linear Vreg drop in replacement

[Mad_Labs]

Hi All,

Now that I am off the grid, I want to replace the linear voltage regs. (like LM7805,LM2940, LM317) with more efficient types. I want a drop in replacement with input voltage range of 12 to 20V, outputs of 3.3V, 5V and an adjustable would be nice. Any recommendations?

Jonathan

[mike90045]

About the only "drop in" is a LVD (low voltage disconnect) version that only needs 1V over the desired output voltage, for good regulation.

I think you want some switching regulator modules, but they are not drop in, they either need a cap or inductor or both, which are not likely present in a linear regulator.  They also may introduce some extra electrical noise, at their switching fq.

Digikey:
http://search.digikey.com/scripts/DkSearch/dksus.dll?vendor=0&keywords=voltage+regulator+IC

National:
http://www.national.com/analog/power/simple_switcher

[BruceM]

For micropower circuits, the LDO linear regulators have much lower quiescent current, down to 0.1ma instead of as much a 5ma.  Linear regulators all just piss away the voltage difference between input and output as heat. Here's a previous post about my favorite 7805 LDO substitute:
http://www.microcogen.info/index.php?topic=188.0

Switching regulators do an inductive power conversion instead of being "heaters", but eat up enough current in the switching regulation process that they often are less efficient for micropower circuits, and the output is "noisy" and not so tightly regulated.  

Tell us more about what you're trying to save power on.  If it's something drawing a meaningful amount of low voltage DC, then perhaps a DC-DC converter is what you need.  These are mini power supplies that would convert your battery bank voltage to a regulated low voltage DC.  They are available in a wide range of voltages and currents.  Lots of good sources- Digikey, Jameco, Mouser, etc.  At 1 watt or more, these would be the way to go.  You could also add an LDO linear regulator on the output (at least 1 volt above what you need) for applications requiring less voltage ripple.

[Mad_Labs]

Bruce,

I have lots of projects to replace the regulators in. For example, I have a Nixie clock I built with a GPS to get the time. Right now I am working on my power system controller/data logger. For new projects I can use whatever the latest and greatest IC is. For old projects I would like a direct drop in replacement.

Most of these projects use very little power, with 500mA being a lot of power.  I found these as a drop in:

http://www.v-infinity.com/adtemplate.asp?invky=931361

and these:

http://www.murata-ps.com/data/meters/dms-78xxsr.pdf

They seem fairly similar. The Murata is much more expensive, but a "name brand".

[Wizard]

Hi all,

PWM high frequency (300KHz plus) regulators are awesome way to do.  The drop in is good and dandy but both links only shows 500mA max where TO-220 linear were able to do 1A with heatsink.  Look into different PWM hi-freq regulator if need is high current.

The noise is not a issue, designing into your circuit is the key.  The waveforms I saw on these drop in regulator is odd, shouldn't have big ripple on input, and sine waveform on other with hash noise (normal for insufficient or missing low ESR with high current ripple capacitors on both sides of regulator or using general purpose capacitor is not good idea except for signal coupling).

A good PWM hi-frequency regulator circuit is capable of 6A just with SOIC 8 pin with solder heat spreader in center of IC on PCB (not much copper area is all needed for cooling) and little noise.  I recommend 220uF-1000uF low ESR with high ripple current capacitors 105C for proper application on both sides of regulator and several smaller low ESR with good ripple current scattered around the circuit board.

What you wish to acquire is book called "Art of Electronics".  Highly recommended and I have it and this cleared up many things of issues and bad designs I see in electronics repair.

http://en.wikipedia.org/wiki/The_Art_of_Electronics

Cheers, Wizard

[BruceM]

Johnathan, Thanks for the good info! I agree, the Murata is much better performing; less ripple and less quiescent current, higher efficiency.  Sorry I missed the boat on your question. I forgot about these types of 78xx switching replacements.  Not all designs will get along with a switcher's EMI, but most digital gear will, especially multilayer boards with a full ground plane.  Some marginal 2 layer boards will become glitchy.

These packaged micro-switchers by Murata could be very handy for some of my future projects, I appreciate the reminder!

I'm using a PIC 16F877 for my battery bank controller, and I found that my total board current draw when active doing SOC calcs once per second but not driving opto-relays and such was 3 ma. This is including the 0.5 ma loss of the LDO regulator, all the analog circuitry for current sensing, temperature, battery voltage scaling, etc., which  use micropower op amps (Microchip).  The 4x20 LCD display (2 ma) (local or remote via plastic fiber) only gets turned by timed push button.  That makes the switchers useless for my application, it would double my current draw.

I"d love to read more on your power system controller/data logger!

Bruce

[BruceM]

Good points Wizzard.  There is a trade off for the higher frequency; higher switching losses, so worse efficiency at very low currents, though this is getting better all the time with new reduced gate capacitance mosfets.   For a new design, you can always optimize for your intended current range, and integrating onto the board lets you do a superior job of filtering inputs and outputs.

My brain is like a bad 2 layer design now- being near even tiny current switchers makes me hurt and glitch.  I avoid them as much as possible as the hassle of shielding and filtering is too great. But sometimes they can't be avoided, or are worth the effort.    

I just modified a crap Newhaven Display serial 20x4 character LCD display with backlight last night;  the moron board designer put a PWM LED current control with the power to the backlight running all around the opposite sides, very outside edge of the circuit board (2 layer, no ground plane).  What a disaster, EMI wise!  You can't do worse than switching a capacitive load like a 150ma white LED, with zero inductive filtering, not even a low ohm resistor, via two traces run the long way around the opposite outsides of a board!  

I disabled the microscopic SOT-3 switching mosfet by cutting the gate trace, grounded it and put a LED current dropping resistor to a ground trace not too far from the +5V.  What a frig!  I had no choice but to run the backlight;  despite the BS from their tech support I found the transflective display is NOT very readable without the backlight. There was no termination of the serial data input line either.  Fixed that too,  then the serial data input was solid.  

I'm switching to a reflective parallel interface display- found some on ebay for $6 each.  Screw it, it's not that much hassle to drive these compared to a serial display, and I was going to add a processor as display/input controller anyways.  

[Wizard]

Micro mA of these stuff?  Regular regulator way overkill with normal TO-220.  Consider redo your project that all drains is used by one source and use very small linear regulator (TO-92 style or small SMD (this one with heat tab soldered onto PCB where necessary at point of use.

Cheers, Wizard

[BruceM]

I've found that packaging (TO-220 or TO-92 or SMD) means little in  linear regulator selection as far as current to ground for very light loads.

In picking a linear regulator you have to decide how tight the voltage regulation must be, as that is inversely proportional to quiescent current (current to ground with no or very low load).  I  spend a whole day selecting candidates and studying their datasheets sheets, ordered my favorite four and bench tested them. I finally picked the LF50AB, 5v 500ma LDO regulator, as I have some precision analog circuits on the board, and I needed tight regulation, with as little current loss as possible for my few milliamps of typical night time load. 

On my battery bank controller board,  peak loads (worst case about 70ma) precluded a TO-92 packaged linear regulators for thermal reasons, as the source is a 12V battery which might be as high as 15.8v. I picked a TO-220FP package for the LF50AB for my convenience in mounting and (minimal) heat sinking.