Choosing diodes for high current power supply

Thread Starter

liquidair

Joined Oct 1, 2009
179
Hi all,

I'm designing a 12VDC 3.5A power supply around a LM2679 switching regulator (datasheet: http://www.ti.com/lit/ds/symlink/lm2679.pdf). I was feeling great about things until I got to the bridge rectifier diodes/reservoir cap.

Everything I've designed thus far has been easy in regards to picking diodes: figure out the DC load current, pick a diode with an Iavg rating of considerably more and things will be fine. However, now with higher currents involved, that low Vf we could usually ignore will make the diode dissipate significant power and therefore heat.

Now I realize the above "shorthand" method is wrong because the diode in a bridge remains off most of the time until the rectified voltage exceeds the voltage of the reservoir cap, at which point the diodes turn on and deliver current. In my sims, this current with a 11,700uF cap is 27A (!!) although this is brief, but seems to average out to be around 1.7A.

So my question is, how do we ensure we are picking diodes that are up for the job at higher currents like this? I'm looking at the 80SQ045N which is rated for 8A with a very low Vf, but it says it is optimized for hf switching. Can a DO-201 really handle a linear application like this, especially at 5-12A?

Thank you in advance!
 

wayneh

Joined Sep 9, 2010
17,143
Heat will follow the average current. The diode specs need to accommodate your peak current but it’s the average current that will dictate the power dissipation. This is because, unlike a resistor, the diode voltage drop is nearly constant and so power is VA with V nearly constant.
 

Thread Starter

liquidair

Joined Oct 1, 2009
179
Thank you wayneh! So if I'm understanding you correctly, the 80SQ045N will be more than OK in this application, since it's average is 8A and peak is 140A (peak repetitive isn't given)? Especially considering the sim didn't factor in the xfmr resistance, so 27A peaks are actually unlikely, which means average current should also end up being lower too?
 

ebp

Joined Feb 8, 2018
2,332
The average current will be exactly the same - it must be because it is dictated by the load (actually it will be very slightly different because the average voltage on the capacitor be very slightly different due to small difference in the forward voltage of the diodes and the switcher has a "negative resistance" input characteristic.

Give your output power and estimate of average current in the rectifier, diodes rated at 3 amperes would more adequate. In a single phase bridge rectifier each diode only conducts for half of the time, so its average current is half of the overall output current of the bridge. You do need to consider the transient current at turn-on with a discharged capacitor, but it is unlikely to be an issue. The transformer impedance will play a significant role in limiting the peak current.

Schottky barrier rectifiers are certainly usable at low frequency. What makes them highly desirable for high-speed switching applications is their nearly-zero reverse recovery time (the time it takes for the diode to go from conducting with forward bias to blocking with reverse bias). This is unimportant at AC line frequency. You do get the advantage of lower forward voltage than with silicon PN junction diodes, but the Schottkys will cost more.
 

Thread Starter

liquidair

Joined Oct 1, 2009
179
Thank you ebp! This is a great reply...definitely makes sense and puts my mind at ease. I think seeing the massive peaks in the sims threw me off!
 

wayneh

Joined Sep 9, 2010
17,143
Thank you wayneh! So if I'm understanding you correctly, the 80SQ045N will be more than OK in this application, since it's average is 8A and peak is 140A (peak repetitive isn't given)? Especially considering the sim didn't factor in the xfmr resistance, so 27A peaks are actually unlikely, which means average current should also end up being lower too?
Sounds good to me, yes.
 
Top