New build - supercharged Atomic 16 - heater questions

[flood]

Hey all, am about 50% done with this build. It also serves as an experiment with part pcb construction. The circuit is pretty much stock atomic 16 with the addition of an extra pentode gain stage for OTT gain as well as a fully buffered effects loop.

I have a question though - my power transformer supplies 3.15-0-3.15 at a whopping 5A, enough to handle the 3 odd amperes demanded with ease. I've decided to heat all preamplifier/dual triode filaments using regulated 6v dc (four 7806 regulators - two for the filaments, one for the LEDs and one for the two relays) by ignoring the centre tap.

Now the important question - can I leave this dc supply unreferenced? I still plan on heating the two EL84 filaments using the ac supply with the ct referenced to the circuit ground. I figure that using a full wave rectifier on a 6.3 VAC would drop the voltage too low to let the regulator work correctly.

So would there be any adverse effects to consider when "floating" filament dc? Thanks in advance. Will post more in a while.

[flood]

Here are the schematics for the preamp and the DC supply I was talking about. I have a board etched, drilled and ready to go....

It actually struck me that I could simply bridge 3 of the regulators and run ALL the filaments all that - by my calculations, i will lose about 30% current by using a bridge rectifier, which still gives me something around 3.5A regulated 6VDC. Any ideas on this one? I wanted to avoid bridging regulators as far as possible so that it would be easy to pinpoint a problematic one if something goes wrong somewhere down the line. i wanted to run the 5 small tube filaments off two regulators so that they'd do fine even with a smallish heatsink, and let the EL84s be supplied with AC.

Will upload the schematics for the FX loop + PI board soon. the FX loop schematic was taken from diyaudio.com, i used a 1u instead of 4.7u though for the cathode coupling cap - should still be sufficient at guitar frequencies.

[Cliff Schecht]

All of the grounds must eventually be referenced to the same point. If you leave separate grounds around the circuit, you end up with floating signals which will induce so much hum that the amp will be unusable. In your schematic, you're keeping the center tap ground and rectifier ground isolated which is a no-no, these need to meet at some point in the circuit or you'll just end up with a bunch of floating voltages because there are no current return paths for the current to flow.

The best way to keep ground noise to a minimum is to keep all grounds separate up until the post-rectifier capacitors. The capacitors are what swallow up the AC noise in the circuit and connecting all grounds only here prevents grounds from crosstalking at all.

I think you also don't realize that you are burning up over 200W through each regulator with your current configuration! These regulators can't dissipate more than 10W max which for 6V at 1A means that the biggest input voltage you can have is 16V. You need to either find a low voltage, high current source to run the heaters from or avoid the DC heater scheme completely, which IMO is more of a bandaid for bad layout and circuit design in most cases. There are hundreds of ultra-high gain amps that run the heaters off of AC so why add the extra complexity, cost and heat of regulated DC heaters? Something else to consider is that linear regulators are only about 40% efficient so you are required to run an excessively large power supply to support the regulators. I'd scrap them if I were you.


Edit: Sorry, I misread your schematic as rectifying the full 230V! Disregard what I said about the 200W number but the rest of what I said still stands..

[paulster]

Even with regulated DC, unless there's a ground reference for your heaters they'll hum like crazy.

[flood]

hmmmmm....

ok, how about this: i power all heaters off the 3.15-0-3.15 supply and use the DC circuits ONLY for the LEDs/relays? there is probably no harm in leaving those unreferenced, i suppose? or will it cause the relay magnets to go mental and switch at warp speed?

[flood]

I think you also don't realize that you are burning up over 200W through each regulator with your current configuration! These regulators can't dissipate more than 10W max which for 6V at 1A means that the biggest input voltage you can have is 16V. You need to either find a low voltage, high current source to run the heaters from or avoid the DC heater scheme completely, which IMO is more of a bandaid for bad layout and circuit design in most cases. There are hundreds of ultra-high gain amps that run the heaters off of AC so why add the extra complexity, cost and heat of regulated DC heaters? Something else to consider is that linear regulators are only about 40% efficient so you are required to run an excessively large power supply to support the regulators. I'd scrap them if I were you.


Edit: Sorry, I misread your schematic as rectifying the full 230V! Disregard what I said about the 200W number but the rest of what I said still stands..

no worries, just goes to show that i really need to label my schematics clearly sigh, premature alzheimers.... that too practically fresh out of engineering school, in germany of all places!

[paulster]

In my experience which includes running AC power tubes, DC preamp tubes and switching off the same power supply, the quietest way to do it is as follows:

Tape off the centre tap.
Run the power tubes off AC so they are essentially unreferenced (for now). If you're using push-pull then ensure that you keep the heater wiring in phase between the tubes so that you'll get some common-mode noise rejection in the OT - that's just a sensible precaution.

Run your preamp tubes off regulated DC and tie the 0Vdc connection to the chassis.

At this point your preamp tubes will be ground-referenced and quiet. Your power tubes will also be ground referenced but where they'd have been alternating between +/-3.15V and -/+3.15V with the centre-tap they'll now be alternating between 0/6.3V and 0/-6.3V. This doesn't cause a problem though as they're not overly sensitive to noise, are comparatively low gain, and you can take advantage of the common-mode rejection we mentioned earlier.

For your switching you need to decide whether to leave it ostensibly floating (it isn't really because there is a path back via the bridge rectifier diodes) or whether to tie its DC ground to the heaters DC ground.

Either will work but I've usually found that tying all the DC grounds together at the same point works best in really, really high gain circuits.

Finally, don't think about using the centre tap if you've got a 0Vdc connection tied to the chassis at all unless you love the smell of burnt silicon, as you'll short out one side of each bridge rectifier each half cycle.

[flood]

that's a fantastic suggestion, paulster. i think i'm going to double check my wiring and give this a shot.

it's amazing how much one learns on these forums. thanks!

here's a preliminary shot of the guts so far:

[IMG:453:604]http://i46.tinypic.com/2rq0z2p.jpg[/img]