, I'd star ground it on principle. It *is* an alternate input to your output stage, so if there's hum or RF on your chassis, it's conducted into the inputs of your output tubes. I'd star the bias - unless actual listening says it works fine without it. 

A star ground example

Here's what I did to my Vox Pacemaker (the tube version, not the solid state one).  It was originally Godawful - "grounds" made by just wiring to the nearest terminal strip terminal that was riveted to the chassis. It hummmmmmmmmmed. It's much quieter now - the hiss of the vintage carbon composition resistors can be clearly heard 8-)