• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

PS idea: isolating transformer from load

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I just saw this idea at another thread. Here is the post

http://www.diyaudio.com/forums/showthread.php?postid=1675297#post1675297

John discusses some further improvements in the further posts of that thread.

I understood the basic idea: isolating the mains and first capacitor from the system in the 10 to 20% of the cycle where the capacitor is being charged... while in the remaining 80% of the cycle both capacitors are connected and the second one can charge.

To me it seems a great idea to finally get rid of the mains induced garbage that gets in the supply (EC8010 just posted this )

What are your thoughts?

Erik
 
I see a couple of problems with this design.

As Rick Miller posted, the xfmr center tap is still connected to the output ground, providing a direct link for common mode noise from the power line. This is fixable using a low CM capacitance split bobbin xfmr.

Also, a common mode choke could be used. However, CM chokes are tricky to get right, since they will have a capacitor to ground after them, so they are series resonant at some frequency. One has to make sure this resonance does not fall into the audio band.

Second, and the biggest problem: the transistor switch is effectively connecting two voltage sources (ie = short between the two caps) when conducting. This is a definate NO-NO. Makes large noisy current spikes of its own. This can be fixed with a series resistor or inductor to smooth out the current conduction. But why bother with this at all. It is hard to get rid of noise once created.

It is far safer to just use proper filtering to get rid of the HF line noise than such a nutty switching approach. The reason most power supplies have HF noise passing thru them is the lack of inductors in the path. And even with standard E-I core inductors, one must still include a low distributed capacitance RF capable gapped ferrite core inductor in the path. Standard E-I inductors have way too much distributed capacitance in them to block RF, they only stop 60Hz or low harmonics of 60 Hz from passing.

Take a look at filters designed for switchmode supplies like Vicor's to see how well an optimal HF filter can work.
 
Hi Don

Thanks for the explanation, it makes sense to me.

The author describes the inclusion of a (small) resistor in the drain of the switching mosfet. It will help the current spikes, but off course I can't comment if they will be lower than current spikes from the diodes themselves.

Thanks for the recommendations and names on the switchmode supplies, I will check them out.

Erik
 
smoking-amp said:
I see a couple of problems with this design.

We are in audio world here, so the results are measured by what
people hear not by what they see or think.

John is tweeking his dac since 18 month averaging about an
update per week. He has tried numerous known and unknown
power supply circuits and this current approach of his seems
to work best.
I find it a little rude to call his findings nutty.
See, at the end what counts is just what has been build and tested
and not what some mind thinks about it.

-ecdesigns- said:
Hi ErikdeBest,

I read the reaction of smoking amp (Don)..........

I wouldn't have developed a "nutty" charge-transfer power supply if I could match, or outperform a battery power supply by just using a conventional power supply with (passive) filters.


stefanobilliani said:
Hello EC ,
I tryed the circuit myself and its very promising , the changes in clarity of sound and dynamics has been evident to me .

So please update the schematic when possible , I will be glad to try it .


Greets,
Klaus
 
"See, at the end what counts is just what has been build and tested
and not what some mind thinks about it."

Has he put a spectrum analyzer on the 2nd cap?
Most switching noise is not hearable directly, so I am not surprised that he doesn't hear it, but his FM radio and TV set will.

Don
 
I'm not saying that the switching scheme can't be made to work.

Putting an inductor or resistor in series with the switch would quiet it down considerably, as would controlling the turn on rate of the transistor too. The Linear Technology quiet switcher control chips use current and voltage feedback to control the switch slew rates.

One could also replace the switch with a tracking current source to transfer charge between the caps. This would be a big improvement over a switch. (Simple version: put a current limiting resistor in series with the Mosfet source terminal, gate always on, rectifier in series with drain. More sophisticated: modulate the gate voltage with the drain to source voltage via an R divider and zener gate protection)

My main point was that the switching route opens the door for noise problems, shielding issues, electrical code conformance rules and mandated independant lab testing to ensure that no one's pace maker will misfire or some radio service be interfered with. You will be running up $10,000 dollar independant lab testing bills trying to get that thing approved. DIYers don't have the test equipment to check for RF problems for home only setups.

A properly implemented RF filter should work fine for this power line noise issue, without all the extra headaches. Not that hard to do. But likely not done properly in many cases, hence giving the impression they don't work.

I've seen sophisticated military equipment, for example, that used gapped ferrite pot cores for RF DC line filters, they did not work. They failed to realize that pot core bobbins cause too much distributed winding capacitance. Single layer gapped toroid windings are required, as any switchmode power book will spell out. Also, RF rated bypass caps are required, beside the electrolytics for the 60 Hz end.

Don
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.