BUF634: Using BUF634 as a mid-rail splitter with unbalanced load

Uri Garbi

Part Number: BUF634
Other Parts Discussed in Thread: TINA-TI, OPA2810,

Hello,

I am using BUF634 to derive a mid-rail virtual ground in a 24V single supply system, thus creating a +12V and -12V dual supply.

I have a constant load of 50mA that draws its current from the positive side of the virtual ground.

When simulating this in TINA-TI - I see a significant decrease (about 0.5V) in the mid-rail voltage. Instead of 12V I get 11.5V.

This can be solved by applying a constant load by a resistor between the -12V and the virtual ground, but I was hoping to avoid having to use such a resistor (0.6W).

To clarify - I use this virtual ground for other parts of the circuit - so need it to be stable and exactly in half point.

Is there another way?

thanks

Uri

over 5 years ago

0 Sean Cashin over 5 years ago

TI__Mastermind 18036 points

Hi Uri,

Can you provide a circuit so I can better understand what you mean? The BUF634 is high current, but it is open loop, which means it will have less reliable load transient response and higher source impedance than a closed loop op amp. You could try putting it inside the loop of another op amp, buffering the output. This will improve your error correction significantly.

Best regards,

Sean

0 Uri Garbi over 5 years ago in reply to Sean Cashin

Intellectual 285 points

Hi Sean,

See the attached circuit. basically follows the datasheet application section example. This circuit draws 50mA from the 24v positive supply rail (sink through the BUF output).

There is additional current draw from the negative supply rail but it is much smaller.

See also attached the simulated voltage between the 24V positive rail and the BUF output. It is 11.58V when the 50mA is drawn, and goes up to about 12V when the 50mA stops.

Can you suggest a closed loop circuit (and op-amp) suitable for this scenario to be looped with the BUF?

Thanks

Uri

0 Sean Cashin over 5 years ago in reply to Uri Garbi

TI__Mastermind 18036 points

I suggest using the OPA2810, like the front page of the BUF634A datasheet, www.ti.com/.../buf634a.pdf, just use gain=+1 instead of +21, and without an input high pass filter. Connect Vin in your circuit to the OPA2810 Vin+, and Vo to the BUF634A output.

0 Uri Garbi over 5 years ago in reply to Sean Cashin

Intellectual 285 points

Thanks Sean.

Looks like this will work out fine.

Regards

Uri

0 Michael Steffes over 5 years ago in reply to Sean Cashin

Guru 48395 points

Hey Sean,

The reason the gain is so high in that BUF634 updated front page is

1. The Buf634 imbedded inside the loop adds a lot of phase shift, essentially turning the OPA2810 into a decomp device - hence the high gain to get back to stability

2. Going gain of +1 in that circuit will almost certainly oscillate

3. You should set this up in TINA for a phase margin sim as described here -

0 Uri Garbi over 5 years ago in reply to Michael Steffes

Intellectual 285 points

Hi Michael,

Do you see this oscillation risk in the following circuit? This is a rail splitting circuit, aimed for creating a +/- 12V supply from a 24V supply, where the positive rail consumption is significantly higher (50mA) than the negative rail.

Thanks

Uri

0 Michael Steffes over 5 years ago in reply to Uri Garbi

Guru 48395 points

Yes, that 10pF cap might help, but really the best thing to do is to set up and run the LG phase margin sim.

Hopefully Sean can do that.

0 Uri Garbi over 5 years ago in reply to Michael Steffes

Intellectual 285 points

Thanks Michael.
Sean - can you help with this or direct me how to perform this simulation?
Regards
Uri

0 Michael Steffes over 5 years ago in reply to Uri Garbi

Guru 48395 points

Hello Uri, had a bit of time from some other active filter work I was doing,

The attached word file will step through a design option

Also, I will attache a LG sim for where I left this circuit,

I have to, I think, reply in a 2nd post to then attach the closed loop sim. Saved both these as V7 Tina files,

Here is a recent VFA stability article I am drawing on a bit,

Mid Supply driver design discusion.docx

CLosed loop LG sim OPA2810 BUF634.TSC

0 Michael Steffes over 5 years ago in reply to Michael Steffes

Guru 48395 points

Well I didn't see that 2nd post with the closed loop output impedance made it - this ckt might have some problems if the mid scale line has a lot of capacitance on it, I am doing this with 1nF there now, but it really fails if I put 1uF there.

CLosed loop Zo test.TSC

0 Uri Garbi over 5 years ago in reply to Michael Steffes

Intellectual 285 points

Many thanks for this Michael. This is quite helpful.
I have very little load capacitance on this mid-line - so I guess my circuit with the 10pF should suffice.
Is that your understanding as well?
Incidentally - the closed loop Zo test did not open on my TINA-TI. Any idea why this is?
Cheers
Uri

0 Michael Steffes over 5 years ago in reply to Uri Garbi

Guru 48395 points

Well, I had intentionally put a 1nF load to improve the source impedance looking back into the circuit.

Darn, I think Kai was telling me I have to close TINA to attach another file - what a pain, I currently have 10 other circuits open that I need to close, I will try that, can you please let me know if you can open it after I try again here.

6138.CLosed loop Zo test.TSC

0 Uri Garbi over 5 years ago in reply to Michael Steffes

Intellectual 285 points

Still can't open this one.

0 Michael Steffes over 5 years ago in reply to Uri Garbi

Guru 48395 points

Maybe I did not save as V7, try this one -

4064.CLosed loop Zo test.TSC

0 Uri Garbi over 5 years ago in reply to Michael Steffes

Intellectual 285 points

Yes. This one opens fine.
Thanks!

0 Michael Steffes over 5 years ago in reply to Uri Garbi

Guru 48395 points

So you can see I changed the circuit some - using something like that dual loop cap load driver in that article,

Also, I powered up the BUF634 as high as it would go shorting that one pin to -Vee. In this circuit, we are still losing the phase margin due to the BUF634 being inside the loop, so very tight layout would be needed. I have another approach in mind that might work better, but this one seems promising also.

0 Uri Garbi over 5 years ago in reply to Michael Steffes

Intellectual 285 points

Thanks Michael. Would be interested in hearing the other approach as well.
Cheers
Uri

Amplifiers

Amplifiers forum

BUF634: Using BUF634 as a mid-rail splitter with unbalanced load