Star grounding problem

[dobrota]

Hi,
I'm building 2x6V6 tube amp, power section is copy of the Fender Deluxe reverb. I use star grounding system for smallest hum, but it isn't work, even hum (i guess not low 50/60Hz sound but "rectified" 100/120Hz tone) is loud. Please help me, what's wrong ?
- main trasnsformer is mounted horrizontaly on chassis, shielded output transformer verticaly, not close about 3" far from main transformer with 90 angle between their axis
- i have one star ground point on the chassis (SGP next), not close to main transformer and wires to tube heaters
- power supply is on PCB and design is: capacitor 10n parallel to HT winding, bridge rectifier 4x 1N4007, first 33u capacitor, wire to OT center tap, 1k resistor, second 33u cap, wires to power tubes G2 pins through 470 ohm resistors, 10k resistor, third capacitor 33u to supply phase inventor. Grounds pins first two caps are connected together and wire from ground pin of the first cap runs to SGP
- next wire runs from ground pin of third cap to SGP
- heaters windig has no center tape, next wire runs from virtual center (two 100 ohm resistors) to SGP
- all wires to tube heaters are twisted
- next wire runs to SGP from bias power supply capacitor ground pin
- cathodes of power tubes are connected together and next wire runs to SGP from this point
- output jack is isolated from chassis
- wires from OT secondary run direct to output jack pins, next wire runs from jack ground pin (sleeve) to SGP
- next wire runs from tip pin of output jack to top phase invertor negative feedback resistors divider
- next wire runs from negative feedback resistors divider ground point to SGP
But, as I wrote, it doesnt work well... I also tried to connect wires from ground pin of third cap and output jack ground pin direct on phase invertor negative feedback ground point and then run only one wire to SGP from here, but nothing didn't improve
I thank You very much for Your advices

[nickb]

There are a couple of possible ground loops that I can see if your output and input jacks are not isolated from the chassis. However, before we get too complex, try to figure out the nature of the problem. How loud is the hum? Do you have a meter with AC volt range on it? Measure the ripple on the power supply. If that is OK try to isolate by pulling all the tubes other than the 6V6's then putting them back in order working back down the signal path starting from the power section.

[dobrota]

Hum

There are a couple of possible ground loops that I can see if your output and input jacks are not isolated from the chassis. However, before we get too complex, try to figure out the nature of the problem. How loud is the hum? Do you have a meter with AC volt range on it? Measure the ripple on the power supply. If that is OK try to isolate by pulling all the tubes other than the 6V6's then putting them back in order working back down the signal path starting from the power section.

Hi, I'm just trying only power amp (2x6V6 + phase invertor), preamp isn't connected to it. Both input and output jacks are isolated from chassis. After posting my thread I tried also ground buss system instead of star ground but all is almost the same, so it seems problem is maybe really in power supply circuit, I'll try to replace capacitors. Unfortanelly I have just only multimetr with 200 and 750V AC range, so i can't rigth measure ripple on the power supply. I also tried to remove phase invertor tube, "rectifed" hum is lower but still noticable enough, so it seems as PSU / ripple? problem again.
But I'm thinking about bias power supply now - it's made from HT winging in "Marshall style" e.g.
http://www.drtube.com/schematics/marshall/3203pwr.gif
Maybe it reguires special way to connect its ground points to other grounds in PSU. I'm also not really sure the way to ground output pin of the output jack, if run it to PSU ground point or connect it to NFB divider ground is better.

[Enzo]

Remember the bias supply is negative, so the filter caps for it must be wired in reverse - their + end is grounded. If you ground the negative end of them like all th other caps in the amp, they won;t work and you get hum.

Your meter only has 200v and 750v on AC? So use it anyway. Set it for 200VAC and see what you get, particularly on the bias suply.

You have described your wiring, but was it successful? MEASURE DC voltage at each pin on the power tubes. Do you get full B+ on pin 3 AND pin 4 on each power tube? DO you get full negative voltage on pin 5? Is pin 8 REALLY grounded?

[nickb]

The 200 VAC range should be OK for this as we're looking for fairly large signals. Good idea on the bias cct, it's a critical area. You can put a bigger cap across the 47uF to reduce hum. I'm guessing your meter is, if you'll 'cuse me, "a less expensive" type so you might get misleading readings if you probe the bias voltage. You can still pull the phase inverter tube.

[Chuck H]

And I would add that replacing components before isolating the problem is always a good way to make more problems before the old one is fixed. Leave the filters alone for now.

[soundguruman]

Hi,
I'm building 2x6V6 tube amp, power section is copy of the Fender Deluxe reverb. I use star grounding system for smallest hum, but it isn't work, even hum (i guess not low 50/60Hz sound but "rectified" 100/120Hz tone) is loud. Please help me, what's wrong ?
- main trasnsformer is mounted horrizontaly on chassis, shielded output transformer verticaly, not close about 3" far from main transformer with 90 angle between their axis
- i have one star ground point on the chassis (SGP next), not close to main transformer and wires to tube heaters
- power supply is on PCB and design is: capacitor 10n parallel to HT winding, bridge rectifier 4x 1N4007, first 33u capacitor, wire to OT center tap, 1k resistor, second 33u cap, wires to power tubes G2 pins through 470 ohm resistors, 10k resistor, third capacitor 33u to supply phase inventor. Grounds pins first two caps are connected together and wire from ground pin of the first cap runs to SGP
- next wire runs from ground pin of third cap to SGP
- heaters windig has no center tape, next wire runs from virtual center (two 100 ohm resistors) to SGP
- all wires to tube heaters are twisted
- next wire runs to SGP from bias power supply capacitor ground pin
- cathodes of power tubes are connected together and next wire runs to SGP from this point
- output jack is isolated from chassis
- wires from OT secondary run direct to output jack pins, next wire runs from jack ground pin (sleeve) to SGP
- next wire runs from tip pin of output jack to top phase invertor negative feedback resistors divider
- next wire runs from negative feedback resistors divider ground point to SGP
But, as I wrote, it doesnt work well... I also tried to connect wires from ground pin of third cap and output jack ground pin direct on phase invertor negative feedback ground point and then run only one wire to SGP from here, but nothing didn't improve
I thank You very much for Your advices

You might think that star grounding is the best way, but it's not.
You will get more noise with a star ground not less.
The power supply and filters should be grounded on one side of the chassis, including the bias filter cap.
The preamp grounds should be on the opposite side of the chassis.
NOW, there is the resistance of the chassis between the two ground points.
Believe it or not, that works much better.
AND the chassis must be grounded to the service ground. (the power outlet supplies the safety ground from the third pin)

[dobrota]

All voltages are OK, each tube is biased about 25mA, bias voltage is about -34V, there's about 380V on center of OT primary. Power amp works with speaker and guitar or external preamp connected to it's input. Yes, it works almost OK.... but this hum !
I have other better multimeter with VAC range from 200mV, but unfortunally it stopped work some days ago, I have to buy new other one. It seems, I'm just in some bad DIY period.

[nickb]

Just to be clear, the point of the temporary additional capacitor test is to isolate the problem to/not to the grid bias section. It's not a replacement.

[R.G.]

You might think that star grounding is the best way, but it's not.

Define "best way".

Star grounding is a means to an end. There are an infinite number of ways to do grounding. Star grounding is the only way I know of that can be shown ahead of time to eliminate ground-induced hum. It does this by eliminating ground currents flowing in common conductors, and thereby converting the current into a voltage in the common resistance.

However, no grounding scheme can guarantee freedom from capacitive-mode noise transfer or magnetic field pickup noise and hum.

You will get more noise with a star ground not less.

That's a pretty strong statement. Can you say why that should be true? Are there some circumstances where it might not be true? Is it on average true, or always?

The power supply and filters should be grounded on one side of the chassis, including the bias filter cap.
The preamp grounds should be on the opposite side of the chassis.
NOW, there is the resistance of the chassis between the two ground points.
Believe it or not, that works much better.

(a) Can you define "better" as you mean it here? How is that measured?
(b) It does not matter at all if there is a resistance of any amount between two points if there is no current flow between them. The certain way to ensure that there is no current flowing between two points is to have one and only one conductor connect them. There is no electrical circuit, hence no current can flow. The resistance can be zero, one ohm or one megohm. If there is no current flowing between them, there cannot be a voltage between two resistively connected points. This is the point of star grounding.
(c) There are many ways hum gets into audio circuits. One is conducted hum, or the hum of a floating ground. Another is electrical field (that is capacitive coupling) and another is magnetic field (that is, inductive coupling, including wire-loop hum pickup and transformer pickup). Another is the whole family of ripple voltages, including insufficient filtering, ripple buzz from RF generation caused by rectifier snap-off at 2x line frequency, improper grounding of the first filter cap, and others. And radio pickup. There's hum on radio signals, too.

Well thought-out star grounding can pretty much eliminate hum from power line grounding issues. It may or may not help with hum from other causes. You can also get into situations where the "grounding" to eliminate hum actually induces hum, but just happens to introduce it in a way that cancels some other hum source, like the pot across a heater supply on some amplifiers.

IMHO, it's overly simplistic to say that star grounding or other grounding means is not the best way, and that some other way is. There exists a host of hum sources. To get truly low hum, you have to circumvent them all.

I'd have to see the OP's layout and wiring, but I suspect he has problems with rectifier/first filter caps. I'll have to read it more closely.

[soundguruman]

Define "best way".

Star grounding is a means to an end. There are an infinite number of ways to do grounding. Star grounding is the only way I know of that can be shown ahead of time to eliminate ground-induced hum. It does this by eliminating ground currents flowing in common conductors, and thereby converting the current into a voltage in the common resistance.

However, no grounding scheme can guarantee freedom from capacitive-mode noise transfer or magnetic field pickup noise and hum.


That's a pretty strong statement. Can you say why that should be true? Are there some circumstances where it might not be true? Is it on average true, or always?


(a) Can you define "better" as you mean it here? How is that measured?
(b) It does not matter at all if there is a resistance of any amount between two points if there is no current flow between them. The certain way to ensure that there is no current flowing between two points is to have one and only one conductor connect them. There is no electrical circuit, hence no current can flow. The resistance can be zero, one ohm or one megohm. If there is no current flowing between them, there cannot be a voltage between two resistively connected points. This is the point of star grounding.
(c) There are many ways hum gets into audio circuits. One is conducted hum, or the hum of a floating ground. Another is electrical field (that is capacitive coupling) and another is magnetic field (that is, inductive coupling, including wire-loop hum pickup and transformer pickup). Another is the whole family of ripple voltages, including insufficient filtering, ripple buzz from RF generation caused by rectifier snap-off at 2x line frequency, improper grounding of the first filter cap, and others. And radio pickup. There's hum on radio signals, too.

Well thought-out star grounding can pretty much eliminate hum from power line grounding issues. It may or may not help with hum from other causes. You can also get into situations where the "grounding" to eliminate hum actually induces hum, but just happens to introduce it in a way that cancels some other hum source, like the pot across a heater supply on some amplifiers.

IMHO, it's overly simplistic to say that star grounding or other grounding means is not the best way, and that some other way is. There exists a host of hum sources. To get truly low hum, you have to circumvent them all.

I'd have to see the OP's layout and wiring, but I suspect he has problems with rectifier/first filter caps. I'll have to read it more closely.

How is it measured?
With a Tektronix 465 O scope.
Why is it better?
The ripple rejected from the power supply will end up in your audio preamp chain, if you don't separate them.
Better means you can actually measure an improvement.
Many manufacturers are using a resistor between the grounds instead of connecting them together. it hums less.
OR we use the resistance of the chassis to separate them.
Hum comes from multiple sources, you have to attack them all.
Yes, I have said that many, many times in this forum. I agree.
I was referring to the grounding of the power supply and the preamp. BUT there are several other sources of hum, like the AC filaments.

Overly simplistic? No not at all, it's reality. Keep your power supply grounds separated from the audio grounds, less noise.

We have been taught that star grounding is better, but in reality, it's not true.

[Alan0354]

It is very hard to trouble shoot by reading description. If you can post the schematic and take a few picture, then it should be easy to trouble shoot. If you have secret circuit, you don't have to show the detail. Just need to know where is the ground of the cathode circuit, the filter cap grounding of the preamp. Also the ground of the tone stack. If you just copy the Deluxe PA, then post the schematic and the picture, I need to look at the ground connection.

Ground can be tricky. Star ground sounds good in theory, but real life circuit might be different. Look at how Fender ground the stages, they use the chassis as ground plane for the preamp. Then they use one point ground on the high current filter cap. the ground of the first big filter cap should be close to the two cathode of the power tube so the path is very short for the high current of the power tube. I actually do it a little different, I actually connect the cathode of the two power tube to a lug on the chassis close to the tube, then I solder the ground of the big filter cap onto the same lug that the cathodes of the power tube connect to. This way, the PA return current never even go through the chassis, straight from cap to cathode. I have NO hum period even with cascade stage cranked up.

Last time when I was helping someone trouble shooting a kit amp of Fender Twin, turn out he put multiple star ground and mistakenly put the ground of one of the preamp tube to the same star of the PI stage. that's cause all the problem. I caught it from the photo. Again, Download an old Fender schematic, it has the layout and the grounding point. Follow them, don't be fancy and create your own ground unless you really know what you are doing.

[Chuck H]

I have to side with R.G. on this. If there is NO current between grounded circuits, AND EACH CIRCUIT GROUND IS ON IT'S OWN LEAD!!!, there will be no current between grounds AND therefor, no grounding issues. BUT... This NEVER happens in reality. There are always daisy chained grounds (except in my amps ). There are NO corners to be cut when designing a ground scheme. If there is even ONE daisy chained ground then it's entirely possible to get quieter performance by separating the PA and preamp ground points on the chassis. Otherwise, if properly done, the star grounding ideology is the best. I HAVE NEVER DONE IT!!! I actually have two or three ground points on my amps. Somewhat as per SGM's description. But I do this for convenience. I am careful otherwise and I don't have any noise issues. But I am very aware that it's not ideal when I do it. There is a little devil in my head that doesn't want the PA ground leads neadr the preamp. Obviously the preamp is sensitive and there is SOME resistance in the lead itself. There is also significant current in the PA as compared to the preamp. This also get's in my head. The important truth is that if I ever have a problem with the ground scheme I've been using I will quickly abandon it for a star ground pattern. Because I know that any ideologies I base my current method on are not based on tested results but star grounding is.

There is NO reason to have any resistance between grounds. In fact it is generally a bad thing. If all circuit grounds are at 0V with 0 resistance that's as good as it gets. And, as I mentioned before, it's the poor ground scheme and corner cutting shared grounds in every design out there that create problems. How many builders even run the actual earth ground from the AC cord to a point near the input??? Of course most builders tack it or screw it to the chassis right where it comes in! That's also where they ground the PA. Same thing I do. Then I have a single point near the input for all the preamp grounds. Same thing everyone else does. But a light came on for me when I was looking at modern uber gain layouts. The grounding can seem confusing at times with the PCB's and all. But it's really a star ground most of the time. So it seems that when the going gets critical, the critical go with a star ground.

JM2C

[R.G.]

How is it measured?
With a Tektronix 465 O scope.

Actually, you miss the point. Not "with my right hand on the volts/cm knob", is better lower volts? What frequency? Power line? 2x power line? harmonics? Spectrum?

Why is it better?
The ripple rejected from the power supply will end up in your audio preamp chain, if you don't separate them.

Actually, you miss the point. You imply that a resistive connection between power ground and signal ground causes lower overall hum in the signal (although you only say "better" not lower hum). I meant - why SHOULD a resistance between the two cause lower hum? How does that work? Does Maxwell's Demon sit there and hold the Hum Demons at bay? Do extragalactic anti-hum particles get attracted? Why should it be "better" as defined someday by the first question?

Better means you can actually measure an improvement.

See question 1, and define "improvement".

Many manufacturers are using a resistor between the grounds instead of connecting them together. it hums less.

And as we know, Mother Nature always follows the voting of many manufacturers, right?

OR we use the resistance of the chassis to separate them.

HOW does that make hum lower?

I was referring to the grounding of the power supply and the preamp.

Here's the rub: preamps use power, so they have to connect to the power supply ground somehow, yes? Is that one wire, many wires, many wires with 100R resistors in them, what? Which is better and what is the technical explanation for which one makes it better?

Overly simplistic? No not at all, it's reality. Keep your power supply grounds separated from the audio grounds, less noise.

We have been taught that star grounding is better, but in reality, it's not true.

Again, describe WHY that happens, not just that "well, I've seen a lot of them and these are quieter."

There are several kinds of ground. There's power ground, signal/reference ground, shield ground and what I call sewer ground. And all conductors are actually resistors. If you *ever* have current running through a ground wire, that current makes a voltage by V = I*R. If that wire is conducting both rippling power ground currents and signal/reference ground, you induce hum. If you take the same setup, split the wire so the signal/reference ground and the power ground currents go through two different wires, the hum goes away. That's a two pointed star.

Tell me how you'd set up the same situation with a resistor between the signal ground and power ground, and make that quieter.

One really, really, really common problem with ripple/hum in amplifiers is incorrect wiring of the first filter cap (-) lead. The negative side of the first filter cap must connect to the chassis, and to signal ground, and to the power return of the output stage. It also has to source large(-ish) pulses of current to the negative side of the rectifiers (which is the CT in many amps). To get minimal power supply ripple noise on this setup, the minus side of the rectifiers must connect directly to the first filter cap negative lead, and no where else. From there, you can take wires to a star point, chassis, whatever. But if you don't make at least this terminal a two pointed star, you will get ripple from the current passing through that wire. The only thing that minimizes the noise is to take two wires, one in with the power pulses, and one or more out with the DC. It's easy to see this on a Tektronix.

[R.G.]

Actually, I'm being a bit ingenuous on you. There is one situation where inserting a resistor in ground loop makes things quieter. That's when the loop is carrying AC currents that would otherwise be inductively coupled to other places. Cutting the current cuts the M-field transmission. But if what you want is quieter voltage differences, it is not a help.