Power Supply Freds, Hexfreds, Ultrasofts, Ultrafasts and Fast Recovery

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

I'm currently putting together a psu based on Kevin Gilmores design:

An externally hosted image should be here but it was not working when we last tested it.


I've read through most of the discussions here on this topic, but I have yet to find anything really explaining the differences between the following types insofar as they're used in fairly low voltage (60 Vdc) linear psu's.

The choices that I know of at this point are as follows:

Standard Recovery silicon
FREDs
HEXFREDS
Fast Recovery Silicons
Schottkys
Ultrasofts
Ultrafasts

I would appreciate any opinions or recommendations you may have on this!
 
Schottky diodes are very nice because they have no reverse recovery characteristic. They simply do not overshoot. Unfortunately they do not have very good reverse voltage ratings, so they can only be used in low voltage supplies. Most of the time, they are not suited.

All other kinds of diodes have a problem with stored charge. When the diode is conducting, it builds up a stored charge. After the diode stops conducting, this charged is dumped downstream. This can happen slowly (standard recovery) or quickly (fast recovery) and it can happen in an uncontrolled manner or it can happen in a controlled manner (soft recovery).

This dumping of charge can have the audible effect of putting high-energy high-frequency componets onto your power rails. What you want are either no recovery Schottky barrier diodes, or ultrasoft recovery diodes, which have the least stored charge and release their stored charge in the most controlled manner.

FRED is simply an acronym for Fast Recovery Epitaxial Diode.

My recommendation is for the diodes I always use: Internation Rectifier HEXFREDs with the smallest stored charge I can get that still meets my specs. Often this is the HFA08TA60C, or others of that series.
 
Consider a diode like a tap (a plumber would say a ****) in line with a pipe. When the water wants to flow in the forward direction the tap says "oh yes" and opens and away the water flows. Then the water decides it wants to try and flow backwards, and this is where most of the difference happens.

In a standard rectifier type tap ;) several litres flow backward before the tap figures out what is going on. Then it shuts off very abruptly causing water hammer in the pipes, comparable to noise and voltage spikes on your voltage rail.

In a fast recovery type diode / tap only a small amount of water gets to flow backward before the tap shuts off, but still very abruptly, with the same problems.

In a fast, soft recovery diode / tap there is very little reverse flow and then the flow tapers off to nothing in a controlled manner, not in the "snappy" manner of the other diodes. "Snappy" is actually the proper word for diodes. http://www.google.com/search?hl=en&ie=ISO-8859-1&q=snappy+diode&btnG=Google+Search

Schottky diodes have a lower forward drop and they turn off virtually immediately the current tries to flow backward. The small amount of initial reverse current flow is the internal capacitance charging. They can only stand a limited amount of reverse voltage. With some exceptions, they begin to get rare above 100v. (Infineon does have a 600v 6A Silicon Carbide schottky but the forward voltage drop is about 2.8 volts or so)

The difference between snappy and soft recovery diodes is like stretching a rubber band till it breaks suddenly :eek: and stretching a piece of chewing gum that eventually breaks but very softly.

On a diode recovery graph there is two time periods, ta and tb. Ta is the time a specific quantity of electrons (measured in micro or nano Coulombs) is being washed out of the diode in the reverse direction (while still fully conducting backwards) and tb is the time it then takes for the current to taper off in a controlled way to a certain low value as the voltage across it rises, also in a nice way. The downside is that the tb period is dissipative, not an issue at mains frequencies though.

Page 3 onward of this doc is relevant http://www.irf.com/technical-info/whitepaper/murdiodes.pdf
 
mbroker said:


Could one not series two or more for increased voltage? Thinking class A bits of a bigger poweramp.

Mark Broker
Can't see why not. As you hit their reverse voltage limit they start to get very leaky so I expect if you have two or more in series then the first one to hit it's limit would stop and the others would then get their fair share. You can definitely do this with avalanche protected diodes like BYV26C for instance. 30 or 40 threaded in series through the length of an old ic tube works well for rectifying the output of an ignition coil. I haven't tried it with zeners though but it should be reasonable. Otherwise, put an equal value resistor say 4k7 across each to balance the reverse voltages. I have since read that IR have some 150v 80 A (dual?) TO-247 schottky's.:cool:
 
Circlotron said:
30 or 40 threaded in series through the length of an old ic tube works well for rectifying the output of an ignition coil. I haven't tried it with zeners though but it should be reasonable. Otherwise, put an equal value resistor say 4k7 across each to balance the reverse voltages.

I prefer to use 1N4007 for HV work, especially if you want to full-wave rectify a HV transformer. Hard to beat $.02 ea :)

I, too, was thinking a higher-value resistor in series would probably help balancing and increase diode life. But I was thinking 100k or more so as not to waste too much power, particularly for a higher voltage, low current supply.

Regards,

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