split supply problem

Thread Starter

philbowles2012

Joined Mar 28, 2013
42
@admin - sorry, I posted this in FB & suggestions by mistake!

I am trying to build a split supply for some op amp circuits. I came across this circuit http://www.instructables.com/file/F4XPE5WH3Z3UFPV

I'm a bit of a n00b to all this and I simply don't "get" it at all.

I tried it in a simulator and then built it. In both cases, I get the requisite +12/-12v, so what's the problem then?

The problem is when I try to USE it! If I connect an led(+ 470R resistor) between 0 and +12V then the -ve side drops to -19 and the "+12" is now +5v. Any "load" on either side totally screws up the "balance"

I have a theory...if it's right, I'll go "Eureka!" and carry on gradually learning electronics...if not, my head will probably explode and I will have to accept I have far more learning to do than I thought. Your help is therefore appreciated:

I'm using a wall-wart, not a battery. I'm thinking that "-ve" here is genuine, earth-potential 0v and that the wart isn't "floating". This is sort of "confirmed" as it has a metal earth lead rather than a dummy plastic one.
Also in my simulator, it won't run until at least something is at ground potential...hence same results in sim?

Hmmm - i tried it with a "floating" supply (plastic earth pin) and a battery pack...my theory is wrong, my head now explodes as any "load" above the 0v rail causes major changes in values of + and - rails no matter the type of value of the input supply

What am I missing?

heeeeeeeeeelp!
 

#12

Joined Nov 30, 2010
18,224
It's a bad circuit for your purposes. The 1k resistors limit it to very low current..not even enough to run an LED! Throw it out and start over.

If you know how much current you want, we can whip something up here.
 

Thread Starter

philbowles2012

Joined Mar 28, 2013
42
It's a bad circuit for your purposes. The 1k resistors limit it to very low current..not even enough to run an LED! Throw it out and start over.

If you know how much current you want, we can whip something up here.
Wow thanks for rapid response (glad to know I'm not going completely nuts!)

Are you saying that the (excessive) load causes a voltage drop on the one side, which the other side then "balances" up?

PS I was thinking of getting round the problem in my existing design by using an ICL7660 -ve voltage generator - is that sensible?

Not sure what kind of current I will need - my overall design is part of a power supply, so "as much as possible" I guess...but realistically enough to drive an op-amp or two...say 150-200mA? My existing +5 is coming out of a 7805 so any more than 1A will pop that anyway!
 

Thread Starter

philbowles2012

Joined Mar 28, 2013
42
use a circuit like this for a split supply


opamp
I tried something like that, but I need the "centre point" to be 0v rather than Vcc/2

I need my differential amp to read 0v when Vin+ and Vin- are the same.

The plan is to use the output from an ACS712 current sensor (Vcc/2 + 100mV per Amp) going into a diff amp with Vcc/2 into the other input, hence the O/P should just be the 100mV/A part which I can then scale up to get a volatge which is exactly the same as the current thru the '712

Hence when current=0 I need to see 0v.

If anyone can tell me a better way or how to do this with single-supply op amps, I'd be eternally grateful!

Phil
 

#12

Joined Nov 30, 2010
18,224
1) Yes, it, "balances up". (badly).
2) the 7660 chip is pretty feeble, too. Limit is 10ma IIRC.
3) "enough to run a couple of op-amps" is pretty small. An LED takes more than an op-amp!
4) These drawings will show you that "0v" is what you declare it to be.
5) I can't understand what you want to do with the ACS712. Please elaborate, and a drawing would help, a LOT!
 

Attachments

Last edited:

#12

Joined Nov 30, 2010
18,224
So...where are you getting this 24 volts? Batteries? Wall wart? Give me the specs and I'll see what I can do with it.
 

Thread Starter

philbowles2012

Joined Mar 28, 2013
42
@ #12 - thanks again. I understand about centre-tapping xformers, but that isn't an option for me, I'm starting from a +12v supply.

I also (kind of) understand that 0v "can be what I want it to be" and I think I understand virtual earth (as in the other reply where the "mid point" of the op-amp is Vcc/2) but I really do want "0v" to be zero in this case. To put it another way, I want the mid-point of my op-amp to be 5v lower than my +5v rail!

The plan is this:


The LHS is a simple divider to get Vcc/2 it is fed into a buffer so that the RHS op-amp doesn't affect it. The Voltage source feeding into the RHS op amp (configured as diff amp with 10x gain) represents the ACS712 with 0.5A flowing through it. The '712 will have Vcc/2 + (0.5 * 100mV) or 2.55v

So the RHS amp gives difference of 2.55 - 2.5 = 0.05 (i.e. just the amount of mV per amp from the '712). The gain of RHS multiplies this by 10 giving 0.5v. Thus the RHS Vo is the same in volts as the current thru the '712 in amps. This then gets fed to a DVM display which will show the current through the '712 in human readable form.

So for two reasons - a) because the '712 Vo swings above and below Vcc/2 (for forward and reverse currents) and b) because when there is no current I want it display 0v not the Vcc/2 (2.5v) that the '712 will be showing.

Effectively the circuit should subtract 2.5v from the '712 and multiply the result by 10.

With a 5v rail, I know I will only ever be able to show the equivalent of (5-headroom) amps. I'm OK with that if its all I can get. I'd prefer a split +-12V if I could get it, but the limitation is I only have a fixed +12v supply.

Without an op amp that can swing below 0v, the lowest figure I will ever see (even with -ve currents) is the +value of the lower rail, which ain't what I want to see!

Does that explain it?
 
Last edited:

#12

Joined Nov 30, 2010
18,224
I got busy working on a timing problem, and now, my real life demands attention. I hope somebody else picks this up while I'm gone...or it will just take a while.
 

tubeguy

Joined Nov 3, 2012
1,157
@ #12 - thanks again. I understand about centre-tapping xformers, but that isn't an option for me, I'm starting from a +12v supply.

I also (kind of) understand that 0v "can be what I want it to be" and I think I understand virtual earth (as in the other reply where the "mid point" of the op-amp is Vcc/2) but I really do want "0v" to be zero in this case. To put it another way, I want the mid-point of my op-amp to be 5v lower than my +5v rail!

The plan is this:


The LHS is a simple divider to get Vcc/2 it is fed into a buffer so that the RHS op-amp doesn't affect it. The Voltage source feeding into the RHS op amp (configured as diff amp with 10x gain) represents the ACS712 with 0.5A flowing through it. The '712 will have Vcc/2 + (0.5 * 100mV) or 2.55v

So the RHS amp gives difference of 2.55 - 2.5 = 0.05 (i.e. just the amount of mV per amp from the '712). The gain of RHS multiplies this by 10 giving 0.5v. Thus the RHS Vo is the same in volts as the current thru the '712 in amps. This then gets fed to a DVM display which will show the current through the '712 in human readable form.

So for two reasons - a) because the '712 Vo swings above and below Vcc/2 (for forward and reverse currents) and b) because when there is no current I want it display 0v not the Vcc/2 (2.5v) that the '712 will be showing.

Effectively the circuit should subtract 2.5v from the '712 and multiply the result by 10.

With a 5v rail, I know I will only ever be able to show the equivalent of (5-headroom) amps. I'm OK with that if its all I can get. I'd prefer a split +-12V if I could get it, but the limitation is I only have a fixed +12v supply.

Without an op amp that can swing below 0v, the lowest figure I will ever see (even with -ve currents) is the +value of the lower rail, which ain't what I want to see!

Does that explain it?
Here's a simple negative voltage supply using the good old 555:
555 voltage invertor.gif
You could regulate both + and - if needed.

Or, as suggested, an op-amp rail splitter could work for the circuit you posted. It appears you need low current. You need a floating 12 VDC wall wart, connect your circuit ground to the center (virtual gnd point) of the op-amp splitter.
The negative terminal of the 12VDC wall wart becomes in essence the -6 volt rail to your circuit and must not be connected to real world ground.

The op-amp splitter works well within it's current range because it uses feedback to hold the voltage with changes in current draw. The simple transistor circuit does not..,
 
Last edited:

#12

Joined Nov 30, 2010
18,224
Back already?
Here is an idea. I think you are educated enough to do this with no more than a block diagram.
If the voltages aren't tight enough, you can use a precision reference to get the 5.0 volts.
 

Attachments

Last edited:

Thread Starter

philbowles2012

Joined Mar 28, 2013
42
Thanks to all for your valued assistance. Today I received a couple if ICL7660s. It was short work (2 caps!) to get my -5v supply.

Still had problems tho - mainly because I'm such a n00b - it took me a while to realise that the REF pin on the instrumentation amp I was using needed to be tied to 0v rather than left floating.

Soon as I got that wrinkle ironed out - hey presto! It works pretty much like I expected it to, so thanks again.

I say "pretty much" because I'm dealing with some very small voltages and there seems to be a lot of fluctuation - think I need some more careful smoothing / regulation in my PSU.

I like the 555 splitter, I might well try that and compare it to the '7660 as I'm getting slighlty less than V+ at the V- terminal (e.g. +4.95 in gives only -4.88 out). I'm guessing this skews the mid-point, but I can trim that out and my final panel meter has only 1 decimal place (It was never intended to be fluke-meter accurate, I just want a broad view of current draw when I run some DC motors) so all in all, I'm getting within 1 or 2% of what my multimeter says once I draw more than 100mA. I'm kinda happy!

SO - what have I learned?
1) What a great place this is and how grateful I am to the body of knowledge here, and the generosity of the members
2) Don't always trust circuits you find on the 'net!
3) I still need to learn a lot
4) I'm not going totally mad, my concept was indeed sound, and with so,e further smoothing / tweaking will do what I want it to do.
 

Thread Starter

philbowles2012

Joined Mar 28, 2013
42
Back already?
Here is an idea. I think you are educated enough to do this with no more than a block diagram.
If the voltages aren't tight enough, you can use a precision reference to get the 5.0 volts.
LOL, I must be more "educated enough" I think (thanks for the compliment) since that is pretty much the way I have my PSU etc.

I have built a mini multi-PSU with 18v-ish wall wart going into a 7812 which feeds a 7809 => 7805 => AMS1117-3.3v. The '12 also feed a 317 hooked up to a 'pot so I have a 2" cube giving me all the voltages I will ever need:



Its got all the decoupling (I think) it should have, but I'm still getting fluctuations - might need to do some more thinking on that or buy a 'scope...(I wi$h!)

Could it be the bird's nest of stray wires, long wiring runs and lack of decoupling on my messy breadboard?
 

#12

Joined Nov 30, 2010
18,224
My instincts say, "Look at your grounding scheme." Conductors that small might need a "star" ground method.
 
Top