ampage Tube Amps / Music Electronics	For current discussions, please visit Music Electronics Forum.

ampage archive

Vintage threads from the first ten years

previous: lion Thank for the replies Dave and Davi... -- 1/6/2000 1:15 PM

Re:Still confused – maybe Randall Aiken?

1/6/2000 2:58 PM
R.G.	Re:Still confused – maybe Randall Aiken? Think about it this way - in your mind, replace every run of "ground" wire with a resistor. The currents through these resistors make voltages proportional to the current through the resistor. The whole trick in grounding is to keep the "ground" voltages from affecting the signal. The most important technique is ground wire separation. If you have two stages that are "grounded" through the same ground wire/resistor, the ground currents mix in the wire/resistor and BOTH stages see their grounds wiggling around with the sum of the two currents. If each stage has its own separate ground wire/resistor, no mixing is possible, and the ground wire/resistor developed voltage only acts to reduce the stage gain, not to introduce interference. This is pure star grounding - every single thing that connects to "ground" is connected through a separate ground wire/resistor. I haven't read Randall's sendup on grounding yet, so I'm only guessing about the two level stuff, but I'd guess he means something like the following. Within one stage or even sometimes two stages, the ground currents are related. Certainly, the grid leak resistors and cathode connections can be connected together because there is almost no grid current anyway, and the cathode ground wire/resistor voltage is proportional to and opposes the grid leak ground wire/resistor voltage, so no interference is possible, just a slight lowering of gain. These can be connected together. A volume control from the plate has the same signal voltage across it, so its ground wire/resistor current could also be added to the grid leak/cathode current without interference, just a change in gain (depending on the physical location of the volume control, we might get 60Hz hum pickup, so we might not want to do this, but then the devil is always in the details.) If there is a second stage that immediately follows the first, and neither stage clips, then the second stage ground current is out of phase with the signal in the first stage, so there is actual cancellation of the ground current AC signals, and combining the two ground currents can make the local collection of grounds even quieter. IF this stage clips, it also introduced hash into the first stage ground, so this must be done carefully. A third stage is a no-no - the gain is too high, and the ground components are certain to be in-phase with something at the first input, so we introduce the possibility of oscillation from ground currents. An unrelated stage is also verboten - the currents are unrelated and don't cancel, so all you're doing is cross-feeding signals between the two sections. What we quickly get to is that you can collect certain grounds locally to a local star-ground if you do this thoughtfully. The local stars can then be collected to the global star ground. A special case is the first power filter capacitor. It's negative side carries the pulsed currents that charge the capacitor. If you ground the transformer CT (or - output of the rectifiers in a bridge) and then connect the other star ground wires there, the capacitor charge pulses appear on the ground wire/resistors as 120 hz buzz. You must take the CT to the - side of the cap and a separate wire which then only carries the DC power to the system star ground. It's in general a bad idea to ground the CT and run a wire to the cap, leaving that as your chassis ground connection, because then the charging pulses appear on the chassis. Instead, isolate the CT, wire it to the first filter cap, run a wire to the star ground without connecting to the chassis and once all the star grounds are properly connected, run one and only one wire to the chassis. This absolutely prevents rectifier hum from getting into things through the chassis. You wind up with a two level star grounding system - Local stars for one or two stages where the currents are related and compatible, and a global star point where the local stars are connected. One of the "locals" is the (-) side of the first filter cap, because it's locally noisy, and one is the chassis, because it's an interference shield and you want it to short out any induced currents within the chassis, not convey them into the circuits. The whole trick in grounding is to mentally think of every ground wire as a resistor and then to follow where the currents go and what they do in those resistors.

1/6/2000 2:58 PM

R.G.

Re:Still confused – maybe Randall Aiken?

Think about it this way - in your mind, replace every run of "ground" wire with a resistor. The currents through these resistors make voltages proportional to the current through the resistor. The whole trick in grounding is to keep the "ground" voltages from affecting the signal.

The most important technique is ground wire separation. If you have two stages that are "grounded" through the same ground wire/resistor, the ground currents mix in the wire/resistor and BOTH stages see their grounds wiggling around with the sum of the two currents. If each stage has its own separate ground wire/resistor, no mixing is possible, and the ground wire/resistor developed voltage only acts to reduce the stage gain, not to introduce interference. This is pure star grounding - every single thing that connects to "ground" is connected through a separate ground wire/resistor.

I haven't read Randall's sendup on grounding yet, so I'm only guessing about the two level stuff, but I'd guess he means something like the following.

Within one stage or even sometimes two stages, the ground currents are related. Certainly, the grid leak resistors and cathode connections can be connected together because there is almost no grid current anyway, and the cathode ground wire/resistor voltage is proportional to and opposes the grid leak ground wire/resistor voltage, so no interference is possible, just a slight lowering of gain. These can be connected together. A volume control from the plate has the same signal voltage across it, so its ground wire/resistor current could also be added to the grid leak/cathode current without interference, just a change in gain (depending on the physical location of the volume control, we might get 60Hz hum pickup, so we might not want to do this, but then the devil is always in the details.)

If there is a second stage that immediately follows the first, and neither stage clips, then the second stage ground current is out of phase with the signal in the first stage, so there is actual cancellation of the ground current AC signals, and combining the two ground currents can make the local collection of grounds even quieter. IF this stage clips, it also introduced hash into the first stage ground, so this must be done carefully.

A third stage is a no-no - the gain is too high, and the ground components are certain to be in-phase with something at the first input, so we introduce the possibility of oscillation from ground currents. An unrelated stage is also verboten - the currents are unrelated and don't cancel, so all you're doing is cross-feeding signals between the two sections.

What we quickly get to is that you can collect certain grounds locally to a local star-ground if you do this thoughtfully. The local stars can then be collected to the global star ground.

A special case is the first power filter capacitor. It's negative side carries the pulsed currents that charge the capacitor. If you ground the transformer CT (or - output of the rectifiers in a bridge) and then connect the other star ground wires there, the capacitor charge pulses appear on the ground wire/resistors as 120 hz buzz. You must take the CT to the - side of the cap and a *separate wire* which then only carries the DC power to the system star ground. It's in general a bad idea to ground the CT and run a wire to the cap, leaving that as your chassis ground connection, because then the charging pulses appear on the chassis. Instead, isolate the CT, wire it to the first filter cap, run a wire to the star ground *without connecting to the chassis* and once all the star grounds are properly connected, run one and only one wire to the chassis. This absolutely prevents rectifier hum from getting into things through the chassis.

You wind up with a two level star grounding system - Local stars for one or two stages where the currents are related and compatible, and a global star point where the local stars are connected. One of the "locals" is the (-) side of the first filter cap, because it's locally noisy, and one is the chassis, because it's an interference shield and you want it to short out any induced currents within the chassis, not convey them into the circuits.

The whole trick in grounding is to mentally think of every ground wire as a resistor and then to follow where the *currents* go and what they do in those resistors.