Power Supply +/- 15VDC for Op Amps

[Kalvin]

Hi,

My apologies if this has addressed before, but at least after fast search I couldn't find any solution to this.

I am looking for a hobby solution for a problem of powering the op amps requiring dual power supply. Although modern rail-to-rail op amps function pretty well with a single power supply, there are tons of older and cheap op amps and schematics available for the hobbyists requiring dual power supply, for example +/- 12VDC.

There are some readily available dual output dc/dc converters from ebay which will take +5V in and produce +/- 15V (isolated) 2W output. When coupled with a LC-filter and optionally linear 7812/7912 regulators, they are great nifty gadgets to create dual power supply for op amp circuits. The dc/dc-converters may exhibit some noise issues, but other than that they are typically good enough for a hobby projects.

Those dc/dc-converters cost around $7 in ebay. It is not much, but in the spirit of DIY hobbyist I was thinking would it be possible make something similar much cheaper using salvaged parts or buy something very cheaply and recycle the components for better use.

There are vary many cheap dc/dc converters available in the ebay. For example, there are tiny solar/usb boost converters which cost less than $1 but has all necessary components to be used to create a positive +15V (okay, maybe the capacitor needs to be changed for higher voltage value). The negative power supply is a problem here.

Of course, one could use a 555 as an inverting charge pump, but let's think other solutions first.

It may be possible to take two of these converters and use the other as a step-up converter and the other as inverting converter. The inverting topology needs some tweaking, but is should be doable. Buying a few of those cheap dc/dc-converters and creating a simple PCB design should produce a bunch of nice (adjustable) dual output power supply for op amp circuits.

Instead of using two dc/dc-converters it may be possible to modify the dc/dc-converter design so that it will use a flyback design and a transformer with dual secondary windings producing +/- 12V after rectifying/filtering/regulation.

I have been searching for a cheap transformer which could be used in flyback dc/dc-converter, converting +5V to +/-15V 20mA / max. 100mA. Typically, the switching frequency using these cheap dc/dc-controllers is in range 50kHz - 100kHz. The transformer may be something like 1:3:3 to 1:5:5 as the dc/dc-converter will regulate the output voltage anyway. However, I haven't been able to find any suitable and cheap parts in ebay. Does anybody have any hints where to look for these suitable transformers?

Thanks,
Kalvin

Edit: Added typical switching frequency 50kHz - 100kHz.

[T3sl4co1l]

First rule of manufacturing: you can't make something as cheap as China, period.

Second rule: if you're making less than one million pieces, you can't make something as cheap as 10 times China, period.

Well, that may be exaggerating the difference, but suffice it to say, the cost-per-piece advantage in significant quantities (well over 100) is huge.  If you pay a designer to create the product, that NRE also amortizes poorly into that quantity (a small project might cost $5k NRE, so accounts for $50/piece, regardless of production costs).

So, if we instead gear this towards hobbyists, where there's just a schematic to follow on the breadboard, and you wind your own transformer... okay, a lot of those parts can be found or purchased for relatively little.  You can buy transformer bits for relatively cheap.  Or you can attempt to salvage and rewind them.  But with amateur build quality, it might not even work as well as the Chinese thing, and that's after significant time for the average hacker to figure out how the circuit works, fixing his mistakes, and not routing the power and ground signals very well.

I like to make circuits like these, from time to time:





but as far as I'm aware, no one else has ventured to build them.  (The main premise is a discrete circuit which performs better than the UC3842 in some respects, though lacking in others.  The good old 3842 circuit is of course easy to build as well.)

As with all switchers, getting quiet operation out of these requires careful wiring and extra filtering.  Not something that can be done on breadboard (functionally, yes: I developed these on the breadboard, with a little care; but not for low noise purposes, no).

I could make PCBs for some of these, but as I have no particular need, I tend to make deadbug one-offs of slightly customized variations.  PCB also forces a very rigid limitation on the transformer: pin spacing, THT or SMT, etc.  You're probably as well off purchasing a dev kit for your favorite controller-in-one chip instead.

Tim

[dannyf]

The best solution, as usual, is to not do it, by going single rail.

If you have to do dual rail, a better solution is to use an inverter. For up to 100ma, you can do that with the good-old 34063, or the newer SimpleSwitch from NatSemi.

[Kalvin]

Thanks Tim and dannyf.

I do agree that it is very difficult to beat the price the chinese are producing the devices in volumes. As I wrote above, there are lots of nice devices that can be purchased very cheaply and converted into more suitable and useful. Like these dirt-cheap step-up-converters which can be bought from ebay for less than $1 "dc/dc solar boost".

I did manage to find this nice application note from Linear Technology:
http://cds.linear.com/docs/en/design-note/dn223f.pdf
which is using a dc/dc-controller as a step-up converter and a charge pump generating +/- 20V @ 4mA.

The given 4mA may be too little, but I guess that the circuit may be scaled up to support higher currents from 10mA up to few tens milliamperes. Anyway, the design is simple and the part count is low.

Possibly the design can be adapted to 34063 or the cheap dc/dc-controllers found in those chinese step-up converters. Thus, ripping the parts from two chinese dc/dc-converters should provide parts for one dual output power supply for less than $2 supplying +/- 15VDC from a single +5VDC.

Br,
Kalvin.

[Skimask]

If you can get +15v, then maybe use one of these to get your -15v:
https://www.tindie.com/products/SpeedyLab/2a-positivenegative-dc-dc-step-down-power-module/
I've got a few of these in various projects.  Needed the same thing, + & - rails.  Dicked around for a few months trying out various topologies, trying to get the right combination of cheap, easy to build, etc.  Said screw it and started using these modules.  Simple, fairly cheap, and plug right in.  Swap out a couple resistors and you've got your voltage.

[langwadt]

you can do split supplies from standard boost converter with just a single inductor

http://m.eet.com/media/1188717/powertip60_01.jpg

[edavid]

If this is for wall wart power, it's usually better to use a single supply with a linear rail splitter to generate "ground".  The disadvantage is that the power jack ground and input/output ground aren't the same, so make sure your circuit can survive a short between them.

[Zero999]

If this is for wall wart power, it's usually better to use a single supply with a linear rail splitter to generate "ground".  The disadvantage is that the power jack ground and input/output ground aren't the same, so make sure your circuit can survive a short between them.

One thing you can do to safeguard against that is to use a floating PSU and make sure it doesn't power any other devices. That way it's unlikely you'll short something out and buy a decent quality unit which is short circuit protected, just in case.

[Kalvin]

Thanks Skimask and langwadt.

It seems that my original idea of using a flyback topology with a transformer may be way too complicated for this particular application of generating dual-rail supply for op amps from a single +5VDC.

As Skimask pointed out, using a separate, inverting dc/dc-converter may do the trick for supplying negative power rail. Actual component changes to a existing step-up design are minor and easy to do.

Langwadt provided nice addition to the charge pump option, which is a simpler solution compared to a separate inverting dc/dc-converter, although its power capability may not be as good as using a real inverting dc/dc-converter.

Meanwhile I was able to find this Maxim's application note "Regulator Topologies for Battery-Powered Systems:
http://www.maximintegrated.com/en/app-notes/index.mvp/id/660
Especially the figure 2 looks interesting:
http://www.maximintegrated.com/en/images/appnotes/660/660Fig02.gif

This reminds me of this circuit design "Build Your Own Ultra-Low-Cost Isolated DC-DC Converter" from Electronic Design magazine:
http://electronicdesign.com/content/14978/60029_fig.gif
The design uses a cheap common-mode choke as a transformer.

Maybe replacing the inductor of the existing step-up converter with a suitable common mode choke would do the trick? Any comments or suggestions?

[Kalvin]

edavid,

I wanted to rule out the solution of using mains transformer as the +5VDC is typically quite readily available, for example in Arduino designs etc. And of course, if the design is portable and runs on (rechargeable) battery, the transformer solution is not possible anyway.

Br,
Kalvin

[Kalvin]

Found this article from Texas Instruments "Generate Auxiliary Voltages at Low Cost", http://powerelectronics.com/site-files/powerelectronics.com/files/archive/powerelectronics.com/mag/801PET21.pdf
which pretty much covers everything proposed in this discussion thread.

It seems that either the coupled inductor design or the charge pump design is applicable to the original problem ie. generating a dual-rail power supply of +/- 15VDC from a single +5VDC power supply using a cheap chinese step-up dc/dc-converter available in ebay for under $1.

The common-mode chokes in ebay seem to be high in inductance, but luckily there are a few chokes in 100uH range, which seem to be suitable to be used with the existing dc/dc-converter designs.

[rob77]

the cheapest solution would be 2x MC34063 - there is nothing cheaper out there (at least i'm not aware of anything)

MC34063 : 15cents
4 x 1206 resistor : 5 cents
2x el. capacitor : 20 cents
1x ceramic cap : 2 cents
inductor: 40 cents
schottky diode: 10 cents
2cmx3cm FR4 : 5 cents

sum: slightly less than 1 euro - the whole DC converter cheaper than the modern ICs alone - and the MC34063 is good enough for hobbyists

use 2 of those for powering op amps - one as boost 5->12V and the second as voltage inverter to make the -12V (the MC34063 can be boost, buck, or voltage inverter).

i hear the comments that 34063 is so '80 technology (or '70  ? ), but that little thingy works , it's still available and is good enough for many applications;) efficiency in the 60-70% range... but who cares

[edavid]

edavid,

I wanted to rule out the solution of using mains transformer as the +5VDC is typically quite readily available, for example in Arduino designs etc. And of course, if the design is portable and runs on (rechargeable) battery, the transformer solution is not possible anyway.

OK, but for a non-battery application I would still consider a single ended DC-DC converter plus a rail splitter, for simplicity.

[kalebris]

I have tried to solve the same problem, i have an 8v DC adapter, so that's what I'm using, but it's pretty easy to modify based on http://www.nomad.ee/micros/mc34063a/.

But this is my first ever project and I'm trying to use the outputs from this to power a few op-amps (ne5532) but I am not sure if the ramp-up and the assymetry of the rampup is an issue or not. So the ramup for based on beta.multisim is around 100ms for the -15V and around 50ms for the +15V. Is this a problem for the op-amps? Also the ripple is around 10mV as the simulator shows, is this going to be a problem?

I appreciate all the help.

[Benta]

I went through this exercise a few years back.
As I only do one-offs or maximum 10 units, price is not really an issue, but ease-of-use is.

My standard setup is a 5 V, 5 W regulated switching wall wart, which powers most of my circuits.
If I need +/-12 or +/- 15 V, I've bought a handful of Traco TMR0522 and TMR0523 SIL modules. +5 V in, dual outputs, input/output isolation and regulated outputs, 2 W total power, outputs can be freely combined. I throw one of these in any circuit that needs it and concentrate on the "real" design.

Not cheap, 15 Euro apiece in low volume, but I have not found anything with better performance that is easier to use.

[Zero999]

I have tried to solve the same problem, i have an 8v DC adapter, so that's what I'm using, but it's pretty easy to modify based on http://www.nomad.ee/micros/mc34063a/.

But this is my first ever project and I'm trying to use the outputs from this to power a few op-amps (ne5532) but I am not sure if the ramp-up and the assymetry of the rampup is an issue or not. So the ramup for based on beta.multisim is around 100ms for the -15V and around 50ms for the +15V. Is this a problem for the op-amps? Also the ripple is around 10mV as the simulator shows, is this going to be a problem?

I appreciate all the help.

You forgot to trim the white space and set the contrast to a reasonable level. The result: a huge page with a tiny faded drawing.

[station240]

This Kit will provide +/-16V DC @ 3A from a single +5 to 24V rail, just assemble it w/out the linear regulators on the output.
https://www.banggood.com/Orignal-Hiland-DIY-USB-Boost-Single-Turn-Dual-Power-Module-Linear-Regulators-Multiple-Output-Power-Kit-p-1022517.html

It should be low ripple (as it is Sepic/CuK converter) but I'm yet to measure the ripple or test the actual max current limit.
Could always increase the feedback resistor so it output 17-18V and add a pair of 7815/7915.

[MagicSmoker]

The circuit used in a lot of those mini dc-dc modules is a (current-fed) Royer; aka, a self-oscillating push-pull. If you use MOSFETs for the switches and the supply voltage is in the range of 5-15V then you can use cross-coupled Schottky diodes from the gate of one MOSFET to the drain of the other with resistors from each gate to the supply to make it self-oscillating.

[MagicSmoker]

Attached is an LTSpice simulation of a current-fed resonant push-pull I modified to get closer to what the OP wants. The voltage overshoot during startup is considered a feature when driving fluorescent lamps - this circuit's most common application - but not so welcome here. Use zeners or increase the value of the output capacitors to deal with it in the real world.

Also note that this type of converter exhibits exceptionally low noise because the voltages and currents are sinusoidal.

[splin]

You could use one of these 5 to 15V 2W isolated convertors for the -ve rail and a dirt cheap non isolated step up for the +ve rail. They aren't very efficient at 60% but are only £17.87 inc shipping for 10 pieces:

https://www.aliexpress.com/item/Free-shipping-10pcs-lot-dcdc-supply-module-5V-to-15V-DC-DC-step-up-isolation-of/32518723762.html?spm=2114.01010208.3.2.xGfYzN&ws_ab_test=searchweb0_0,searchweb201602_1_10152_10065_10151_10068_10130_10136_10137_10157_10060_10138_10155_10062_10156_10154_10056_10055_10054_10059_10099_10103_10102_10096_10148_10147_10052_10053_10142_10107_10050_10051_10084_10083_10119_10080_10082_10081_10110_10111_10112_10113_10114_10179_10181_10037_10182_10032_10078_10079_10077_10073_10070_10123_10120_10124-10120,searchweb201603_16,ppcSwitch_5&btsid=f68fe3d6-54cd-44a1-8c39-4a2521ded38d&algo_expid=9d114ace-7f5e-4c3f-ae7f-3c4292aee417-0&algo_pvid=9d114ace-7f5e-4c3f-ae7f-3c4292aee417

or 12V versions £15.73 for 10 pieces. Search for B0512s-2W

1W versions (B0512S-1W)  are even cheaper at £10.68 for 10. (There is one sellor at £9.15/10 but feedback rating of 92.6% isn't good).

Smaller quantities are a bit more expensive but how much is your time worth?

[BravoV]

If the current requirement is tiny <= 50mA, and moderate price is tolerable, imo LTC3265 does make life easier. 

[Zero999]

Use a 30V wall wart and the TLE2426.

[Wimberleytech]

That looks like a good solution and a good price.  If +/-15 is really desired here is a similar one from China
http://www.ebay.com/itm/High-power-Simulation-OP-AMP-Power-Supply-Module-3V-5V-12V-15V-200mA-Output-/171267793486?hash=item27e05b224e:g:OowAAOSwHaBWhQCF

My first stop for this kind of product is Ebay.

[Buriedcode]

This topic comes up a fair bit, and I know its an old thread but...

Most have covered the standard DC/DC modules (easiest) but I haven't seen much mention of my two preferred methods - if I'm going for versatility rather than convenience/cost/size.

1) For <50mA I just use a charge pump on the switching node of a boost converter to create the negative voltage. It isn't regulated, but I rarely use the negative rail as a 'reference' anyway.  Whilst it isn't as efficient, you can use a linear regulator after the boost for the positive rail, and a regulator for the negative rail, it greatly reduces the ripple on the output of the boost, and on the negative side limits the 'sag' from changing current draw.  Costs two schottky diodes, and two caps.  + regulators and associated circuitry if you want regulated.

2) Coupled inductor, used for either a dual output buck, or dual output boost.  Requires an actual coupled inductor, which is harder to find/more expensive, but has higher current capability on the second rail than the charge pump.  Its pseudo-regulated as it follows the regulation of the positive output.  Also has the advantage that the second output is isolated, so can provide a negative voltage, or if hooked to the positive rail, double the output voltage (handy for single AA cell ->5v ->9V boosts for low power audio stuff that need a PP3).

As always, depends on the application, current requirements, efficiency, space, noise, versatility etc..  Switching regulators do radiate noise of course, its not that hard to filter it out on the rails, but its the radiated noise that becomes a problem.  A of-the-shelve module generally, as stated by another poster, uses a royer oscillator which has an almost sinusoidal output - meaning it doesn't kick out much noise, and when using a post regulator can make quiet dual rails easy.

[splin]

That looks like a good solution and a good price.  If +/-15 is really desired here is a similar one from China
http://www.ebay.com/itm/High-power-Simulation-OP-AMP-Power-Supply-Module-3V-5V-12V-15V-200mA-Output-/171267793486?hash=item27e05b224e:g:OowAAOSwHaBWhQCF

My first stop for this kind of product is Ebay.

I guess you didn't finish reading the OP:

Those dc/dc-converters cost around $7 in ebay. It is not much, but in the spirit of DIY hobbyist I was thinking would it be possible make something similar much cheaper using salvaged parts or buy something very cheaply and recycle the components for better use.

[I see Buriedcode got in before me whilst I was editing this but I'll post it anyway]

Providing you aren't after a lot of power, you probably don't need a switching regulator at all - especially if you have a clock available, of at least a few kilohertz. You you can easily and cheaply create the -ve rail using an inductor, diode, pnp or pmos transistor and a filter cap. Use a resistor/zener shunt regulator to provide a constant load and caclulate the component values accordingly. Probably wouldn't cost more than £.2 /$.25.

Even if you use a switch mode IC you will still need the inductor, filter cap and probably the diode. Efficiency won't be great over a wide range of loads (but can be pretty good at the deisgned load current), but that's not likely to be much of an issue in a typical op-amp circuit with fairly constant loads - epecially on the -ve rail.

You could do the same the same for the +ve rail, or use a cheap ebay step-up module and use the switch signal to provide the clock for the -ve convertor. Regulation or efficiency wouldn't be as good, as the regulator varies the switch signal mark/space ratio according to load.

[EDIT] As blueskull pointed out, Q2 was the wrong way round - corrected.

[EDIT 2] Added D5 to protect Q2 if clock voltage exceeds Vin