DC - DC modules with LM2576HV - adj?

[Pommie]

I need a 48V to 5V 3A buck converter and found this on ebay
**broken link removed**

I find it hard to believe that this little module can handle 3A without additional cooling. Has anyone got any experience with this particular module/chip? Opinions?

Thanks,

Mike.

 

[ronsimpson]

We don't really know what is on the board.
Bucking 48 to 5V you will have the diode on most of the time. Watch the diode temperature.
They clearly say you will need cooling at 3A!

 

[Pommie]

Thanks Ron, I missed the additional cooling requirement. I've sent a question asking about the diode.

As it's a one off job I'm tempted to just order the board plus a few TO220 versions of the chip and a schottky diode. I'm guessing from the pictures that the chip on the board won't be very easy to cool.

Mike.

 

[simonbramble]

This is just a standard buck converter. You should be able to design a unit in half the board area that can handle twice the current with no cooling. It is only because the efficiency of the LM2576 (and its Schottky diode) is so bad that you need a heatsink. This device will do it:
https://www.linear.com/product/LTC3891

It has external high and low side FETs which are much more efficient

 

[Ian Rogers]

I use the fixed voltage version.... I have tested this device up to 2.6A without a heat sink.. It performs very well.

Two things I found.... The schlocky diode has to be a real decent one, and if you use any kind of pre filter, design carefully... These things run at 52Khz ish, and when they start oscillating, they don't half go with a bang!!

 

[dr pepper]

The catch diode is what takes the hammer in this circuit, the input being 48v will nearly be 10x less than the o/p current so the switching trans wont be working that hard, the diode will be handling for short periods more than the 3a o/p current.

 

[ronsimpson]

The transistor (inside the IC) will have high AC loss with 49 volts across it. 48v + diode
The current will be about 3A for 10% of the time so the DC losses will be low.

The diode will have about 3A for 90% of the time. DC loss is high. There is AC loss to.

I have soldered small copper fins on diodes and ICs to help with heat.

 

[Mosaic]

For diodes and TVS units I have soldered 0.25" spade terminals (onto the diode leads) parallel & close up to the diode body to remove heat. That makes it standard PCB assembly capable and a cheap heat sinking method. For additional cooling you can stick female connectors onto the male spades. For additional heavy cooling the female connectors can be crimped/soldered onto bare, stranded floating #14 AWG copper wire unraveled into a 'fan' shape.

 

[Tony Stewart]

The loss is related to the load regulation which related to the RdsOn losses and Eddy Current losses and Diode losses.
Since load regulation is rated at 1%Vout @0.5A, or 50mW ideal RdsOnly losses per volt @0.5A

You might expect 6%Vout@3A thus 180mW per volt minimum.
But from datasheet. the fixed 5V version is 77% efficiency with 12Vin at 15W or 23% of 15W=3.5W

Does this help?

 

[Mosaic]

Maybe a higher frequency unit can reduce the ripple currents thru the parts and keep your cooling in line.
**broken link removed**

Edit:
Also one of these:
https://www.newark.com/webapp/wcs/s...estType=Base&partNumber=17T4209&storeId=10194

Will remove your diode heating worries. I tested it at about 0.5Vf at 3A, dual diodes active as I have them on hand.