Greetings,
Can we regress back to the first couple pages of this thread when the discussion mentioned using a CRC filter and the benefit of a little impedance/resistance in the PS. Resistance ~ 0.22 ohms was a suggested value to place series between filter caps. I'm fairly sure this was in reference to Class A/B.
The only modeling tool I have is Duncan's PSUDII. Inputting different resistor values into a CRC filter in PSUDII I see “relatively” little difference in the ripple voltage when the resistance is varied from 1 ohm to 0.01 ohm. Of course what really smoothes out the ripple is a CRCRC filter. If PSUDII is accurate modeling these low value resistors is there some other reason to use higher resistance in the filter if it is not needed to smooth the ripple? How much resistance is too much in a Class A/B PS and how would it effect the sonic character of the amp? If 0.22 ohms is acceptable would a CRCRC with 0.1 ohm resistors be a better choice (less ripple)?
Scott
Can we regress back to the first couple pages of this thread when the discussion mentioned using a CRC filter and the benefit of a little impedance/resistance in the PS. Resistance ~ 0.22 ohms was a suggested value to place series between filter caps. I'm fairly sure this was in reference to Class A/B.
The only modeling tool I have is Duncan's PSUDII. Inputting different resistor values into a CRC filter in PSUDII I see “relatively” little difference in the ripple voltage when the resistance is varied from 1 ohm to 0.01 ohm. Of course what really smoothes out the ripple is a CRCRC filter. If PSUDII is accurate modeling these low value resistors is there some other reason to use higher resistance in the filter if it is not needed to smooth the ripple? How much resistance is too much in a Class A/B PS and how would it effect the sonic character of the amp? If 0.22 ohms is acceptable would a CRCRC with 0.1 ohm resistors be a better choice (less ripple)?
Scott
Hi Scott,
I am surprised that RCRC with 0r22 ohms in the second R is showing significantly poorer results than RCRCRC with 0r1 for the second and third R.
The first R is the resistance in the wire connections and the transformer windings. PSUD2 takes this into account if you correctly specify the transformer.
I have found that the final C value should be as big as a C only PSU. This ensures that the amp sees a PSU source of lowish impedance and thus fairly low signal induced ripple on the rails.
I am surprised that RCRC with 0r22 ohms in the second R is showing significantly poorer results than RCRCRC with 0r1 for the second and third R.
The first R is the resistance in the wire connections and the transformer windings. PSUD2 takes this into account if you correctly specify the transformer.
I have found that the final C value should be as big as a C only PSU. This ensures that the amp sees a PSU source of lowish impedance and thus fairly low signal induced ripple on the rails.
scottw said:
Hi Scott,
The CRC approach is not strictly for ripple reduction per se. It is to better filter out the high frequency components of the ripple, and HF noise. The small resistance of 0.22 ohms also helps us to force the garbage currents to flow the way we want them to. Take the case of a 0.22 ohm series resistor and a 10,000 uF final capacitor. It will form a corner at about 80 Hz, so it will not do much for the fundamental of the ripple at 120 Hz; but it will really round the edges and attenuate the harmonics. Remember, well designed amplifiers have very good power supply rejection at low frequencies, but the PSRR of almost any amplifier begins to degrade at higher frequencies.
Bob
I was wondering where that first R came from that you mentioned. Entered the values actually measured from the tranny so I think it is right or at least close (not that I really know what I'm doing). From what I entered, PSUD2 calculated a source resistance of 0.138 ohms for a 400va toroid with 20vac secondaries.
That's why I was wondering if PSUD2 was accurate modeling the low value resistance. I don't have any big caps handy or I'd breadboard something and measure the ripple with different series R.
Scott
Not to throw too big a wrench into this discussion (which I have found to be very enlightening, BTW), but how would the use of 4 pole caps influence the sort of arrangement (CRC, RCRC, etc) that should be used? Or, IOW, what would the optimal circuit arrangement be pre- and post- regulator if using 4 poles?
Bob, look here
http://www.jensencapacitors.com/audio/products/electrolytic.html
About halfway down there is the description of their 4 Pole caps.
http://www.jensencapacitors.com/audio/products/electrolytic.html
About halfway down there is the description of their 4 Pole caps.
tonyptony said:
Thanks for the tip. I looked it up and did a little research. These are electrolytic capacitors with two inputs and two outputs, and the supply currents flow through the capacitor itself. In a sense, it appears that it forms a distributed RLC low pass filter. As a result, suppression of transmission of HF hash etc is greatly reduced, much superior to just having a conventional 2 pole capacitor in shunt with a very good ESR.
They are made by Jensen Capacitors, and I did not find any other current supplier of four pole capacitors. They were originally developed for switching power supplies, and appear to be based on Patent 3,988,650, granted 10/26/76 and assigned to Siemens.
They are expensive, however. The Jensen caps are targeted specifically at high-end audio, and their dealers are high-end audio parts sources. A 4700 uF, 63V cap costs $37. A 10,000 uF, 80V cap costs $53. A 1000 uF, 200V cap costs $51.
I'd love to use them if they were a little less expensive. The concept seems clean and elegant. It is an integrated topological solution to a difficult problem. The first place I would use them would be as the smaller local reservoir capacitor located near the output transistors. Indeed, using one as the rail-to-rail X capacitor and passing both rails through one of these caps would help isolate HF hash in BOTH directions. For this, one would probably use the 1000 uF, 200V cap.
Cheers,
Bob
Mr Cordell,
Frako produced the 4-pole electrolytics for only a couple of years in the 1980s.
The Frako capacitors didn't look like much from the exterior, not much of a market for them back then, and even fewer people knew what a 4-pole capacitor was at the time.
The 4-pole caps were not excessively higher priced than regular caps, from the top of my head i think i paid about twice as much for Frako 4-Ps compared to regular good quality lytics.
e.g., Siemens Sikorels i bought at the same wholesaler in Germany were more expensive.
The point i was trying to make is that it would not surprise me if the original manufacturer starts production again if the market is ready for a 4-pole electrolytic cap this time.
Frako capacitors has gone through some corporate restructuring over the last decade, it's big enough for it.
Frako produced the 4-pole electrolytics for only a couple of years in the 1980s.
The Frako capacitors didn't look like much from the exterior, not much of a market for them back then, and even fewer people knew what a 4-pole capacitor was at the time.
The 4-pole caps were not excessively higher priced than regular caps, from the top of my head i think i paid about twice as much for Frako 4-Ps compared to regular good quality lytics.
e.g., Siemens Sikorels i bought at the same wholesaler in Germany were more expensive.
The point i was trying to make is that it would not surprise me if the original manufacturer starts production again if the market is ready for a 4-pole electrolytic cap this time.
Frako capacitors has gone through some corporate restructuring over the last decade, it's big enough for it.
Bob Cordell said:
Given this, what would be the closest discrete equivalent to what a four pole can do?
jacco vermeulen said:
But jacco, taking price out of the discussion, I guess my curiosity is based on whether these 4 poles can do things in a way better than their discrete equivalents. Any thoughts?
Hard for me to say, i'm not a RF guy and even had to visit a friend for access to a scope during the Frako 4-P years.
Constructing Hiraga/Kaneda and serious class A Hitachi MOSFET circuit amplifiers was very popular overhere.
CLC/CRC powersupply setups and extensive foil cap bypassing has been a regular routine since the mid 80s in these parts.
Personally, i was into building very high bandwidth class A amps at the time.
For output stage powersupplies the 4-poles were too small, i tried them on the front end PS.
(up to date i've not constructed a PP MOSFET design with a single rail for front end and output stage)
By ear, i settled for the Siemens Sikorel, which has also been designed for switching PS use and the reason for mentioning it earlier. Like the Kwak clock guy, my hearing has Veto privileges, Elso told me he also likes the Sikorels.
I've had a personal briefing by Dieter Burmester on the importance of RF shielding, keeping power supply capacitors as close as possible to the circuits and paying much attention to wiring.
Nelson Pass, whom i've been eyeballing as a fetishist hawk for over 2 decades, has demonstrated that it is possible to design marvellous sounding products with mediocre devices such as MJ15011/MJ15012s in his Threshold SA amps and continues to do so with vertical MOSFETS in his PLabs stuff.
Technically, yeah. But my guess is that you need to be one heck of a designer to fully take advantage of the merits of 4-pole capacitors.
If you're not, you'll be paying a fools price for hardware that doesn't deliver and you'll be an audio parts fetishist. Bottomline, if someone such as the likes of Mr Cordell does the design including the layout, 4-poles might be the way to go. Just sticking it in, as demonstrated by the average diyer with so much exotic hardware, is a waste of cold hard cash.
Constructing Hiraga/Kaneda and serious class A Hitachi MOSFET circuit amplifiers was very popular overhere.
CLC/CRC powersupply setups and extensive foil cap bypassing has been a regular routine since the mid 80s in these parts.
Personally, i was into building very high bandwidth class A amps at the time.
For output stage powersupplies the 4-poles were too small, i tried them on the front end PS.
(up to date i've not constructed a PP MOSFET design with a single rail for front end and output stage)
By ear, i settled for the Siemens Sikorel, which has also been designed for switching PS use and the reason for mentioning it earlier. Like the Kwak clock guy, my hearing has Veto privileges, Elso told me he also likes the Sikorels.
I've had a personal briefing by Dieter Burmester on the importance of RF shielding, keeping power supply capacitors as close as possible to the circuits and paying much attention to wiring.
Nelson Pass, whom i've been eyeballing as a fetishist hawk for over 2 decades, has demonstrated that it is possible to design marvellous sounding products with mediocre devices such as MJ15011/MJ15012s in his Threshold SA amps and continues to do so with vertical MOSFETS in his PLabs stuff.
Technically, yeah. But my guess is that you need to be one heck of a designer to fully take advantage of the merits of 4-pole capacitors.
If you're not, you'll be paying a fools price for hardware that doesn't deliver and you'll be an audio parts fetishist. Bottomline, if someone such as the likes of Mr Cordell does the design including the layout, 4-poles might be the way to go. Just sticking it in, as demonstrated by the average diyer with so much exotic hardware, is a waste of cold hard cash.
jacco vermeulen said:
I tend to agree with all that you have said, although I don't think it would be as hard to properly apply the 4-pole capacitors as you suggested (but thanks for the compliment
).I also don't yet have a good feeling for how close we can come to the same behavior by discrete means. In other words, is the four-pole capacitor really just a convenience that saves parts count and space, and is elegant? Or does it, by its integrated, distributed filtering nature, offer performance that cannot be approached by a reasonable number of interconnected discrete components?
Cheers,
Bob
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.