Powering an Op-Amp with a split supply

Thread Starter

greenisland

Joined Jan 7, 2020
63
I am new to this forum and a novice, so take it slowly. I have received much help previously on a bench power supply query, this query is along the same lines. I have been away from this site for a while but I have continued experimenting on breadboards with op-amps during this time.
I am using a 741 op-amp and I know a lot of people don't like this old chip but I am using it because I have many and if I break one, it's not a problem and it seems to work ok for my requirement at the moment ie. learning about op-amps.

I built a power supply previously using an op-amp in an Non-Inverting Configuration using a single rail supply of 23v, I had a voltage drop problem which I solved with the help of people on this forum, happy days!.

I was still confused about powering an op-amp with a dual supply, why go to the trouble (I know for audio you use both + and - waveforms). I now know if I want to lower the voltage down to zero, I need a negative rail.
I don't have a centre tapped transformer with @15-0-15v output. I read about using an op-amp to create a virtual ground, I seen circuit marked "A" online which outputs enough current to power circuit "B", it does work.
The pot P1 sets the + or - balance.

My question:

When I power the virtual ground op-amp with a 30v supply, I can set P1 to balance the rails very evenly or unevenly.
If I balance the output +15v -15v (slightly under this), I get a very limited range output in "B" circuit.
If I set the output to a very uneven output +17v -12v I get a an increased range in "B", regulation is good to about 12v.

I know the 741 is limited to 30v +/- 15v, but this 30v, can it be so heavy unbalanced?.
I have noticed if I go further and unbalance it more, +20v -10v, there is nothing to be gained, it peaks at around 12v, so is the max for the 741, 30v/2=15v minus 3v( as close as it can get to the rail) =12v?? or have I got this completely wrong?.
 

Attachments

Last edited:

dl324

Joined Mar 30, 2015
12,684
I know the 741 is limited to 30v +/- 15v, but this 30v, can it be so heavy unbalanced?.
There's nothing that says the positive and negative rails need to be symmetrical.

This is your B circuit drawn more conventionally to make function more obvious:
clipimage.jpg
With a 17.65V positive rail, the maximum guaranteed output voltage from the 741 is 13.65V. After subtracting the two B-E votage drops, you're at about 12V (17.65V - 4V - 1.4V).

If you use typical output voltages, you might get 17.65V - 1V - 1.4V = 15.25V; but your adjustment range limits you to 13V.
 

Audioguru again

Joined Oct 21, 2019
2,875
All opamps have a spec called "Common Mode Input Voltage Range" which for the uA741 is +/-12V when powered from +/- 15V.
That means the inputs do not work if they are within 3V from a supply voltage. In the 30V/3A power supply project I used TLE2141 opamps that have inputs that work down to and including the negative supply. That project originally used TL081 opamps that have inputs that go crazy if they are within 4V from its negative supply.

The TLE2141 opamp will allow your new power supply project to have its output go down to 0V without using a negative power supply voltage.

Your new circuit will work fine with an old 741 opamp if its negative supply is -3.5V or more.
 

Thread Starter

greenisland

Joined Jan 7, 2020
63
There's nothing that says the positive and negative rails need to be symmetrical.

This is your B circuit drawn more conventionally to make function more obvious:
View attachment 231347
With a 17.65V positive rail, the maximum guaranteed output voltage from the 741 is 13.65V. After subtracting the two B-E votage drops, you're at about 12V (17.65V - 4V - 1.4V).

If you use typical output voltages, you might get 17.65V - 1V - 1.4V = 15.25V; but your adjustment range limits you to 13V.
Thanks for your help Dennis,

If the 741 is limited to 30v, and can be balanced very unequally as you say, how far can you make it unbalanced? +28 -2v ?
 

Audioguru again

Joined Oct 21, 2019
2,875
A lousy old 741 work with no negative supply (0V) if its positive supply is from about 10V to +36V, its inputs are from +3V to 3V less than the positive supply and its output can go no lower or higher than about 3V to 5V away from a supply voltage.
With a single supply, a powerful virtual ground is not needed for an opamp since the current fir its = reference input is extremely small. Two high resistance resistors as a voltage divider and a filter capacitor makes its virtual ground.

With a split supply, an opamp does not have 0V (ground) on any part of it.
 

Thread Starter

greenisland

Joined Jan 7, 2020
63
Thanks for your help Dennis,

If the 741 is limited to 30v, and can be balanced very unequally as you say, how far can you make it unbalanced? +28 -2v ?
Thanks Audioguru for that clarity.
I have TLE2141CP op-amps from the previous power supply, I was unaware they could allow the voltage to go to zero without a negative supply.

I did not know if unbalancing the supply so much was safe practice.
But to understand this correctly, the 741 if powered by +25v and -5, it could achieve 21v using my circuit?
 

Thread Starter

greenisland

Joined Jan 7, 2020
63
Thanks Audioguru for that clarity.
I have TLE2141CP op-amps from the previous power supply, I was unaware they could allow the voltage to go to zero without a negative supply.

I did not know if unbalancing the supply so much was safe practice.
But to understand this correctly, the 741 if powered by +25v and -5, it could achieve 21v using my circuit?
Ok a lot of good food for thought to go on with from both Dennis and Audioguru, thank you both
 

Audioguru again

Joined Oct 21, 2019
2,875
Most opamps use NPN input transistors that need their emitters to have a negative supply if the inputs need to work at 0V.
But the TLE2141 opamp uses PNP input transistors to allow the inputs to work with no negative supply and go to 0V.

The opamp does not have a 0V connection so it does not know that you are "unbalancing" the 0V of a split supply.

If you want your 741 opamp and two transistors as a darlington circuit to produce 0V then the output of the opamp must be as low as about +1.2V which a 741 output can barely do so it needs some negative supply voltage.
For the output of the project to be 0V then the opamp input must also be about +1.2V but a 741 opamp input cannot go within 3V from its negative supply so its negative supply must be at least 3V - 1.2V= -1.8V.

For the circuit to have a max loaded output of +21V the transistors might produce a 3V loss and the output of the opamp might produce a 4V loss then the positive supply must be at least 21V + 3V + 4V= +28V.
 

MrChips

Joined Oct 2, 2009
23,090
No need to have unbalanced supply voltages. If you have a single 30VDC supply that will be fine.
All you need now is a virtual ground that is halfway between the supply rails.
Use one of your LM741 to create a virtual ground. This becomes your new common for your circuit.

1614220654723.png
 

dl324

Joined Mar 30, 2015
12,684
Thanks for your help Dennis,

If the 741 is limited to 30v, and can be balanced very unequally as you say, how far can you make it unbalanced? +28 -2v ?
Yes, but to have an output down to 0V, you need to have a negative supply of -3V; so 27 and -3V would work.

Why are you using a power transistor for Q1? A general purpose transistor would have a higher beta.
 

Thread Starter

greenisland

Joined Jan 7, 2020
63
No need to have unbalanced supply voltages. If you have a single 30VDC supply that will be fine.
All you need now is a virtual ground that is halfway between the supply rails.
Use one of your LM741 to create a virtual ground. This becomes your new common for your circuit.

View attachment 231356
I have made a virtual ground circuit on a breadboard with a pnp/npn current booster,see attachment photo at the first post. I can vary the imbalance with pot1.
 

Thread Starter

greenisland

Joined Jan 7, 2020
63
I have made a virtual ground circuit on a breadboard with a pnp/npn current booster,see attachment photo at the first post. I can vary the imbalance with pot1.
This is all very interesting and yet confusing for me at the same time, I have re-read the comments on this post to make sense of it, the datasheets are very confusing to a novice.

I did try using a general transistor first a 2n3904 then a 2n2222 instead of a TIP41a at the op-amp output at first, but did not see any difference (I only have two multimeters). Is there a differerce I am incapable of seeing?.

Another question, I used two transformers in this test, I didn't explain this at the start sorry, you probably noticed it, if you look at the photo you will see +23v at the collector of the 2sc5200, I did not want much current going through the op-amp splitter.
I also tested the splitter using a 2n2222/2n2907 and 2n3904/2n3906, the 2n2222/2n2907 did not work for long, but the 2n3904/2n3906 did work but were hot to the touch, tip41a/tip42a, no problem. Current seems to be around 50mA.

I tested the circuit with a car bulb, drawing 2amps at 12v. The circuit as presented operated from 0-12v fine, no drop on load, after 12v, 30mv then quickly 200mv drop.
 

Attachments

Last edited:

dl324

Joined Mar 30, 2015
12,684
the datasheets are very confusing to a novice.
What part is confusing you?
(I only have two multimeters).
You only need one.
Is there a differerce I am incapable of seeing?
2N3904 and 2N2222 have more current gain than TIP41, so there will be less load on the opamp. Less load lets it swing closer to the rails.
I used two transformers in this test, I didn't explain this at the start sorry, you probably noticed it
If you're using transformers, why don't you use one of them to generate your negative rail?
 

Thread Starter

greenisland

Joined Jan 7, 2020
63
What part is confusing you?
You only need one.
2N3904 and 2N2222 have more current gain than TIP41, so there will be less load on the opamp. Less load lets it swing closer to the rails.
If you're using transformers, why don't you use one of them to generate your negative rail?
2N3904 and 2N2222 have more current gain than TIP41, so there will be less load on the opamp. Less load lets it swing closer to the rails.
Ok, I can comprehend that, point taken, (more re-testing will commence) and these transistors can supply enough base current for the 2sc5200?

If you're using transformers, why don't you use one of them to generate your negative rail?
Both are not centre tapped transformers, I am dedicating one to the op-amp generating the neg. rail.

Trying to understand specifics, powering an op-amp, single/dual, current required by op-amp to function, amount of current op-amp can supply, rail to rail etc.. I'm sure it's all there on the datasheet if you can understand the terminology, but complicated for a novice/hobbyist.
 

dl324

Joined Mar 30, 2015
12,684
Both are not centre tapped transformers, I am dedicating one to the op-amp generating the neg. rail.
You can generate positive and negative voltages from a secondary without center taps if you use half wave rectification and you're okay with lower available current.

EDIT: Like this
clipimage.jpg
current required by op-amp to function
Most of the current consumed in an opamp will be in the output stage. From the positive rail, that will be limited by R9; 0.7V/25ohms = 28mA. From the negative rail is a bit more complicated. They usually incorporate that in an earlier stage due to PNP performance issues.
clipimage.jpg
amount of current op-amp can supply
It's in the datasheet:
clipimage.jpg
clipimage.jpg
rail to rail
LM741 is an early design so the input and output can't get within 3-4V of the rails.

If you use rail to rail opamps/comparators, read the datasheet. Some have R2R inputs, but not outputs, some have both. Some let you get close to the rails, some let you exceed the rails. Most modern opamps probably don't suffer inversion when you violate the common mode input voltage range; some non R2R will.
 
Last edited:

Audioguru again

Joined Oct 21, 2019
2,875
Your new schematic has no load and no set output voltage then the circuit is useless to measure and to talk about.

1) Set the output to +5V.
2) Use a 1 ohm load that draws 5A.
Then the 0.1 ohm resistor has 0.5V across it. The 2SC5200 has (23V - 5.5V=) 17.5V across it and a current of 5A so it heats with 17.5 x 5= 87.5W so it will need a huge heatsink and a fan.
3) The maximum base current of the 2SC5200 is 5A/35= 143mA.
4) The TIP41 has an emitter voltage of about +6.25V and has the collector current of 143mA so it heats with 6.25 x 0.143= 0.9W which would fry a little 2N3904 or 2N2222.
 

Thread Starter

greenisland

Joined Jan 7, 2020
63
2N3904 and 2N2222 have more current gain than TIP41, so there will be less load on the opamp. Less load lets it swing closer to the rails.
Ok, I can comprehend that, point taken, (more re-testing will commence) and these transistors can supply enough base current for the 2sc5200?

If you're using transformers, why don't you use one of them to generate your negative rail?
Both are not centre tapped transformers, I am dedicating one to the op-amp generating the neg. rail.

Trying to understand specifics, powering an op-amp, single/dual, current required by op-amp to function, amount of current op-amp can supply, rail to rail etc.. I'm sure it's all there on the datasheet if you can understand the terminology, but complicated for a novice/hobbyist.
Your new schematic has no load and no set output voltage then the circuit is useless to measure and to talk about.

1) Set the output to +5V.
2) Use a 1 ohm load that draws 5A.
Then the 0.1 ohm resistor has 0.5V across it. The 2SC5200 has (23V - 5.5V=) 17.5V across it and a current of 5A so it heats with 17.5 x 5= 87.5W so it will need a huge heatsink and a fan.
3) The maximum base current of the 2SC5200 is 5A/35= 143mA.
4) The TIP41 has an emitter voltage of about +6.25V and has the collector current of 143mA so it heats with 6.25 x 0.143= 0.9W which would fry a little 2N3904 or 2N2222.
Again the response and help received from people on this forum is amazing, it comes in so heavy and fast I have to sit back and re-read comments over again when you're all busy elsewhere, thanks.
 

Thread Starter

greenisland

Joined Jan 7, 2020
63
Again the response and help received from people on this forum is amazing, it comes in so heavy and fast I have to sit back and re-read comments over again when you're all busy elsewhere, thanks.
Speaking of large heat-sinks, I think I will have to bolt one to the side of my head after this.

Ok.....AG.When selecting transistors, the listed power dissipation on the spec. sheet is a way of determining how much current the transistor can carry at various voltages?, see attachment screenshot of 2n2222 specs, maximum collector current 800mA; I always knew I was miss-interpreting this figure, so high for such a small transistor.
So when looking at the specs of a transistor, the maximum current is only the maximum current the transistor can supply at only one ideal specific voltage, how do you calculate if it can carry the current you require at the voltage you require?.

I'm really going to take off the dunces hat and put on the big dunces hat now!!!.

Referring to the voltage splitter/virtual earth circuit "A", Pot1 is a voltage splitter, it is reducing or increasing the level of positive voltage to pin 3, the op-amp is constantly making a comparison with the voltage in the feedback loop to it, but it is always positive, isn't it?, it is all dc after all. In the regulation the voltage will surely always go one way, down upon load requiring the op-amp to increase the output voltage?.
But at the output of the op-amp there is two transistors one npn and the other pnp, push-pull config. if I remember, this increases/amplifys the current output of the op-amp, but how is it turning on the pnp transistor when the output is positive?, or is the output a waveform, partially positive and negative?, or have I oversimplified it and got it wrong as usual.
Sorry a lot of question marks.
 

Attachments

Last edited:

BobTPH

Joined Jun 5, 2013
3,006
o when looking at the specs of a transistor, the maximum current is only the maximum current the transistor can supply at only one ideal specific voltage, how do you calculate if it can carry the current you require at the voltage you require?.
No, the supply voltage has nothing to do with the max current. A 2N2222 can conduct 800 mA with a supply of 1.5V and it can with a supply of 30V.

Why? Wouldn’t the power be much higher at 30V?

The answer is, that the power dissipated in the transistor is the current times the Vce, i.e. the voltage across the collector and emitter. Notice the supply voltage is not in the equation.

In order to safely conduct the full max current, the transistor must be in saturation. When it is, the Vce is low. At 500 mA the PN2222 lists a saturation voltage of 1V. Then the power dissipated will be 1/2W which is the max the TO92 package can handle without a heat sink. For higher current, you would need a heat sink, but the PN2222 maxes out at 600mA anyway.

Bob
 

Audioguru again

Joined Oct 21, 2019
2,875
The 2N2222 "survives" 800mA but like most transistors it works poorly above about 0.4 times its absolute maximum "do not exceed" rating. At 320mA some of them barely work properly. All of them work well at 200mA if they are allowed to stay cool.

I did not know the antique 2N2222 is available in a surface mount package, it started in a metal can.
 
Top