cool dc/dc converters

We're going to be doing several instruments that need isolated signal channels, so I've been researching low power dc/dc converters. I'll need floating +-5 or maybe +-12 volts.

These are really cool

formatting link

They're tiny and cost around $4 at 100 pieces. There's at least one second source, Mornsun. One gotcha is that the CUI parts short-circuit protection looks tricky.

Anyhow, I want low noise injection into the isolated-common side, so I breadboarded it to play with it:

formatting link

I added a cute little Bourns 500 uH common-mode choke on the input side to theortically reduce spikes kicked across from ground to the output common.

So, the output common showed 10 volts p-p noise relative to ground! Square wave! Adding a 1 nF cap from out common to ground hardly changed it at all. I measured gnd-com capacitance on another part and it's like 12 pF. Made no sense.

Well, pin 10 on the data sheet says NC, but it should say DNC. It seems to be one end of the transformer primary, and it is 10 volts p-p at 100 KHz, and I soldered it to the output common copper on my little board, just a mechanical thing I thought. Lifted, the noise on the floating common is tiny.

For $4, one could use these to generate new power rails even if isolation isn't required. There are parts that convert practically anything to +-anything else. They are proportional, not regulated.

The Bourns inductor doesn't seem to affect noise much, so I'll probably leave it out. The CUI part may need a polyfuse on the input or something, in case something on the channel side shorts out. Or maybe not.

--

John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com
Reply to
John Larkin
Loading thread data ...

I was always taught to not connect NCs, they might connect to who knows what inside.

NT

Reply to
tabbypurr

I may use them in places that don't need isolation. Mornsun makes some of them that invert, +5 to -5 for example, and that could be handy.

--

John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com
Reply to
John Larkin

I wouldn't on a PC board, but it was convenient to solder pin 10 down on the Dremeled proto. Now I know!

Technically, NC means no internal connection, and DNC means Do Not Connect. There orughta be a law.

--

John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com
Reply to
John Larkin

We use the 5V-5V Cui to make -5V, this one.

formatting link
It worked and a I never looked at the noise that closely.

George H.

Reply to
George Herold

Yeah, that's the same series that I noted.

I busted one open partially. The filled epoxy is about like granite; band saws and soldering irons don't have much effect. One side is a tiny toroid and the output rectifiers and caps. I couldn't get into the other side with reasonable effort, but it must be an IC.

In the past I made my own isolated supplies, or use the big SIP things, but these are sure cute.

The Mornsuns are 3-pin non-isolated, higher power.

formatting link

Those people have a world-class-bad web site. I hope their products are better.

--

John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com
Reply to
John Larkin

Well, manufacturers don't seem to stick to that notation. I can't remember why some NC pins are connected, but they are.

NT

Reply to
tabbypurr

Grab it with some vice grips and apply slowly to the belt or disk sander.

George H.

Reply to
George Herold

Some LN2, a Kevlar bag, and a 2-lb sledge. :)

Cheers

Phil Hobbs

Reply to
Phil Hobbs

Den fredag den 19. juni 2015 kl. 19.14.46 UTC+2 skrev Phil Hobbs:

or xray:

formatting link

-Lasse

Reply to
Lasse Langwadt Christensen

I started with a band saw, which threw sparks and hardly cut, as if it were granite. After grooving it all around, probably ruined the blade, I squeezed it in a vise and shattered things. The tiny transformer had an inner silicone potting, so it came out sort of intact. The other side, with the primary-side drive, is still a little chunk of granite.

It would be fun to have an x-ray machine.

How can anybody make something like this and then sell it for $3.88? Digikey must pay them $3 for it or something.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com
Reply to
John Larkin

On Friday, June 19, 2015 at 10:54:09 AM UTC-7, John Larkin wrote: [about a DC/DC isolated converter]

The old Pico designs (scroll down to the bottom of this page)

used transformer saturation; so, you could get away with a couple transistors, couple diodes, couple of capacitors, and tune the sucker by grinding away at a C core. That design, though, has some magnetic leakage (saturation, remember), and frequency varies with load. You might want to check before putting 'em into a critical location.

Once your robot get three windings onto a bobbin, nothing much of any cost comes into it. Making it compatible with reflow soldering, though, requires that high-temperature tough encapsulant. If you want to see the parts, it might be easiest to ask for a factory tour.

Reply to
whit3rd

John Larkin schreef op 06/19/2015 om 04:19 AM:

In my experience the short circuit protection doesn't work at all with DC-DC converters in this form factor. I'm more inclined to put a switcher chip on the board.

Reply to
N. Coesel

I played with the CUI converter some more. Frequency is proportional to voltage, so I expect it's a saturating-magnetics multivibrator kind of circuit, like the first Pico diagram. It runs around 120 KHz. If I short the output, the output current hangs at about 200 mA, and the frequency goes up to 7 MHz. Really! It barely gets warm shorted. There's such a small volume on the driver side, I can't imagine what's in there.

I used to make high-voltage inverters for motorcycle CD ignitions with pretty much that circuit. I did have a clever bias circuit that had distinct startup and run base bias currents, so you could short it and not have it blow up. Two resistors and a diode.

Pico makes nice stuff, if you don't mind paying 20x the going rate.

--

John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com
Reply to
John Larkin

I've used some SIP dc/dc converters that claimed to withstand a short for one second. They in fact blew up in about 100 ms. I used an LM317L, with the ADJ pin open, as an input current limiter. That worked great.

The dual-output CUI specs unlimited short protection for all but the

24V input version. I just tested the 5V to +-5 part, and it barely gets warm when the output is shorted, and recovers nicely. As luck would have it, we have 24 volts available. Polyfuse? Just let it blow up?
--

John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com
Reply to
John Larkin

[snip]

Don't you think it's a Royer converter?

Cheers

Klaus

Reply to
Klaus Kragelund

Looking at pin 10, apparently one side of the primary, I see a 0 to

+10 volt square wave, not very Royer. A Royer needs another big inductor anyhow, and there's no room for that.
--

John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com
Reply to
John Larkin

That suggests the core material is lossy at 7 MHz, so it (1) decouples the primary and secondary and (2) acts like a primary-side current limit resistor.

Reply to
whit3rd

I wish I could decapsulate it enough to figure out the circuit, but I don't think I can. The classic problem with these cheap dc/dc bricks is short-circuit protection, and they seem to have fixed it here.

My old motorcycle CD ignition circuit did pretty good. It was beta limited at startup, to some safe power, and kicked in more base current after oscillations built up.

formatting link

Something like that maybe.

--

John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com
Reply to
John Larkin

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.