WELCOME!!

[green]

Been thinking of trying a TLE2426 for my detector. Did a circuit in spice to try. Change in volts equals change in current times R1. Used a LT6234 to simulate, need to find an available 0Pamp that can sink and source ground current and doesn't oscillate with a low R1. Probably there is a better way.

[Davor]

OK, so you think you need a virtual ground at about 1 ohm impedance? Well, Sijosae splitter was not intended for such low impedance, and perhaps you actually need it that low, but with a little care most probably not. In case you are a noise freak, a slight complication of a basic Sijosae circuit may do precisely what you need, like this

It runs with a quiescent current you may set with a trimmer, and at 1mA it gives below 1 ohm, and below 5nV/sqrt(Hz) noise.

[Davor]

Can you say more?

This is a basic idea:


And you may find a practical design for powering something much more powerful here: http://www.mtukrc.org/download/clark...paper_2009.pdf (1.1MB)

There are also a single capacitor versions that are just as good. Benefits of such balancing is more capacity.

[Teleno]

Here's a regulated rail splitter with 3.4mA quiescent current and a ripple of 0.25mV for a current of 130mA.

[Thomas]

Like the current mirrors

[Monolith]

Why not use a dedicated rail-splitter? like the one from TI, link below?

http://www.ti.com/lit/gpn/tle2426-ep

http://www.electronicsweekly.com/mar....kBpRFQ6N.dpuf

[Davor]

It is not a common part, that's all.

[Achillion]

The TLE2426 can only source or sink 20mA which is not enough for some applications. With the LT1210 as a buffer it can do about 1A.

[Old cart]

That is why I like the use of common regulators as the rail splitters. Available everywhere, capable of a lot of current, and rugged. The design is also simple and requires no adjustment. The rail splitter would be fine also except for availability. I do believe that the whatever we use should be capable of sourcing and sinking significant current because we have no idea what the rest of the design wil look like in terms of required peak currents. I can see audio for example easily imposing highly peak current requirements. Even the peak currents in the input protection diodes can exceed 100 mA so would modulate ground in some of these schemes 2 volts p-p.

One of the main reasons of going with a rail splitter is to hold down supply noise. If the output is soft, has high impedance, then asynchronous events like audio could easily introduce the very noise we are trying to get at back into the system.

It is probably best to just build a couple of these proposed designs up and measure them. If we agree this is a good idea then we need to devise a test protocol. This need not very complicated. Maybe just vary each of the supplies independely and separately to simulate the effects of pulsing and other noises. Likewise, the output load could be pulsed. I think you could do the output test just using a good 50 ohm generator. For the input test we might also need a transformer to couple the signal source into the supply.

[Old cart]

Actually, I just looked at the specs for the TI rail splitter and they are very good. Especially is you use the 8 pin version with a noise filtering pin. Likewise they waste very little power doing their job. The main limitation is they can only sink or source about 20 mA of current.

[Old cart]

Can any of the members here comment on the availability of parts in their region? Is the problem that the parts are too expensive or that you can just not get them?

[Thomas]

No comments on Telenos rail splitter? I sure like it.

[Old cart]

Actually I did comment that I liked it-the main problem being parts count.( never showed up) I do think the spitter must handle>100mA with minimum spiking due the current the protection diodes dumping current into ground during transmit and fly back times. The ground voltage has to recover within maybe 5 uS after the peak of the flyback to allow accurate sampling. I don't know how close to ground this must be.

Any input?

[Old cart]

Here's a regulated rail splitter with 3.4mA quiescent current and a ripple of 0.25mV for a current of 130mA.



[ATTACH]34516[/ATTACH]
[ATTACH]34515[/ATTACH]

Have you built this?

[Old cart]

Another alternative is to just do a supply like was in another project. This apparently works quite well. Note the battery in this case is about 12 volts. The battery alarm can be eliminated. You could also use 4 and 8 volt fixed regulators. The coil drive would have to be 8 volts (+/-4 volts) in this case.

Can anyone tel me how to attach an image? It shows fine in the message preview.