Cheap LM2596 based adjustable lab supply

[casper.bang]

Hey guys,

As a newbie I do not have a dedicated lab supply, so usually play (breadboard and stripboard/veroboard) using LiPo's (I fly RC heli). Not sure when I get a hold of a real lab supply, so when I saw this tiny and cheap adjustable voltage/current with CC and CV from Shenzhen/China (yeah I know the stuff from there is low-quality and frowned upon by Dave) I had to order one. I plan to mount it permanently to a large breadboard, add 2 multi-turn 10K pots for better control of voltage/current and use an external 12V DC adaptor to feed it (meaning a 1.3V - 10V output range), which would make it possible for me to "play" even while sitting in the living room relaxing at night:
http://www.ebay.com/itm/LED-Voltage-and-current-display-LM2596-DC-DC-Step-Down-CC-CV-Adjust-Power-Supply-/150878659741

It's based on TI's LM2596 DC-DC conveter and appears to be an implementation of the reference design outlined in the data sheet:
http://www.ti.com/lit/ds/symlink/lm2596.pdf

Obviously I am concerned about ripple and noise; the data sheet does outline how to filter most of this ripple using a post-ripple LC filter but this reference design does not appear to contain one.

- For the L in the LC filter, the sheet mentions a low ESR cap. This internal resistance thing is a bit new to me, what kind of caps would I typically have to go for here? It looks like plain electrolytes on the diagram but on reference boards it looks like a super-cap.

- What kind of noise trouble could I run into even when applying this LC filter? I assume for general digital circuits there would be no problem, what about op-amps and other analog work?!

- Would there be any difference looking at the output with a scope, depending on whether the circuit would be fed via mains (transformer, rectifier bridge and cap) or a battery?

- Last but not least, I noticed that in one of Dave's video (USB power supply?) he uses a linear IC, but somehow lowers the voltage feeding this, in order to minimize power draw and heat dissipation. Could I somehow do the same; such as to let an effecient but noisy switching frontend feed a less effecient but cleaner backend such as an lm317 or better? (Obviously doing this, means the voltage/current display as well as protection mechanisms of the product I linked to above will mean nothing, but I'm still wondering about the theory.)

Thanks in advance,
/Casper

[Christopher]

May as well use a 317 with a pot and big heatsink for up to 26V at 2A..

CC could be implemented with a current shunt on the input, diff amp, and transistor on the feedback pin, emmiter to ground

I measured about 30mV/mA p-p average, with the max being about 50-70mV/mA. Load regulation is quite poor.


PS  LM2596 is pretty crap, had three of those cheap £1 Chinese boards (similar quality to this) blow up on me for no reason... at 8W load (3v in-out differential). Dodgy silicon perhaps as the actual chip is more than a £1 at Farnell.

Don;t expect any more than 70-75% efficiency with these cheapo inductors... Piss poor for a smps

[knarf.be]

I bought 3 of these http://www.ebay.com/itm/1pcs-DC-DC-Buck-Converter-Step-Down-Module-LM2596-Power-Supply-Output-1-23V-30V-/251066005460?pt=LH_DefaultDomain_0&hash=item3a74b337d4 a week ago.
I'm going to test them first (24 H usage, different loads etc.) and write a blog about them within a mount.
But really, believe me: just buy a €/$ 70 lab supply from a local store, it's so handy and you could rely on it. Also a good lab supply will last for years, but these things are just for tiny projects and are not good enough to last ages. For this purpose, it's a waste of money.

[casper.bang]

Thanks for the replies guys. I am pretty sure I will be getting a good linear PS eventually (probably the Agilent E3610A+U1272A bundle when resellers have the E3610A back in stock), but $24 was just too cheap* to pass on. I realize and accept this is not lab quality, but it would offer me a way to build up a small portable PS for bread-boarding fun in the livingroom at night (that's when I have time to play) while the wife knits or whatever. I'm pretty sure the wife would frown at a 5+ Kg humming PS! Unfortunately Dave Jones portable 3080 based linear PS is not production ready yet I believe. :/

Question still remains; could I somehow stabilize the output further using a 317 or 3080 linear step? What would happen if I added such a step, apart from a forward voltage drop and a requirement to tweak the voltage/current meters? Would the short-circuit/CC mechanism still function or would I wreck the 317/3080 frontend?

/Casper

* I could barely take the wife to a McDonald's for $24!! (Disclaimer; I hate McDonald's, but it's the best international monetary reference I could think of.)