Question about Merlin's grounding

[dwinstonwood]

I've been re-reading Merlin's chapter on grounding: http://www.valvewizard.co.uk/Grounding.pdf
And, I've gone back and re-read Willabe's advice on one of my previous builds here on this forum.

What's still not clear is how the reservoir cap is connected to chassis ground. I understand that the PT HT CT (what Merlin calls a rectifier lead) is connected directly to the reservoir cap's negative terminal, along with the speaker ground (and, the bias circuit, if there is one) creating a power section ground star.

My question is, does the reservoir cap's negative terminal (power section ground star) then have a wire running to its own chassis ground lug, or does this ground star daisy chain to the other stars and ultimately connect only to the input jack ground? Merlin seems to state that there should only be one ground point: at the input jack. But, my mind tells me that the power section ground star should have its own separate ground point.

Thanks for any help clearing this up!

[tubeswell]

In Merlin’s galactic ground system, there is only one single grounding point to the chassis for the ground buss wire - at the input Jack. The reservoir cap and screen supply cap grounds (along with the output tube grounds and the heater ground reference) are attached to the other (‘floating’) end of the ground buss wire.

[dude]

As Tubeswell said "floating" end of the buss grounded at input jack. IMO of the wizard's grounding, the Reservoir filter's negative side is the CT, star ground point. It makes no difference if you connect the PT ground to the negative "leg" of the reservoir filter or to a point on the "chassis near that filter" as long as both are connected to the same star ground on the chassis.


I always used two grounding points, one for the PT, Reservoir, and other HT filters, heaters, bias, speaker grounds. On the other end of the chassis I ground the floating buss to the input jack, picking up the board grounds (filter for preamp too) on the buss. For good measure I always ground the first preamp cathode to the input ground. I don't rely on the input jack for my ground alone, I put a screw and nut in the chassis near the input and ground the preamp end of the buss with input jack ground along with the cathode of 1st preamp tube, never had any noise issues, correct me if I'm wrong. 


BTW, does anyone know if a larger nut is available for a Carling on/off switch? That thin nut never seems wide enough for my faceplates...? 

[dwinstonwood]

Thanks tubeswell and dude. That clears it up.

I've usually used the two ground-point method:

PT CT, heater CT, reservoir cap and screen supply cap, speaker, power tube cathodes, and bias all to one bolt on the chassis, near the PT. And, preamp filter caps and cathodes, pots, and PI to the input jack. I can see why it's a good idea to then run a wire from the input jack ground to a nearby bolt on the chassis for good measure.

But, if I've got it right, I could also connect the power amp star to the end of the preamp buss farthest away from the input jack and eliminate its chassis ground point. That's the part that wasn't totally clear from reading Merlin.

Thanks.

[tubeswell]

This is how the two pictures in Merlin's Grounding chapter go together (fixed bias example for the power amp)

[dwinstonwood]

Thanks, that's great tubeswell. When I see the entire picture the single ground point concept falls into place. It's still a bit challenging trying to visualize the path that ground takes. Maybe I need to think of ground more like a 0V reference point and not as a closed circuit? More reading to do...

[tubeswell]

... trying to visualize the path that ground takes...

While the ground return is at 'zero' voltage potential, the ground return is still (nevertheless) a current return path. The same number of mA that starts out in the HT goes through the ground return. (Same goes for the filaments'n'all)

[pdf64]

… Maybe I need to think of ground more like a 0V reference point and not as a closed circuit?...

I suggest the opposite, ie problems can arise if we view the chassis as a big solid 0V common reference, which can be connected to willynilly and used as a conductor for 0V common without any consideration.

[dwinstonwood]

I suggest the opposite, ie problems can arise if we view the chassis as a big solid 0V common reference, which can be connected to willynilly and used as a conductor for 0V common without any consideration.

Yes, I agree with that. Choosing the optimal ground point(s) is what I'm after. The issue I'm having is trying to visualize the return path of currents through the amp. It's probably obvious to most here, but I'm missing it. 

[tubeswell]

... trying to visualize the return path of currents through the amp...

This picture is another way of looking at it. The ground returns for each part of the circuit (within each grey box) relate to the filter cap from which the current is sourced. The current coming from each respective filter cap B+ goes through the amp's various component pathways for that part of the amp, and end up tying back to that respective filter cap ground lead. The ground leads are joined together on the ground buss (so that they are all at the same potential), but the most noise-sensitive parts of the circuit (the V1 pre-amp stage where S:N ratio is highest) is at the closest end of the buss to the 'real ground'  (i.e. a fixed ground reference point), so it remains relatively unaffected by current from the high-current parts of the circuit in the ground buss.


(Edit Picture contains error in bias supply - fixed later in this thread. Thanks Franco)

[dwinstonwood]

tubeswell, that's it! That's exactly what what I was after, so thank you. I always tend to grasp things better when I can "see" them. Which is, in reality, very difficult with electricty. 

Merlin's diagram of current loads at the bottom of this page is another one that went a long way in helping me chose PT's for projects: http://www.valvewizard.co.uk/smoothing.html

[PharmRock]

This is all great information, as I have been trying to figure all this out myself, attempting to incorporate best practices into my layouts.

How does the use of multi-section can capacitors affect this grounding approach?  Would the best approach be to use one can cap for the power amp section (e.g. plates and screens, and possibly the phase inverter), and another can cap for the preamp stages?

[dwinstonwood]

...Would the best approach be to use one can cap for the power amp section (e.g. plates and screens, and possibly the phase inverter), and another can cap for the preamp stages?

I think Merlin says as much (in the link I posted above). I've started using cans only for the power tubes, and putting smaller preamp caps on my boards. But, I might switch to using two caps for the plates and the screens. It would be easy to make a small cap board and install it inside the chassis where the can usually goes in amps like the Princeton, etc. Didn't ken Fischer do that?

Thanks, and I'm glad you found my questions useful!

[Willabe]

My question is, does the reservoir cap's negative terminal (power section ground star) then have a wire running to its own chassis ground lug, or does this ground star daisy chain to the other stars and ultimately connect only to the input jack ground? Merlin seems to state that there should only be one ground point: at the input jack. But, my mind tells me that the power section ground star should have its own separate ground point.

Merlin wants to eliminate as many chassis grounds as possible, random chassis grounding. So he goes with only 1 chassis ground. You can't control/predict the ground path through the chassis. So he wires his ground path, so does Kevin O'Connor and others.

The separate wired ground stars are isolated from each other and wont modulate a different ground star. Because the different random chassis grounds don't cross each other in the chassis looking for ground.

With normal (lower gain) amps, you don't have to make a separate ground star for every preamp e-cap and it's tubes ground connections, you can run a preamp ground buss wire. Then ground the buss wire at the input jack chassis ground.     

So using a power amp chassis ground and preamp chassis ground, 2 chassis grounds, can work great too. Guys do it here all the time. Again, it works because there's no other 'random' chassis grounds in the path from the preamp chassis ground and the power amp chassis ground. 

With the reservoir cap ground, doesn't matter if your using the 1 chassis ground wiring or the 2 chassis ground wiring, DO NOT wire the screens and PI grounds directly to the reservoir cap's negative terminal and PT HT CT.

Wire the reservoir cap's negative terminal and PT HT CT together, then run a wire from there to the power amp ground star, ie, the screens e-cap negative terminal, the power tubes K, -bias, and PI e-cap (if the amp has -FB loop) PI tube ground. If the amp does not have a -FB loop, give the PI and it's e-cap it's own ground star.

That wire from the reservoir cap's negative terminal and PT HT CT isolates the charging currents from the reservoir cap and the rest of the grounds. It forms it's own loop and will not modulate the other grounds.

The reservoir cap's negative terminal and PT HT CT ground has the largest charging current in the amp, it's also the lest filtered B+ node in the amp, so it's the nosiest B+ ground node in the amp.

look at Melins drawing that tubeswell posted and tubeswells drawing.

I wrote all this out for you before with a hi-lighted drawing.   

[dwinstonwood]

DO NOT wire the screens and PI grounds directly to the reservoir cap's negative terminal and PT HT CT.

Thank you Willabe!

My issue is that I used a 32uF/32uF cap can, so there's no way to separate the reservoir cap ground from the screen cap ground - it's shared on the can. I'm thinking the idea method would have been to use two single caps instead.

Here's a PDF layout showing the closest I can get using the cap can. There's no NFB, so I gave the 12AX7 its own star, and grounded the speaker at the power amp chassis lug.

But... since the cap can has a shared ground, I moved the PT CT to the chassis lug in order to isolate it from the screen cap. Does this look OK? Thanks.

[sluckey]

I'm very fond of using a dual cap can for the plate and screen nodes. Works very well.

[pdf64]

DO NOT wire the screens and PI grounds directly to the reservoir cap's negative terminal and PT HT CT.

Thank you Willabe!

My issue is that I used a 32uF/32uF cap can, so there's no way to separate the reservoir cap ground from the screen cap ground - it's shared on the can…

That’s not a problem. Output valve anode and screen grid current both share the cathode, it’s not like mixing the 0V commons of later and earlier stages.

[dwinstonwood]

That’s not a problem. Output valve anode and screen grid current both share the cathode, it’s not like mixing the 0V commons of later and earlier stages.

I'm very fond of using a dual cap can for the plate and screen nodes. Works very well.

Thanks both of you. With this rebuild I got some 60 cycle hum that wasn't there in the original build. 

Another thing I'm doing is putting a Hammond 194A 4H/50mA choke inside the chassis on the PA side. The original build had a huge 15H/100mA choke right behind the EF86 with its leads running back and forth under the board. The little 194A will sit right inside the chassis far away from any preamp stuff with very short leads. But, this requires shifting the choke to be after the OPT node since it's only rated at 50mA (the 15H/100mA choke was positioned before the OPT as it is in the original Vox schematic).

[pdf64]

60Hz hum can’t be coming from the HT circuit; that would be 120Hz.
So it’s most probably coming from the heater circuits.

[dwinstonwood]

Well, it could be 120. I have trouble discerning the two unless I hear both back and forth with each other. At any rate, optimizing the HT wiring was still worthwhile and really easy to change.

When I rebuilt the amp I didn't even touch the heater wiring. Nothing was changed or rearranged with that. As soon as the choke arrives I can install it and see what happens.

If it does turn out to be the heater wiring would adding two 100 Ohm balancing resistors be worthwhile? Or, maybe a hum pot? Can you leave the heater CT connected to ground when doing this?
Thanks.

[sluckey]

Using resistors or a humdinger pot on the heaters probably won't be any better than just connecting the real center tap to ground. BTW, you use real CT or the artificial CT but not both. Connecting the real CT to the EL84 cathodes may improve the hum a bit.

I would start pulling tubes to see if this kills the hum. Pull the EF86 first, then the 12AX7.

[dwinstonwood]

Thanks sluckey. I'm hoping that the hum was mostly due to having that big choke right behind the EF86 and its leads running almost the entire length of the board.

I do your test when I get the new choke (I already took the old one out).

[Willabe]

My issue is that I used a 32uF/32uF cap can, so there's no way to separate the reservoir cap ground from the screen cap ground - it's shared on the can. I'm thinking the idea method would have been to use two single caps instead.

We went through this before. I wouldn't use a can, Sluckey and many others love them. 

Merlin says right in his book and on line web site and KOC says in his books not to use them unless you use both caps in the can in || .

Here's a PDF layout showing the closest I can get using the cap can. There's no NFB, so I gave the 12AX7 its own star, and grounded the speaker at the power amp chassis lug.

That's good.

But... since the cap can has a shared ground, I moved the PT CT to the chassis lug in order to isolate it from the screen cap. Does this look OK?

I'd wire the PT B+ CT at the cans ground lug. That would at least isolate the power tubes K and -bias ground. You might not hear any difference from the way you have it now or putting the PT B+ CT back on the cans ground lug. I would.   

I wire it like Merlin and KOC to nip any potential ground problems in the bud, before they happen.

Many amps work fine without doing some/all of Merlins/KOC's wired ground scheme.

The higher the gain of an amp, more gain stages, the more the noise gets amplified with the signal. So lower gain amps don't always need the absolute most careful of grounding schemes. 

[dwinstonwood]

Thanks Willabe.

I can move the PT CT back onto the cap can ground tab. I'm sure it all adds up; one thing here, another there. I saved a lot of money on this amp by using a readily available pre-drilled Princeton chassis and a standard Princeton cabinet. And, I didn't have to wait a long time for custom work. The Fender chassis more or less determined where I placed the iron and the tubes, etc. I wish the Hammond 1444 series chassis were just a bit thicker. They're very affordable and come in so many sizes. They're just a bit flimsy under the weight of transformers. But, they would allow you to move things around and create an optimum layout. I'm leaning that way for my next project.

[trobbins]

Imho there are three aspects to this grounding topic.  One aspect is that few people 'hear' ground loop related hum in their amps, no matter what grounding scheme they use, and so the use of one particular scheme over another just depends on what people have experienced - and if hum is heard then it's often an error on the part of the diyer, or a poor tube or part, or related to the first-filter node of the power supply.

Another aspect is to do with an external connection introducing a hum loop, and that may relate to how the input signal ground passes through to the chassis ground and then through the mains earth back to the external equipment.  A key aspect of linking the amp's signal ground to chassis at the input end of any distributed star ground scheme relates to avoiding using the distributed star ground path to conduct such external hum current.

The last aspect is typically insignificant but nonetheless worth appreciating as it could show itself in some situations.  It relates to signal currents that loop through some major parts to the chassis by way of capacitive coupling.  Most noticeable is capacitance between power transformer and output transformer and choke-input windings and the chassis, and how that 'parasitic' current loops back through the chassis, and then through chassis to ground link (or links) and back to the originating transformer windings.  All parts will exhibit a titch of this form of coupling capacitance to chassis, but luckily it normally doesn't have any impact on a guitar amp, but may become an issue for amps with feedback.

[bmccowan]

Dude -

BTW, does anyone know if a larger nut is available for a Carling on/off switch? That thin nut never seems wide enough for my faceplates...?

I do not like that thin knurled nut that comes with Carling switches - I even bought those stupid special pliers. I found both hex nuts and beauty washers to fit. I think I bought them from Mouser, but not sure as I bought 10+ a while back. This place has washers and nuts for Carling Switches. https://www.musicalinstrumentshoppe.com/

[kagliostro]

Ciao Tubeswell

Seems that there is something missing on the bias supply .....



Franco

[sluckey]

Ciao Tubeswell

Seems that there is something missing on the bias supply .....

That's called the anti-ground. Quietest scheme of all!   

[kagliostro]

 

[tubeswell]

Ciao Tubeswell

Seems that there is something missing on the bias supply .....



Franco

Oops yes - bad schematic - should be an AC-coupled bias supply - fixed 

[dwinstonwood]

So, trying to figure this out... with a bridge rectifier (PT with no CT) you need a capacitor before the diode in the bias circuit?

[tubeswell]

So, trying to figure this out... with a bridge rectifier (PT with no CT) you need a capacitor before the diode in the bias circuit?

Yes - see bottom of page in this link


http://www.valvewizard.co.uk/bias.html

[dwinstonwood]

Thanks tubeswell. That's a pretty involved circuit. I think I'll just keep buying PT's with center taps for fixed bias amp builds. No reason I can think of not to.

[dwinstonwood]

A big thanks to everyone who helped out here. I now have a very quiet amp. I have to get down close to the cabinet to hear any hum at all. So, the issue is solved. 

Here is the final layout that I used. The choke is actually rotated about 30 degrees counter-clockwise, but I couldn't rotate my drawing in DIYLC (I made the choke from shapes).

Thanks to all who posted!