Voltage regulator recommendations

[newbie666]

Hello,

I need to generate following voltages for a chip:
+7V@10uA (max)
-7V@10uA (max)
+5V@5uA (max)

from +-15V supplies

I only have experience with lm317/337 and I was thinking about using them at first, but it seems like a terrible waste of power considering that minimum current at which lm317 starts to regulate is around 10mA if I remember correctly.

Can you recommend me a micropower linear voltage regulator that would take this relatively high input voltage (15V).
Would be nice if it came in the package that is possible to hand-solder.

Thanks!

[james_s]

How accurate do those voltages need to be? I think I would look at using a zener diode or if you want something more precise try a reference like the TL431. There are lots of precision low power voltage references available intended for things like DAC references.

[newbie666]

5V supply is for powering digital part of the chip so doesn't have to be accurate (datasheet specifies 4.5-5.5V)

+-7V supplies for analog part can be in a wide range of 4.5-7V actually, but I'd like them to be as close to 7V as possible (for headroom) AND symmetric as this will determine dc offset at the output.

I'm going to look into zener diode regulation, in the meantime if you have any other suggestions shoot

[oPossum]

Maybe a voltage divider from the +/- 15V rails buffered by a (quad) op-amp or emitter follower.

Is this for audio switching using CMOS switches like 4066, 4051, 4052, 4053?

[newbie666]

hmmm, quad opamp + a voltage reference may be a great idea. I'll get the parts and try it on the breadboard.

[Kjelt]

Seeing your allowable tolerances is resistor voltage divider from the +/- 15V not adeqaute enough ?

[Alex Nikitin]

hmmm, quad opamp + a voltage reference may be a great idea. I'll get the parts and try it on the breadboard.

Yes, REF3325 +LT1496 + some high-ohm resistors can get you all the voltages with about 10uA current consumption.

Cheers

Alex

[schmitt trigger]

Is your project battery powered? If it is line powered 10 mA from each rail should be of minor consequence.

Otherwise, the quad opamp with a single voltage reference plus a a few high valued resistors should get you there. There are quad micro-power opamps available if a LM324 seems power hungry.

[Cliff Matthews]

Nothing fancy but "cheap-as-cheese" 78/790x's can be used as variable.. (Vref inside is just higher than lm317)
eg: https://www.eleccircuit.com/dual-adjustable-power-supply-circuit-diagram/

[Zero999]

Nothing fancy but "cheap-as-cheese" 78/790x's can be used as variable.. (Vref inside is just higher than lm317)
eg: https://www.eleccircuit.com/dual-adjustable-power-supply-circuit-diagram/

The LM78/79xx series are also not specified to work without a load, note the data sheet only gives the output voltage range from I_O = 5mA to 1A, but they mostly do work  at 0mA without any problems. They also have a quiescent current of 4.2mA typical, 8mA maximum.
https://www.sparkfun.com/datasheets/Components/LM7805.pdf

Internally, the LM78/78xx are very different to the LM317/LM337 and will perform relatively poorly if used as a floating regulator, with a higher output voltage, than their rating. Making a 10V regulator from an LM7805 will not give as gooder performance as one made from an LM317.

[David Hess]

With those low current levels, I would use operational amplifiers.  Instead of a reference, I might even have them track the positive or negative 15 volt supply voltage.

If low AC output impedance is required or if they have to drive ceramic decoupling capacitors, it may be necessary to frequency compensate them to handle a capacitve load.  One counter intuitive way to do this is to add like a 10 to 100 microfarad electrolytic capacitor directly from their outputs to ground but this should be tested for the operational amplifier selected.  Lower power operational amplifiers are more picky about driving output capacitance although some parts are designed to have unlimited capacitive load driving capability and they would be ideal in an application like this.

[newbie666]

Thanks Davis,

I think I'll go with this solution. I'll use a reference rather than track power supply rails though.

Do you have any suggestions for the opamps that would be good at driving capacitive load?

I've started googling opamps driving capacitive load and I found this nice writeup with scope screenshots and all:
http://www.analog.com/en/analog-dialogue/articles/techniques-to-avoid-instability-capacitive-loading.html

Posting here in case somebody else has the same problem.

I found this part from TI: LM8272, however it's pricey at over 3 bucks in single quantities.

[james_s]

Try something cheap to begin with, if that doesn't work then look toward fancier parts. Something like the TL084 would probably work.

[David Hess]

If you compensate it right, then any operational amplifier can be used.  Linear Technology has the LT1351/LT1352/LT1353 which might be ideal but they are not cheap.  I did a search for TI parts and did not find any suitable ones.  The three alternatives using frequency compensation are:

1. Swamp the output with like a 10 to 100 microfarad aluminum or tantalum electrolytic to ground resulting in dominant pole compensation.  This can provide about the best performance but it should be tested.  Do not use a polymer electrolytic (or ceramic or film) part because low ESR will defeat the purpose of using this method.

2. Use the capacitive load driving circuit shown in the last example here.  I have actually had better results with method 1 above.  It is not that this does not work but it has a mid-band rise in impedance which may be a problem with some loads.

3. Use an operational amplifier which provides external compensation so its capacitive load driving capability can be increased and then use 1 above.

There are some power buffers like the LT1010 which can also be used but an operational amplifier is more economical and better at low power.