Better rectifier for the SIT 1

[2 picoDumbs]

The SIT-1 has ±28V supplies at 3.6 amps (200 watts). Any of the usual suspects having (Icontinuous >= 8 amperes) and (Vreverse >= 150 volts) and (trr < 75 nanoseconds) would be just fine.

I'm not too concerned about that.
Will the device be up to the task of sustained large ripple currents during the charging cycle using a supply with say 120,000uF on the positive rail?

They don't appear to have an easy way of mounting a decent heatsink. Will this be a problem under the conditions used above?

 

[Mark Johnson]

I'm not too concerned about that.
Will the [FBS 10-06SC silicon carbide integrated bridge mentioned in post #32] device be up to the task of sustained large ripple currents during the charging cycle using a supply with say 120,000uF on the positive rail?

They don't appear to have an easy way of mounting a decent heatsink. Will this be a problem under the conditions used above?

IXYS specs the max surge current of that SiC bridge at 12 amps, which is lower than the expected peaks of the current waveform when powering a single channel of SIT 1 class A amplifier. This screams Danger! Much better to use rectifiers whose datasheet max surge current is 120 amps or greater.

Perhaps this IXYS bridge can use the Aavid 5301 heatsink, which has been successfully applied to GBU2510 bridge rectifiers (25 amps) in the very similar GBU package. The rectifier bridge package is compressed to the heatsink by means of a spring clip. Its theta CA in still air is 6.3K/watt. Still, the very high Vfwd of SiC diodes, and its positive temperature coefficient (!), are worrisome.

 

[2 picoDumbs]

Hi Mark
I purchased your article. Very nice by the way.
If you're still interested in doing further work in this area, I'd be happy to donate some other hexfred diodes for testing.

 

[Mark Johnson]

... your [diode] article [in Linear Audio V.10] Very nice by the way.
If you're still interested in doing further work in this area, I'd be happy to donate some other hexfred diodes for testing.

Thanks for the kind offer but I'm out of the diode testing business for good. 48 diodes at two currents apiece, was more than enough. Maybe someone else is secretly (or publicly) assembling a stockpile of diodes, right now, for a new round of testing; if so, they haven't told me about it. Maybe YOU are thinking of testing some diodes?

You could make contact with the Editor of Linear Audio and ask whether anyone has expressed interest in writing a follow-on article with test results on a bunch of new diodes. You could volunteer to help acquire diodes, or help with testing, or write up, or other tasks & duties.

 

[2 picoDumbs]

Thanks for the kind offer but I'm out of the diode testing business for good. 48 diodes at two currents apiece, was more than enough. Maybe someone else is secretly (or publicly) assembling a stockpile of diodes, right now, for a new round of testing; if so, they haven't told me about it. Maybe YOU are thinking of testing some diodes?

You could make contact with the Editor of Linear Audio and ask whether anyone has expressed interest in writing a follow-on article with test results on a bunch of new diodes. You could volunteer to help acquire diodes, or help with testing, or write up, or other tasks & duties.

Yeah I could certainly do that. Probably follow the same testing procedure as yourself for consistency. I certainly won't have the time to do the number of diodes you tested, probably 10 at the most, unless people were to nag me to do more, in which case I'd consider doing more.

 

[Mark Johnson]

If you do decide to test some diodes for yourself, I strongly recommend that you don't omit the Variac from the test fixture. It provides a "soft start" because your hand/eye can't wind up the knob in less than ~ 500 milliseconds, and oh by the way that is 25 full periods of the AC mains waveform. Whether you're doing so intentionally or not, you're smearing the inrush current over many cycles of the mains. The Variac + output meter also gives you an early warning when you've misconfigured the test gear (most likely error: diode connected backwards). When you turn the Variac up past 15% and the output voltage doesn't rise, that means: Yikes! Something ain't right!

Another thing I would suggest is: spend the money on eBay to buy some super high wattage load resistors, and never burn your fingers again. here is a 300 W resistor while over there is a 50 watt resistor.

Finally, I recommend you tackle the diodes in a certain order, so you can train yourself to perform the measurements reliably and quickly. First do all of the low current measurements (200mA in my case) on all diodes. These have very little self-heating, so you can move slowly and deliberately while you're building up "muscle memory" of what to do to test a diode. Next, do the high current measurements on the low-Vfwd diodes first. Why? Because they have the least self-heating. (PwrDissipated = Iload * Vfwd). Learn how to take the data quickly before the diode warms up. Finally, do the high current measurements on the high-Vfwd diodes last. You'll notice that plenty of the ultrasoft recovery rectifiers have Vfwd > 2 volts at 2 amperes of average current. That's 4 watts! You can burn your fingers and scorch your PCB with that kind of heat. So learn how to do it quickly and how to store the scope trace image quickly, before everything becomes a fireball. Good Luck!!

 

[dantwomey]

When it come to DIYAudio rectifier happiness is low Vfr and low Trr the key?

IXYS VBE 17-06NO7
Vfr 2.09V
Trr 35ns
600 V 27A
Full Bridge

MUR3060W
Vfr 1.7V
Trr 60ns
600V 30A
Pair of Diodes

APT100S20B
Vfr .89V
Trr 70ns
200V 120A Single Diode

FES8DT
Vfr .95V
Trr 35ns
200V 8A
Single Diode

STTH6110TV
Vfr 1.3V
Trr 42ns
1000V 60A
Pair of Diodes

Regards,
Dan

 

[Zen Mod]

voltage drop at currents of interest , not max

 

[Mark Johnson]

My interpretations of the measured data are summarized on page 105 of the Linear Audio article. There's a rule-of-thumb that predicts whether a diode is likely to be "excellent" or "terrible" based on datasheet parameters. Naturally it was right 48 times out of 48 on the diodes in the paper.

I installed two key pieces of take-away information at the very beginning (Abstract) and the very end (Summary) of the paper, namely

1. "They all rang, including Schottkys and HEXFREDs."
2. "It was also found that CRC snubbers eliminate ringing completely, even with the poorest diodes."

 

[dantwomey]

My interpretations of the measured data are summarized on page 105 of the Linear Audio article. There's a rule-of-thumb that predicts whether a diode is likely to be "excellent" or "terrible" based on datasheet parameters. Naturally it was right 48 times out of 48 on the diodes in the paper.

I installed two key pieces of take-away information at the very beginning (Abstract) and the very end (Summary) of the paper, namely

1. "They all rang, including Schottkys and HEXFREDs."
2. "It was also found that CRC snubbers eliminate ringing completely, even with the poorest diodes."

According to Mark's guidelines this would be an example of the perfect rectifier. Sjöström Audio - RFB03 The high current ultra fast rectifier bridge

Regards,
Dan

 

[Michael Bean]

The current rating on that device is not very high.

And the currency rating is too high for my poor wallet...$43.29600 each sold in lots of 25 for $1,082.40! Oh, and not a stock item, must be special ordered too.

Mike

 

[2 picoDumbs]

Hehehehe.

 

[Mark Johnson]

... this would be an example of the perfect rectifier: {MUR3020 by ON Semiconductor}

No, that is an example of a diode that fits the rule-of-thumb which identifies excellent diodes. Not perfect diodes, excellent diodes.

Draw yourself a Venn Diagram including U, the universe of all diodes, E, the set of all excellent diodes, and P, the set of all perfect diodes.

MUR3020 is in U and is in E but is not in P. As you would expect, P is a proper subset of E.

A perfect diode would have infinitely good specifications, and infinitely low price, and infinite quantities in stock "ready for despatch" as they say on UK websites. But perfection is difficult to find in real life. I have settled for merely excellent.

 

[Mark Johnson]

Venn

_

 

[dantwomey]

Venn

_

Venn-interesting. 😎

 

[dantwomey]

I'm going to try some snubberized FES8DT's in my upcoming VFet2. Thankfully Digikey has some.

FES8DT-E3/45 Vishay Semiconductor Diodes Division | Discrete Semiconductor Products | DigiKey

Regards,
Dan

 

[Mark Johnson]

Its brother device, the FES8GT, did quite well in the Linear Audio tests: it finished in the top 20. But the 8DT has a much lower Vfwd which is probably a very good thing.

 

[dantwomey]

Its brother device, the FES8GT, did quite well in the Linear Audio tests: it finished in the top 20. But the 8DT has a much lower Vfwd which is probably a very good thing.

Thanks!

 

[WBS]

Since the STTH6110TV is an ultrafast recovery diode with datasheet guaranteed softness factor S>1 (!), this article in LA might be peripherally useful:

Soft Recovery Diodes Lower Transformer Ringing by 10-20X​

...

Mark, I'm wondering what you think of what Nelson had to say in his 2001 article on power supplies:
"I don't like the fast recovery types that some audiophiles have raved about. Fast recovery means that they withstand many amps and volts in a tenth of a few nano-seconds, something we don't see very often on the old 60 Hz AC line. They are essential element in switching power supplies, but for regular "linear" power supplies, I much prefer SLOW diodes, and we create them by placing small capacitor circuits across the diodes, which greatly reduces radiated noise."

 

[Mark Johnson]

... what Nelson Pass had to say in his 2001 article on power supplies:
"I don't like the fast recovery types that some audiophiles have raved about. Fast recovery means that they withstand many amps and volts in a tenth of a few nano-seconds, something we don't see very often on the old 60 Hz AC line. They are essential element in switching power supplies, but for regular "linear" power supplies, I much prefer SLOW diodes, and we create them by placing small capacitor circuits across the diodes, which greatly reduces radiated noise."

I am indebted to diyAudio member Hikari1 who posted the Pass Labs Xs amplifier sales page (Dec 2014) in another thread: LINK . I have highlighted one piece of it.

Xs

These are Pass Laboratories ultimate expression of Power Amplifiers the perfect blend of science and visceral emotion.

The successor to our already lavishly praised and well regarded Supersymmetry patent. Every technical effort has been made to assure exceptional specifications, among these:

• Separate chassis for lower electromagnetic noise
• Power supplies with greatly enhanced storage capacitance
• Banks of redundantly parallel high speed / soft recovery rectifiers
• Improved high frequency noise filters
• Larger / quieter transformers
• ...

I think it's perfectly okay for anyone to change their opinion over the course of thirteen years (2001 - 2014). Certainly I've changed my opinions plenty of times, on lots of topics, in the last 13 years.