Grounding

[Paul P]

I split this off into another thread so as not to clutter up The Captain's Princeton debugging thread.

http://music-electronics-forum.com/showthread.php?t=4174

dai.h, you were saying :

I think you misunderstood. The circuit ground should be connected to chassis prob'ly at the preamp end. The reason for not putting that PART (i.e. that doesn't mean not putting the cct.grnd down on the chassis at all) is to keep the noisiest portion of the ground (pulsing AC ripple--100 120Hz--still there before it has gone through some of the RC or choke C filtering stages) off the chassis (so it doesn't go through the chassis) and keep that noisy loop closed (wired so it least effects the ground line and thus the other returns esp. the sensitive ones detrimentally and physically so the physical loop area is small). You seem to have a strong interest in this so you might want to check out this article and the links listed:

http://ampgarage.com/forum/viewtopic.php?t=220

in addition to the links listed, the Rane site has some good stuff as well as ax84 (try searching for Randall Aiken and Paul Ruby posts--they are not the only people with good info of course, but these should lead to some good discussions re: grounding). Also epanorama IIRC.

Thanks dai.h for these articles. They plugged some holes in my view of things.
A real eye-opener was Tetsu Kimura explaining that a decoupling capacitor
is in parallel with the tube stage that it's decoupling. He also provides some
interesting ideas on the placement of decoupling capacitors, though they
don't seem easy to do with conventional amp layouts.

There is still one thing I'm not clear on and that is where the chassis ground
should be.

Randall Aiken says :

"One of the best amplifier power supply grounding schemes is a "star"
ground system, where all the local grounds for each stage are connected
together, and a wire is run from that point to a single ground point on the
chassis, back at the power supply ground. Even better is a two-point star,
where the power supply grounds (PT center tap, first filter cap ground) and
output stage grounds (output tube cathodes for fixed bias, or cathode
resistors for cathode biased, and output transformer secondary ground) are
connected together and to the chassis at a single point, right at the
ground of the first filter capacitor. The ground of the second filter
capacitor, after the choke or filter resistor, is the star ground point for the
preamp stage grounds. Use a local common point for each preamp stage
ground, and run a wire from this common point back to the second star
point."

and

"The power transformer center-tap wire should be soldered directly to the
ground lug of this cap [first filter cap], and a very short, heavy wire should
run from there to the chassis ground connection (if there is one). Do not
connect the power transformer center-tap to the chassis and the first cap
ground to the chassis at a different spot; this will cause heavy ground
current flow in the chassis. Also, do not tie the AC mains safety ground to
this point; it should be connected to the chassis with a very short length of
wire right at the chassis entry point."

What does he mean by "if there is one" ?

It looks like Randall Aiken wants the single chassis connection at the power
transformer.

Paul Ruby says :

". One and only one connection from circuit ground to the chassis."
. No Ground Loops. Only one path should exist for any ground current.
. Tight, isolated loop for PT, recto and 1st filter cap.
. Ground circuits to their respective B+ filter cap negative terminal.
. Ground your OT secondary and the OT itself."

and :

"The best physical place to make the connection from the signal ground
to the chassis is debatable. But, to avoid noise getting into your signal
path, the best place is at the input jack."

Do we really want the input ground and the PT ground on the same point ?
Since the PT transformer is as far from the input as physically possible
should we run a wire from the "Tight, isolated loop for PT, recto and 1st
filter cap" all the way to the input jack ? Wouldn't this create an enormous
loop ? If we don't run a wire then we end up with the complete opposite
of what Randall Aiken proposes with the 'river' flowing from the PT to the
input jack, to the chassis, to earth ground.

Paul P

[MWJB]

There's undoubtedly more than one grounding scheme that can be made to work, personally I'd use a combination of the 2 approaches you mention:

AC cord is grounded to a PT bolt close to entry point to chassis.

PT centre tap, main & screen supply filter caps, bias supply, power tube cathodes are all grounded to the same PT bolt (not the AC cord PT bolt).

PI & preamp filters, preamp cathode grounds all have sperate wires (a cathode resistor and its bypass cap can share a ground wire) running to a star ground at the input jack. Any pot mounted components can be grounded to the back of the pot they connect to, a buss wire from the back of the pots also terminates at input jack ground.

Do not link all the grounds together on the circuit board and then run a single wire to the input jack star ground.

[dai h.]

What does he mean by "if there is one" ?

if there is a center tap on the B+ winding

1st
filter cap" all the way to the input jack ? Wouldn't this create an enormous
loop ?

by 1st filter cap he means the one directly at the rectifier output. I think you are confusing that one with the one at the extreme preamp end.

[Paul P]

I just read a Rane article that links Randall Aiken and Paul Ruby together at
opposite ends of the discussion.

From http://www.rane.com/note151.html :

There are two schools of thought on where to connect the signal ground to
the chassis. They are both versions of the star ground scheme mentioned
above. The first connects a trace (or wire) directly from the audio power
supply ground terminal and connects to the chassis ground point.
...
Another school of thought on where to connect signal ground to the chassis
simply moves the center of the star ground to the input jack's ground. This
scheme makes the most sense for unbalanced units and balanced units
equipped with 1/4" connectors where use of mono plugs is possible.

I think I'll pull an Enzo on this one and wire things so that the signal ground
(suitably star-grounded in each stage) can be connected to the chassis at
the PT, or the input jacks, or both.

Paul P

[dai h.]

yeah, that's a good article. I think you can do either. For example, on the Soldano SLO100, the connection to chassis is towards the power input end and not the signal input end. And the footswitch supply was left floating for some reason I don't quite remember.

[raiken]

I just read a Rane article that links Randall Aiken and Paul Ruby together at
opposite ends of the discussion.

That's the problem with publishing internet papers - people think you know what you are talking about.

My article is misleading, so I'll correct it when I get a chance. Here's what it should say:

(1) You can connect your grounds to the chassis (and you should), but only at one end, either the power supply end (main star) or right at the input jack.

(2) That article assumes you are using isolated jacks and are connecting the ground at the power supply. If you do this, you *must* have a low impedance path for AC signals (mainly high frequency AC signals) at the ground side of the input jack, otherwise your amp will be susceptible to RF interference. The way to do this is to put a capacitor (typically a 0.01uF) directly from the shield side of the input jack (which must be isolated) to the chassis with as short leads as possible.

(3) If you are using non-isolated input jacks, you can instead directly connect their shield to chassis directly with a short wire (don't depend on the nut to make contact, because they corrode over time). If you do this, you cannot also ground the main power supply star node to the chassis, or you will likely develop massive hum.

I tend to favor a combination of star and buss ground. I always star the main power supply - you have to run the center-tap of the power transformer B+ winding (or the bottom of the B+winding if you are using a bridge rectifier) directly to the first filter cap without going through the chassis or any other part of the ground buss, or you will get buzz noises from the high-current in the return path. Also run your output transformer common wire directly to the output jack shield connection, to keep high currents from flowing in the chassis.

If using a choke, I run the second filter cap close to the first one, connecting it's ground to the star point. This node can have some large ground currents as well (but not near as high-energy as the first filter cap's ground return), so we want to keep it from causing mischief in the circuit. Note that choke wires themselves can radiate a lot of noise, so keep them away from sensitive areas of the preamp.

The preamp grounds I will either fully buss, or I'll have little "islands" of ground for each stage, and then at time of layout, I'll decide how the islands connect back to the main star point. Sometimes I'll use an entire ground plane on top, or I'll cut the plane into several copper pours for individual "star-type" returns. You have to apply a knowledge of circuits and signal flow to figure out what is critical and do your layout accordingly, so there is no one way that will always be correct, because the circuit paths may be intertwined. Proper circuit layout can minimize the risk and make the grounding easier.

I also tend to put the preamp filter caps in the circuit where they are used ("local" bypassing). For example, the first preamp tube's filter cap will be physically placed in the area of the first preamp tube, with the ground and B+ connections made right to the bottom of the cathode resistor/cap and to the top of the preamp tube resistor. If I decide to instead group all the filter caps together, I'll be sure to also locally decouple each filter cap node with a smaller capacitor, typically a 0.1uF/400V) directly from the top B+ node of that stage to the ground node at the bottom of the cathode resistor. You'd be surprised how many brand new filter caps have very high reactances at frequencies within the range of a distorted guitar. Sometimes you can take a "bad" filter cap and bypass it with a 0.1uF cap and it will sound fine. It never hurts to have good high-frequency decoupling at all nodes.

I hope this clarifies things a bit.

Randall Aiken

[Paul P]

Thanks Randall for the clarifications. I didn't realize you were a member here.
I must say, there is a lot more to grounding an amp than you can deduce from
looking at a schematic !

I think I have a pretty good understanding now of the issues, I just have to
incorporate them into my design. Getting a big filter cap close to the circuit
it decouples is not an easy task (unless you have the room to fit it on the
circuit board). It's starting to look to me that the best place would be to
stick it to the underside of the circuit board. I already have grounds on one
edge and power on the other so the filter caps could bridge the two pretty
easily.

Paul P

[raiken]

Getting a big filter cap close to the circuit
it decouples is not an easy task (unless you have the room to fit it on the
circuit board).

That's why local decoupling with 0.1uF caps for higher frequencies is nice. It allows you to put the lower-frequency, large, bulk filter caps further away, because those frequencies aren't as critical of layout and long traces as are the high frequencies, which can be locally decoupled with smaller capacitors.

Randall Aiken

[dai h.]

Randall, could I trouble you to explain a bit about the correct place to put the filament center tap? This wasn't covered in your article (which I read many times when searching out literature on the subject) so I tried to determine the correct spot empirically first putting it at the preamp end (got noisier), then towards the power end (quieter) where the higher current points incl. the power tube cathodes. (I also know that on the Soldano SLO100 the false filament center tap connects to the cathodes.) I don't quite understand why that was the better place to put it though. Is it something to do with heater cathode leakage or??

[raiken]

Randall, could I trouble you to explain a bit about the correct place to put the filament center tap? This wasn't covered in your article (which I read many times when searching out literature on the subject) so I tried to determine the correct spot empirically first putting it at the preamp end (got noisier), then towards the power end (quieter) where the higher current points incl. the power tube cathodes. (I also know that on the Soldano SLO100 the false filament center tap connects to the cathodes.) I don't quite understand why that was the better place to put it though. Is it something to do with heater cathode leakage or??

You should make a voltage divider with two resistors to elevate the filaments above ground. This will make them almost as quiet as DC filaments, as far as filament-induced hum is concerned (if your hum is coming from other sources, such as wire routing or ground loops, it may not help at all, or may even make it worse!).

If you have a relatively high plate voltage, use something like a 470K resistor in series with a 22K resistor from B+ to ground, and add a 10uF/100V cap across the 22K resistor (very important - if you don't add this your filaments will be very noisy!). Connect your filament center tap (or a "virtual" tap using two 100 ohm resistors) to the junction of the 470K/22K.

This will elevate the filaments 18-20V or so above ground.

If you have a cathode-biased output stage, you can connect the center-tap to the output tube(s) cathode resistor. They will already be 10-30V above ground, so you will get a "free" bypassed DC voltage to connect to.

Randall Aiken

[dai h.]

Thanks. Sorry if I'm not articulating my question clearly enough, but I'm not really needing anything quieter but want to learn why the correct placement of the filament center tap seems to be where the power tube cathodes are (the reasoning, the rationale). For example, the B+ center tap going to the primary filter cap negative makes sense because of the high amount of ripple there making it desirable to keep that circuit in a closed loop isolated yet still connected (avoiding a common resistance with a sensitive ground point, say) to the ground line, or the spk. output ground being grouped with the power tube cathodes makes sense because the spk. out return is high current, so group high current with high current. Or grouping preamp grounds in a star makes sense since the currents are low or very very low, plus a star configuration avoids common impedances/resistances. But with the filament center tap, I don't quite understand why the power tube cathodes seem to be the correct (quieter) place to connect it to the ground line, since apparently the current flow is low (and tying it to the preamp end resulted in more buzz when I experimented). Maybe something to do with heater to cathode leakage? (Leakage current is from the heater so needs to go back to the source to the heater through the cathode?)

[raiken]

But with the filament center tap, I don't quite understand why the power tube cathodes seem to be the correct (quieter) place to connect it to the ground line, since apparently the current flow is low (and tying it to the preamp end resulted in more buzz when I experimented). Maybe something to do with heater to cathode leakage? (Leakage current is from the heater so needs to go back to the source to the heater through the cathode?)

Oh, sorry for not answering your real question!

Yes, if you don't have elevated heaters you want to connect the HT CT to the main PT ground, or better yet, the second filter cap ground if it is different. This generally gives a quieter result.

Theoretically, there is no actual "real" current flow in the CT wire because it is isolated from the other windings, so it shouldn't matter where you ground it, because you are just establishing a "reference" point for the filament winding relative to the preamp tube cathodes. If you ground it at the preamp side, there should be the same amount of heater-cathode leakage inside the preamp tube as there would be if you ground it at the power tube end, because the ground is at very nearly the same potential in either case, although there is a small difference due to preamp current flow in a non-ideal, non-zero ground buss impedance.

However, there is also some capacitive and inductive coupling of noise (particularly the high-current cap charging spikes that occur at a 120Hz rate) that can flow between the windings of the PT. You don't want these currents to flow in your sensitive preamp ground buss, so it is quieter if you ground the CT at the power supply end.

There is an easy test to determine which of the two is causing your problem: Is the noise 60Hz hum or 120Hz buzz? If it is 60Hz hum, it would be more likely heater-cathode leakage. If it is 120Hz buzz (assuming you have a full-wave rectifier), it is PT HT coupling noise.

If you get a chance, make the test and determine the frequency of the part of the noise that improves when you ground it in different places to determine the real cause. You may have both 60Hz and 120Hz noise components, but one of them should improve when you ground your filament CT at the power side.

Randall Aiken

[dai h.]

Oh, sorry for not answering your real question!

thanks! it's no problem!

However, there is also some capacitive and inductive coupling of noise (particularly the high-current cap charging spikes that occur at a 120Hz rate) that can flow between the windings of the PT. You don't want these currents to flow in your sensitive preamp ground buss, so it is quieter if you ground the CT at the power supply end.

so if I understand correctly the buzz (when I grounded the filament center tap at the preamp end) was coming from the B+ winding inducing and capacitively coupling 120Hz(full-wave rect.) harmonics into the filament winding inside the PT. Would that mean if you used a separate filament transformer (with a filament cnt. tap) you could place the filament center tap at the preamp end? (Trying to think about this a bit more--so I guess the same logic for the ground side on a bias winding tap in the 100W Marshall--i.e. should be grounded at the power supply end since that winding will have on it induced noise from the B+ winding from proximity and maybe the higher the frequency and bigger the input cap the worse the noise?)

(BTW I don't really have a noise problem with the amp. It's mainly a theory question since I didn't have much of an idea what was going on specifically with the fil. cnt. tap.)