Dual supply ICs that aren't microscopic?

[paulca]

I want to make a dual +-12V supply from a single input, Ideally 12V, but I can live with 5V.  I can find switching regulator chips which do this from Linear, Maxim and TI, but they are all, so far, seem to be in the smallest little packages they can get.  QFNs etc.

Does anyone know of examples or more managable devices?

The other option is a boost converter and an inverting boost converter as a pair.

Of course this is where you tell me I'm barking up the wrong tree and these don't do what I expect anyway.

[Tommy1984]

ICL7660 for max 10 V
ICL7660A for max 12V 


OBS! I thought you need it to power an op amp!

[Mr.B]

You don't mention your power requirements...

muRata have a range of modules:

MTU1D1212MC (+/-12vdc out, 10.8 to 13.2 in, 1W)
https://nz.mouser.com/ProductDetail/Murata-Power-Solutions/MTU1D1212MC?qs=sGAEpiMZZMvGsmoEFRKS8EQfQWTYwRjY13tljw%252bH9s8%3d

NCM6D1212C (+/-12vdc out, 12v in, 6W)
https://nz.mouser.com/ProductDetail/Murata-Power-Solutions/NCM6D1212C?qs=sGAEpiMZZMvGsmoEFRKS8NGQWIL5m%2fkiSj4lVxi5VomTcB%252bcVdWoLw%3d%3d

[Neomys Sapiens]

LT8471 in a 20-Lead TSSOP?

[Buriedcode]

In addition to the previous replies:

For lower current (<20mA) you can use a simple charge pump added to any boost converter - output won't be regulated though.  Costs two diodes and two caps.
For medium current you can use a coupled inductor on a buck converter - pseudo regulated
If you can find those inductors with multiple windings, I forget the brand, but they have 5-6 separate windings all with the same inductance to mix'n'match you can use a boost converter with a center tapped secondary - pseudo regulated.

QFN is indeed a bit of a bugger to prototype with (unless you use hot-air and adapter boards) but I'm sure there are MSOP/SSOP about which can be hand soldered fairly easily (flux'n'drag).

[paulca]

Sorry power requirements are to power a set of dual opamps, probably 7 or 8 of them with only 2 of them producing any power for headphones.

So less than 100mA, call it 200mA to be safe if I add LEDs or some such.

As this is audio, I suppose a further requirement is that the switching frequency is as high as possible.  Certainly not a close multiple of audio frequencies like 60Khz.

[paulca]

LT8471 in a 20-Lead TSSOP?

Looks good until I get to the exposed base pad that would need hot air.

I have no hot air gun.  Also I will be prototyping with DIL/DIP adaptor boards.

[Zero999]

Is the +12V rail already well-regulated enough for your requirements? If so, then how about adding a buck-boost converter to get -12V?

It can be done with the cheap MC34063, but there are better ICs available.
http://www.ti.com/lit/ds/symlink/mc34063a.pdf

Have you considered a virtual ground? The easiest way is to use the TLE2426.

Do the LEDs really need to run off the regulated output?

Can the circuit be redesigned to use a single rail?

[paulca]

I 'could' use a virtual ground.  However I have run into problems before with virtual grounds on the output.  I don't like connecting a 6V GND to another random mains powered device, it makes me uncomfortable.  For example I connected my headphone amp input to an Arduino and it browned out and crashed.  I measured the potential between the grounds and it was 8.5V.  Arduino was running off a bench supply, I think the brown out was just the time it took the various capacitance to shift the ground, but I'm probably talking out of my hat here.  Maybe my discomfort is unnecessary.

As to regulation, ideally I would like a rather broad input spec.  The typical auto-motive spec of 9V to 15V so I can run it off an unregulated 12V battery or solar.

I 'could' use a linear regulator to drop that to a fixed, stable input voltage, but that is unlikely to be 12V, more practically 9V or 6V.

I can limit the LEDs to a power LED which I can put a 1k on and it will run on the unregulated input.  A level meter strip of LEDs post pre-amp stage would be a nice addition, but not really necessary.

[Benta]

For low volume runs, I don't bother building a PSU myself, but use the Traco TMR 2 and TMR 3 modules. Available with 9...18 V input and +/-12 V regulated, isolated outputs. Not cheap, but dead simple and available at most of the catalog distributors..
No affiliation.

[paulca]

For low volume runs, I don't bother building a PSU myself, but use the Traco TMR 2 and TMR 3 modules. Available with 9...18 V input and +/-12 V regulated, isolated outputs. Not cheap, but dead simple and available at most of the catalog distributors..
No affiliation.

Hmmm.  £12-15, but the self build solution for a single run will cost about as much.  Nice wide input range, +-12V output, 125mA, perfect.  The TMR-3 has 75mV vpp ripple, but I'm not dealing with mic level, only line level inputs so this should be fine?? 

The reveals my problem in that I don't yet understand audio design criteria at the same time I'm not an audiophool and so it will probably be grand.

[paulca]

muRata have a range of modules:

Sorry I missed this.  Just as good as the TMRs.  I missed it as the Mouser links never seem to work, they ask me to select my country and dump me to the homepage

Thanks for the suggestion.

[tszaboo]

For low volume runs, I don't bother building a PSU myself, but use the Traco TMR 2 and TMR 3 modules. Available with 9...18 V input and +/-12 V regulated, isolated outputs. Not cheap, but dead simple and available at most of the catalog distributors..
No affiliation.

Those are really really noisy for analog, which means you need to filter them. Designing a proper filter means measuring the unit, and in the end, you put the same effort in the filter, than the power supply. There is no free lunch.

The reason that these are in QFN and similar packages is easy. If you have an IC which needs to deliver 200mA +/-12V, that is almost 5W. Even with 80-90% efficiency, you have 0.5-1W dissipation in the IC. And the volume application for this is for LCD, portable devices, so as small as possible.
I would suggest the LT1945. Or hack a typical boost converter to work with negative output, and use the same IC for the positive.

[paulca]

Those are really really noisy for analog, which means you need to filter them. Designing a proper filter means measuring the unit, and in the end, you put the same effort in the filter, than the power supply. There is no free lunch.

The reason that these are in QFN and similar packages is easy. If you have an IC which needs to deliver 200mA +/-12V, that is almost 5W. Even with 80-90% efficiency, you have 0.5-1W dissipation in the IC. And the volume application for this is for LCD, portable devices, so as small as possible.
I would suggest the LT1945. Or hack a typical boost converter to work with negative output, and use the same IC for the positive.

How noisy is noisy?  Are we talking about obviously noticeable on an op amp output amplifying Line level 500mVpp by gain of 4 or 5 or are we talking about only noticeable on a scope while doing frequency analysis and characterization?

[Benta]

Those are really really noisy for analog, which means you need to filter them.

Not really, when we talk audio. Peak ripple is at the switching frequency (and harmonics), which is around 200 kHz upwards.

I wouldn't worry about it, and it's easy to filter out.

[tszaboo]

Those are really really noisy for analog, which means you need to filter them. Designing a proper filter means measuring the unit, and in the end, you put the same effort in the filter, than the power supply. There is no free lunch.

The reason that these are in QFN and similar packages is easy. If you have an IC which needs to deliver 200mA +/-12V, that is almost 5W. Even with 80-90% efficiency, you have 0.5-1W dissipation in the IC. And the volume application for this is for LCD, portable devices, so as small as possible.
I would suggest the LT1945. Or hack a typical boost converter to work with negative output, and use the same IC for the positive.

How noisy is noisy?  Are we talking about obviously noticeable on an op amp output amplifying Line level 500mVpp by gain of 4 or 5 or are we talking about only noticeable on a scope while doing frequency analysis and characterization?

Depends on the PSRR of your Opamp. These work in the 50-150KHz region, and they have no regulation whatsoever. Meaning that the output voltage will change if you change the input voltage or the output current. They require minimum load and they dont like large capacitors on the output.
https://www.tracopower.com/products/tmr2_application.pdf
Just look at the curves and see the 100mV changes for a load change.

[Benta]

So you're saying that building one yourself will give better results than the modules? Allow me to doubt that.

[Benta]

Just look at the curves and see the 100mV changes for a load change.

Yes, that corresponds to 1% of the output voltage, which is actually very good (and according to specs, which is +/-1.0% for a 0...100% or 100%...0 load step). I challenge you to do better with a homespun DC/DC converter.

[Buriedcode]

Also note that, for opamp circuits - often the signal route isn't directly coupled to either rail, and all opamps have power supply rejection ratio (PSRR) in the order of >50dB.  This means that, unless you have a resistor going from either rail to the signal path, opamps can be reasonably tolerant of power supply noise.

What tends to be more problematic in my experience is radiated noise - EMI from switching converters coupling to high impedance inputs.  This is why I lean towards either charge pumps  - which are hard to get for >100mA working at less than -5V - or royer converters that have a more sinusoidal switching - narrower band of noise, easier to filter out.

With that said I have made a headphone amp using a boost converter and coupled inductor for a bipolar power supply, that has no discernible noise, as the switching frequency is so high.  This isn't a partocularly 'good' design' just an example that switching converters don't always cause problems.

[Kalvin]

Have you checked these already:

https://www.ebay.com/itm/DC-DC-Converter-Step-Up-Boost-3-3V-5V-12V-to-5V-9V-12V-15V-24V-Dual-Power/172765565537

I have no idea how well these perform or how noisy they are, but they seem to be able to supply quite high positive and probably sufficient negative voltage from a single input supply. The price is not too bad either.

Edit: The controller chip is XLSEMI XL6007E1: http://www.xlsemi.com/datasheet/XL6007%20datasheet.pdf

[Benta]

Input voltage is too low and they're not isolated or regulated.

[tszaboo]

So you're saying that building one yourself will give better results than the modules? Allow me to doubt that.

Yes. You dont need to doubt it, it is how it is.
It is really easy to understand, if you think about it.
The isolated converters are just a PWM signal fed into a transformer. The transformer is driven to saturation (EMI everywhere). There is absolutely no regulation. Meaning that if you do something on the output, draw current, your voltage will go down. And if your input voltage is only 4.5V for a 5V model, then your output is 10% lower. If your input has a 1KHz ripple, it goes to the output.
Meanwhile if you take a boost-invert topology, there is immediately feedback if the output changes. The inductors are driven with just enough current and it is nice and silent.
Also, you have 100mV noise where your signal is about 1 Volt. In frequency ranges where the opamp doesnt have any PSRR.
Yes, you can filter it, but it is not smaller task than designing a proper (and not isolated) DC-DC.
A homespun DC-DC doesnt have to be coming from the junk bin. I personally can solder QFN package with a chinese hot air station.

[Benta]

The isolated converters are just a PWM signal fed into a transformer. The transformer is driven to saturation (EMI everywhere). There is absolutely no regulation.

I'm certain that cheap converter modules exist that are built as you say.

But I think you're underestimating  the modules I suggested. They are indeed regulated, check the data sheet.
Like I said, I've used them myself to power opamps at +/-12 V and have had absolutely no noise problems or EMI. The high frequency ripple is killed in the end by the decoupling caps at the opamps.

But you go to your church, I'll go to mine.

[schmitt trigger]

I also suggest the use of charge pumps for opamp negative supplies.

[David Hess]

So less than 100mA, call it 200mA to be safe if I add LEDs or some such.

That is high current for a simple charge pump.  Many switching regulator topologies can produce a bipolar output by using either a dual inductor or an inductor to drive a charge pump.  There is no requirement to use a specialized dual output switching regulator IC.  The last examples in the LT1616 datasheet and Linear Technology design note 232 show this.