Tube Amp connections : 0V vs EARTH vs HEAT vs CURRENT

[systemtruck]

Hey everyone,
Somewhat unusual mindset here, since I’m doing a number of projects that involve external power supplies and want to make sure i have some important principles lined up.

I learned from one (successful thus far) project that the PT output HV, let’s say for example 360/0/360, needs to have it’s 0V connected to the rectifier tube’s output negative. So let’s say 320V / 0V. And then, that pair of 320V / 0V wires physically should go, after some length since this is an external supply, as directly as possible to the output power tubes GND and the output transformer GND.

But by default, that stuff is all connected to the chassis. So, although some distance is involved, it is still connected to mains earth. But what happens if we delete the entire chassis in concept? Does our HV 320V / 0V output DC need to make it’s 0V connection to mains earth somewhere? If so, where would be the best place in principal to make this connection? Back on the power supply at the rectifier tube’s 0V? Just a short wire from that to the mains earth input?

And again, if our chassis is deleted and the entire imaginary point to point system is just floating mid air… where do we now need to install some heat sinks? I believe that a chassis also provides heat sinking, albeit in messy fashion. So if that’s gone in principal.. maybe add some heat sinks at the output tube grounds, and also the output transformer ground (which is possible the same point) ? And also back on the power supply.. on the PT’s lugs (which are earthed) ? And the rectifier 0V (which may or may not be earthed, depending on question above) ?

I have one old Fisher amp that physically has the OT’s negative be brought back via a long wire to the phase inverter’s ground at the cathode bias point. So maybe i need to be tying together all grounds in the phase inverter and output tubes and output transformer, and send the power supply’s 0V DC to that conglomeration as one input point on audio segment.

Thank you!

 

[systemtruck]

Replying to myself.
I found @ruffrecords ‘s wonderfully helpful grounding scheme guide for tube circuits.
It certainly answers the question of what to do with the 0V DC signal.. and the answer is to connect it back to the mains earth along with the rest of the star system.

I’m still wondering what items in a typical tube amp are best all getting a ground from this 0V DC, and how heat is being handled at those spots. So I believe this is the Phase Inverter, the Power Tubes, and the Output Transformer. All one common target for the 0V DC signal to directly connect to. But if there is no chassis there, do we heat sink the living daylights out of that ground point?

 

[dmp]

This is a must read
https://www.valvewizard.co.uk/Grounding.html
Addresses what you are asking
I own the book this chapter is from. very good

 

[Brian Roth]

We're lucky that Merlin pops in and out of GDIY!

Bri

 

[dmp]

Current should not flow to the chassis, it should be flowing to 0v at the rectifier. So deleting the chassis should not change the current flow. Current is conserved so you can't hear sink the current if you don't have a return path. With no return path nothing happens. Energy transfer (heat) happens as current flows across a voltage drop. A heat sink doesn't need an electrical connection but should be thermally conductive to the place where heat is being generated.

 

[systemtruck]

I read the article. It is terrifically written and answers nearly all my questions! This includes confusion about output transformer ground and the NFB ground relationship. Amazing. Maybe I need to buy the whole book.

It does confuse me on my very first question though.. about the negative of the reservoir capacitor.

Before reading this article and after reading @ruffrecords ’s grounding article, I was coming to my own conclusion that the DC negative of the power supply, so the reservoir capacitor and the center tap of HV output from PT, should be locally connected to earth. I was thinking that this would be the only place DC negative was earthed, and all else down stream would be floating in principle. I was then going to ultimately downstream connect the input jack ground to the bus bar which upstream finally connects to reservoir negative which is connected to earth as stated in power supply.

But this article suggests that the negative of reservoir cap should not ever be connected to earth in the power supply section, but instead, its bus bar which follows the circuit downstream will be directed to earth at a strategically chosen point. I understand the complexities of all that. In the most simple case, this will be at the audio input jack to chassis.

So if that’s true, the PT’s high voltage AC center tap, which feeds the rectifier, has no connection to earth close by at all. This is the opposite of what I was thinking for my external power supplies, which is that I would put a short heavy gauge wire from reservoir negative to earth at mains.

Which is the best methodology for handling the power supply section’s DC negative/ground? Earth it locally? Or leave it floating entirely through the circuit and earth it at the audio input chassis?

 

[ruffrecords]

For tube power amplifiers with unbalanced inputs, I prefer to follow the RCA scheme. This uses a copper bus bar for the 0V. The diagram below shows how the HT capacitors and transformer centre tap are connected to it at one end and the input connector is connected to it at the other end. At the input connector is a single connection of the 0V bus bar to the chassis. This and the connection from the mains IEC connector safety earth to chassis are the only connections to the chassis.


0V for the preamp, phase splitter and power amp stages are connected to the bus bar in order of signal level. This avoids high output currents, from power tubes for example, flowing in the grounds of sensitive input circuits. The copper bus bar us usually a length of 16AWG copper supported at each end by and insulated stand-off

Cheers

Ian.

 

[systemtruck]

For tube power amplifiers with unbalanced inputs, I prefer to follow the RCA scheme. This uses a copper bus bar for the 0V. The diagram below shows how the HT capacitors and transformer centre tap are connected to it at one end and the input connector is connected to it at the other end. At the input connector is a single connection of the 0V bus bar to the chassis. This and the connection from the mains IEC connector safety earth to chassis are the only connections to the chassis.
0V for the preamp, phase splitter and power amp stages are connected to the bus bar in order of signal level. This avoids high output currents, from power tubes for example, flowing in the grounds of sensitive input circuits. The copper bus bar us usually a length of 16AWG copper supported at each end by and insulated stand-off

Cheers

Ian.

Ok, got it. Understood.
As a new guy to tube stuff, what’s most confusing is this.. if someone told me to keep the power supply ripple away from the audio circuit, id immediately sink that crucial rectifier / reservoir 0V to earth within the tight proximity of the power supply. Basically I’d follow the bus bar pseudo star method in the article just linked precisely, except I’d connect 0V to earth within the power supply so that all that power current was sunk immediately. And then I would NOT connect the input jack to the chassis locally, instead I would connect it to the preamp 0V bus bar point. Then there’s still only one chassis connection, and it’s steered away from the sensitive gain stages.

So to learn that the best method is to instead not let that intense 0V point see earth until it’s passed all the way upstream to the input signal jack is just backwards to what I’d expect. But clearly this screams a lack of experience to all of you haha.

 

[ruffrecords]

So to learn that the best method is to instead not let that intense 0V point see earth until it’s passed all the way upstream to the input signal jack is just backwards to what I’d expect. But clearly this screams a lack of experience to all of you haha.

If the input connector is floating then there is no reason for you not to do it the "normal" way by connecting 0V to chassis at the PSU end of the bus bar. The important concept is maintaining the separation between the chassis (as a screen) and analogue 0V (as a bus bar) except at a single defined point.

I have seen the RCA mthod used in several of the power amplifiers from the 1940s or 1950s. In those days metal jack and phono sockets were the norm so the input connection of ground to chassis could not be avoided.

Cheers

Ian

 

[Wavelength]

ST,
Why external? This makes it really more complex because now you have HV cable, shielding and connector issues. Especially if you are new to Tubes. Your best bet is follow Ian's lead with a single chassis format.
Localized star grounding is best... by that I mean look at each sections grounds tie those together then hit the ground bar from there. If your project is spread out this works better. If it's really simple then yes a single star would work as well.
Thanks,
Gordon

 

[Matador]

So to learn that the best method is to instead not let that intense 0V point see earth until it’s passed all the way upstream to the input signal jack is just backwards to what I’d expect. But clearly this screams a lack of experience to all of you haha.

Search here for "hierarchal ground" and you find many posts by myself and other discussing it at length.

Chassis is a safety consideration, and should play no role in the circuit. Intermingling 0V and the chassis is a Mistake (TM), even though a lot of famous designs got away with it. Your circuit should function perfectly well in a plastic chassis, so there is effectively no place to "earth" the 0V line.

 

[ccaudle]

where do we now need to install some heat sinks?

From what do you want to dissipate heat? In a tube amp it is the tube which generate the heat, and those dissipate the heat primarily through the glass, so there is not much heat transferring through the socket pins.
If you have solid state rectifiers or any transistors as part of the power supply or output stage then those need heat sinks, both because the maximum allowable temperature is lower and because the smaller package increases the heat density at the same time it decreases the surface area for dissipating the heat.

if someone told me to keep the power supply ripple away from the audio circuit, id immediately sink that crucial rectifier / reservoir 0V to earth within the tight proximity of the power supply

As soon as you start using terms like "sink...to earth" you indicate that you are engaging in magical thinking, and are not actually analyzing what causes voltage differences and where current is flowing.
The reason you connect your chassis to the protective earth is so that in the event of a power fault the chassis does not become energized from the power line and electrocute you. The reason the protective conductor of the power line is connected to the literal earth and not just to the neutral side of the building power transformer is to limit the common mode voltage of the power connections, and to divert lightning induced energy away from the building interior.

Note that dealing with your power supply ripple is explicitly not one of the purposes of the protective earth connection.
Follow the current in the power supply: it originates in the secondary winding of the power transformer, through the rectifier tubes or diodes, causes displacement currents in the smoothing capacitors, which then find their way back to the secondary of the power transformer.
Note that the upper part of the planetary crust is not part of that loop, so if you connect the power supply loop to the earth, no noise is going to "sink" into the earth because none of the voltage difference around that loop was generated from the earth, so the earth cannot complete the current loop.

(Just for completeness, there is one path which can actually be completed through the earth connection, and that is noise on the power entry which gets onto the power transformer secondary through capacitive leakage, but that is orders of magnitude smaller than the power supply ripple current, and is typically diverted using capacitors back to the primary side power connections.)

And then I would NOT connect the input jack to the chassis locally, instead I would connect it to the preamp 0V bus bar point.

The problem with that is now the cable shield is not a shield, but an antenna directing external RF energy into your audio circuitry. Some designs can get away with that, especially if there is just one external cable connection, but the more connections you have going in and out of the chassis the harder it becomes to reason about where the RF currents are flowing and where they might get demodulated into audio frequency noise, so it is generally safer to have the shield remain a shield, i.e. the metalic covering shielding the cable signal conductor connects directly to the metalic covering shielding the audio circuitry. There can be tradeoffs around power line related noise vs. RF induced noise, and tubes are typically not as prone to demodulating RF as solid state circuits, so you may be able to get away with a more flexible approach.