Power supply?

[jamboard]

I need to generate -5V/0/5V for my synth project but I don't want to mess with 230V / transformers - indeed I want to build something small I can connect to one of the numerous 12V/500mA bricks I have around.

I need something like 150mA on the +5V rail (backlit display, a lot of LEDs + the audio circuitry), 50mA on the -5V rail (used only for the audio circuitry). Currently, I have a 7810 -> 2.2k/2.2k divider -> 741 -> BD135+BD136 push-pull (741 + push-pull replacing a high-output current op-amp), but I am not sure it is well suitable for the unbalanced load.

If you have already built something like that, your advice is appreciated! There are plenty of virtual ground circuits out there but I am just lost with all these options, none of them really matching my use case.

[blue hell]

jamboard.

Does the backlight have to be ground referenced? If not, the current could be better balanced ...

[jamboard]

Hi,

Good idea, if the backlight could be individually powered by the -5V/0V, this would shift ~50mA from the + rail to the - rail and balance the thing. But I'm using this:
http://www.sparkfun.com/commerce/product_info.php?products_id=813

which uses a common ground and supply for the back light, the controller, and the LCD itself. So the ground and +5V supply have to be the same as that of the microcontroller connected to the display (and pretty much all the digital section of the thing).

[blue hell]

Ah yes, I see .. a smart backlight, nice thing in itself ... but the pulse width modulation in combination with a virtual ground may not be the best idea indeed.

Can't come up with a smart idea for avoiding a real double supply tho.

At least make sure to have some proper AC decoupling for the virtual ground to filter out the PWM stuff (to the + and - supply lines).

[jamboard]

[blue hell]

Looks reasonable, but would check with scope ... re. how they do it ... probably a switching power supply.

[bod]

what about the ICL7660 (i think thats what its called), you could run 12v into that, create a -12v supply, regulate down to -5v with a 7905, and a 7805 for the +5v from the same 12v that was supplying the icl7660?

i've been using this for my cem3394 experiments and a variety of other things. the only issue with not regulating the negative supply, is that there is a slight voltage drop, so instead of -12v you get something like -11.6.

maybe there is a way round this though if any one knows?

[spippoli]

Don't know if a icl7660 would do, but anyway max1044 is a better chip. They're pin equivalent but with icl7660 you get a 10KHz ripple that will result in noise in the audio, while with max1044 (pins 1 and 8 bridged) the ripple is well out of audio frequencies

[jamboard]

Two questions about this switched capacitors thing:
- I checked the datasheet of several 7660 likes and they all work at 10kHz. Would this create an audible whine/buzz? If you have a scope, can you observe this? Do you know some equivalent of the 7660 with a higher switching frequency? The MAX1044 (a 7660 clone) has a "BOOST" pin to enable a faster switching frequency outside the audio range, but it's not available in a DIP package...
- What would be the respective advantages/disadvantages of:
* Getting -12V with a 7660 and regulating each branch with a simple 7x05.
* Getting +5V with a 7x05 and inverting it with a 7660?

[jamboard]

[jamboard]

Ahh, I think that would do:
http://uk.farnell.com/microchip/tc1044scpa/dc-dc-volt-converter-1044-dip8/dp/1196840

The datasheet answer my question... There's a voltage drop equivalent to a 70 Ohm resistance. Taking 50mA at 5V would cause a massive 3.5V drop, so it's better to get -12V first and regulate each rail.

So let's go for a 1044, and a pair of 7x05 on each rail. I'll compare this with the "virtual ground" approach, I have a found a cheap power op-amp that could be used for it: http://uk.farnell.com/jsp/search/productdetail.jsp?sku=1094203

[bod]

i agree that you do get ripple noise with the icl, and it bugged me endlessly for many moons... however, theres a thread in the MFOS forum that covers this, and someone there, who's name escapes me, found that connecting pin one of the icl7660 to the positive supply worked and removed the noise, at least at audio levels!

it work for me as well with the ones i had bought from Rapid in the uk. a nice wee hack!!

[jamboard]

[bod]

http://electro-music.com/forum/topic-32855.html

thats the thread, i really made my day reading that and then trying it! i'm waiting on a new scope (if SWMBO takes a hint for my christmas poresent.... ) to see just how bad it is, but i cant hear it any more, and i works fine for my sound lab and filter modules i've built recently. i'm going to take them down to my friends recording studio to see if we can pick anything up on highend recording equipment.

the data sheet for the icl7660a (the 12v model) shows pin one as no connection, but maybe these are all just the same chips with different branding and different datasheets?
they can only pump out 100ma, but thats enough for most of my projects and modules, and cheap enough to build into each one.

[jamboard]

Thanks all, it worked perfectly with a 1044!