Star grounding problem

[Steve Conner]

Indeed, check the wiring of the rectifier and first filter cap. A strict application of the star grounding principle passes the first filter cap charging current through the star point, which causes dreadful hum. At least it did in one of my early builds. You have to separate the charging current loop from the rest of the ground system, as RG explained at length.

SGM is right in a way. Grounding the power supply to one end of the chassis and the preamp to the other end would achieve this separation. But his explanation obscures what's really going on.

[woodyc]

The most common problem I see is preamp tubes that hum. I'd try tube swaps before re-doing grounds. Its good to keep a few solid NOS tubes around for checking.

[soundguruman]

Indeed, check the wiring of the rectifier and first filter cap. A strict application of the star grounding principle passes the first filter cap charging current through the star point, which causes dreadful hum. At least it did in one of my early builds. You have to separate the charging current loop from the rest of the ground system, as RG explained at length.

SGM is right in a way. Grounding the power supply to one end of the chassis and the preamp to the other end would achieve this separation. But his explanation obscures what's really going on.

Yes I am obscure and oblique, inscrutable and vague, so hard to pin down... I am thinking of my own trials with this.
In a very high gain (absurd infinigain) (with DC heaters) amplifier, I have tried every possible configuration, measuring each time the actual noise on the output, with an O scope.
And it turned out that separated was actually better by a substantial amount. Cut the hum in half, or more.
But in a normal lower gain amp, you may not notice or appreciate this reduction at all.
And as it has been said many times, it's not just the grounding config that makes an amplifier hum.

ANY hum from any source, that enters the first stage of the preamp audio chain will be amplified thousands of times.
And that is where I concentrate the effort, to prevent it from entering to start with.

[Chuck H]

So, like me, there is at best a vague understanding but plenty of actual trial and error. That's entirely valid. But in such circumstances it's possible to THINK you understand something while the truth is something different. That's why I try to be careful when giving advice on subjects that are like that for me. I'll say "What works for me is..." and "What I think is happening is..." I'll sometimes even include the tag line "corrections accepted". Just because I know something works doesn't always mean it works for the reasons I think it does. I prefere to be wrong and learn to being right only in my own head but wrong in actuality. There's still plenty I think I know!?! Corrections accepted.

[woodyc]

...And it turned out that separated was actually better by a substantial amount. Cut the hum in half, or more...

Any chance you could post a chassis pic of an amp you grounded like this?

[R.G.]

I am thinking of my own trials with this.
In a very high gain (absurd infinigain) (with DC heaters) amplifier, I have tried every possible configuration, measuring each time the actual noise on the output, with an O scope.

There are, literally, infinite number of configurations, so this is logically impossible. And perhaps more importantly, it's on one amplifier (if I read that correctly), so by definition, it is logically impossible to say it's every possible combination. Further, it's faulty reasoning to say that from trials, however extensive, that it's better on one set of trials, therefore it's always better on different amplifiers. For an entertaining read, I suggest "Why People Believe Weird Things" by Schermer. He explains some of the pitfalls of small trials, anecdotes, and logical fallacies.

And it turned out that separated was actually better by a substantial amount. Cut the hum in half, or more.

I do believe on your trials that it might have been. The issue I'm after is **WHY???** If one can't understand how it changed, one cannot say how to make that same change in a different amplifier...

But in a normal lower gain amp, you may not notice or appreciate this reduction at all.

... as you say.

ANY hum from any source, that enters the first stage of the preamp audio chain will be amplified thousands of times.
And that is where I concentrate the effort, to prevent it from entering to start with.

And I agree with you enthusiastically about this.

Can you imagine how very different and much easier solving hum issues in guitar amplifiers would be if the high price of tubes had not made an input differential amplifier out of the question for the designers in the Golden Age? One duotriode in the front end as a diffamp amplifying a shielded twisted pair from the guitar? This kind of circuit offers some amount of power supply and input noise rejection. The single triode input has essentially zero ground noise rejection, so when you go to hyper-gain amplifiers with clipping, the noise sensitive grows hugely.

[Alan0354]

Post the schematic and post the pictures of your grounding. You can't trouble shoot by talking. You don't have to post the detail, white out the part you don't want us to know. Show the +HV and the ground of the preamp tube, show the ground of the and the power section. Then post the picture. I am sure I or other people can spot the problem if it is grounding problem. This is not that hard.

[soundguruman]

Post the schematic and post the pictures of your grounding. You can't trouble shoot by talking. You don't have to post the detail, white out the part you don't want us to know. Show the +HV and the ground of the preamp tube, show the ground of the and the power section. Then post the picture. I am sure I or other people can spot the problem if it is grounding problem. This is not that hard.

but, i like to troubleshoot by talking.
It's effective quite often.

[soundguruman]

"There are, literally, infinite number of configurations, so this is logically impossible. And perhaps more importantly, it's on one amplifier (if I read that correctly), so by definition, it is logically impossible to say it's every possible combination. Further, it's faulty reasoning to say that from trials, however extensive, that it's better on one set of trials, therefore it's always better on different amplifiers. For an entertaining read, I suggest "Why People Believe Weird Things" by Schermer. He explains some of the pitfalls of small trials, anecdotes, and logical fallacies."

You're right, there are a crap load of possibilities. And I've tried them all.

[R.G.]

You're right, there are a crap load of possibilities. And I've tried them all.

There are, literally, infinite number of configurations, so this is logically impossible. And perhaps more importantly, it's on one amplifier (if I read that correctly), so by definition, it is logically impossible to say it's every possible combination. Further, it's faulty reasoning to say that from trials, however extensive, that it's better on one set of trials, therefore it's always better on different amplifiers. For an entertaining read, I suggest "Why People Believe Weird Things" by Schermer. He explains some of the pitfalls of small trials, anecdotes, and logical fallacies.

[Alan0354]

Grounding is a very involved subject. A whole category of engineer called Signal Integrity engineer were created in the last 15 years that specialized in just dealing with grounding to avoid noise emission, susceptibility and ground loop. This subject is a branch off from electromagnetic, you want to learn and study about grounding, get a book on EMC and study the whole thing. At least those EMC books try to bypass the EM and get into more of a cook book approach. There are so many ways you can go wrong, there is no way to cover in this kind of discussion on a forum AND particular cover all posibilities. This takes years of experience in system grounding, pcb layout and testing to learn. You want to learn, buy a book to start with, you want to solve a problem, post the schematic, take a few pictures and we'll work on it. Experiment is really not a good way to do, as some very small variation that don't seem apparent can change the whole thing and you think you nail down a formula just to be proven wrong later..........Just like people get burnt by the "star ground" over and over.

The biggest mistake engineer or anyone make is to think the circuit is important, then grounding is after thought and fix it later. If you ignore grounding and power filtering ( grounding), the circuits in the book actually work!!!! Why circuit don't work when people build it? Mostly is back down to grounding and filtering ( AGAIN, GROUNDING). All the hum, motor boating, oscillations..........And then the kicker is people starting to make is as black magic, start having all the believe and snake oil.

If you don't want to go through the pain of studying the EMC book, look at the layout of the BF or SF Fender, they have good tubes layout ( which drive the grounding) and the simplest scheme of grounding. Read my first post here, they use the chassis as ground plane for the preamp, they use a copper strip solder at multiple points onto the chassis 2/3 the length of the chassis to serve as ground strap. They avoid injecting large current from the power amp by grounding the PI and power tube on the other end of the chassis and start using star grounding with the filter caps. I made further improvement that I pull up the cathodes of the power tubes and ground onto one single lug together with the low side of the big filter caps. With this, the large current of the power tube never even enter the chassis and you'll never see the ground current. Study the Fender, they did it right, it is simple and it work. Modify the PA grounding like I described, I just did it on my Bassman 100 and is quiet like anything with cascade gain crank up. Only noise I get is the front end thermal noise type of hissing sound. Don't get fancy, do it simple, save the difficult grounding for more sensitive circuits, which, this is not.

I think the way Fender putting the filter caps in a separate box and wire the two ends of each caps to the points they want is ingenious. You put the ground and bypass exactly where you want.

[jpfamps]

When I first started getting (back) into analogue electronics I built a couple of amps with dreadful hum problems, so I resolved to try understand how to eliminate this.

A couple of articles that really helped my understand the problem where one by Randall Aiken (which I think you can't now access from his website), and the section in Douglas Self's Power Amp book on grounding.

The two principals that I found useful, is that you should consider current flow as it is current flow through common impedances that cause ground loop hum, and that valves amplify the difference between the grid and the cathode.

Regarding the second point, this means that the bottom of the grid reference and bottom of the cathode resistor of a gain stage should be connected together with as little wire between them before you connect them to any ground reference. Thus any hum on the bottom of this will be common to the grid and cathode and not amplified.

The system I use is thus:

The heaviest currents in the amp are the charging currents to the first filter cap. I always connect the bottom of the main filter caps to either the bottom of the bridge rectifier, or the HT centre tap for amps with full wave rectifiers. This confines to the heavy ripple currents, and prevents them encroaching in to the audio. Call this star 1

The power valve cathodes, and bias supply filters are connected to the bottom of the screen supply filter cap. Call this star 2. I run a wire from star 1 and star 2, and then a wire from star 2 to the star ground point.

For the next stage (usually PI) I connect the bottom of the grid resistors to the bottom of the cathode resistor supplying that stage. These are connected to the filter cap supply power to that stage (star 3), and then I run another wire from star 3 to the star ground.

Essentially you will have a "mini" star at the bottom of each filter cap to which all the ground references for the gain stages supplied from that cap are connected, and then I run a reference wire to the star ground from there.

It's a lot easier to do that describe! I can probably generate some form of diagram if you want.

I expect you could simplify the above system somewhat by using a combination of star grounding and buss bar wiring, but I haven't tried this.

[woodyc]

...And it turned out that separated was actually better by a substantial amount. Cut the hum in half, or more...

Could you post a chassis pic of the amp where you observed this? Please?

The complete lack of specifics is impairing my ability to reproduce your result.

[Alan0354]

When I first started getting (back) into analogue electronics I built a couple of amps with dreadful hum problems, so I resolved to try understand how to eliminate this.

A couple of articles that really helped my understand the problem where one by Randall Aiken (which I think you can't now access from his website), and the section in Douglas Self's Power Amp book on grounding.

The two principals that I found useful, is that you should consider current flow as it is current flow through common impedances that cause ground loop hum, and that valves amplify the difference between the grid and the cathode.

Regarding the second point, this means that the bottom of the grid reference and bottom of the cathode resistor of a gain stage should be connected together with as little wire between them before you connect them to any ground reference. Thus any hum on the bottom of this will be common to the grid and cathode and not amplified.

The system I use is thus:

The heaviest currents in the amp are the charging currents to the first filter cap. I always connect the bottom of the main filter caps to either the bottom of the bridge rectifier, or the HT centre tap for amps with full wave rectifiers. This confines to the heavy ripple currents, and prevents them encroaching in to the audio. Call this star 1

The power valve cathodes, and bias supply filters are connected to the bottom of the screen supply filter cap. Call this star 2. I run a wire from star 1 and star 2, and then a wire from star 2 to the star ground point.

For the next stage (usually PI) I connect the bottom of the grid resistors to the bottom of the cathode resistor supplying that stage. These are connected to the filter cap supply power to that stage (star 3), and then I run another wire from star 3 to the star ground.

Essentially you will have a "mini" star at the bottom of each filter cap to which all the ground references for the gain stages supplied from that cap are connected, and then I run a reference wire to the star ground from there.

It's a lot easier to do that describe! I can probably generate some form of diagram if you want.

I expect you could simplify the above system somewhat by using a combination of star grounding and buss bar wiring, but I haven't tried this.

Yes, I forgot to mention about the bridge rectifier, the low side connect to the point where the cathode of the power tube and the big cap connect together. Point is all the large current come to that one point and never even touch the chassis.

It is all about current, the preamp tube draw low current and it is not as bad. One think against conventional believe in higher frequency but not as prominent in audio ( but definitely applies), ground current don't spread out on the metal chassis as people think. They follow the power wire above it to a big extend. There is a saying in EMC: Return current follow the path or least resistance at very low frequency. Return current follow the path of least IMPEDANCE in moderate and high frequency. They don't spread out. By a well planned tubes layout, you can avoid a lot of problem and just simply ground on the chassis using a copper bar like Fender. Star ground is a double edge sword, I trouble shoot for a guy here on a Kit Twin, the problem turn out to be he accidentally put the ground of one of the tube in a channel onto the star of the PI tube and cause all the problem. With a star, you are forcing all the ground current of the tubes to go to one place and they cross talk. With spread ground, by understanding the concept of current follow the path of least impedance or even resistance, you literally isolate the different parts of the circuit just by the proximity effect. I told the person to just follow the Fender SF and it was dead quiet and he was a happy camper.

there is no black magic in grounding what so ever, it is very well researched and written in EMC books.

People should not forget about the power, B+ through capacitor to ground induce ground current. It is important the B+ from rectifier connect to the center tap of the transformer ( push pull) and the hot side of the big cap at one point also. The return of the filter cap after the choke should stay away from the preamp stages. The return of the filter caps for different stages of the preamp should go to their cathode circuit grounding point if possible so the current noise has the shortest path return to the circuit. Don't force the return of the filter caps in a cascade stages to go to one start as you'll be forcing their current together and start interacting. This is not the right way to do grounding.

I had issue in cross talk on the HV lines in the cascade high gain stages from the B+ of PI or even the power stages. The result is actually like having tremolo when the volume is cranked up. It needs more RC filtering than I expected.

[jpfamps]

Yes, I forgot to mention about the bridge rectifier, the low side connect to the point where the cathode of the power tube and the big cap connect together. Point is all the large current come to that one point and never even touch the chassis.

The bottom of the bridge rectifier shouldn't be connected to the power valve cathodes.

The power valve cathodes should be connected to the bottom of the screen supply filter cap.

The SF or indeed BF grounding scheme is nothing to write home about, and is hardly a model example of grounding.

Yes it works, by these are fairly low gain amps.

The grounding scheme seems more to facilitate the construction than optimize the amp for low noise.