Capacitors in mains supplied rail splitter

[aneevuser]

I want to make a 18V to 9/0/-9V rail splitter along the lines of the op-amp buffered version from here:

https://tangentsoft.net/elec/vgrounds.html

with a push-pull backend for additional current capacity.

However, I will be using a mains powered 18V supply, whereas these designs all assume a battery powered supply. Given that, is there any need for the front end 220 uF capacitance used in these circuits?

[TimFox]

So long as the mains-powered 18 V DC supply is floating (and transformer isolated from the mains for safety), it can replace the floating battery in the single-battery to split-power-rails versions shown in your reference.
The author does not discuss the 220 uF capacitors, but surely your mains-powered supply will have at least that much capacitance at its output (filter capacitance).  Is your power supply regulated by a 7818 or similar device?
In general, op amps require capacitance between their positive and negative power pins for stability, usually something like 100 nF close to the device along with the power-supply filter capacitance.
The main thing about rail splitters is to ensure that the "gnd" output (whether a power ground or virtual ground) can supply enough current in the correct direction to handle any imbalance in current between the positive and negative outputs.  If you know you have a substantial difference in those currents, you can use a output topology that only supplies current in that direction.

[aneevuser]

Tim, thanks for the response. However, I've just discovered that I'm going to have to do some voltage regulation too (I managed to buy an unregulated supply - misread the spec) so I need to think about my precise requirements a bit more.

As for your questions re: unbalanced supply currents, I'll need to measure that. Should you be interested, the circuit that I'm ultimately going to be powering is the mic. amp that I put together from advice on this thread:

https://www.eevblog.com/forum/beginners/measuring-amplifier-noise-with-scope/25/

[TimFox]

In that circuit as drawn at the beginning of the thread, the imbalance in supply currents will be dominated by the output current into ground.  The rail splitter will not be very efficient in reducing hum and noise from the power supply, so make sure that the filtration and regulation are sufficient to achieve your noise requirement.

[aneevuser]

For this project, I have bought one of these:

https://uk.rs-online.com/web/p/ac-dc-adapters/1392270/

It's supposed to be an unregulated 18V supply, but having taking a look at the o/p on my scope, it seems to provide a very clean 26V, unloaded i.e. the 18V looks to be very nominal indeed, and I can see no measurable ripple. Is this to be expected from one of these devices? And is it an artefact of looking at the o/p without a load?

[TimFox]

A simple rectifier with filter capacitor C  will show very little ripple voltage without load current.  With load current I, the approximate peak-peak ripple voltage will be I T/ C, where T = 10 ms for 50 Hz or 8.33 ms for 60 Hz (for a full-wave rectifier).

[mikerj]

For this project, I have bought one of these:

https://uk.rs-online.com/web/p/ac-dc-adapters/1392270/

It's supposed to be an unregulated 18V supply, but having taking a look at the o/p on my scope, it seems to provide a very clean 26V, unloaded i.e. the 18V looks to be very nominal indeed, and I can see no measurable ripple. Is this to be expected from one of these devices? And is it an artefact of looking at the o/p without a load?

Now load it up until you get to the nominal 18v output and see how much ripple you get.

[aneevuser]

A simple rectifier with filter capacitor C  will show very little ripple voltage without load current.

I really ought to think before posting. That is in fact blindingly obvious, isn't it? With no current, the cap. can't discharge...

[aneevuser]

For this project, I have bought one of these:

https://uk.rs-online.com/web/p/ac-dc-adapters/1392270/

It's supposed to be an unregulated 18V supply, but having taking a look at the o/p on my scope, it seems to provide a very clean 26V, unloaded i.e. the 18V looks to be very nominal indeed, and I can see no measurable ripple. Is this to be expected from one of these devices? And is it an artefact of looking at the o/p without a load?

Now load it up until you get to the nominal 18v output and see how much ripple you get.

In fact, I think that I need to think about how much ripple I can tolerate, since I'm going to be driving a lowish noise preamp. The requirements for the whole thing may be more demanding than I initially thought.

Having a had a chance to consider the 26V that I'm seeing, and the fact that it's an unregulated supply, I guess that's a design feature, to allow sufficient dropout voltage headroom for any regulator that sits in front of it?

[TimFox]

You can lower the ripple substantially with a 78xx series three-terminal regulator after you measure the voltage and ripple when you load it by your required current.  If the ripple is high, make sure that the voltage at the bottom end of the ripple waveform does not go below the dropout voltage for the regulator input.

[aneevuser]

You can lower the ripple substantially with a 78xx series three-terminal regulator after you measure the voltage and ripple when you load it by your required current.  If the ripple is high, make sure that the voltage at the bottom end of the ripple waveform does not go below the dropout voltage for the regulator input.

Why a 78xx rather than something like a 317? Do the fixed voltage regulators give better ripple rejection?

[TimFox]

Check the data sheets.  The fixed ones are easier to use.  If you want adjustability, the 317 is better (since the ways to adjust 78xx are kluges).  The data sheets show the ripple rejection as a function of frequency.

[David Hess]

However, I will be using a mains powered 18V supply, whereas these designs all assume a battery powered supply. Given that, is there any need for the front end 220 uF capacitance used in these circuits?

The input capacitor is included in the battery powered circuit to lower the AC impedance which is relatively high in a 9 volt battery.  The same applies to a DC power supply if it is remotely located because of inductance in the wiring.

The operational amplifier has to supply relatively high current in a rail splitting application and this also requires better than normal decoupling of the power supply making the input capacitor even more important.