Hi guys,
I am looking for a Fast/Soft rectifier for my F5...a "block" type package, with either screw mounts or tabs...something 35A plus.
I have heard great things about the IXYS bridges:
http://www.partsconnexion.com/PDF/l363.pdf
Problem is, they are PCB/solder pin mount. I really don't want to attach the big, solid-core secondaries from my transformer to those.
So what do you guys know? What is the preferred rectifier these days? Any recommendations?
I am looking for a Fast/Soft rectifier for my F5...a "block" type package, with either screw mounts or tabs...something 35A plus.
I have heard great things about the IXYS bridges:
http://www.partsconnexion.com/PDF/l363.pdf
Problem is, they are PCB/solder pin mount. I really don't want to attach the big, solid-core secondaries from my transformer to those.
So what do you guys know? What is the preferred rectifier these days? Any recommendations?
If you absolutely insist upon 35A soft recovery diodes, you won't find them in a bridge rectifier assembly, either in package type GBPC (lugs) or GBPC-W (wire leads). You're stuck with standard silicon non-soft-recovery diodes in a bridge.
I had a little article in Linear Audio showing scope photos of a power supply using one of those bridge rectifiers rated 35A +1000V, with enormous oscillatory ringing (before), and then completely damped out to zero ringing (after). The difference was a simple 2C+1R snubber across each secondary. So if you find yourself stuck using a standard silicon diode bridge, the happy news is that you can get just as perfect a non-ringing waveform with your lumbering old-technology bridge, as if you had used four of the best of the best, newfangled discrete Fairchild Stealth-II soft recovery diodes. On the other hand, a proud and fastidious perfectionist would naturally want BOTH snubbers AND soft recovery diodes; and if that means four discretes instead of one monolithic bridge, so be it. Perfection and convenience don't always go hand in hand.
article link
_
I had a little article in Linear Audio showing scope photos of a power supply using one of those bridge rectifiers rated 35A +1000V, with enormous oscillatory ringing (before), and then completely damped out to zero ringing (after). The difference was a simple 2C+1R snubber across each secondary. So if you find yourself stuck using a standard silicon diode bridge, the happy news is that you can get just as perfect a non-ringing waveform with your lumbering old-technology bridge, as if you had used four of the best of the best, newfangled discrete Fairchild Stealth-II soft recovery diodes. On the other hand, a proud and fastidious perfectionist would naturally want BOTH snubbers AND soft recovery diodes; and if that means four discretes instead of one monolithic bridge, so be it. Perfection and convenience don't always go hand in hand.
article link
_
When building DIY gear for hobby purposes, I generally include both soft recovery diodes and snubbers. I don't mind the extra expense (about the cost of a ticket to the cinema), and I do take comfort in the knowledge that the enemy has been slain twice, with both a bazooka and a flamethrower.
To name one example, I've got a power supply design for chipamps (not F5s!) here on diyAudio, which uses 30A, 200V soft recovery diodes AND 2C+R snubbers. The circuit design and PCB mfg Gerber files are freely downloadable; look for "RingNot" in the thread title. I'm able to sell these power supplies, stuffed and soldered and fully assembled and tested, for USD 50. Whereas other pin- and function-compatible PCBs (designed for ordinary non-soft bridge rectifiers, and which don't have snubbers) cost USD 50 for the bare board only (!). An internet search for "Power-86" shows more.
To name one example, I've got a power supply design for chipamps (not F5s!) here on diyAudio, which uses 30A, 200V soft recovery diodes AND 2C+R snubbers. The circuit design and PCB mfg Gerber files are freely downloadable; look for "RingNot" in the thread title. I'm able to sell these power supplies, stuffed and soldered and fully assembled and tested, for USD 50. Whereas other pin- and function-compatible PCBs (designed for ordinary non-soft bridge rectifiers, and which don't have snubbers) cost USD 50 for the bare board only (!). An internet search for "Power-86" shows more.
I do take comfort in the knowledge that the enemy has been slain twice, with both a bazooka and a flamethrower.
Perfectionist overkill and convenience don't always go hand in hand.
(see RingNot for an example of a PCB that uses 30A discrete diodes rather than an integrated diode bridge. It's the same size as the $50 bare PCB which uses an integrated bridge)
Last edited:
I soldered two snubbers across transformer secondaries in my MC phono preamp PS which uses 4A ordinary bridges in dual bridge configuration. I also put a small switch for experimenting. Difference in reproduced sound(which is a purpose if we not forgot it)is a big zero.My flatpack transformers have very high leakage inductance of 25mH for 24VA, 24V.Lower resonance freq.?
Might one simply use CRC filtering to kill the ringing from the diodes turning off?
If you install them after the diodes (where filters are usually installed), it won't help. Oscillation occurs in the transformer secondary itself. If you install a CRC filter between the transformer and the diodes, you'll dissipate quite a lot of heat in the resistor and you'll drastically lower the input voltage to the diodes... therefore you'll also drastically lower the DC output voltage from the diodes.I soldered two snubbers across transformer secondaries in my MC phono preamp ... I also put a small switch for experimenting. Difference in reproduced sound is a big zero
Now all you need to do is capture oscilloscope photos of the secondary waveforms with the switch in both positions. Maybe your transformer has unusual amounts of core losses that contribute self damping, or maybe your implementation of transformer+snubber+switch+rectifiers, doesn't achieve a favorable damping ratio -- i.e. your snubber isn't properly tuned to the rest of your circuit. You could try going in the other direction, to see whether there is still no audible effect when you deliberately increase the ringing. Remove the snubbers, add a new high-Q inductor in series between the transformer secondary and the rectifiers, then listen. Choose a high-Q inductor whose inductance is about ten times the (estimated) value of your transformer's leakage inductance.
I bought and read the article. It's very well done.
In the end though, it appears that if ringing is the issue with "standard" rectifiers then the hyper fast/soft/FRED rectifiers and magnitudes better. But a cap and resistor can be just as effective...and the FRED/hyper diodes also benefit from a CR filter.
If so, why would anyone use the Hyper/FRED diodes when a simple cap/resistor is just as effective?
No listening test in the LA article...would listening tests reveal a preference for the Hyper/FRED diodes that doesn't show up in the ringing measurements? That's the real question.
In the end though, it appears that if ringing is the issue with "standard" rectifiers then the hyper fast/soft/FRED rectifiers and magnitudes better. But a cap and resistor can be just as effective...and the FRED/hyper diodes also benefit from a CR filter.
If so, why would anyone use the Hyper/FRED diodes when a simple cap/resistor is just as effective?
No listening test in the LA article...would listening tests reveal a preference for the Hyper/FRED diodes that doesn't show up in the ringing measurements? That's the real question.
Yes, the diode measurements article contained no listening data. As they say in Ph.D. theses, that has been left as an Opportunity For Further Research. Maybe a supermotivated person will purchase a set of 48 different diodes, install them into one or two pieces of audio equipment, listen to them, and report the (subjective) listening results where everything is kept the same except the choice of diodes. You, perhaps?
You probably won't need an appendix full of equations, or several big waveform pictures that each occupy 1/3rd of a page, so your listening tests article could end up considerably shorter than the measurement article's 14 pages.
The good news is, you won't need to design and build a test fixture (PCB + lab equipment) to run the listening tests. The bad news is, you will need a LOT longer than 10 minutes per diode type, to swap new diodes into the audio gear and then listen to it play music.
You probably won't need an appendix full of equations, or several big waveform pictures that each occupy 1/3rd of a page, so your listening tests article could end up considerably shorter than the measurement article's 14 pages.
The good news is, you won't need to design and build a test fixture (PCB + lab equipment) to run the listening tests. The bad news is, you will need a LOT longer than 10 minutes per diode type, to swap new diodes into the audio gear and then listen to it play music.
Hey, I said the article was good!
Obviously, I am not going to test all those diodes. It was an honest question...why use FREDS/hyperfast diodes when a CR filter seems to accomplish the same thing? It seems like all the high-end manufacturers have gone to these diodes in power amps (even Pass), must be a reason other than marketing. Maybe it's something other than ringing?
Obviously, I am not going to test all those diodes. It was an honest question...why use FREDS/hyperfast diodes when a CR filter seems to accomplish the same thing? It seems like all the high-end manufacturers have gone to these diodes in power amps (even Pass), must be a reason other than marketing. Maybe it's something other than ringing?
Why did person P perform action A? ask them!
Why did my RingNot design use both snubbers and FFPF30UP20 soft recovery diodes with datasheet-guaranteed softness factor (tb/ta)? Because I take comfort in the knowledge that the enemy has been slain twice, with both a bazooka and a flamethrower. And I judged the extra cost to be acceptably low.
_
Why did my RingNot design use both snubbers and FFPF30UP20 soft recovery diodes with datasheet-guaranteed softness factor (tb/ta)? Because I take comfort in the knowledge that the enemy has been slain twice, with both a bazooka and a flamethrower. And I judged the extra cost to be acceptably low.
_