Other Parts Discussed in Thread: TLE2426,
TLV930x looks like a nice candidate as a rail-splitter as it seems to have relatively high output current.
Any issues using it as a rail-splitter?
This thread has been locked.
If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.
Hi,
rail splitters are really tricky because of the heavy capacitive load they have to drive.
What capacitive load?
The decoupling capacitors from VCC and VEE to virtual ground.
I would recommend to use a special virtual ground generator chip like the TLE2426, e.g.
Kai
TLE2426 is really old, big, and expensive. I hope TI refreshes it soon.
Also, it is really hard to find on the web site.... it's filed under "Video Amplifiers"? Weird.
This is just used for a reference point on an isolated bipolar bias supply; no capacitors from VEE to GND or VCC to GND.
If I put caps from those rails to GND, I would put a resistor (~20-30 Ohms) in series with amo output, and compensation zero in the feedback to stabilize it.
It would make a nice app note for this part.
-Steve
Steve,
Here is a circuit with a Zout test for the VGND. The impedance is very low for most frequencies but peaks smoothly at 5.3 ohms near 720 Hz.
Here is the same circuit setup for phase margin. Result was 83 degrees.
Simulation looks OK, but I have not built this circuit.
Thanks, looks like there are signs of life.
It would be great if you all could take the core out of this TLV930x and update the TLE2426 rail-splitter with it. Seems like a pretty easy project?
In this application, I don't really need the caps to terminate at VGND. Per the original diagram, the voltage source would be from an isolated secondary winding and rectifying diode; a single cap spans from VCC-VEE. The output of the rail-splitter simply "pins" the circuit to the external reference potential.
Another op-amp in this package then transmits a signal relative to this virtual ground, so the load on the rail splitter is actually resistive, not capacitive.
I'll mess around some more with this; the cap value you used might be pretty large (10u) for the application and the compensation zero is set down at ~318 Hz, so probably room for optimization there.
I like the TLV930x as it seems to be able to source more current and is relatively inexpensive.
Thanks, Best, Steve
Steve,
If you asked for a mid supply voltage reference then, I'd wouldn't have added any capacitance (except for the new cap I'd put on the non inverting pin to ground). The idea for the big capacitance is that I expected 0.1uF bypass caps all over the place plus a little more to get 10uF.
So add a cap on non-inverting input. Pull the load caps off of VGND. Make C5 open, short R6.
Kai-
Thanks; I am familiar with it; it is a very nice part I believe built on the TI Excalibur process.
However, it is only available in a now giant SO8 package, and its price starts about $0.70.
The TLV9301 is in a SOT23 ,$0.18, has decent output current capability, and is also a 40V device.
Furthermore, the TLV930x is a family of parts, and what is more likely to happen is 1 amp in a package of 2 or 4 will be devoted to this rail splitter, while the others are processing signals. From a system perspective, this will be the smallest, most economical solution.
-Steve