Author Topic: Rail splitter circuit (looking for approval :D)  (Read 2127 times)

0 Members and 1 Guest are viewing this topic.

Offline MaruloTopic starter

  • Newbie
  • Posts: 5
  • Country: sk
Rail splitter circuit (looking for approval :D)
« on: May 04, 2019, 04:34:30 pm »
Hey!

So, I've been trying to make a symmetrical supply from an unsymmetrical for quite some time now, and I think (or at least hope) that I got it!
Problem is, I don't currently have any MOSFETs and ordering them will take some time so I thought, "Why not just ask?".
So I attached the schematic, so please check it out. My train of thought went like this (hopefully you can correct me if I went wrong somewhere):

I can make a normal resistor voltage divider using and buffer it with an op amp, but the problem is, the op amp can't source/sink enough current for, for instance, a modular synthesizer.
So, can I buffer the buffer op amp? I looked online and found a circuit similar to this, briefly simulated it and all seemed good :-+.

Thanks for the help guys!
 

Offline MrAl

  • Super Contributor
  • ***
  • Posts: 1666
Re: Rail splitter circuit (looking for approval :D)
« Reply #1 on: May 04, 2019, 05:35:30 pm »
Hi,

I dont think that is a good idea because both mosfets will be turned on and thus conducting a lot of current.
 

Offline MarkF

  • Super Contributor
  • ***
  • Posts: 2763
  • Country: us
 

Offline MaruloTopic starter

  • Newbie
  • Posts: 5
  • Country: sk
Re: Rail splitter circuit (looking for approval :D)
« Reply #3 on: May 04, 2019, 06:02:47 pm »
Thanks!

I'm guessing I'm mostly interested in this part of the circuit (see attachment).
I know I may sound kinda dumb, but could you explain why using BJTs is better in this situation?

Thanks man! Really helped!  ;D ;D ;D
 

Online Zero999

  • Super Contributor
  • ***
  • Posts: 20357
  • Country: gb
  • 0999
Re: Rail splitter circuit (looking for approval :D)
« Reply #4 on: May 05, 2019, 09:32:28 am »
Hi,

I dont think that is a good idea because both mosfets will be turned on and thus conducting a lot of current.
Actually no, the MOSFETs are configured in class B, not class A.

The problem will be poor transient response, as the op-amp's output will have to change very quickly, when the direction of the output current changes.
Thanks!

I'm guessing I'm mostly interested in this part of the circuit (see attachment).
I know I may sound kinda dumb, but could you explain why using BJTs is better in this situation?

Thanks man! Really helped!  ;D ;D ;D
Better but it's prone to oscillation. Look at the end of the thread linked to by MarkF, for some circuits which have fixed the problem.
https://www.eevblog.com/forum/beginners/virtual-ground-push-pull/msg2374209/#msg2374209
 

Offline MrAl

  • Super Contributor
  • ***
  • Posts: 1666
Re: Rail splitter circuit (looking for approval :D)
« Reply #5 on: May 05, 2019, 02:11:41 pm »
Hi,

I dont think that is a good idea because both mosfets will be turned on and thus conducting a lot of current.
Actually no, the MOSFETs are configured in class B, not class A.

The problem will be poor transient response, as the op-amp's output will have to change very quickly, when the direction of the output current changes.
Thanks!

I'm guessing I'm mostly interested in this part of the circuit (see attachment).
I know I may sound kinda dumb, but could you explain why using BJTs is better in this situation?

Thanks man! Really helped!  ;D ;D ;D
Better but it's prone to oscillation. Look at the end of the thread linked to by MarkF, for some circuits which have fixed the problem.
https://www.eevblog.com/forum/beginners/virtual-ground-push-pull/msg2374209/#msg2374209

Hello there,

Oh ok i see it now it looked like the sources were connected to the rails but that's not the case.

Yes this circuit will suffer from bad crossover distortion which for this circuit equates to bad transient response.

The key to getting it right is to use TWO op amps driving each mosfet individually.  See if you can figure out the configuration without me drawing it up.  I'll draw it up if need be though.
« Last Edit: May 05, 2019, 02:14:42 pm by MrAl »
 

Online Zero999

  • Super Contributor
  • ***
  • Posts: 20357
  • Country: gb
  • 0999
Re: Rail splitter circuit (looking for approval :D)
« Reply #6 on: May 05, 2019, 04:15:35 pm »
Yes, if the MOSFETs were connected the other way round, they would form another inverting stage and the circuit would just latch up and sit with its output saturated at either the positive or negative rail.

I don't see how using more than one op-amp would solve the problem of crossover distortion. It's not something I've seen in an amplifier circuit before.
 

Offline MaruloTopic starter

  • Newbie
  • Posts: 5
  • Country: sk
Re: Rail splitter circuit (looking for approval :D)
« Reply #7 on: May 05, 2019, 05:13:25 pm »
Yes, if the MOSFETs were connected the other way round, they would form another inverting stage and the circuit would just latch up and sit with its output saturated at either the positive or negative rail.

I don't see how using more than one op-amp would solve the problem of crossover distortion. It's not something I've seen in an amplifier circuit before.

Hello! I've done a few spice simulations and it looks pretty good.
Also, I have to say I've seen the dual OpAmp configuration multiple times, mainly in rail splitters similar to this one. I'm still wondering if there's something I could improve, and mainly, how...

Thanks again lads  :-+
 

Offline MaruloTopic starter

  • Newbie
  • Posts: 5
  • Country: sk
Re: Rail splitter circuit (looking for approval :D)
« Reply #8 on: May 05, 2019, 05:24:11 pm »
Also I tried to make a simulation with some "parasitic" properties, still looks pretty good.
 

Offline MrAl

  • Super Contributor
  • ***
  • Posts: 1666
Re: Rail splitter circuit (looking for approval :D)
« Reply #9 on: May 05, 2019, 09:47:51 pm »
Yes, if the MOSFETs were connected the other way round, they would form another inverting stage and the circuit would just latch up and sit with its output saturated at either the positive or negative rail.

I don't see how using more than one op-amp would solve the problem of crossover distortion. It's not something I've seen in an amplifier circuit before.

Hi,

I came up with the configuration back in the 1980's when i wanted to create a near crossover distortion free amplifier for audio.  Back then mosfets were just becoming popular and i realized that they offered a big advantage over bipolars when driven as voltage followers.  Namely, the control was smoother to begin with.
Then i realized that i needed to drive the output both positively and negatively in order to keep the impedance low for each section of a test sine wave.  So i got the idea to use two amplifiers in parallel, one driving high and one driving low.  Then it was just a matter on how to adjust the quiescent current because they both have to be on at all times at least a little bit.
That works out pretty well because each op amp controls it's own mosfet.

As you can see the post that follows your shows such a configuration although that is a bit more basic then we actually have to use.  We have to be able to drive each mosfet with gate voltage enough even with higher amplitude output (for audio work).  For a rail splitter it may not matter as much though.
The basic circuit should have some small value resistors in series with each source.  That helps to mitigate the op amp offset voltages.  We also have to make sure we have both turned on slightly with no load.
So a few little adjustments and it will be great.
 

Offline MrAl

  • Super Contributor
  • ***
  • Posts: 1666
Re: Rail splitter circuit (looking for approval :D)
« Reply #10 on: May 05, 2019, 09:50:53 pm »
Also I tried to make a simulation with some "parasitic" properties, still looks pretty good.


Hi,

That's nice but you may want to add small value resistors in series with each mosfet source.
Check the current from the power supply with no load and with load too.  It should be low if this is done right.
 

Online iMo

  • Super Contributor
  • ***
  • Posts: 5570
  • Country: va
Re: Rail splitter circuit (looking for approval :D)
« Reply #11 on: May 05, 2019, 09:54:49 pm »
The best rail splitter is none rail splitter :)
Readers discretion is advised..
 

Offline exe

  • Supporter
  • ****
  • Posts: 2647
  • Country: nl
  • self-educated hobbyist
Re: Rail splitter circuit (looking for approval :D)
« Reply #12 on: May 05, 2019, 09:55:58 pm »
Hello! I've done a few spice simulations and it looks pretty good.

I don't see component values for opamps and fets. Without them ltspice uses unrealistic models, which are not representative for real circuits.
 

Online Zero999

  • Super Contributor
  • ***
  • Posts: 20357
  • Country: gb
  • 0999
Re: Rail splitter circuit (looking for approval :D)
« Reply #13 on: May 05, 2019, 10:59:55 pm »
Yes, if the MOSFETs were connected the other way round, they would form another inverting stage and the circuit would just latch up and sit with its output saturated at either the positive or negative rail.

I don't see how using more than one op-amp would solve the problem of crossover distortion. It's not something I've seen in an amplifier circuit before.

Hi,

I came up with the configuration back in the 1980's when i wanted to create a near crossover distortion free amplifier for audio.  Back then mosfets were just becoming popular and i realized that they offered a big advantage over bipolars when driven as voltage followers.  Namely, the control was smoother to begin with.
Then i realized that i needed to drive the output both positively and negatively in order to keep the impedance low for each section of a test sine wave.  So i got the idea to use two amplifiers in parallel, one driving high and one driving low.  Then it was just a matter on how to adjust the quiescent current because they both have to be on at all times at least a little bit.
That works out pretty well because each op amp controls it's own mosfet.

As you can see the post that follows your shows such a configuration although that is a bit more basic then we actually have to use.  We have to be able to drive each mosfet with gate voltage enough even with higher amplitude output (for audio work).  For a rail splitter it may not matter as much though.
The basic circuit should have some small value resistors in series with each source.  That helps to mitigate the op amp offset voltages.  We also have to make sure we have both turned on slightly with no load.
So a few little adjustments and it will be great.
I did have a brief play with LTSpice afterwards and the result was as I predicted: crossover distortion. The two op-amps still have to swing two gate thresholds, when the output current changes direction. Using two op-amps in that configuration doesn't solve anything. Something to ajust the quiecent current, to bias it into class AB is required.

I'll post the .asc file when I get time, if no one beats me to it.
 

Offline MrAl

  • Super Contributor
  • ***
  • Posts: 1666
Re: Rail splitter circuit (looking for approval :D)
« Reply #14 on: May 06, 2019, 12:16:24 pm »
Yes, if the MOSFETs were connected the other way round, they would form another inverting stage and the circuit would just latch up and sit with its output saturated at either the positive or negative rail.

I don't see how using more than one op-amp would solve the problem of crossover distortion. It's not something I've seen in an amplifier circuit before.

Hi,

I came up with the configuration back in the 1980's when i wanted to create a near crossover distortion free amplifier for audio.  Back then mosfets were just becoming popular and i realized that they offered a big advantage over bipolars when driven as voltage followers.  Namely, the control was smoother to begin with.
Then i realized that i needed to drive the output both positively and negatively in order to keep the impedance low for each section of a test sine wave.  So i got the idea to use two amplifiers in parallel, one driving high and one driving low.  Then it was just a matter on how to adjust the quiescent current because they both have to be on at all times at least a little bit.
That works out pretty well because each op amp controls it's own mosfet.

As you can see the post that follows your shows such a configuration although that is a bit more basic then we actually have to use.  We have to be able to drive each mosfet with gate voltage enough even with higher amplitude output (for audio work).  For a rail splitter it may not matter as much though.
The basic circuit should have some small value resistors in series with each source.  That helps to mitigate the op amp offset voltages.  We also have to make sure we have both turned on slightly with no load.
So a few little adjustments and it will be great.
I did have a brief play with LTSpice afterwards and the result was as I predicted: crossover distortion. The two op-amps still have to swing two gate thresholds, when the output current changes direction. Using two op-amps in that configuration doesn't solve anything. Something to ajust the quiecent current, to bias it into class AB is required.

I'll post the .asc file when I get time, if no one beats me to it.

Hi,

Well then something isnt being done right.  I'll take another look too when i get a chance.
Remember each mosfet must stay on all the time at least a little or there will be crossover distortion.  That means each op amp stays in the linear mode and never saturates in either direction.

 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf