Strange Vox AC30 problems

[R.G.]

Well I will have a few questions while reorganising the grounding throughout the amp, but one thing I can say now that you probably forgot is that the amp is now tube rectified(GZ34) so there's no ground in the rectifier.

Oops. Concentrating on the circuit. In the case of a tube rectifier, the negative side of the "rectifier" is the CT on the transformer. So the CT on the transformer must go to the negative side of the first rectifier, no where else.

Also maybe you missed what I found out during the last tests: "Something strange here: I tested the small electrolytic caps and they are grounded on both sides, I mean the positive side is grounded too.... is this normal? Or is it because I grounded the shielded wire between Pin 7 V12 - 3.3M resistor?"

Again, ooops. No electro - belay that, no cap of any kind - should be grounded both sides. Which "small electrolytic caps" do you mean on the diagram? I don't see any electros which would be grounded both sides from what we did, although some of the non-electros would be.

[Jimmy74]

I mean the small radial electrolytic caps which on the schematic are shown as being 25uf 25v but in the amp are 47uf 25v off V4B, V1A and V2A. Do you mean that the HV CT from the PT should go to the negative lug of the first filter cap? On that lug I've actually got the bias ground hooked up.

I'll put back in the 3.3M resistor and the other parts and retest that grounding.

[Jimmy74]

Well something is really strange here, it's not as though both sides of the caps are grounded in the same way, there's a higher resistance on the positive side to ground. I'm getting ground off both sides of the small electrolytic caps + bias cap and resistor and pin's 3+8 V4 and pins 3+8 V3.

From what I can see in the amp, all pre-amp grounding and input jacks are grounded to the same point on the V4 tube socket.

[Jimmy74]

False allarm on getting grounding off both sides of the caps, it was my tester giving me the wrong readings. So I'll get back to reorganising the grounding throughout the circuit. It is strange however that Vox actually wired this amp up like this, Do also think that the choke being so close to V11 and the OT could be causing hum. Also I am trying to get bias current down by increasing resistance, at the moment I've the 5w 47 ohm + 5w 56ohm resistors in series and I'm getting around 29-34mA.

[R.G.]

From what I can see in the amp, all pre-amp grounding and input jacks are grounded to the same point on the V4 tube socket.

Actually, that make it easier. Lift that one ground point so it doesn't attach to chassis, and run a wire from it to what becomes point B in my sketch, the main star point for the amp. An important test for star grounding is whether you can open the single wire you **think** attaches to the chassis, and have that make the chassis read open circuit to any signal ground.

False allarm on getting grounding off both sides of the caps, it was my tester giving me the wrong readings.

No problem. I've done a lot of , er, creative use of meters before.

It is strange however that Vox actually wired this amp up like this,

Star grounding is a huge amount of work in a hand-wired amp, so there are almost no examples of it. Manufacturers try to figure out cheaper ways that are almost as good. I could use real star grounding because I was doing it on a PCB, and it came out the same way every time, no more manual labor involved. The kind of tech that automatically dismisses PCBs because Fender and Marshall do a bad job of it won't want to talk about what PCBs can do that's good.

Anyway, star grounding is not the only way to build a quiet amp. It's just the only way you can be sure will result in a quiet amp without experimenting with which wire goes where after you build the amp.

Do also think that the choke being so close to V11 and the OT could be causing hum.

I did think that before you shorted out the reverb pot signal and it made no difference in the hum you heard. It is possible that it makes a difference in what hum you'll find once we nail down the big hum you're tracking. Eliminating hum is like peeling an onion. You remove it a layer at a time, and you cry a lot.

Also I am trying to get bias current down by increasing resistance, at the moment I've the 5w 47 ohm + 5w 56ohm resistors in series and I'm getting around 29-34mA.

In cathode biased amps, I like to make the cathode resistor too large, making the current too small, then trim the resistor down and the current up by paralleling higher value but lower resistance resistors to the main resistor.

Also, you want that main cathode resistor which generates a lot of heat to be somewhere it does not overheat any capacitors and cook them into a shorter life.

[Jimmy74]

Would it be ok to place the CT for the heaters and the shield for the PT at point B too?

[R.G.]

I had already taken out all the previous shorts before placing the shielded wire from pin 7 to the 3.3M resistor. Anyways hum is still there. I also left out the 10pf cap for this test. Hum is still there.

This narrows it down a lot. Something about the V12B triode section is picking up hum.

It is very likely, since you have swapped tubes already, that it is a socket or a grounding problem. Our previous tests have eliminated everything before the grid of V12B, and after its plate. That leaves the grounding, the cathode resistor, the plate resistor and the power supply feeding that section. Of these, the grounding is by far the most likely.

Would it be ok to place the CT for the heaters and the shield for the PT at point B too?

Yes, in almost all cases.

[Jimmy74]

Things to note:
1. The earth/safety ground connects to the chassis. This is exactly one, no more, wires. The connection to the chassis from the AC power line ground must be at a bolt on the chassis used for this purpose, and no other, and there must be a star washer under the wire's ring terminal (and the wire must have a crimped ring terminal!) to bit into the chassis and terminal.
2. The star ground point at B connects to the chassis with exactly one, no more wires. This is the multicolored wire between B and the earth ground. If you remove that one wire, there must be no continuity to the chassis. Although I show it going to the earth ground point, it may be better to connect the input jack bushing to ground and make this the only signal ground to chassis ground point for RF noise reasons.
3. The negative side of the rectifiers does NOT connect to the chassis. It has a single wire to the negative side of the first filter cap. This is quite important for minimizing rectifier buzz. The positive side of the rectifiers goes on one wire to the first filter cap.
4. The negative side of the first filter cap goes on a single wire to the star point B. You must not have the other signal wires connect to the negative side of the first filter cap, nor to the negative side of the rectifiers. This will positively cause hum issues.
5. The two red octopi at the two input channels show what you should do with local grounds. It is important that local grounds NOT be tied to chassis out at the tube. You collect the local grounds for a tube or two which are heavily connected, and make up that center dot. The center dot has a wire which goes to the star point B.
6. Decoupling caps should have their own wire to B.
7. On V5, V12, etc particularly:
V12's local grounds gather at point E, which is then run with a single wire to the star point B. V5's local grounds do the same at D. It's important that even if the tube socket offers you a tempting chassis ground point, you not connect those local grounds to chassis, but instead run them back to the star point.

The controls' outer cases are attached to the chassis by their bushing, and a common mistake is to "ground" lugs to the control's metal case. This can sometimes create hum where there was none. The grounded lugs of the pots go back to the local star point they're most involved with, or to the star point B.

This is a lot of work. And it's not the only way to get a quiet amp. But it's the only way you can say ahead of time that will for sure remove ground hum and noise.

In your case, you have to get the following done (I believe):
1. Get the local grounds off those tube socket center pins which connect to the chassis.
2. As part of that make a star point like B. Run the wire from B to the first filter cap negative as shown. Run the wire between the first filter cap negative and the minus side of the rectifiers, and neither of these wires can touch chassis.
3. Run the local grounds for V5 and V12 as shown, with local collection points at D and E respectively, then a single wire from each local collection point to B.


I've probably confused things. Questions?

I've started grounding the amp as you suggested, taking out the local ground for input jacks and circuitry off the tube socket and moving them to point B. My confusion is: if point B and all points A,C,D and E are all connected to the chassis, even if I take out the wires which connect them to point B, this will be grounded anyways.

You said this "The star ground point at B connects to the chassis with exactly one, no more wires. This is the multicolored wire between B and the earth ground. If you remove that one wire, there must be no continuity to the chassis." but I don't see how this can be if all points are connected to the chassis which grounded from the earth ground.... or am I missing something here??

thanks

[R.G.]

I've started grounding the amp as you suggested, taking out the local ground for input jacks and circuitry off the tube socket and moving them to point B.

It is best to find a local terminal somewhere near the tube socket center pin to reconnect all the Local Star Points(Note 1) to. Verify that the Local Star Points do NOT connect to chassis unless you run the single wire over to B.

My confusion is: if point B and all points A,C,D and E are all connected to the chassis, even if I take out the wires which connect them to point B, this will be grounded anyways.

It is confusing. We have all been taught that ground is ground is ground. It's not. This is a voltage-centric view of the electronic universe. There is a current-centric view that needs used in thinking about grounding.

The whole point of special grounding schemes is summed up in the question "Where does the current flow, exactly?" This is important because all ground conductors are really resistors, not the zero-resistance wires we like to think of them as. If you use the chassis as a ground, that seems to be fine. In actuality, the current flows through the chassis in paths that do not spread over the whole chassis. At low frequencies, the currents flow where the resistance is lowest, that being usually the straight-line path between where it comes from and returns to. The current is not a single path like a wire, but instead a broad hump of current density, most dense in the straight line path, falling off smoothly to the sides as the path length for the sides makes the resistance bigger, but never falling to truly zero.

That's true for a single source/single return in a conductive sheet, like a chassis. If there are other sources and returns along the way, they inject current into the river that's already there, and so there are humps and such in current density. If the other circuits along the way use the current-river for a reference voltage, they sample the local voltage, which may be elevated by the flow of current through the chassis. The chassis is probably steel, which has 5? 7? times the resistance per unit of copper. So the humps and deep places in the flow of current back to the return place has corresponding humps and deep places in voltage as well.

If instead we give each of the possible sub circuits its own return wire, that wire may have its own voltage losses, but (1) the wire is probably copper, and lower resistance and (2) more importantly, it's not "contaminated" with the current flowing from somewhere else. The idea that you can force current to flow in its own wire, not shared with other parts of the circuit, is star grounding. Everything else comes out of that idea if you contemplate electrons and your navel long enough.

I classify grounds into several different types of "grounds". One is a reference ground. I am going to use this not to cause current to flow, but to tell an input circuit "This is zero point zero, zero, zero volts. Base all of your amplifying on this as zero." Another is shield ground. In effect, this is a metallic cage whose function is to intercept all the radio waves and electric power line fields that permeate our world, and force their energy to be expended without confusing the circuits inside the cage. The third is what I call "sewer ground". It's the return path through which the used electricity which has made some section of the circuit work gets back to the power supply to be recycled. Keep these definitions in mind.

I don't see how this can be if all points are connected to the chassis which grounded from the earth ground.... or am I missing something here??

And this is the classical problem people have with thinking about "ground". Yes, you're missing two things. First, there can only be one point in the circuit that really, no fooling, no messing about, has a voltage of zero volts. You're free to designate any place you like as this point of zero volts, but then you must measure every other voltage with respect to it. If you have other places you want to be as close to zero volts as possible, then you have to be careful what you call "ground". Second, if there is only one wire to a metallic object, then no current can flow through the wire.

What that means is that you can make a reference ground have no voltage across it - or a vanishingly small voltage - by not letting current flow through the wire which leads to it. This is very, very good for referencing inputs. You can force the "input ground" on the input jack to be arbitrarily close to the same voltage as the One True Ground by running a wire from that jack to the One True Ground. No current flows (ideally, if you have a high impedance input and all input current returns out the cord that brings it into the input jack), so there can be no voltage across the wire to the One True Ground, and so the voltage difference from the One True Ground is zero. You have a good reference.

Contrast this with what happens if you are running a speaker. You're sending amperes of current from the power supply to the speaker, and that current has to get back to the power supply. There are wires and perhaps a chassis carrying back the return current from the speaker jacks. These amperes of current flow through the milliohms of resistance through the chassis/wires/etc to the power supply from whence they came. Along the way, the amperes cause milliohms of voltage to appear across the conductors which are carrying the sewer of the used speaker current.

If you happen to connect your input jack reference wire to the chassis near where the speaker return amps are coming back, the input circuits think that the speaker current's I*R voltage on the return sewer path are signal, and they amplify it. This can - and does! - easily end up in runaway oscillation.

Exactly the same thing happens if you "ground" the CT of the power transformer to the chassis, then pick up "ground" off the chassis somewhere else. The path through the chassis from where the CT enters it and where the "ground" is picked off wiggles around with the current pulses feeding the fiirst filter cap. These pulses may be amperes, even in a tube amp, because of the way that rectifiers work. If the reference path to your input jacks, or anywhere else a tube grid is tied to "ground" happens to intercept the current in that sewer ground, you will hear it as a hum. Period.

So the way to absolutely guarantee that this will not happen is to force the currents to flow only in the wires you want them flowing in. For the power transformer to rectifiers to first filter cap, back to the CT, the wires are all contaminated with amperes of pulse current. You don't want any of this in your audio path, so you do not connect any of these wires to your audio path. The first filter cap is a reservoir, filled with + pulses from the rectifiers and - pulses from the CT. You do want the DC voltage to get to your circuit, so you take a wire from the - side to your One True Ground, which is point B in my diagram.

We want the chassis to act like a shield, but not a return sewer or reference voltage. So we take one and only one wire from the metal chassis and connect it to point B, our One True Ground. Now the RF and radiated stuff can beat itself senseless on the metal shield, but any voltages they cause across the chassis cannot be picked up by our inputs because our inputs don't connect to the chassis, they connect to the One True Ground.

It would be best if every single part that has a ground symbol on one terminal of it had its own wire back to the One True Ground. I've actually done this, and it works fine. Very quiet. But it's a major PITA, what with the hundreds of ground wires snaking around. What works almost indistinguishably from this is to gather related circuit grounds into Local Star Points ( I finally found my way back to those!) and then take one ground wire from each Local Start Point to B, our One True Ground. Generally these are low current circuits, so the overall voltage drop on each ground wire is small, but more importantly, if you choose which circuits connect to each Local Star Point well, the currents are all +/- the same signal; they tend to cancel, or at least be smaller representations of what you are trying to amplify anyway, so they don't result in either runaway oscillation or noise pickup.

And that gets us finally back to the concept of breaking one wire and testing a ground. To test if your chassis is properly grounded, you should be able to open the one wire to the chassis from the One True Ground and meter an open circuit. Then reconnect the wire. You have it right. If this does not result in an open circuit, you did something wrong. There's another current path to find and kill. Likewise, opening the one wire from a Local Star Point and the One True Ground should result in an open circuit as measured from the Local Star Point to chassis, One True Ground, and the power supply minus as well as speaker return. If it doesn't there's another circuit path to find and kill. Likewise speaker return, likewise first filter cap.

If you do this as explained, you will not have ground-induced noise within the amp. There are other ways to get hum, but ground noise will be eliminated.

This is NOT the only way to get zero ground induced hum. However, all other ways to get rid of it wind up controlling where the ground current flows for some signals, and using fortuitous cancellation of currents in the shared ground paths to keep noise low. There is a lot of trial and error. Most non-star-ground tube amps are done this way to keep down the labor cost of doing star grounding. If you can invest the time to find a cheaper alternate way to wire once, then copy that good-enough way many times, you save money. Can you say "Leo Fender"?

I do seem to have run on. Did that clear it up or muddy it up?

[Jimmy74]

Wow----- what a lot of info there, I had to read it a few times before actually getting the idea of what you're telling me.

So a chassis should be considered an actual electronic source and not just a big open space where we ground our caps and resistors. Channeling specific ground wires to be grounded where we want them to will help get rid of hum. And therefore the local ground points must not be grounded to the chassis directly but they will be grounded with wires going to point B, and this goes for points A,D and E too.
From what I see in the amp:
- The input jacks are grounded to the same tube socket (not the socket centre hole) but to the same the tube socket V4 mounting screw as all other pre-amp grounding wires including the metal bar which connects the ground lugs pots and tremelo switch (none of the pot lugs are soldered to the pot cases). Therefore the pot cases are grounded but the lugs are not when the ground wires at V4 are disconnected.
- The 32uf +32uf caps are grounded to the same ground screw by themselves.
- The main filter cap is grounded to the same ground point as point C
- The CT for the heaters is grounded to the same point as the power transformer shield wire.
- The CT for the HV is grounded all by itself
- The reverb section board, footswitch and tank are grounded to the same ground point as input jacks and pre boards, the reverb section board ground connects to the footswitch/tank on the V12 socket centre hole which you can see in the photos.

So as you say, lifting the ground wires at V4 tube socket mounting screw takes out 95% of the amps grounding which I can channel right to the point B and then work backwards to isolate and divide the reverb section and then take care of the HV caps.

Just one question, I am using a 50+50uf cap therefore the first 2 filter stages do share the same ground point A, this shouldn't be a big problem right?

thanks

[R.G.]

So a chassis should be considered an actual electronic source and not just a big open space where we ground our caps and resistors.

In some ways, yes. Currents going through the chassis can create voltages which can be picked up as interfering signals.

At very high radio frequencies, you have to shift back to using chassis for ground - well, in most cases a ground plane, but chassis can be used - because the frequencies involved make the inductance of a wire too big for the wire to be a "short". The lower inductance of a broad, flat conductor makes it attractive again, and the field effects of high frequency signals make them "want" to run in the plane right under the signal wires. But this is not an issue with guitar amps.

Channeling specific ground wires to be grounded where we want them to will help get rid of hum. And therefore the local ground points must not be grounded to the chassis directly but they will be grounded with wires going to point B, and this goes for points A,D and E too.

In an ideal setup, yes.

From what I see in the amp:
- The input jacks are grounded to the same tube socket (not the socket centre hole) but to the same the tube socket V4 mounting screw as all other pre-amp grounding wires including the metal bar which connects the ground lugs pots and tremelo switch (none of the pot lugs are soldered to the pot cases). Therefore the pot cases are grounded but the lugs are not when the ground wires at V4 are disconnected.

So this point might be a good place to leave alone, making it be "point B". Verify for me - it attaches to the chassis through V4's socket mounting screw?

Does V12's grounded components get into this, or are they attached to ground some other place? How about the shields of the reverb tank cables?

- The main filter cap is grounded to the same ground point as point C ...
Just one question, I am using a 50+50uf cap therefore the first 2 filter stages do share the same ground point A, this shouldn't be a big problem right?

In this case, it is important that the power transformer CT go to the negative side of the 50+50 cap, nowhere else. From there, a wire can go to point C, or to a newly created point B. Neither B nor C should go to chassis except through the single wire to chassis.

- The 32uf +32uf caps are grounded to the same ground screw by themselves.

Potentially bad. Each should have a wire from its minus side to the star ground point B. But probably an inner layer of the hum onion.

- The CT for the HV is grounded all by itself

Bad choice IMHO. This needs to go only to the minus side of the first filter cap.

- The CT for the heaters is grounded to the same point as the power transformer shield wire.

Neutral to slightly bad. This is not critical, but it could be either attached to point B or to even an elevated DC level. But it must be connected to signal ground some way and not floating; this could cause hum all by itself if it floats.

- The reverb section board, footswitch and tank are grounded to the same ground point as input jacks and pre boards, the reverb section board ground connects to the footswitch/tank on the V12 socket centre hole which you can see in the photos.

I would make this have its own wire to point B.

So as you say, lifting the ground wires at V4 tube socket mounting screw takes out 95% of the amps grounding which I can channel right to the point B and then work backwards to isolate and divide the reverb section and then take care of the HV caps.

Yes. However, notice that all of the hum issues we've found so far stem from V12b. So in fact, the rest of the amp may be OK-ish, with one of those cheaper but useful grounding setups which I referred to. It may be that by only cutting all the grounds for the reverb loose from the chassis, and running a ground wire for them to either the V4 point or to point C will eliminate a lot of hum, maybe enough. It's certainly worth a try before you go rewiring the whole thing.

I would still get that transformer CT thing fixed anyway.


thanks

[Jimmy74]

Yes that ground section is connected to the chassis at the V4 tube socket mounting screw together with the shield wire from the input jacks and it is the ground point also for the whole reverb section therefore when this grounding point is lifted there is no ground to the reverb section including V12, the whole preamp section is not grounded and the input jacks are not grounded.
The only grounds left are in the HV section, CT's, the pot cases and the small 4.7uf 450v cap which I added from HV to the reverb transformer. The reverb tank shielded cables are not grounded directly, that is.... the shields from both footswitch and output from the tank are tied together and then grounded at the centre hole of the V12 tube socket. The input for the reverb tank is not grounded at all, but I think that when both input and output are connected to the tank ground is shared. The other thing I noticed is that there is no 22k trim pot for the output from the reverb tank, the "hot" wire is connected directly to pin 2 and the shield is soldered to the center hole on V12.

[R.G.]

At this point it would be helpful for you to annotate the schematic again with what's connected up into which ground points. We can then work on how to mod it.

[Jimmy74]

Ok it took me a while but I managed to do an almost 100% accurate design of the original ground points for this amp, as you can see I put V1, V2, V3, V4, V6, input jacks and pots all on one main ground point, there are actually 3 of 4 ground wires for the pre which all end up on the same mounting screw for V4. The reverb section ground is tied to the this main ground section using the shielded wires and joins the reverb board to the switch/tank at the center hole of V12:

Flickr Photo Download: ac30 original grounding points

the HV CT is grounded by itself
The 4.7uf cap on the reverb HV path is grounded by itself
the heater CT and power tranformer shield are grounded together

Just one question: in your sketch there should be no grounding on any of the local points or points A, B, D or E, these all follow through to the mains safe ground point and when this coloured wire is lifted, only point C should result in being grounded right?

thanks

[R.G.]

Ok it took me a while but I managed to do an almost 100% accurate design of the original ground points for this amp, as you can see I put V1, V2, V3, V4, V6, input jacks and pots all on one main ground point, there are actually 3 of 4 ground wires for the pre which all end up on the same mounting screw for V4.

Notice how the pattern of the ground wires makes a (ragged!) star? They were on this back at the factory. You're probably 90% of the way to a good star grounding setup if this is true.
However:

The reverb section ground is tied to the this main ground section using the shielded wires and joins the reverb board to the switch/tank at the center hole of V12:

It's important to the usefulness of star grounding that you NOT have multiple paths between places. If the reverb section is tied to chassis at the center hole of V12, this is a fruitful place for chassis hum to get in.

In the diagram you have shown, there is only one True Ground. It's the mounting screw for V4. That is the only place in the entire amp where the voltage is really, truly zero volts (for our purposes). And it must be the only place where signal ground connects to the chassis. So to make it better:
Find out exactly how the center hole of the V12 socket contacts ground. Through the socket mounting ears? Then insulate the socket from the chassis at the mounting ears. Use fiber gaskets and nylon screws, for instance. Then when you can verify that the V12 center hole does not contact chassis with an ohmmeter, run a wire from there to the V4 "star ground". This may, possibly, could be what cures the hum from V12b.

the HV CT is grounded by itself

Bad choice. This must go to the minus terminal of the first filter cap. Nowhere else will do for lowest hum. This is another place where it might cure the hum if you corrected it.

The 4.7uf cap on the reverb HV path is grounded by itself

Not a particularly good choice. This should go to the V4 star point.

the heater CT and power tranformer shield are grounded together

This is OK-ish. I'm assuming that you mean "attached to the chassis in the same place" when you say "grounded together". Chassis is not ground. Chassis is not ground. Chassis is not ground. It helps to repeat it. The chassis is a shield which happens to be attached to ground. The PT shield being attached to chassis is OK. This is to prevent noise carried in on the AC power lines from getting into the secondary. In some very critical applications, the PT will have TWO shields, one which is attached to the AC power safety ground, the other which is attached to signal ground. They do different things.

The heater CT just needs to be tied to a DC voltage which is referenced to the signal ground. This can be through some resistance, and it's *probably* OK for it to be tied to chassis.

Just one question: in your sketch there should be no grounding on any of the local points or points A, B, D or E,

Remember **chassis is not ground**. These points are connected to the One True Ground, point B in my diagram, but the center hole of V4 socket in your amp. They are grounded, but they do not connect to chassis except through the one wire which attaches the One True Ground star point to the chassis. Chassis is not ground - it is a shield which happens to be connected to ground.
Your amp should be wired so that if you removed all connections to the chassis, it would still work. Chassis is not ground. It is a shield which happens to be connected to ground.

these all follow through to the mains safe ground point and when this coloured wire is lifted, only point C should result in being grounded right?

I was not properly thorough in my diagram. Point C should have one of those red arrows from point B running over to it. For point C, there are actually two choices. You can take point C to point B, or you can take it to point A. These have different characteristics, based on where the current flows. At really low frequency, connecting it to B, then B to A is OK. At high frequencies, the wire self inductance says that you should connect C directly to A. There is some possibility of high frequency oscillation for the C->B->A connection, although it will be the lowest hum.

I also left off the connection of speaker common to ground. Strictly speaking, the output transformer common terminal doesn't have to be connected to ground at all. It works OK floating. However, this may let the coils in the transformer pick up noise. The proper way to connect this is to make the speaker common go directly to the speaker jack and NOT THROUGH THE CHASSIS. Then the speaker jack common can be connected to the chassis, or to the star point, or through a resistor to either one of these. Hum comes from many different places. Each of these connections may be better or worse for a specific hum source. But running speaker common through the chassis so the speaker current flows in the chassis can cause intractable oscillation in some situations.

In a nutshell:
Make one star ground point. In this case, most of the work has already been done for you. It's the center point of the V4 socket. Make sure that if you disconnect that point from the chassis, then NO ground point connects to chassis. There is one and only one connection of the One True Ground to chassis. This forces no signal current to flow through the chassis, which is what we're trying to do. It makes the chassis a grounded shield, not a current conductor. Test that all is correct by disconnecting the star point from chassis and verifying that the chassis shows no continuity to signal ground any more.

The power transformer secondary is a floating source of high voltage and high current noise. It must go to the rectifier(s) and from there to the first filter cap. The rectifier wires and CT wire from the transformer carry high currents of hum and noise. These wires must go to the first filter cap, which filters them. The minus from the first filter cap is then connected to the One True Ground by a single wire. This forces the rectifier noise to NOT flow in the wires to the signal ground.

If you accidentally get multiple connections to the chassis, or share part of the wiring to/from the power transformer/rectifier/first filter cap, then all your other work will be in vain.

I may have mentioned this before, but I'll say it again. Chassis is not ground. Chassis is a shield which happens to be grounded.