LM338 output ripple

[Steve_UK]

Hi all, I am building a 24v 5A supply using an LM338 and following the data sheet recommended design for an adjustable output voltage. I am able to run with about a 3A continuous load (much higher and the regulator eventually enters thermal shutdown) but there is approximately 100mV ripple on the output. I have attached a scope image showing the output (purple) and the input at the filter capacitor node (yellow). Adding capacitance to the output (I have a 1uF tantalum currently, tried paralleling 2200uF electrolytic) doesn't reduce the spike, and adding additional capacitance to the input (currently 4700uF, tried 8000uF) makes the output waveform more distorted. Any ideas why this is happening and any potential things I could try to reduce the noise? Thanks in advance!

 

[living sounds]

Have you measured the drop (difference between input and output voltage)? Is there a stable output at a higher or lower output voltage (if you adjust it)?

 

[Rob Flinn]

The more voltage you drop between in & out on the regulator the less current capability it has. There is normally a graph on the data sheet to describe the function. To get 5A you will need it on a good heatsink too.

 

[JohnRoberts]

Wiring, PCB trace layout matters. If the 0V (ground) reference for the regulator is corrupted by reservoir capacitor charging current. It might not be clean.

JR

 

[abbey road d enfer]

Any ideas why this is happening and any potential things I could try to reduce the noise? Thanks in advance!

Without additional info about the xfmr, rectifiers, reservoir caps, layout, it is impossible to offer anything better than suppositions.

 

[Audio1Man]

You have a grounding issue, i see it when you said adding a larger cap the HUM increases.

Hand draw the schematic as it is built. Each componet and connecting each wire to lead, and Post it.

 

[Matt Syson]

The reference manual for these regulators tells you how the circuit should be arranged as even small 'errors' (taking connections to slightly different parts of a circuit trace (or wiring) can ruin a good design. Expecting 5 Amps from a single LM338 is a bit tough and you have not said if it is the Steel can (TO3 version or TO220 plastic which may result in different issues. There is a very good reason why commercial power suply manufacturers only expect around 4 Amps maximum form a TO3 device IF they wand a long reliable life. The heatsink insulator is also an important choice but the ripple you show is poor implrementation of the design.

 

[merlin]

You have a grounding issue, i see it when you said adding a larger cap the HUM increases.

This. The spikes clearly coincide with the rectifier ripple current, so you're somehow getting ripple current in your output wiring.

 

[jokeramik]

Send us a schematic an a picture of the amplifier setup.
Best regards!