Quiet low power 5v -> +-15v ?

[max_torque]

At the moment,mostly for convenience, i use a small DC/DC converter (TES 1-0523) to get a +15 and -15v rail from a 5v supply, to drive some op-amps and comparitors.  However, being small and cheap, that device is pretty noisy, particularly with regard to common mode noise.  I put the output through a LC filter, which cleans up most of the noise, but it's still got 10's of mV of high frequency ringing that can cause triggering jitter with my high speed comparitors.

So, is their an easy way to generate a +15 and -15 rail from 5v, without some sort of switching device, i've not thought of.  1W would be plenty of power, it just needs to be simple, easy and small...  How hard can that be?   

[Buriedcode]

However, being small and cheap, that device is pretty noisy, particularly with regard to common mode noise.

... it just needs to be simple, easy and small...  How hard can that be? 

There's the rub.  Simple/easy - small - cheap? - low noise - efficient.  You'll have to compromise between these somewhat.  As far as I know you can't avoid using "something switching" to get +/-15V from 5V.  However, there are varying degree's of "noisy" both in terms of conducted and emitted EMI. 

Royer converters can be quite low noise (but often aren't designed to be) with reasonable filtering, and only require a handful of components, but finding an off-the-shelf-transformer for the job is hard, and often requires custom magnetics.  Simple, small, quiet, not easy.

A boost converter with either a tapped secondary or a dual charge pump can be reasonably small, efficient, simple to work with, but you'll probably want higher frequency for easier filtering, and careful layout.  A tapped secondary isn't easy to find, but easier than the above royer because the converters regulation allows for a range of turns ratios. a 1:1:1 to 1:4:4 is fine.  Not so simple, relatively easy, not so quiet, small.

Charge pumps themselves tend to have lower EMI (no ferrites), and higher efficiency.  But for a 3x boost @1W I haven't seen a single chip solution capable of that, which is why I suggested using them with a boost converter, either to create the negative rail, or for true symmetry, both rails running off the switch node. of a 5->10V converter. 

In the past I've tended towards a small boost converters + charge pumps, but that was for mostly opamp circuits where the CMRR came in handy.

[David Hess]

Linear Technology makes some low noise switching regulators for this type of application.  They include current and voltage slew rate limiting.

Another alternative which I would consider is to drive a high frequency step-up transformer with a sine wave.  This potentially has low enough noise that no secondary side regulation is required.

[Kleinstein]

The charge pump chips tent to very noisy - from switching they produce quite some current spikes that are hard to filter.

The ferrites are not causing EMI problems - it's more like parasitic inductance that produces trouble. Against common mode noise (interference) a common mode choke can help this can also be on the input side (only 2 wires and thus easier than the +-15 V). However the effect of a common mode choke is limited.

[SiliconWizard]

Have you tried putting LDOs after the DC/DC converter and LC filter?
They wouldn't get rid of all the noise but they would certainly help. Of course you'll have some voltage drop but at a few tens of mA (the converter you used gets you +/-35mA max) the drop-out will be pretty small with good LDOs.

[max_torque]

I've not had a chance to play around with the circuit yet, as i've been busy trying to get the basic functionality to work.  What i need to do is make up a test pcb, with some suitable footprints so i can start measuring the noise vs different filter layouts.  I've temporarily "fixed" the comparitor jitter by soldering a 15puff cap right across the +ve and -ve input pins of the SOIC8 packaged IC, which seems to shunt the noise reasonably equally into each input and hence seems to negate the noise some what on.  More experimentation needed!   


I though about doing a sinewave + transformer, but it ends up as a chunk more complex and significantly bigger with off-the-shelf magnetics, compared to the pretty well optimised DC/DC.

[aldi]

I agree with SiliconWizard, i've done this in the past with a LT3032 (https://www.analog.com/en/products/lt3032.html). Switchter -> LC filtering-> LT3032( + Careful layout of course).

[Simon]

If it's just about noise on the comparator supply i guess secondary regulation will help?