Looking for a fast recovery Schottky

Thread Starter

Robin66

Joined Jan 5, 2016
275
Hi. I'm trying to minimise loses in a boost converter. Switching period ~10us, Iavg~15A,, Ipk~20A, L=10uH, Vout = 45V(max). I have a highside FET across the flyback diode, and so the diode is only conducting for ~500ns per cycle between switches. However the flyback diode is getting surprisingly hot.

When designing the circuit I focused on the diode's voltage drop and choose a Vishay SS36 (http://www.vishay.com/docs/88751/ss32.pdf). But now that I've done further research perhaps I should have considered the reverse recovery characteristics too.

...ultra-fast recovery diodes or silicon carbide Schottky Diodes with extreme low Qrr are needed for CCM mode.

But the Vishay datasheet doesn't comment on this at all so should I presume that this isn't a selling point for the SS36. Speed isn't mentioned in the headline "FEATURES". Looking on RS uk, RS3A is called "Fast" and mentions a test recovery time of 150ns (http://docs-europe.electrocomponents.com/webdocs/0dae/0900766b80dae6e2.pdf). Would RS3A be a better choice, or should I not assume it's any better than SS36?

Since the highside FET takes the bulk of the current perhaps I should choose a smaller device that suffers a higher voltage drop for the 500ns that it's needed, but has better reverse recovery.

Any advice appreciated. Robin
 

crutschow

Joined Mar 14, 2008
34,454
You want a higher current Schottky to minimize the forward drop at the 15A peak current going through it.
From Wikipedia: "Schottky diodes do not have a recovery time, as there is nothing to recover from (i.e., there is no charge carrier depletion region at the junction)."
So any Schottky you use should be sufficiently fast for your purposes.
 
Last edited:

DickCappels

Joined Aug 21, 2008
10,180
upload_2017-6-25_23-31-15.png
The rectifier is just fine at 60°C. If it can be exposed to skin contact or is in danger of damaging other components then you should worry. Just think (I know this is an over-simplification): How hot does sand need to get before it melts?
 

Thread Starter

Robin66

Joined Jan 5, 2016
275
Great thanks. So the RS3A datasheet makes a point of discussing reverse recovery because it's not a schottky and so would be expected to be slow.
 

Bordodynov

Joined May 20, 2015
3,180
Schottky diodes can have a complex structure. For example, they may have a guard ring. This guard ring has the structure of a conventional silicon diode. If you do not exceed the passport current of the Schottky diode, then this parasitic diode is practically not active. But if you exceed this current, the diode of the guard ring will open and the recovery time will be long. So if you want to use a Schottky diode, then take a diode with a current of 20A.
 

Thread Starter

Robin66

Joined Jan 5, 2016
275
Ah ok, the plot thickens. Since the Vishay datasheet considers the "Maximum Non-Repetitive Peak Forward Surge Current", you'd think they'd add a graph demonstrating that the fast-recovery assumption of a schottky is invalid in this regime.

Brute force approach it is then!
 

crutschow

Joined Mar 14, 2008
34,454
So basically you want the Schottky to operate at at a peak current that will keep the forward voltage below about 0.6V (as shown on the data sheet).
 

Thread Starter

Robin66

Joined Jan 5, 2016
275
Well the datasheet (fig.3) shows Vdrop=0.6V @20A & 150c. At 25c I can expect Vdrop=0.9V. That's fine for the 500ns switch over time. It may not be fine though if the reverse recovery is now jeopardised, but the datasheet doesn't give me any info on this.
 

ronv

Joined Nov 12, 2008
3,770
Well the datasheet (fig.3) shows Vdrop=0.6V @20A & 150c. At 25c I can expect Vdrop=0.9V. That's fine for the 500ns switch over time. It may not be fine though if the reverse recovery is now jeopardised, but the datasheet doesn't give me any info on this.
Actually your okay where you are if you don't mind it being a little toasty.:)
 

Bordodynov

Joined May 20, 2015
3,180
Ah ok, the plot thickens. Since the Vishay datasheet considers the "Maximum Non-Repetitive Peak Forward Surge Current", you'd think they'd add a graph demonstrating that the fast-recovery assumption of a schottky is invalid in this regime.

Brute force approach it is then!
Maximum Non-Repetitive Peak Forward
 

Thread Starter

Robin66

Joined Jan 5, 2016
275
Isn't "Non-Repetitive" a misleading term? My understanding is that sufficient gap between repeated events must be allowed for thermal dissipation. It doesn't mean that the event can't be repeated again for the life-time of that device. In which case it's reasonable to routinely use the device in that way, and so if this impacts negatively on other device assumptions then it'd be useful to see how.

I've ordered some 15A schottkys (50p a pop!). It'll be interesting to see how this affects the heating in the lower FET which may have been suffering from the guard-ring recovery charge.
 

MrAl

Joined Jun 17, 2014
11,487
Approx 60c. It's only 3amp continuous but I'm using it on a 5% duty cycle

Hello there,

Well, 5 percent may not sound like much but in terms of RMS current that equates to about 22 percent of the peak current for a rectangular wave, and RMS current is normally used as a guide in selecting diodes.
To make a short table at 5 percent duty cycle:
20 amps=4.5 amps RMS
15 amps=3.4 amps RMS

so for either current the RMS current exceeds the rating of the device which you say is 3 amps.
Add to that the other facts presented in this thread especially the forward voltage drop increase with higher peak currents and you have yourself a way to fry pigeon eggs on a cold winter's day :)

To keep the peak forward voltage down, i'd go with at least a 10 amp device.
Diodes do have capacitance and so this may or may not work better than a 20 amp device.

Good luck with your project.
 

Sensacell

Joined Jun 19, 2012
3,449
One additional thing to consider- hot schottky diodes tend to get really leaky, this can lead to thermal runaway in higher voltage situations.
 
Top