Quote:
Originally Posted by Llama16 /img/forum/go_quote.gif
I did some googling about floating ground, but didn't find anything that I really understood.
Seems like a floating ground is insulated from a real ground and you shouldn't touch it.
But I don't understand the following, is a floating ground a 'ground' on a certain voltage above the normal ground? I read that when creating a floating ground on 5V, you can create a supply of 7V by using a 12V source with he 5V ground used as a normal ground (uch, I can't follow this myself).
If so, is the floating ground you mean here at 48V? As this is what the power supply gives.
And would you have an idea of what would happen when still connecting the normal ground of the amp to this floating ground? (If we take in account that the ground in the amp stays insulated from the chassis, else probably ***PZZZ***)
Dries
|
What you've referenced is actually talking about floating the zero reference relative to creating a + or - voltage for analog signals.
All sound signals in audio electronics are sine waves, so the signal varies constantly between + and -. The speed at which that wave varies gives you the frequency - in our interests, the value between 20 - 20,000 Hz.
In the Starving Student, for instance, the supply voltage is set at 48V to the tubes, but the grid (signal ground) is referenced at 19V, i.e., the ground is "floated" to 19V. At the same time, the amplification factor of the 19J6 tube is 38 at 100V. We're running it at 48V, or essentially one-half the rated plate voltage. So, the amplification factor is probably around 19. This means the tube can swing a 1V signal to + or - 19V at the strongest musical peaks. With the reference (or floated ground) set at 19V, the voltage can supply a -19V (down to 0) or a +29V (48V) - but it actually only uses 19V on that high side, too. Actually, it will begin to clip on the low side of the wave, but that clipping will be delayed on the high side until that 29V is reached.
Anyway, this means the voltage at output is 19V (zero signal or "floated ground"), as soon as you turn it on and the tubes warm up. Hence, we have to use output coupling caps to block that DC voltage level from our headphones. That's a big digression, actually, but what was really being explained in your reference. They were looking at 5V as the reference ground, so that a +7 and -5 would be available to supply analog signals.
On the other hand, when we talk about the SSMH's power supply as having a floating ground, it's an intermittent error condition. The power supplies should always be referenced back to ground zero in the wall outlet. When you use a BantamDAC, it's referenced that way all the way through the PC back to the wall outlet. Same thing for most other power supplies - even ones that are not plugged into the ground at the wall outlet (they are isolated in that case). However, one would expect that the zero point of the supplied voltage would never change.
In the case of the SSMH's power supply, I've measured as much as 0.49VDC (1/2 volt) difference between the negative terminal on the Cisco's 2.5mm supply plug and the ground reference measured on a BantamDAC that's connected to a PC's ground. That's 0.49VDC that gets sent back through the BantamDAC's ground. The PCM chip is connected directly to ground in several places, but the output coupling caps in the BantamDAC only sit on the signal lines, not the ground. So, there is no protection from this offset that may potentially develop.
BTW, that amount of voltage reference difference is trivial to most circuits with output protection, output buffers, opamp buffers, etc., which most sources already have. However, the Alien and BantamDAC are special DIY configurations to obtain the "purest" sound from the DAC chips without any intervening circuitry. (A Swenson mod on a CD player may suffer the same vulnerability.) There's not an issue using an AlienDAC or BantamDAC with any other amp that I'm aware of - but there's not an issue with a power supply moving its zero reference around, either (unless all switching supplies operate this way). That's why the problem is unique with the Starving Student.
This may also be the same reason that several of us have measured offset on the RCA jacks upon power up with the SSMH. It's the Cisco power supply "moving around" the voltage supply relative to 0VDC ground. We thought perhaps it was a back emf (something in the circuit that was causing a reverse voltage for a time), but it's the power supply itself moving the zero voltage reference around.
I'm not the best in having theoretical insight to electronic circuits, so you guys correct me if I've stated something wrong.