Finished and working - 12 valve all DHT PP 26-71a-45

[Nick]

Not sure that would stop it working as long as the transformers have sufficient max isolation voltage. On the other hand, not sure what it would gain. But try it and see would be the way I guess. Personally I would at least link to ground with a resistor and bypassed cap, but thats just me.

[simon]

My GM70 amp has separate HT txs for the driver and output, and I have separate signal grounds for each which are tied to safety earth by slightly different resistors i.e. slightly different signal ground potentials. It seemed to work well I recall, but it was a long time ago when I did it so memory's not so great!

[RhythMick]

Simon, where in the PSU circuit for your output valve do you ground it?

My circuit joins all the PSU to ground at the same point, which is the bottom of the LTP, effectively the bottom of the 2nd cap.

The signal circuit for each stage involves the preceding stage transformer secondary, valve grids, cathodes and part of the LTP tail.

[simon]

I'm a bear of simple brain and build SE amps .

From the CT mains trafos I have 2 damper diodes followed by a pure choke input supply IIRC so 10H choke, first main cap (really guessing now, 22uF???), then a split to 2 10H chokes, 1 per channel, and a final 80uF motor run per channel which I use as the local star earth for each channel of the output to minimise current loop.

I do similar for the driver but with a small first tuning cap.

I'm struggling to remember exactly how I earth the local star earths to chassis ground now and I'm away from home at the mo. But I think I ground at the first main cap via ground lift resistors. The idea is to have the driver and output grounds at slightly different potentials to improve isolation. It seems to work anyway.

[RhythMick]

I struggle to understand the reasoning behind putting a resistor in the ground connection. Could someone enlighten me?

The circuit...

TX - LCLC - PP TX - VALVE PAIR - LTP - BACK THROUGH LCLC

... is self contained and galvanically isolated. A connection to star ground (and thence to safety earth) gives a 0v reference. Whether one is needed is another matter.

As far as I can see, no current should be flowing down the ground wire? If that's the case, what's the value of the resistor? It can only lift the circuit to a higher reference potential if there is current flowing through it?

This is I think one of those questions which shows up a building block I've missed on my self - taught journey.

[simon]

It's a generally easy way of avoiding ground loops. If you have two bits of kit both referenced to true earth you may get ground loops if signal connections are not thought through.

In the case above it goes a little further to provide a little more isolation. I suspect it's not quite that simple though as the potential difference of the ground in each circuit will depend on the current flowing through the earth lift resistor.

[RhythMick]

It's a generally easy way of avoiding ground loops. If you have two bits of kit both referenced to true earth you may get ground loops if signal connections are not thought through.

In the case above it goes a little further to provide a little more isolation. I suspect it's not quite that simple though as the potential difference of the ground in each circuit will depend on the current flowing through the earth lift resistor.

Yeah thanks - but there shouldn't be any current flowing through that earth lift resistor should there ? And if there's no current there's no lift - that's what I can't get my head around.

OK if there's a fault (failure or design fault) then there could be current. There could also be capacitative build up on the secondary windings of the transformers, which can drain to earth via that resistor (or straight wire). But aside from those considerations, no current and no lift - right ?

[RhythMick]

OK - having finished Geoffs amp (small buzz still to track down, but otherwise ...) it's time to move on to my own build of this amp.

Initial 3D designs attached. Changes will include;

• DC heater boards for all valves including the 45s, so 12 DC boards total and 6 toroidals (thanks Andrew) - no LCRC circuit for the 45 heaters
• 4 of the boards (for the 71As) are the small type with onboard heatsinks
• The 8 large boards will mount transistors to a large copper sheet which will rise up through the amp to heatsinks mounted on top
• Going to try the 01A valve versus the 26
• The bias will be taken off a separate 100k trimmer pot, still within the cathode tail and below a resistor which sets the minimum bias
• The larger pots will be used to balance the current through the valve pair, by adjusting the cathode up/down from the centre point thus reducing/increasing bias for each valve
• All components in the LTP moved to an LTP board mounted at the front behind a removable front panel, giving access to all measuring and adjustment points

Here the front panel has been moved out of the way to show the LTP boards

View from beneath showing Chokes, Transformers, Caps

The 3D modelling is courtesy of Tinkercad.com (which is free and I've found very easy to use)

[Cressy Snr]

It's a generally easy way of avoiding ground loops. If you have two bits of kit both referenced to true earth you may get ground loops if signal connections are not thought through.

In the case above it goes a little further to provide a little more isolation. I suspect it's not quite that simple though as the potential difference of the ground in each circuit will depend on the current flowing through the earth lift resistor.

Yeah thanks - but there shouldn't be any current flowing through that earth lift resistor should there ? And if there's no current there's no lift - that's what I can't get my head around.

OK if there's a fault (failure or design fault) then there could be current. There could also be capacitative build up on the secondary windings of the transformers, which can drain to earth via that resistor (or straight wire). But aside from those considerations, no current and no lift - right ?

It might help if you thought of the ground wire as the return path of the circuit. A circuit has to exist for current to flow and work to get done. Therefore If you stick a resistor in that path, the resistance will develop a voltage across itself according to how much current it draws. That is your lift voltage above zero.

[RhythMick]

The circuit is complete without that 0v reference wire though. The bottom of the LTP returns current to the bottom of the 2nd LC. Without the ground wire the cucumber would work perfectly, but would not have a 0v reference. I don't think that matters as the valves are linked by transformers and I'm going to try it without in fact.

So the "ground" wire is totally separate from the "psu return" wire. At least in my circuit. If they were the same wire then yes I get where the lift comes in, in fact the PSU return part of my circuit includes 2 choke coils, so the mains tx is lifted above "ground"

[Nick]

It might help if you thought of the ground wire as the return path of the circuit.

But I think that’s where the problem is, a circuit between two transformer stages with its own power supply doesn’t need a return path. Personally I think all the resistor will do is provide a 0v reference to the stage to prevent it floating to the driver/driven stage HT, and to provide a loop for the inter winding capacitance (which will pass signal as well as just floating).

I can imagine a situation where the resistor may interact with that capacitance to provide a CR filter, exactly how much it would matter or happen would have to be tried to know.

It would be interesting to measure the signal voltage (if any) across that resistor.

[RhythMick]

It would be interesting to measure the signal voltage (if any) across that resistor.

I agree. I'm going to build it without 0v references first, then add them in with perhaps 100R inline. I can experiment with what difference it makes being in vs not, also which connection to the 0v reference sounds best and as you say measure current across that resistor.

Fascinating.

[Mike H]

I have had high Voltage on the secondary side of something, that is 'floating', due to inter-winding capacitance maybe, but was a long time ago so can't remember the details. I mean relative to mains earth.

Ideally you want to connect the secondary side ground to mains earth, and have some sort of overload protection, e.g. a fuse, in case you get a short between HT rail (or whatever) and the case, so the case needs to be mains earthed.

However as Simon said this can cause an audible hum loop because there is more than one earth connection between circuit ground and mains earth, i.e. 1 = the mains earth wire to itself, 2 = the ground connection of audio lead(s) connected between it and another 'box' whose circuit 0V is also directly mains earthed.

A resistor of around 10 Ohms or maybe more placed in series between the secondary side or circuit ground, and actual mains earth, at the input to the PSU, is usually enough to kill the hum. But if there's a fuse or whatever to protect the HT or whatever in case of a short to case (which ideally should remain directly connected to mains earth), then you need to bear in mind the short circuit current will go through this resistor so it has to at least survive until the fuse blows.

For some years a lot of stuff I've made includes such a resistor as a matter of course.

HTH

[RhythMick]

I have had high Voltage on the secondary side of something, that is 'floating', due to inter-winding capacitance maybe, but was a long time ago so can't remember the details. I mean relative to mains earth.

Ideally you want to connect the secondary side ground to mains earth, and have some sort of overload protection, e.g. a fuse, in case you get a short between HT rail (or whatever) and the case, so the case needs to be mains earthed.

However as Simon said this can cause an audible hum loop because there is more than one earth connection between circuit ground and mains earth, i.e. 1 = the mains earth wire to itself, 2 = the ground connection of audio lead(s) connected between it and another 'box' whose circuit 0V is also directly mains earthed.

A resistor of around 10 Ohms or maybe more placed in series between the secondary side or circuit ground, and actual mains earth, at the input to the PSU, is usually enough to kill the hum. But if there's a fuse or whatever to protect the HT or whatever in case of a short to case (which ideally should remain directly connected to mains earth), then you need to bear in mind the short circuit current will go through this resistor so it has to at least survive until the fuse blows.

For some years a lot of stuff I've made includes such a resistor as a matter of course.

HTH

I think that's where I've come to as far as understanding. Next will be to build it and test with and without that 0v reference - and also to test the effect of placing that 0v reference at different locations in the circuit. I'll include 100R resistors and measure what if any flow there is across it. I'm expecting it to 0 or very very low - drain for the inter-winding capacitance is all.

[Nick]

Ideally you want to connect the secondary side ground to mains earth, and have some sort of overload protection, e.g. a fuse, in case you get a short between HT rail (or whatever) and the case, so the case needs to be mains earthed.

Though of course, if the secondary side is floating then if it comes into contact with the case nothing happens.