Power Supply Resevoir Size - Page 121 - diyAudio
Go Back   Home > Forums > Amplifiers > Power Supplies

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 16th September 2012, 12:13 AM   #1201
gootee is offline gootee  United States
diyAudio Member
 
Join Date: Nov 2006
Location: Indiana
Blog Entries: 1
Quote:
Originally Posted by Nico Ras View Post
Harrison, would you quantify their individual results so that it can be used as a measure here, I think the first oscilloscope traces from Routhun at SYMEF amplifier explains a whole lot, if there are others that could provide the same a very definite conclusion can be drawn.

Besides that I have been lurking in the dark shadows on your thread to see what is brewing, out of pure interest for where it is going and I would not miss the all revealing formula for anything in the world.

I am not joking - there are those who can successfully market any product and this may be one golden method so I am not laughing at the back of your head. Not yet anyway.
Does anyone know what the Volt-Amp and Output Vrms ratings were, of the transormer used for that square wave test?
  Reply With Quote
Old 16th September 2012, 12:28 AM   #1202
diyAudio Member
 
Join Date: Mar 2008
Tom @ #1172,

Quote:
Originally Posted by gootee View Post
Even with planes, the problem that can partially ruin the improvement from paralleling is when the conductors are no longer completely separate, i.e. the currents have to share some length of conductor. Then there would be MUTUAL inductance. And that wrecks the algebra somewhat so that the inductances don't fully reduce from paralleling.
Tom, thats not actually correct - there is mutual inductance regardless of whether or not they share some length of conductor, as long as the flux of one current-carrying loop cuts another current-carrying loop (think loop antennas).

If the currents share some length of conductor then there is more mutual inductance - this forces the flux of one loop to cut the other - it has to, as they share a common conductor.

This shared-conductor method is actually quite common in small loop antennas, and is used for impedance transformation. The transmitter drives a small (often rectangular) loop, which shares a segment with the main transmitting loop, making what is basically an air-cored transformer (hence the impedance transformation)

I should have pointed this out some time ago, but forgot. sorry, my bad.

Unfortunately this kind of ruins your "n parallel sets of wiring" approach - in order for that method to work in practice, you must ensure that there is little or no magnetic coupling between the parallel sets of wiring.

This also occurs when you parallel a pair of film capacitors - the overall ESL does go down, but it does not halve. see the attached plot. I paralleled a pair of 5.9nH 100nF Panasonic ECQE caps, and the resultant inductance was 4nH, not 3nH as you might expect. the caps were hard against each other (6mm between pins), and I calibrated out the test fixture PCB they were mounted on.

interestingly enough this doesnt really happen when you parallel electrolytics (on a PCB set up as a parallel-plate transmission line) as most of the ESL is inside an AL can, which does a pretty good job of eddy-current shielding (what with ESL effects occurring at high frequencies). If, however, you have a horrible wiring layout (ring terminals and widely spaced wires) rather than a well laid out DS-PCB, then all the wiring inductances mutually couple (IOW with electrolytics the overall ESL is dominated by the wiring/pcb layout)
Attached Images
File Type: jpg paralleled ECQE capacitors.jpg (272.7 KB, 229 views)
  Reply With Quote
Old 16th September 2012, 12:47 AM   #1203
diyAudio Member
 
Join Date: Nov 2010
Location: The City, SanFrancisco
Quote:
Originally Posted by gootee View Post
"Parallel decoupling capacitances", taken to one of its logical conclusions:
Gootee

With the common tie at both ends of the wire pairs your relying on mutual inductance to persuade the current paths, much like parallel planes would.
Would be interested in seeing the effectiveness of this approach for audio.
Too much math for a weekend, will follow up though.

Thanks
-Antonio
  Reply With Quote
Old 16th September 2012, 12:53 AM   #1204
diyAudio Member
 
Join Date: Mar 2008
Folks,
in theory you can indeed play with the wiring and get things to cancel out. In practice however you cannot - believe me I've tried. unless you can accurately model the entire physical assembly (eg PCB parasitic extraction) you will have little or no idea exactly what mutual couplings you have where. Even better is to be able to accurately measure the overall impedances in-situ.

I spent about 3 months working on this a couple of years ago - it was a DM filter for a PFC flyback smps. the first 8-10 weeks were spent fiddling around with various stupid layouts, measuring the results with EMI scans, and not getting anywhere. then I spent a bit of money and bought an HP-35676 50R signal divider for my HP3577A network analyzer, bought/made some calibration standards and started to accurately measure the actual impedances on assembled PCBs.

Once I could accurately, repeatably measure down below 1nH (up to 200MHz) I made significant progress in 2 weeks. Until then, I got nowhere - even though I've got 15yrs experience in this area, guessing my actual parasitics just wasnt good enough.

See the attached plots. The red trace on "EMI filter 1.jpg" was where I started, and I managed to get to the black trace after 8 weeks - all I managed to do was make a 20dB improvement between 600kHz and 2MHz, but all my efforts had no effect whatsoever above 2MHz.

Once I could actually measure the in-situ impedances though, I got to the red trace on "EMI Filter 2.jpg" in 1 PCB revision. I took those plots using the same inductors and capacitors - I de-soldered them from the rev 1 PCB and soldered them onto the rev 3 PCB. t took 2 weeks, but 1 week was waiting for PCBs, and about 0.5 weeks learning how to make the precise measurements. it really only took 1 day to get an extra 60dB attenuation at 30MHz.

This is why I harp on about parallel-plate transmission lines for multiple paralleled caps. all you need to do is not chop great big holes in your V+(V-) & 0V planes and voila, you get an excellent, broadband layout. try pulling clever tricks without measuring anything and you just wont (think wheelstands and endos at 100mph with your eyes shut. how well is that likely to turn out)
Attached Images
File Type: jpg EMI filter 1.jpg (195.7 KB, 218 views)
File Type: jpg EMI filter 2.jpg (252.5 KB, 210 views)
  Reply With Quote
Old 16th September 2012, 12:53 AM   #1205
gootee is offline gootee  United States
diyAudio Member
 
Join Date: Nov 2006
Location: Indiana
Blog Entries: 1
Quote:
Originally Posted by magnoman View Post
Gootee

With the common tie at both ends of the wire pairs your relying on mutual inductance to persuade the current paths, much like parallel planes would.
Would be interested in seeing the effectiveness of this approach for audio.
Too much math for a weekend, will follow up though.

Thanks
-Antonio
I simulated it. See links near beginning of this thread.
  Reply With Quote
Old 16th September 2012, 12:54 AM   #1206
gootee is offline gootee  United States
diyAudio Member
 
Join Date: Nov 2006
Location: Indiana
Blog Entries: 1
Quote:
Originally Posted by gootee
Even with planes, the problem that can partially ruin the improvement from paralleling is when the conductors are no longer completely separate, i.e. the currents have to share some length of conductor. Then there would be MUTUAL inductance. And that wrecks the algebra somewhat so that the inductances don't fully reduce from paralleling.
Quote:
Originally Posted by Terry Given View Post
Tom, thats not actually correct - there is mutual inductance regardless of whether or not they share some length of conductor, as long as the flux of one current-carrying loop cuts another current-carrying loop (think loop antennas).

If the currents share some length of conductor then there is more mutual inductance - this forces the flux of one loop to cut the other - it has to, as they share a common conductor.

This shared-conductor method is actually quite common in small loop antennas, and is used for impedance transformation. The transmitter drives a small (often rectangular) loop, which shares a segment with the main transmitting loop, making what is basically an air-cored transformer (hence the impedance transformation)

I should have pointed this out some time ago, but forgot. sorry, my bad.

Unfortunately this kind of ruins your "n parallel sets of wiring" approach - in order for that method to work in practice, you must ensure that there is little or no magnetic coupling between the parallel sets of wiring.

This also occurs when you parallel a pair of film capacitors - the overall ESL does go down, but it does not halve. see the attached plot. I paralleled a pair of 5.9nH 100nF Panasonic ECQE caps, and the resultant inductance was 4nH, not 3nH as you might expect. the caps were hard against each other (6mm between pins), and I calibrated out the test fixture PCB they were mounted on.

interestingly enough this doesnt really happen when you parallel electrolytics (on a PCB set up as a parallel-plate transmission line) as most of the ESL is inside an AL can, which does a pretty good job of eddy-current shielding (what with ESL effects occurring at high frequencies). If, however, you have a horrible wiring layout (ring terminals and widely spaced wires) rather than a well laid out DS-PCB, then all the wiring inductances mutually couple (IOW with electrolytics the overall ESL is dominated by the wiring/pcb layout)
Terry,

Thanks for the important corrections. Bummer.

What if the paralleled pairs were each tightly twisted (but not with any of the other pairs)? And what about using shielded twisted pair cables?

I guess the logical extension of the n pairs is a pcb like you showed. That seems like the way to go. I just thought it would be nice to also have a point-to-point-wiring method that could achieve something similar. Sounds like it can't work very well. Maybe we should try it, just to see what happens.

Tom

Last edited by gootee; 16th September 2012 at 01:12 AM.
  Reply With Quote
Old 16th September 2012, 01:05 AM   #1207
diyAudio Member
 
Join Date: Mar 2008
here is a parallel-plate transmission line pcb layout.

As you can see the PCB layout is very, very simple. the top layer is +Vdc (+24V in this case), the bottom layer is 0V. and each layer is almost entirely solid Cu, only having holes in it where unconnected component legs are.

this is much, much easier than trying to use parasitics as design elements - with or without fancy measuring gear or unbelievably expensive parasitic extraction software.
Attached Images
File Type: jpg cap bank top layer.jpg (360.2 KB, 219 views)
File Type: jpg cap bank bottom layer.jpg (333.7 KB, 207 views)
  Reply With Quote
Old 16th September 2012, 01:15 AM   #1208
diyAudio Member
 
danielwritesbac's Avatar
 
Join Date: Nov 2007
Quote:
Originally Posted by gootee View Post
Argh!! Where does this idea of always throwing in a parallel film cap keep coming from? Please read the thread about paralleling electrolytics and film caps. It is NOT, repeat NOT, a good idea, especially way out at the PSU reservoir caps. It MIGHT, but only might, be safe-enough, and would even have a chance of being effective, right at the load, for decoupling. Similar "discouraging" comments are waiting for the next mention of placing a film cap (or any cap, by itself) across a rectifier diode.
When I do this, it is with very high loss polyester dip caps, and with much care with comparisons to see if there's any benefit to it. I'm quite likely to accept filters that give BOTH higher resolution audio AND cooler heatsink. Thus, although you'll catch me doing it, just know it wasn't done arbitrarily. Considering the ESR of the little dip caps in use, assume an RC exists.

Actually, I'm afraid that some people will find my designs and corrupt the filters by using high efficiency ceramic/polypro/box caps. This really happens when some people just feel the need to pay extra for more decorous looking capacitors. And it goes badly.
In fact the only caps that work appropriately in my filters are polyester dip (bubble, not box) cap and tiny value electrolytic. Sometimes, a very high voltage electrolytic can be used but that is on a case-by-case basis.

On my new TDA7294 point-to-point design, I wish I could translate the 3n3 vs 2u2 vs 3n3 capdiv power filter into RC's. I'm sorry that I forgot to mention that those are all hi-loss filter purpose polyester caps. The 2u2 from rail to rail is particularly hard to select for adequate ESR and would be much easier as an RC, instead.
__________________
Tools, Models & Software for DIYClipNipper boostLM1875 TurboPowerful TDA7293 kitTDA7294 pt2pt ♦ My post has opinion.

Last edited by danielwritesbac; 16th September 2012 at 01:34 AM.
  Reply With Quote
Old 16th September 2012, 01:35 AM   #1209
gootee is offline gootee  United States
diyAudio Member
 
Join Date: Nov 2006
Location: Indiana
Blog Entries: 1
Quote:
Originally Posted by danielwritesbac View Post
When I do this, it is with very high loss polyester dip caps, and with much care with comparisons to see if there's any benefit to it. I'm quite likely to accept filters that give BOTH higher resolution audio AND cooler heatsink. Thus, although you'll catch me doing it, just know it wasn't done arbitrarily. Considering the ESR of the little dip caps in use, assume an RC exists.

Actually, I'm afraid that some people will find my designs and corrupt the filters by using high efficiency ceramic/polypro/box caps. This really happens when some people just feel the need to pay extra for more decorous looking capacitors. And it goes badly.
In fact the only caps that work appropriately in my filters are polyester dip (bubble, not box) cap and tiny value electrolytic. Sometimes, a very high voltage electrolytic can be used but that is on a case-by-case basis.

On my new TDA7294 design, I wish I could translate the 3n3 vs 2u2 vs 3n3 capdiv filter into RC's. I'm sorry that I forgot to mention that those are all hi-loss filter purpose polyester caps. The 2u2 from rail to rail is particularly hard to select for adequate ESR and would be much easier as an RC.
Well, if your heatsink gets cooler, at least it's probably not oscillating at high frequency.

I guess you could probably save the unsuspecting folks from messing up the designs with low-loss caps by just using low-loss caps with a small resistance in series.
  Reply With Quote
Old 16th September 2012, 01:40 AM   #1210
gootee is offline gootee  United States
diyAudio Member
 
Join Date: Nov 2006
Location: Indiana
Blog Entries: 1
Quote:
Originally Posted by Terry Given View Post
here is a parallel-plate transmission line pcb layout.

As you can see the PCB layout is very, very simple. the top layer is +Vdc (+24V in this case), the bottom layer is 0V. and each layer is almost entirely solid Cu, only having holes in it where unconnected component legs are.

this is much, much easier than trying to use parasitics as design elements - with or without fancy measuring gear or unbelievably expensive parasitic extraction software.
So, if we had a board like that (or one for each power rail, I guess), which stretched from the rectifier bridge to the active devices and load ground point of an audio power amplifier, we wouldn't need much in the way of decoupling capacitors, right? I think you said that the impedance stays very low to quite-high frequencies, right? That should do it. Coool.

I guess we could just build such planes and caps into a larger two-sided board that also contained the amplifier, eh? Or what WOULD be the best configuration? Put the power boards parallel to the amp board, maybe? Even with 2-sided amp boards, it's very difficult to do a good layout if power also has to be there. (I still also want to try making a four-layer, or more, with sandwiched boards, that is suitable for being assembled and soldered by hand, with mostly through-hole devices.)

Last edited by gootee; 16th September 2012 at 01:59 AM.
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Valve power supply - How to size transformer? SanderW Power Supplies 25 4th January 2013 04:12 PM
How do you calculate choke size in a power supply? Original Burnedfingers Tubes / Valves 25 5th January 2012 12:23 AM
power supply bypass cap size BigE Power Supplies 11 5th July 2011 02:59 PM
Power Supply Case Size diymixer Power Supplies 1 10th October 2010 05:47 AM
What size power supply should I get for repair work? spooney Car Audio 3 6th December 2007 11:50 PM


New To Site? Need Help?

All times are GMT. The time now is 01:42 PM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2