Hello,
I recently purchased a 500W DC-DC converter to charge a 12V lead acid battery using a 72V battery. The converter is non-isolated and works fine.
I recently observed that when the input was kept open and output was connected to the 12V battery, I could measure 12V on the input terminal.
Manufacturer acknowledged this issue as the device was not meant for battery as load. Is there anything that I can do externally to avoid this from happening? Like use a diode either on the input or output side? The output current is approx 38A

 

Output side schottky diode is the answer but you'll lose 0.5 volts approx so will need to increase the output voltage slightly. Also at 38A the diode will dissipate around 18W so is going to get hot. A stud-mounted type is necessary e.g. on a reasonably sized heatsink eg Wakefield 303N - will stlll be a case temperature of 80degC (at 30C ambient) so some care in mounting required. Both available from Mouser, but not cheap...$60 - 70 approx. all in.

 

Why is it a problem? It is caused by the source-drain diode in the top MOSFET of a buck regulator.
I would suggest a 72V coil relay which disconnects the output when the input is not present.
https://www.albrightinternational.com/

 

I would suggest a 72V coil relay which disconnects the output when the input is not present.

Or a lower voltage eg 24v and some voltage conditioning. Will need 60A contacts which could be expensive...

 

Or a lower voltage eg 24v and some voltage conditioning. Will need 60A contacts which could be expensive...

Albright SW60 is about £13. I think 72V is a coil option. 24V and 48V coils are common. It is important to make sure it shuts off at 12V.

 

Is there anything that I can do externally to avoid this from happening?

Why do you consider that a problem?

If you really need that blocked, another solid-state alternate is to use a low on-resistance P-MOSFET as an ideal diode, to reduce the operating dissipation well below that of a diode, while blocking the reverse voltage when the 72V battery is not connected (LTspice sim below);

Note that initially, when the 72V is zero (yellow trace), there is no reverse current (green trace) from the battery being charged.
When the 72V is applied, Q1 turns on, connecting the MOSFET gate to ground and allowing it to conduct the charging current with a low drop (MOSFETs conduct equally well in either direction when ON).

The MOSFET dissipation will be its on-resistance times the square of the charging current, which will be a few watts for a low on-resistance MOSFET, that will likely require an appropriate small heat-sink.
Or you could use more than one MOSFET in parallel to reduce the dissipation in each to below what would be required for a heat-sink (<1W each).