Voltage regulator startup question

[sal_park]

Hi Guys,

as previously discussed on the forum I've been playing with an LM2576 voltage regulator, and now after advice here an LM2576HV version given my voltage source (battery) is 42v. I've now built up a new circuit on a breadboard which seems fine, however the startup behavior is a bit odd - I'm generally seeing a startup voltage spike of around 40 - 60v, however sometimes (1 in 20 tries) The voltage get's limited to the output voltage correctly. No changes to the circuit between the two trace outputs below and I can't find any obvious bad connections. FYI the VR is driving a 50 ohm resistor to simulate the arduino. I've attached the traces so you can see what I mean.

Which of the two traces would you describe as correct behavior of the VR ?

Any idea why I'm getting these two different startup behaviors ?

thanks

sp

edit: I've attached the diagram of the circuit I've built and the load is a 50 ohm resistor.

[Neilm]

This looks like either a layout issue or a dry joint.  I would suggest checking that the current loop is as small as possible or as low resistance as possible, see page 13 of the datasheet. Also, have you correctly checked you inductor and capacitor values and are you using a good diode. If you have "just grabbed what was lying around" you might find that this could be the issue if (for instance) the inductor is saturating. Are there any problems when it is running?

Neil

[Bored@Work]

built up a new circuit on a breadboard

As in solderless breadboard?

which seems fine,

Ah, so you think the section in the datasheet about proper layout is just for entertainment purposes and because the datasheet writer had nothing better to do than writing some more stuff?

[blackdog]

Hi, Bored@Work

As always, you hit the nail in one time.
When will they learn to read the datasheets...

sal_park
NO Breadbord!!!
Use just a piece off copper clad board, build Manhattan style.
Try using a starground as much as posible.
Use good quality components, switching powersupply's eat bad components for lunch!

Picture of a 5 to 35V Switching regulator 5A, LT1074, experimenting with Coils and feedback i am testing @ the moment.
The small coil in the middle and big JH cap = extra filter, coil on the right side = commonmode filter


Kind regarts,
Blackdog

[ElektroQuark]

Is this thing alive?

[Mechatrommer]

Ah, so you think the section in the datasheet about proper layout is just for entertainment purposes and because the datasheet writer had nothing better to do than writing some more stuff?

how unsupportive and offensive!

[marshallh]

Get some perfboard, ditch the breadboard, seriously

[c4757p]

Yeah, definitely ditch the breadboard. Switchers do not like breadboards. The first and last one I tried to build on a breadboard gave a perfectly fine output - with less efficiency than a linear regulator for two minutes before it popped the transistor. Same exact circuit done Manhattan style? 87%, cool to the touch. Something was getting capacitive coupling where it wasn't good - I think between the feedback and compensation pins of the controller. The datasheet's not just being overly cautious by claiming that the layout is critical.

[blackdog]

Hi c4757p

If you look @ my Manhattan setup, you see pin 1 & 2 bend away from the rest of the pins.
There is not only capacitive coupling but also a lot of magnetic energy from the wiring.
If your building Switchhing regulators, you have to deal with both to keep it stable.
This is also explained in various application of National Semiconductor (TI) en linear Technology.


Kind regarts,
Blackdog

[c4757p]

Yep, I always do that too. Thought I could get away with a lazy breadboard test once, and like most things containing the phrase "get away with" or the word "lazy", I couldn't.    I should have known better, too - it wasn't even a simple boost or buck or something, it was a continuous flyback...