Hi stefanobilliani,
D1 ... D4 are Schottky diodes. D1 and D3 were already in the existing power supplies to minimize rectifier diode switching noise. D3 was added to create a pulsed DC signal for switching the MOSFET. Other advantage is reduced voltage drop. D4 was added to reduce switching noise on the gate when the 12V zener-diode D3 limits.
However, the circuit should work fine with plain silicon rectifier diodes (1N4007) as the diode switching noise is being blocked by the charge-transfer circuit. FET T1 switches-ON when the rectifier diodes are no longer conducting. When the diodes conduct, the FET has already been switched-OFF.
Parts for the posted schematic:
D1 ... D3 = 11DQ10
D5 = BAT85
D4 = 12V zener-diode 400mW
T1 = 2SJ380 (or comparable P-channel MOSFET)
C1, C2 = 2200uF
Drain series resistor 0.5 ... 4.7 Ohms (reduces charge-transfer surge current).
I connected a capacitance multiplier between both charge-transfer circuit and voltage regulator. The capacitance multiplier blocks the low-frequency ripple current introduced by the charge-transfer circuit. The connected voltage regulator now runs on a clean DC input voltage.
D1 ... D4 are Schottky diodes. D1 and D3 were already in the existing power supplies to minimize rectifier diode switching noise. D3 was added to create a pulsed DC signal for switching the MOSFET. Other advantage is reduced voltage drop. D4 was added to reduce switching noise on the gate when the 12V zener-diode D3 limits.
However, the circuit should work fine with plain silicon rectifier diodes (1N4007) as the diode switching noise is being blocked by the charge-transfer circuit. FET T1 switches-ON when the rectifier diodes are no longer conducting. When the diodes conduct, the FET has already been switched-OFF.
Parts for the posted schematic:
D1 ... D3 = 11DQ10
D5 = BAT85
D4 = 12V zener-diode 400mW
T1 = 2SJ380 (or comparable P-channel MOSFET)
C1, C2 = 2200uF
Drain series resistor 0.5 ... 4.7 Ohms (reduces charge-transfer surge current).
I connected a capacitance multiplier between both charge-transfer circuit and voltage regulator. The capacitance multiplier blocks the low-frequency ripple current introduced by the charge-transfer circuit. The connected voltage regulator now runs on a clean DC input voltage.
-ecdesigns- said:
Hello
I don't have this amp but I have a photo of the Eagle 7A circuit, I had red circle a part who seem to be a triac in the power supply.
Bye
Gaetan
An externally hosted image should be here but it was not working when we last tested it.
Hi JC951t,
Thanks for the compliment.
Well I don't know if I am that smart, I still have a lot to learn.
Electronics has been my hobby, since I was very young. I am 47 years old now, so I should have had plenty of time to gather some experience.
During my work at a well known Dutch copier company, I had to follow a creativity training, perhaps this has learned me to think "out of the box".
Back to the charge-transfer power supply, Here is what I figured out so far, the mains transformer has an extended bandwidth (comparable to a tube power amplifier output transformer), so frequencies well above 50 or 60 Hz are passed with little problems. I built a small tube amplifier (2 x EL84) using a mains transformer for the output transformer, and almost the entire audio frequency range passed. I was able to get an output signal up to approx. 100 KHz (the frequency response wasn't flat of course).
This means that depending on transformer properties, higher frequencies are just as easily transformed as the 50 or 60 Hz mains frequency. That's one way how the interference signals end up on the secondary winding and pollute the derived power supply voltages.
This pollution is injected in bursts with 16.6 ms or 20 ms intervals. These bursts are created during the time the diodes conduct (closed circuit). These (RF / HF) signal bursts are attenuated by decoupling capacitors (electrolytic capacitors). The presence of these (RF / HF) interference currents may well explain the effect of decoupling caps on sound quality.
But that's not all, there is also a capacitance between both primary and secondary transformer windings. I checked some transformers with my capacitance meter and measured values between 45pF (small transformer) and 524pF (large transformer). This basically means that the mains voltage is directly connected to the secondary through a 45 ... 524pF capacitor, this enables a low impedance current path for both RF and HF frequencies.
When you interconnect 2 devices (DAC and power amplifier) using an RCA interlink for example, a RF / HF path is created: mains > transformer #1, ground + power supply rails, ground + signal (RCA interlink), ground + power supply rails > transformer #2 > mains. These interference currents are then added to the audio signal and distort it, even when the DAC produces a perfect audio signal.
So the sound quality of a DAC may well depend on the power supply properties of connected device(s) like power or control amplifiers. Basically any galvanic or capacitive connection to other devices may introduce unwanted interference.
The charge-transfer power supply prevents the injection of transformed mains noise during the time the rectifier diodes conduct. But the problem of the capacitive coupling needs to be solved too (the electronic switch needs to have lowest possible capacitance). Measurements showed approx. 350pF between source and drain of a 2SJ380, that's in the same magnitude as the capacity of the mains transformer. So I added a Schottky diode in series with the drain, now the capacitance dropped to approx. 1.5pF.
And that's exactly what I did with the charge-transfer power supplies in the DI4T, I also modified the 200V anode supply for the tube amplifier. So both stepped passive volume controller and entire DI4T are now running on improved charge-transfer power supplies.
Thanks for the compliment.
Well I don't know if I am that smart, I still have a lot to learn.
Electronics has been my hobby, since I was very young. I am 47 years old now, so I should have had plenty of time to gather some experience.
During my work at a well known Dutch copier company, I had to follow a creativity training, perhaps this has learned me to think "out of the box".
Back to the charge-transfer power supply, Here is what I figured out so far, the mains transformer has an extended bandwidth (comparable to a tube power amplifier output transformer), so frequencies well above 50 or 60 Hz are passed with little problems. I built a small tube amplifier (2 x EL84) using a mains transformer for the output transformer, and almost the entire audio frequency range passed. I was able to get an output signal up to approx. 100 KHz (the frequency response wasn't flat of course).
This means that depending on transformer properties, higher frequencies are just as easily transformed as the 50 or 60 Hz mains frequency. That's one way how the interference signals end up on the secondary winding and pollute the derived power supply voltages.
This pollution is injected in bursts with 16.6 ms or 20 ms intervals. These bursts are created during the time the diodes conduct (closed circuit). These (RF / HF) signal bursts are attenuated by decoupling capacitors (electrolytic capacitors). The presence of these (RF / HF) interference currents may well explain the effect of decoupling caps on sound quality.
But that's not all, there is also a capacitance between both primary and secondary transformer windings. I checked some transformers with my capacitance meter and measured values between 45pF (small transformer) and 524pF (large transformer). This basically means that the mains voltage is directly connected to the secondary through a 45 ... 524pF capacitor, this enables a low impedance current path for both RF and HF frequencies.
When you interconnect 2 devices (DAC and power amplifier) using an RCA interlink for example, a RF / HF path is created: mains > transformer #1, ground + power supply rails, ground + signal (RCA interlink), ground + power supply rails > transformer #2 > mains. These interference currents are then added to the audio signal and distort it, even when the DAC produces a perfect audio signal.
So the sound quality of a DAC may well depend on the power supply properties of connected device(s) like power or control amplifiers. Basically any galvanic or capacitive connection to other devices may introduce unwanted interference.
The charge-transfer power supply prevents the injection of transformed mains noise during the time the rectifier diodes conduct. But the problem of the capacitive coupling needs to be solved too (the electronic switch needs to have lowest possible capacitance). Measurements showed approx. 350pF between source and drain of a 2SJ380, that's in the same magnitude as the capacity of the mains transformer. So I added a Schottky diode in series with the drain, now the capacitance dropped to approx. 1.5pF.
And that's exactly what I did with the charge-transfer power supplies in the DI4T, I also modified the 200V anode supply for the tube amplifier. So both stepped passive volume controller and entire DI4T are now running on improved charge-transfer power supplies.
Hi gaetan8888,
When using a power amplifier with +/- power supply, and implement a charge-transfer power supply, you will need at least 2 switching elements and 4 large smoothing capacitors. So the Eagle 7A doesn't seem to use this type of power supply. The triac might be part of a voltage regulator circuit that stabilizes the amplifier power supply. Or it could just be part of some kind of protection circuit.
When using a power amplifier with +/- power supply, and implement a charge-transfer power supply, you will need at least 2 switching elements and 4 large smoothing capacitors. So the Eagle 7A doesn't seem to use this type of power supply. The triac might be part of a voltage regulator circuit that stabilizes the amplifier power supply. Or it could just be part of some kind of protection circuit.
Hi John,
At work I strive to acheive similar objectives as your approach
but as we all know when it comes to dollars & cents, we can
only attain a certain percentage of it.
Regards to transformers maybe that's one of the reasons that
I found that I prefer EI trans over Toriods. I remember reading
somewhere that Toriods are considered wideband. Also would the use of common mode chokes provide the same advantage
as your approach ?
Thanks
At work I strive to acheive similar objectives as your approach
but as we all know when it comes to dollars & cents, we can
only attain a certain percentage of it.
Regards to transformers maybe that's one of the reasons that
I found that I prefer EI trans over Toriods. I remember reading
somewhere that Toriods are considered wideband. Also would the use of common mode chokes provide the same advantage
as your approach ?
Thanks
Hi Telstar,
I think it's similar:
mains > transformer #1 > power supply & GND > XLR interlink (GND + signals) > power supply & GND > transformer #2 > mains.
The advantage over RCA interlinks is that the common noise on the XLR interlink (especially when using long interlinks), caused by external electromagnetic fields (hum) is attenuated.
The mains interference could still pass through both interconnected devices, because they are likely to have conventional mains power supplies installed. You usually have to connect at least 2 devices (DAC and power amplifier for example), so the problem is already created. The following example could be a typical configuration:
Computer > USB > DAC > pre-amplifier > power amplifier, or CD transport > SPDIF coax > DAC > pre-amplifier > power amplifier.
Note that every device needs a power supply, and this is usually a conventional mains power supply. When multiple devices are interconnected without galvanic insulation, multiple unwanted interference paths are created. In other words, the more equipment you hook-up, the worse it might get.
I think it's similar:
mains > transformer #1 > power supply & GND > XLR interlink (GND + signals) > power supply & GND > transformer #2 > mains.
The advantage over RCA interlinks is that the common noise on the XLR interlink (especially when using long interlinks), caused by external electromagnetic fields (hum) is attenuated.
The mains interference could still pass through both interconnected devices, because they are likely to have conventional mains power supplies installed. You usually have to connect at least 2 devices (DAC and power amplifier for example), so the problem is already created. The following example could be a typical configuration:
Computer > USB > DAC > pre-amplifier > power amplifier, or CD transport > SPDIF coax > DAC > pre-amplifier > power amplifier.
Note that every device needs a power supply, and this is usually a conventional mains power supply. When multiple devices are interconnected without galvanic insulation, multiple unwanted interference paths are created. In other words, the more equipment you hook-up, the worse it might get.
Hi ec,
Looking at you schematic it seems that the negative side of the supply is connected directly to the center tap of the secondary of the transformer. Would not this be a path way for common mode noise? The plus side of the circuit is isolated by the 2SJ380, but the minus side has no isolation to noise coupled across the transformers primary to secondary capacitance.
Rick
Looking at you schematic it seems that the negative side of the supply is connected directly to the center tap of the secondary of the transformer. Would not this be a path way for common mode noise? The plus side of the circuit is isolated by the 2SJ380, but the minus side has no isolation to noise coupled across the transformers primary to secondary capacitance.
Rick
I thought about that a few years ago... never had time to actually implement it though. It's been patented a few times like everything else... there are even a few boutique shops selling stuff like that (dont' remember the name, is it offline PSU or never-connected PSU or flywheel PSU...) who cares, if it sound good, just build it, and enjoy it ! I'm curious about your implementation, though.
Hi ampliton,
Thanks for the reply, I did look at that post and the schematic shows the same thing that I brought up before, the center tap is ground. If one uses the center tap all the common mode noise that passes thru the transformer is on the signal ground. I prefer to use two separate secondary windings and two diode bridges to reduce common mode noise that a center tap design can produce. Of course I am not using the charge coupled design that ecdesigns is using. There should be a way to use two separate secondary winding in his design and avoid the center tap. I like his use of another diode in series with the output to reduce the capacitance for less noise on the output.
Thanks for the reply, I did look at that post and the schematic shows the same thing that I brought up before, the center tap is ground. If one uses the center tap all the common mode noise that passes thru the transformer is on the signal ground. I prefer to use two separate secondary windings and two diode bridges to reduce common mode noise that a center tap design can produce. Of course I am not using the charge coupled design that ecdesigns is using. There should be a way to use two separate secondary winding in his design and avoid the center tap. I like his use of another diode in series with the output to reduce the capacitance for less noise on the output.
Hi JC951t,
I tried chokes up to approx. 100 Henry in order to solve the power supply problem (using a battery power supply as reference), they appear to be rather ineffective when compared to the simple charge-transfer circuit.
It seems to be very important to interrupt the direct charge current loop between both transformer and load, and that the electronic switch has lowest possible capacitance when switched-OFF.
I tried both EI core, and toroidal transformers with similar rating with the charge-transfer power supply, no detectable change in resulting sound quality. Without the charge-transfer power supply there was definitely a change in sound quality.
I also measured 2SJ380 MOSFET d-s capacitance, it's around 350pF. That's why I added a 11DQ10 Schottky diode to lower this capacitance. I later measured the capacitance of various diodes I had laying around:
BAT42 small signal Schottky diode, 8.8 pF (limited peak and continuous current)
1N4937 high-speed silicon switching diode, 13pF
1N4007 general purpose silicon diode, 13.6 pF
BYV27-400 fast-slow recovery silicon diode, 74.6 pF
11DQ10 Schottky diode, 77 pF
Then it seemed like I picked the wrong diode, but Schottky diodes produce less switching noise, that's a fact. So why not use the best of both worlds and combine a Schottky diode with (a number of) diodes that have lower capacitance.
I was able to get down to approx. 2pF (measured with a HAMEG capacitance meter) when combining a 11DQ10, and some 1N4937 diodes. The decreased capacitance also resulted in better subjective sound quality (can be easily checked by bypassing some diodes).
I was curious what would happen when GND was interrupted as well, and designed, built and tested a circuit that should provide full isolation (3 MOSFETs and one transistor). But the HF interference at the output was worse than when using the simple version with only one MOSFET.
I tried chokes up to approx. 100 Henry in order to solve the power supply problem (using a battery power supply as reference), they appear to be rather ineffective when compared to the simple charge-transfer circuit.
It seems to be very important to interrupt the direct charge current loop between both transformer and load, and that the electronic switch has lowest possible capacitance when switched-OFF.
I tried both EI core, and toroidal transformers with similar rating with the charge-transfer power supply, no detectable change in resulting sound quality. Without the charge-transfer power supply there was definitely a change in sound quality.
I also measured 2SJ380 MOSFET d-s capacitance, it's around 350pF. That's why I added a 11DQ10 Schottky diode to lower this capacitance. I later measured the capacitance of various diodes I had laying around:
BAT42 small signal Schottky diode, 8.8 pF (limited peak and continuous current)
1N4937 high-speed silicon switching diode, 13pF
1N4007 general purpose silicon diode, 13.6 pF
BYV27-400 fast-slow recovery silicon diode, 74.6 pF
11DQ10 Schottky diode, 77 pF
Then it seemed like I picked the wrong diode, but Schottky diodes produce less switching noise, that's a fact. So why not use the best of both worlds and combine a Schottky diode with (a number of) diodes that have lower capacitance.
I was able to get down to approx. 2pF (measured with a HAMEG capacitance meter) when combining a 11DQ10, and some 1N4937 diodes. The decreased capacitance also resulted in better subjective sound quality (can be easily checked by bypassing some diodes).
I was curious what would happen when GND was interrupted as well, and designed, built and tested a circuit that should provide full isolation (3 MOSFETs and one transistor). But the HF interference at the output was worse than when using the simple version with only one MOSFET.
Hi peufeu,
I didn't have much time either, still I designed, built, and tested the circuit within a few days (that happens when I get curious). Seems like most electronic circuits have already been invented, so there is a relative big chance of re-inventing them.
I wished I thought about these power supply improvements much sooner, but it seems that the power supply always comes last when designing something, so it will usually get least attention.
If you are curious about my charge-transfer power supply version, just build some and try them. The schematic is simple, and the parts are cheap.
I modified my DI4T power supplies within half an hour using spiderweb construction. My DI4T now looks like a mess, but it sounds better than ever.
Don't forget the diodes and current limiting resistor in series with the MOSFET drain.
Schematic for negative voltage can be easily derived, using a N-Channel MOSFET like the 2SK1350 for example, and carefully observing both diode and capacitor polarity.
I didn't have much time either, still I designed, built, and tested the circuit within a few days (that happens when I get curious). Seems like most electronic circuits have already been invented, so there is a relative big chance of re-inventing them.
I wished I thought about these power supply improvements much sooner, but it seems that the power supply always comes last when designing something, so it will usually get least attention.
If you are curious about my charge-transfer power supply version, just build some and try them. The schematic is simple, and the parts are cheap.
I modified my DI4T power supplies within half an hour using spiderweb construction. My DI4T now looks like a mess, but it sounds better than ever.
Don't forget the diodes and current limiting resistor in series with the MOSFET drain.
Schematic for negative voltage can be easily derived, using a N-Channel MOSFET like the 2SK1350 for example, and carefully observing both diode and capacitor polarity.
Hi Rick Miller,
Yes, the minus is still connected, using a fully insulated charge-transfer power supply didn't seem to provide expected improvements. Hooking up multiple mains-powered High-End audio devices is rather problematic, the interference currents appear to be all over the place. It's very difficult to measure these multiple interference paths using an oscilloscope that also runs on a conventional power supply. The oscilloscope will interact with the interference signals itself, making it very difficult to perform reliable measurements.
It seems that the mains power supplies are now much more polluted than say 20 years ago, thanks to modern switch-mode power supplies and wireless applications. It would be very revealing to count all electronic devices that have either a switch-mode power supply, or emit electromagnetic radiation in one's house. Most electronic devices like television sets, computers, and satellite receivers will continue to pollute the mains power supply when placed in stand-by mode. The cheap (and usually poorly filtered) energy saving lamps are also highly effective in polluting the mains voltage.
Yes, the minus is still connected, using a fully insulated charge-transfer power supply didn't seem to provide expected improvements. Hooking up multiple mains-powered High-End audio devices is rather problematic, the interference currents appear to be all over the place. It's very difficult to measure these multiple interference paths using an oscilloscope that also runs on a conventional power supply. The oscilloscope will interact with the interference signals itself, making it very difficult to perform reliable measurements.
It seems that the mains power supplies are now much more polluted than say 20 years ago, thanks to modern switch-mode power supplies and wireless applications. It would be very revealing to count all electronic devices that have either a switch-mode power supply, or emit electromagnetic radiation in one's house. Most electronic devices like television sets, computers, and satellite receivers will continue to pollute the mains power supply when placed in stand-by mode. The cheap (and usually poorly filtered) energy saving lamps are also highly effective in polluting the mains voltage.
John
The charge-transfer power supply seems like a fantastic implementation. Enthousiastac as I am
I just posted your explanation on the charge-transfer power supply in the tubes forum, to see what people over there think about it.
http://www.diyaudio.com/forums/showthread.php?postid=1681187#post1681187
Erik
The charge-transfer power supply seems like a fantastic implementation. Enthousiastac as I am
I just posted your explanation on the charge-transfer power supply in the tubes forum, to see what people over there think about it. http://www.diyaudio.com/forums/showthread.php?postid=1681187#post1681187
Erik