Point to point star ground returns (Routing)

[lux01]

Hi, my 1st post here, searched archives some but stillno answer. I'm building my 1st mic pre from scratch ( Lot's of guitar amps ) and have never done a star ground
inside a chassis. Here's my question: How to route the individual returns from each stage to the star point? From a neatness point of view I'd like to twist them but something tells me this is not the most sensible thing to do noise wise. There are 4 stages 1st is ef86, second and third are 12ay7, then 12au7 output (Pultec MB1 cathode follower output stage).
Thanks

 

[abbey road d enfer]

First, do you really need a star ground for audio?
Grounding a piece of audio equipment involves reconciling two potentially conflicting necessities.
In order to minimize internally generated noise, you must make sure the ground connection between two subsequent stages are as short as can be.
In order to minimize externally induced noise, you must make sure the input and output ground connections are as firmly connected to ground as possible.
For a balanced in/balanced out unit, it is not that difficult, because the input and output ground connections do not need to be connected to the audio ground.
But for an unbalanced unit, it is rather difficult to achieve, that's why that kind of gear needs some tweaking for best noise performance. Since your mic pre is of the former category, you should concentrate on a ladder-type ground, one that is as close as possible to the schematic, IN PARTICULAR, the smoothing caps grounds MUST be wired in the same order as their positive pins, i.e. chassis to negative of 1st smoothing cap, then to negative of 2nd smoothing cap, to cathode resistor of last tube, to 3rd smoothing cap, to cathode of 3rd tube, to 2nd stage, to first stage,...
Then you connect pins 1 of both in and out XLR's to chassis with the shortest path between them. That's where you want your star ground.
And anyway, it is always recommended to make the low-level stages as compact as can be, with the shortest ground path possible.

 

[radiance]

I see ground paths as branches of a tree. Eventually they're all connected at one point (the trunk), but they're not a star ground.

 

[rotation]

I don't think you really need to make "real star grounding". In my builds i'm latelly making buss grounding; this means that all ground wires are connected to one solid wire which goes from input to output stage and then to your central (star) ground. This way you only have two connections to central ground, "buss wire" (solid wire that goes from input to outptut stage) and power supply minus (taken from the last smoothing cap).
You can read more about grounding here:
http://www.aikenamps.com/StarGround.html
But don't make what i did once because of reading too much git amp stuff; separate anode resistors and decoupling caps from the rest of the circuit by puting them on another board. You only need to separate psu from the rest of preamp.

Good luck!

 

[JohnRoberts]

The important concept to understand is that ground is not a voltage or node, but a circuit or series of connected conductors (bus?) that will have different voltages at different points due to simple current flow times series resistance.

To maintain signal robustness or integrity, the signal must always be looked at differentially, or relative to it's local ground  wherever it is..

Professional signal wiring uses three conductors, two for audio and a third for ground so current flowing in the ground doesn't corrupt the signal.

Inside a modest sized chassis, ground management should be a combination of brute force to keep ground potential variations small, and proper differential treatment of signals so when sent from one internal location to another, the two signals are referenced to their individual local grounds.

In theory with proper differential treatment all ground potentials are cancelled. In real life we never get perfect rejection so a mix of both keeping potentials low and differential treatment delivers best results.

JR

 

[Svart]

I've always found that a proper ground planes and layout can trump any attempt at star grounding.  Sometimes simply taking a poorly grounded unit(thin ground traces, etc) and grounding all of it's remote grounding areas to the chassis(forming a large ground plane out of the chassis) can fix all kinds of problems. 

What you are trying to avoid is large loops of ground which carry current.  That's what star grounding is attempting to do but generally when you star ground something, you are changing ground between grounding areas that are usually fixed in place, like units in racks.  When you work inside of a unit, star grounding *can* help you if you can't change the way that the PCBs are grounded.  If you can, then I suggest grounding directly to the chassis.  A wide/thick, low impedance ground is better than having a bunch of grounds return to one point.

 

[JohnRoberts]

I've always found that a proper ground planes and layout can trump any attempt at star grounding.  Sometimes simply taking a poorly grounded unit(thin ground traces, etc) and grounding all of it's remote grounding areas to the chassis(forming a large ground plane out of the chassis) can fix all kinds of problems. 

What you are trying to avoid is large loops of ground which carry current.  That's what star grounding is attempting to do but generally when you star ground something, you are changing ground between grounding areas that are usually fixed in place, like units in racks.  When you work inside of a unit, star grounding *can* help you if you can't change the way that the PCBs are grounded.  If you can, then I suggest grounding directly to the chassis.  A wide/thick, low impedance ground is better than having a bunch of grounds return to one point.

It is pretty common practice among DIY to just try stuff to see what works and I doubt I will change that.

It is pointless to argue about hypotheticals. It may be worth inspecting the two approaches. Star ground (use of multiple ground paths returned to a single point). Presumably reduce ground errors by segregating the ground currents so they don't cross contaminate each other.  Brute force, just attempts to make the ground resistance so low that voltage drops will likewise be small.

Contrary to how it may appear using a chassis as ground has it's own problems. I have seen chassis used a ground bus in some power amps to deal with the healthy ground current flows involved, but not only are there still significant current related voltage drops, but transformer magnetic fields induce voltages across the (steel) chassis. 

One final thought, ground loops will always be with us, but they are only problematic, when the signal is referenced to two different points on that ground circuit without correction. That ground circuit, loop and all, can serve as two different local ground references with proper differential treatment between them.

JR 

 

[Svart]

True.

If I find a unit that I've built to be hum rich, I'll take a thick wire from the earth ground input and then place it around the PCB's grounds until I find a spot that reduces the hum.  I then investigate the cause of the hum, usually inadequate grounding in that area.  If I can't fix it, I'll either bolster that trace or connect that area to the chassis with a short, thick wire.  It's worked in all cases that I've had hum problems.

In the RF/highspeed world, we don't simply star ground.  Thick, unbroken ground planes are the norm.

 

[Winston OBoogie]

In the RF/highspeed world, we don't simply star ground.  Thick, unbroken ground planes are the norm.

As I wrote in another thread, this may be fine for RF but I don't believe a planar approach (either on a PCB or using the chassis) is the best for audio.

Draw the circuit out with all the current loops and equiv. R in place and lay the circuit out accordingly.

 

[lux01]

O.K., thank you all and, back to the original question...  perhaps I need to clarify a bit. ;D I intend to ground each stage locally at the filter for that stage and the run a separate return wire back to ONE  central ground point in the power supply area which is the only place that will be tied to the chassis.( except for  pin 1 of the xlr's  at in and out).
My question was concerning the layout of the individual leads that return each stage's local ground node ( negative side of filter for that stage) back to the central point.
As they are not negative and positive sides of an identical signal - what would the consequences of twisting them together and running them down the center of the chassis be - for instance? (Very neat aesthetically).Or should they be run separately and an attempt be made to keep them away from each other?
But Wait... There's  More....  Ideally I'd rather use an isolated  (from chassis except at input) buss wire . This method would seem much cleaner and simpler but I'm not sure what to
do at the pwr spply end. Would I just tack the main filter caps on the end of that and not the chassis... or tie them to the chassis? And if I did the latter don't I  have 1 Giant Ground Loop?  
Thanks,
Ian

 

[bockaudio]

In the RF/highspeed world, we don't simply star ground.  Thick, unbroken ground planes are the norm.

As I wrote in another thread, this may be fine for RF but I don't believe a planar approach (either on a PCB or using the chassis) is the best for audio.

true.



{EDIT: tangled Quote corrected -PRR}

 

[Winston OBoogie]

Hey D,

Your reply got mingled with my quote, but I believe your reply was:

.true.

In which case, thank you 

I'll add that I respect your opinions regardless...

 

[Winston OBoogie]

...running them down the center of the chassis be - for instance? (Very neat aesthetically)

I see some designs that look like they are swayed by what looks nice on the schematic jpg: "I can fold my complimentary diff. amp schematic in half and it lines up exactly...".  Same thing with PCB layouts, real pretty so you can frame it and hang it on the wall.

If your layout ends up looking a mess but is an optimized one and works well...so be it.  Go that route.



Edit: missing word.

 

[PRR]

In most low-level vacuum tube circuits: your currents are low, your impedances are high, your runs are short. The audio impedance of any hookup wire fat enough to handle is very-very-very low. You can take great liberties.

As you know from gitar amps, many different schemes work. Even when there are real theoretical objections. You can even run rectifier pulses through 4 inches of chassis, double-ground your system at input jack and PT bolt, ground-loop the whole pot panel, and the buzz can be acceptable. Some well-regarded gitar amp layouts make me cringe, yet when you work out the consequences, the crap is negligible.

A prime "problem" in tube work is that power return and signal reference are the same place. (In op-amps, they may be separate.) So it is a good thing that any practical wire is far lower impedance than the tube.

Unlike a gitar amp, a mike amp usually has a transformer input. We may treat the external line as "hostile", take XLR ground and transformer case to our overall garbage shield (case). Begin our work at the transformer secondary, which often may be returned to our convenience.

Just run a bus. Start at input transformer secondary. Then take your first tube cathode. Then your first-stage power filter cap return. Then pot returns, cap returns, and more cathodes in the order of signal flow, to your output stage, and then to the power supply negative.

If your output is via a transformer, and there are no intermediate in/out taps, then there is no strong reason to have your amplifier ground bus tied to chassis. Many rack-amps have been built with all internal circuitry floating from chassis. A link allows the guts to be tied to chassis for simple systems, or floated and tied to an isolated studio ground bus for more complex systems.

However, unless you already know you need floating guts, I would tie the input end of the bus to chassis.

Unless you use a main filter cap which solders to chassis. Then your buss must go to chassis at the power supply end, and the input end left off-chassis.

If your output is not transformer isolated, then you will probably want to tie your 1/4" output jack to chassis for safety, and let the input end of the bus float.

Star systems have uses. But in a string of tube stages, it is not important that they all reference the same point. It is important that each stage references the stages before and after it. A bus is natural. A star just adds 0.01 ohm under each stage, and risks injecting large output signals directly to sensitive input areas.

Svart's RF work is different. Audio is mostly under 100KHz and often over 1K impedance. In this world, any happy hookup wire has "zero" impedance, near-enough to swallow most sins. Svart's work is above 1MHz, and often below 100 ohms. In this range, current does not flow "in" a wire, it flows only "on" the surface. The inside of a round wire is a total waste. What matters is skin area. Techniques include silver-plating (small but sometimes useful improvement on 10MC tank coils), tubing, flat bars (very common in power distribution, both for skin effect and for heat dissipation), and PCB layers supplemented by case-metal. These are essential for low impedances and high frequencies, and may be applicable to transistor audio, but are not needed and usually "wrong thinking" in tube audio.

And do NOT over-think in advance. Nor lock your grounding in stone. Build the stages. Tack the grounding together. Try it. Be prepared to experiment. However if you have a holistic plan, then any trouble will usually be one misunderstood connection.

 

[Winston OBoogie]

...essential for low impedances and high frequencies, and may be applicable to transistor audio, but are not needed and usually "wrong thinking" in tube audio.

Agree.  Except, for the squalid-state stuff I sometimes play with, it is probably wrong thinking for that too.

Others may have a different experience?

 

[JohnRoberts]

I may be offering a contrary viewpoint here, but I am in favor of over thinking, or at least rigorously considering the layout but also looking at grounds as integral to the circuit design. While we all react to ground issues we discover at the prototype level, in my experience, after you see the ground issue you will (hopefully) also see what you did wrong that caused that problem, and not repeat old mistakes.

I will make a sweeping statement that it is impossible to make a ground layout where ground is perfectly 0V everywhere.. so I don't try. A decent layout with proper application of differentials will maintain signal integrity.  I think of grounds like a sewer pipe who's job is to carry away unwanted currents and not leave too much smell behind. There will always be a little stink, so work to keep what little stink is left out of the audio.

JR

PS: Guitar  amp layouts are an odd duck.. in many cases crosstalk and ground corruption between unbalanced high impedance paths is what gives a certain amp it's unique character.

 

[Winston OBoogie]

... A decent layout with proper application of differentials will maintain signal integrity.  I think of grounds like a sewer pipe who's job is to carry away unwanted currents and not leave too much smell behind. There will always be a little stink, so work to keep what little stink is left out of the audio.

John,
I, for one, enjoyed your analogy  ;D
May I ask if you are in favour of a planar approach for audio in PCB layout?
To continue with your theme, an albeit diluted reservoir at the treatment plant...

 

[Svart]

You guys have to remember that "RF" is never just RF.  It's really mixed signal.  The only way to get all of the switchers, switches, LDOs, LNAs, opamps, FPGAs, DDSs, PLLs, and whatever else is crammed into a small board is to have low impedance grounding through ground planes.  Digital ground currents can noisy and significant in current.  So can switching supplies and power amps.

I don't think that it's really needed in audio work but I do find that some of the things I've done in RF have fixed issues in the audio world where nothing else seemed to help.

 

[bockaudio]

I was agreeing with John about the planar ground. Blindly hit the "quote" key and......
Bear in mind I wasn't talking about anything digital audio, not my field.

 

[ioaudio]

nice to see so many experts around!