# How does a split rail supply work and how to design one?

I need a split supply for a project I am working on and there's no way around it, however this is very unknown to me.

How does one even design one and how does it work? I find it confusing to do a KCL on it just so I can see the analysis but I am failing on that as well

The goal is use a 12VDC wall wart and split it into +/-6 V I would be really happy with that. It will not be driving anything heavy besides OP-AMPS and buffers.

Sample:

simulate this circuit – Schematic created using CircuitLab

This is the current problem Iam facing:

simulate this circuit

This is what I believe is the solution:

simulate this circuit

EDIT:

I am sorry the confusion I have laid on to people but here is my XY problem I am having.

Goal Design: To create an amplifier with auto gain control using the Arduino and a rheostat to achieve this.

How I achieved it: Firstly I had my audio signal going into the INPUT of the precision rectifier to convert the audio signal in a "DC Line" with a magnitude of the peak of the audio signal. From there I took the output of the precision rectifier and put it into A0 (Analog pin) of the Ardunio so it can read the value. From there I have the code that the Arduino executes that finds the suitable R2 value from the feedback equation of: (Gain = 1+ (R2/R1)), where Vout can be adjusted via code and Vin(Vp) was measured on the A0 line. Once found a R2 I do some basic R2/RS, where RS is the steps I had on the rheostat and then do some SPI stuff. The code then tells the rheostat (Which is in the NFB loop of the amplifier) To change the rheostat resistance and get the desired Vo regradless of Vin. All in terms of Voltage peaks

The flaw: I found a big flaw recently when I was doing some input/output Impedance stuff for some filtering for the output of the amp. I realized Depending on the Rs( Output impedance) of the output source (iphone or laptop) the Vin that the op - amp sees is a different Vin from the ardunio sees. I realized I should capture the Vs from the V+ terminal and not from the audio source. But the V+ node has a DC offset that is unwanted for the ardunio to read I guess.

Theoretical if this works, this can defy the source resistance and can be disregarded due to the rheostat compensating for the voltage loss due to the loading of the stages from the input.

simulate this circuit

EDIT: EDIT: My solution ? Will this work? Should Make node A the same voltage as Node B assuming 1khz at 0dB.

simulate this circuit

• Why do you think you NEED split rail supply and there is no way around it? Do you need DC coupling for some reason? Other than that, a split supply would be easier from 12VAC. – Justme Nov 21 '18 at 19:02
• Its hard to explain, but it comes down to having remove DC offset from a Audio signal going into a precision rectifier however if I add a Capacitor it nothing happens, the precision rectifier is adding DC back to the cap, so the only way I can think is adding a Split Supply to a buffer and have a cap go into. Ill add a schematic to show it. – Pllsz Nov 21 '18 at 19:08
• so, you've just reduced your "has to be a split rail supply" to "I think a split rail supply is the solution (but I don't understand what it is)", which might indicate you should be asking a question with a title like "How to remove a DC offset from an audio signal?" and explain what you've figured out so far - you could, for example, suggest a split rail supply. But honestly, if your DC blocking cap doesn't block DC, you've designed things incorrectly. – Marcus Müller Nov 21 '18 at 19:15
• I don't understand why you think you need a split supply, here. You need a power supply. – Marcus Müller Nov 21 '18 at 19:42
• I'd like to see a description of the "signal" that is driving $C_1$ and a discussion about why it should be rectified and averaged, as well as things like the sample rate and what the ADC values will be used to achieve, afterwards. – jonk Nov 21 '18 at 19:59

The actual problem is that you precision rectifier is measuring the negative cycle of your input waveform (which, by the simple fact that it seems to be somehow working must be an actual centered AC). During the positive cycle the opamp is saturated, not just by the presence of the diode but by the signal itself, this can be considered either a bug or a design feature in this case. By giving it a positive DC you have simply put it out of range.

You already have a split rail in your design. You created one for biasing node "A" at the input of the first opamp. You can do exactly the same thing for biasing the positive terminal of your super diode opamp, but this means that you will lose half your range as now the DC will determine your new minimum output value.

If, after you set your mid-supply bias, you AC-couple your super-diode it will work and behave as it should, but now the negative cycle of your input waveform defines the bias point of the op amp and keeps it saturated. This is what is known as a DC-recovery circuit. However, this bias point is the same as the rectified value.

The only thing missing, after AC-coupling, is to provide a discharge path for the output capacitor when the OP amp is saturated. You can do this by adding a resistor to ground in either the input or the output of the super diode.

But BTW: the way that AGC circuits normally work is by looking at the peak output of your amplified signal and trying to keep it constant. That is, setting the gain via a feedback algorithm not a feedforward one.

• Ahh I see, yeah I am doing it vis feedback algorithm I just dont understand how can my source - sink the DC offset so well, but when I have the op amp buffer sourcing the precision rectifier the rectifier isnt sinking the current? – Pllsz Nov 21 '18 at 23:59
• @Pllsz The rectifier does not sink any current in your configuration. All of the sinking was done by the source. – Edgar Brown Nov 22 '18 at 0:46
• Thats what i mean, why is my op amp not sinking like my source did? and what I mean by an op amp is I have a buffer going into the input of the rectifier instead of the source now – Pllsz Nov 22 '18 at 0:54
• I think I am just going to add a buffer between the Input audio signal and the input to the amplifier. That should do it. I should be able to use the Vin single again into the rectifier – Pllsz Nov 22 '18 at 1:10
• @Pllsz that won’t work. The opamp will clip and you won’t have negative excursion to rectify. – Edgar Brown Nov 22 '18 at 1:35

Because a picture is worth a thousand words. I will show this:

Is there any additional comment needed?

• But how does one pick the desired values? And is it save to us a two prong wall wart with it – Pllsz Nov 23 '18 at 18:25
• The voltage divider current should be at least ten times larger than the expected virtual ground current. – G36 Nov 23 '18 at 18:41