Component shortage: You now need a TO220 diode for a 0.5W power supply?

[Faringdon]

Just doing a 5V, 100mA  Buck from 400VDC using LNK302.
Needs to be all PTH components.
LNK302 datasheet says that when in CCM, a 35ns diode is needed.

…Since Fsw is  fixed at 66khz for LNK302, and since vout is just 5V,  and since the minimum on time of LNK302  is 475ns…….the part will indeed be in CCM much of the time.  (actually a mixture of cycle skipping and CCM operation)

As such, I searched for a  cheap PTH diode with <35ns……there are virtually none anywhere. There are so few that it would be unwise to plump for one, since you would be in danger of nil-stocking at some point. Therefore, the only way we can conform to <35ns is by using….yes…..a SiC TO220 diode!...(there’s loads of them in stock everywhere) for a 0.5W Power supply. There's pretty much nothing SiC in less than TO220   for PTH.
Have i gone wrong here somewhere?

LNK302 datasheet
https://www.mouser.com/datasheet/2/328/lnk302_304-306-179954.pdf

[dmills]

I would have guessed UF4007 but it is not quite fast enough.
This is the downside of doing a PTH design in a SMT world, you cannot get the modern stuff.

[bdunham7]

Have i gone wrong here somewhere?

LNK302 datasheet
https://www.mouser.com/datasheet/2/328/lnk302_304-306-179954.pdf

A quick read of the datasheet indicates nowhere that I can find that the LNK302 is capable of 100mA output.  How are you figuring it can?  In any case, the datasheet also tells you that the recommended design choice in this instance is to move up to the LNK304 which will not go into CCM @ 100mA and can use the bog-standard UF4005.  Also, I didn't see that small TH-package <35ns 600V rectifiers were 'out of stock', I didn't see them listed at all.  Are there any?

I've never used these HV non-isolated converters, so what is in the datasheet is all I know--so take that for what it is worth.

Edit:  can these handle 400VDC?

[pigrew]

Reading the LNK302 datasheet, it suggests BYV26C.

What am I missing, or is the BYV26C too expensive? (I can't imagine a SiC diode would be cheaper...)

Its plastic version SBYV26C seems cheaper, though is marked as not for new designs.

31GF6-E3 may also be an option?

[bdunham7]

What am I missing

Just that some people, like me, struggle with the Mouser search functions.  I'm not sure why.  A second look shows SF28G for $0.16 and LNK304 in a PDIP for $0.57.  Or if the status of the SF28G is too scary, the ER106 for somewhere between a nickel and a dime in quantity.  But even with a <35ns diode I don't see that the LNK302 works.  Am I missing something?

[Jay_Diddy_B]

Hi,

Power Integrations has done a very poor job in specifying the diode.

Trr is function of many things:

Ifwd
Irev
Tj
-di/dt

Different manufacturers test the diodes for Trr in different test circuits.

OnSemi for the UF4005 uses Ifwd = 0.5A, Irev = 1A (limited by the test circuit)



The pulse generator needs to be powerful, something like the HP214B is needed.

I would contact Power Integrations and ask them for recommendations.

Jay_Diddy_B

[Faringdon]

Thankyou for all these replies.
I had noticed a few axial diodes, but so few that i worried they may go future nil-stock.
I think you hit it when you said "PTH in an SMD world"

Interesting what you say about 100mA being too much for LNK302.....thing is , ipk is 165mA, so as long as the L is big enough, i should be able to squeeze 100mA out of it....my vout, at 5V, is so low, that the L doesnt discharge down quickly....but youre correct, the top of the datasheet says its "63mA max".......i have a qu on Power website forum about it but no reply as yet.
-Because its odd that in no single power integrations App Note, is there any mention of doing a 5V supply with any of their HV Bucks. And yet, 5v is a pretty standard vout.

But you have a point because to get 100mA out of a LNK302  you do need an L  which is higher than the recomended 1.5mH.....and maybe they are thinking there will be issues with inter-winding capacitance or something?
Also,
It seems a bit odd that D2 on page 4 of the LNK302 datasheet is just a slow 1N4005.....but maybe they think it will have zero current in it by the time the cap C3 snaps back up to 300V or whatever is Vin.

[Jay_Diddy_B]

Snip ..

Also,
It seems a bit odd that D2 on page 4 of the LNK302 datasheet is just a slow 1N4005.....but maybe they think it will have zero current in it by the time the cap C3 snaps back up to 300V or whatever is Vin.

D2 is operating at an average current of 1mA and discontinuous conduction. Since Qrr and therefore Trr, is a function of Ifwd. There will be very little Qrr in D2. They will be able to get away with a 1N4005 in this location.

Jay_Diddy_B

[bdunham7]

Interesting what you say about 100mA being too much for LNK302.....thing is , ipk is 165mA, so as long as the L is big enough, i should be able to squeeze 100mA out of it....my vout, at 5V, is so low, that the L doesnt discharge down quickly....but youre correct, the top of the datasheet says its "63mA max".......i have a qu on Power website forum about it but no reply as yet.
-Because its odd that in no single power integrations App Note, is there any mention of doing a 5V supply with any of their HV Bucks. And yet, 5v is a pretty standard vout.

I did notice that despite mentioning that the current specs were valid only at 12V, there is no graph or description of how they might vary with voltage.  But since the LNK304 is about a nickel more in quantity and comes in the same PDIP package, I'm not sure how much thought is really merited.

[Faringdon]

Thanks....i am going to bottle out of doing 5vout...and do 12Vout...and down-convert afterwards......because i can get  DCM operation with the higher vout, and DCM is less noisy......i am using all PTH, so i will be more noisy anyway......also, i dont like handling SiC.....i also cant garantee that all the guys will handle it with full ESD protection...and i am convinced that SiC has higher ESD susceptibility than normal Si.