Clarification request: Aiken's "grounding" tutoria

[deiseldave]

Could someone clarify this concept that I read in Randall Aiken's "grounding" tutorial? What it says is:
"While the star ground is excellent for eliminating ground loop hum, it is not always the best scheme for preventing radio-frequency interference (RFI). Fortunately, there is a simple addition to the star ground scheme that will make for a very quiet amplifier with no RFI. Simply add a 0.01uF capacitor from the chassis to the ground lug of the isolated input jack using very short leads. "
My question is: Would this be in place of the wire that connects input jack ground to its star ground point? Or, in addition to it?
Probably a stupid question. Just want to make sure I'm understanding.
Thanks for any help.

[Firestorm]

In addition. That said, star grounds are overrated. You can build an amp where the chassis connects to nothing except the safety ground and all circuit grounds are done with wire. This can be a lot of wire, so the chances of ground loop hum increase. We are surrounded by 60Hz hum. There's one room in my house where anything plugged in picks up a stray oscillation from nearby power lines.

[deiseldave]

Thank you.

[caps8419]

I found this helpful...

http://el34world.com/charts/grounds.htm

Basically you don't want to connect all the ground points on the same rail as it will increase noise. You take individual sections of the amp and ground the to 1 main point on the chassis but you do so independently not just in one big chain. So the preamp/ preamp filters are one wire, the power amp filter caps are another wire, output jack another, center tap from PT another etc...

-M

[brewdude]

I think the OP's question is more about the .01uF cap atatched to the input jack.

I don't really know the answer, but I suspect it would be in addition to the connection to the star ground. (?)

I think Merlin suggests using a similar cap, but he has it going from the input grid (after the grid stopper) to ground.

[R.G.]

And it's time for some electromagnetic field theory.

OK, OK, not very much, and no math.

Electricity is lazy - it goes where it's easiest to go. At low frequencies, "easiest" means "lowest resistance" and so it follows the path with least resistance, and divides itself among resistive paths by how resistive they are.

This is why star grounding works for DC, power line frequencies and audio, these all being "low frequencies" in the fields-sense. If the choice electricity is presented is going down its own wire or running through free space, it will choose the wire. Giving each return its own wire means it doesn't get to pollute other users' wires.

When frequencies get above that, into RF, the fields thing comes into play and things get weird. At higher frequencies, free space is no longer as offensive to the electricity as it was at DC, and conductors experience their own inductance/capacitance effects. Inductance in particular becomes the enemy. Frequency related effects will make the current follow paths that are the lowest inductance, and on signal-trace/ground plane setups, the ground-side current will run right under the wire, held there by field effects.

This is why an electrical setup ought to be a metallic shell enclosing all the wiring. At low frequencies - power line! - the grounded shell can eat up all the current from a fault and conduct it to the safety ground wire without frying the operator. At RF the metallic shell is a "Faraday shield", and all the RF stays on the outside shell, not going into the circuits inside unless you let it.

And now we're coming to the RF capacitor thing. You want the RF to stay on the shell. It will go into the circuits if there is a low-inductance path, like a wire penetrating a hole into the insides (i.e. a jack). If you put a low-value cap from the signal wire to the outside shell, the RF would much rather stay on the outer shell, and all the RF is siphoned off to the shell, where its own field effects make it want to be anyway.

That brings up the issue of where to attach the bushing of the input jack.

You have two good options and many variations with shades of worse.

The good options are (1) Ground the shell to the input jack ONLY and (2) ground the shell at the power supply reference ground and NOT at the input jack.

The worse options are all variations of multiple connections to signal ground, speaker ground, sewer ground and safety ground. There are RF and power line currents running around on the outer shell, and how big the voltage differences are at different places on the shell is hugely complicated. Sometimes it is so small that you just get away with it. Sometimes it bites you, well, you know.

One way to really wipe off the RF from an input jack is to take a small ceramic cap, 10pF to 100pF from signal to bushing at a grounded input jack, and then put a ferrite bead on the wire to the circuits, introducing some of that deadly inductance. The RF really, really wants to stay out of that. A small resistor, 100ohms or so, can help too.

It's far, far more effective to keep the RF outside the circuits and on the Faraday shield than to try to easter-egg solutions down in the circuits themselves.

[deiseldave]

I found this helpful...

http://el34world.com/charts/grounds.htm

Basically you don't want to connect all the ground points on the same rail as it will increase noise. You take individual sections of the amp and ground the to 1 main point on the chassis but you do so independently not just in one big chain. So the preamp/ preamp filters are one wire, the power amp filter caps are another wire, output jack another, center tap from PT another etc...

-M

Thanks caps8419. That was interesting info.

I think the OP's question is more about the .01uF cap attached to the input jack.

I don't really know the answer, but I suspect it would be in addition to the connection to the star ground. (?)

I think Merlin suggests using a similar cap, but he has it going from the input grid (after the grid stopper) to ground.

Thanks brewdude. I believe Merlin’s recommendation is adding a pico cap from grid to ground. I think Randall Aikens suggestion is to add the .01uf from input ground to chassis, not from signal to ground.

If you put a low-value cap from the signal wire to the outside shell, the RF would much rather stay on the outer shell, and all the RF is siphoned off to the shell, where its own field effects make it want to be anyway.

Thanks for the response R.G. I think Randall Aikens suggestion is to add the .01uf from input ground to chassis, not from signal to ground. However, I am aware of adding a pico cap to ground close the the grid stopper, as Merlin has suggested. Is this the same as what you are suggesting? Also, how does Aiken’s .01uf from input ground to chassis work?

One way to really wipe off the RF from an input jack is to take a small ceramic cap, 10pF to 100pF from signal to bushing at a grounded input jack, and then put a ferrite bead on the wire to the circuits, introducing some of that deadly inductance. The RF really, really wants to stay out of that. A small resistor, 100ohms or so, can help too.

I have put a ferrite bead at the signal side if the input jack. But, had thought the pico cap was supposed to be past it, towards the grid. I read that here: http://ampgarage.com/forum/viewtopic.ph ... ht=ferrite Did I misinterpret what I read?

It's far, far more effective to keep the RF outside the circuits and on the Faraday shield than to try to easter-egg solutions down in the circuits themselves.

The chassis will be mounted in a combo cab, do you think it would make a big difference to shield the underside of the top of the cabinet, and ground that to chassis?

Thanks again for the help everyone. Dave

[Phil_S]

...sewer ground...

I enjoyed reading that one. Thanks.

Though I don't play on the same field as the giants who have already posted, I've built a few amps and tinkered with ground schemes a bit. Someone once gave me some very practical advice, which seems to serve me well. I use a buss ground to cover the input jack all the way to the PI.

Think of the buss as a linear star. There are several ground stages, as Aiken discusses. Gather the elements of each stage and group the ground wires together on the buss -- basically a modified star point.

Not meaning to start a ground war here, but I have never been able to wrap my head around making the chassis ground for the buss near the input jack. Many swear by it, and I'm not suggesting they are wrong. After all, who am I to argue with success?

As I understand it, each ground stage essentially ends with a cathode. So, you work across the schematic grouping up the ground wires on this basis. The 1st tube has the lowest ground potential and the PI has the highest. (Honestly, I don't know if that's right...) I ground each B+ cap with it's relevant stage. This means I collapse the number of stages, typically for both cathodes of a 12AX7. The PI ground is last. From the PI end of the buss, I run a wire to the single circuit ground on the chassis, which also has all the high potential grounds, including the OT, 1st B+ cap, screen cap, PT CT, power tube cathodes, probably other stuff that isn't coming to mind. I've always thought of this as a modified star topology and it seems to work. Like many people, I tend to stick with what works for me.

[JazzGuitarGimp]

Hi Dave,

I belive also, Aiken is suggesting the cap be connected from input jack ground terminal to chassis. At first glance, this doesn't make sense, since, from a DC perspective, you have ground at both sides of the cap (yes, I kind of just answered your question: the cap is not in lieu of, but in addition to wiring the input jack ground terminal to circuit ground). However, borrowing from R.G.'s post, RF will take the path of least impedance, which means through the cap, and directly to chassis. It's important that the cap at the input tie directly to chassis (short as possible), and that no other circuit ground wire be connected at the same point as the cap. At least, that's my interpretation...

[Ken Moon]

Many modern amps (JCM900) and some Fender RI's (Bassman, Reverb Unit, maybe more) use this type of earth lift circuit, connected right between the star ground and the chassis.

On the Fender reissues, the star ground is located right at the ground lug of the input jack. I'm not sure about the physical location on other amps.

Merlin discusses the earth lift concept in his grounding pdf (linked to on his website).

I think the cap does the same thing for RF that Mr. Aiken was mentioning, and the resistance dissuades any ground-loop currents from entering the circuit via the input jack.

[img:921:456]http://i.imgur.com/aPHFv.png[/img]

[R.G.]

Aiken's suggestion of a cap from the input jack bushing to the chassis only makes sense if the input jack bushing is isolated from the chassis. In that case, yes, it's a great way to shunt the RF to the Faraday cage/chassis.

And in that case, I would still use a few pf to the chassis from the signal line, as well as some series impedance like a small resistor or ferrite bead.

We had to actually derive Maxwell's equations from first principals in a midterm test in my fields class. I still get the shakes. But one of the results of Maxwell's field equations is that for high enough frequencies, the fields inside any volume completely enclosed by a conductive exterior will do a combination of cancelling inside and forcing themselves out to the conductive exterior. For an ideal conductor enclosing a volume, there are *no* EM fields inside.

Of course, there are no ideal conductors so far that we can use. So fields do get inside because of non-ideal metal conductors, and especially due to holes, slots and seams. An EM field cannot "see" a hole that is smaller than 1/4 of the field's wavelength. The wavelength of an RF wave is the speed of light divided by the frequency. So 100MHz radio has a wavelength of L = 3*10^8 m/s divided by 100*10^6 = 3 meters, about 10 ft. FM radio then can't leak into a hole with a largest dimension smaller than 29.5 inches. At least it can't leak in very well.

Aircraft radio at 270MHz and such can get in holes 12-15inches long. Notice that an imperfectly closed seam, or cover sheet constitutes a hole, and in fact a slot of 1/4 wave can act like an *antenna* and actually suck RF in.

You're generally OK with holes of under an inch as long as you don't go sticking wires through the holes to carry RF in. Ooops, that's what we do with every jack. We can make it hard for the RF to come in on the signal line by putting a few pf to the metallic shell and some impedance in the wire. If we have been clever and isolated the jack from the enclosing metallic shell, we need to RE-ground the bushing by putting a good low impedance cap from the ground/bushing of the jack to chassis. These two things shunt the RF out of the signal wire and the shield/ground to the shell. If we don't do this, the wire carries the RF through the hole it could not otherwise get through and down to the circuit.

Once the RF is inside the circuit, we have some options, although they're not as good as keeping it out entirely. Tubes do not have infinite frequency response. In fact, a 12AX7 has a hard time with AM radio. But if the RF is already inside the enclosure, you can do two things: (1) shunt it to ground at the input of the tube with a small cap from grid to ground or grid to emitter, or (2) make the tube even more immune to RF than it already is by putting a cap from plate to grid, or plate to ground.

[catalin gramada]

I tried to imagine using this cap preventive to suprime posible oscilation in first stage which is most sensitive. the signal grounds wire I consider more important than supply ground return and in normal way the decoupling cap(usually not too good in HF range) should be as close as possible over the stage what never happen in real life. all long path signal ground return are prone to oscilations special in HF. there should a way to drain hf components to a low impedance path(like chassis is) through this small cap keep[n away from supply rail

[deiseldave]

Thanks for the input, everyone. I will have to study some of the responses to understand better....
Nevertheless, I implemented the "Aiken" input ground cap, even though I don't understand how a cap from input ground lug to chassis could be lower impedance than the straight wire right next to it.
I also lowered grid resistor to 10K and ran 470pf cap from grid to ground.
Noticeably less hiss.
I wish I would remember to only make one change at a time, so I can see what works best.
Anyway, since I already had the D'lite grounding topology in place, I decided to stick with that. However, I will reflect on some of the great ideas shared here on my next build, when putting together my grounding scheme.
Thanks again for all the input.
Dave

[catalin gramada]

even though I don't understand how a cap from input ground lug to chassis could be lower impedance than the straight wire right next to it.
Dave

The chassis is "0" potential for the inside circuit as time main star ground is tied to it .the buss wire have some "resistance" for certain frequencies due to its residual resistance and inductance

[jaysg]

We had to actually derive Maxwell's equations from first principals in a midterm test in my fields class. I still get the shakes. But one of the results of Maxwell's field equations is that for high enough frequencies, the fields inside any volume completely enclosed by a conductive exterior will do a combination of cancelling inside and forcing themselves out to the conductive exterior. For an ideal conductor enclosing a volume, there are *no* EM fields inside.

I too did not enjoy my electro-masochism class. What I see is that no major amplifier manufacturers put any effort into these concerns. I used to work in medical devices, and yes, we were putting beads and other devices on inputs and supplies.

In my experience, a series chain of grounds works out fine. Either a preamp buss on the pot backs or a flying one, and a line from the PI ground back to the AC and output ground near the PT.