Dual supply from single 0v-24v input with zeners

Thread Starter

leogomide

Joined Dec 19, 2019
8
I see some circuit exemples using oscilators (555 and 74HC14), and some others using isolated supplies or the ICL7660 IC.

But all of them have the same problem, current.

After some search on youtube, i come to THIS VIDEO , where Wagner uses this circuit:
1.png

Using the simulations, it works really well, but again it have current limits, because all the current are passing on the 100Ohm resistors

After some google, i see this other circuit (See this link)
2.png

And finally after some more time of research, i see THIS VIDEO and the following circuit:
*really awesome explanation from the author by the way, big thanks to M Caldeira :D
3.png
As showed in the video, the voltage at the regulated point will be HIGH_Voltage - (zener_voltage + VGS_voltage)
*time of this explanation is 17:00

So finally, i came to the idea of mixing all this circuits together, to have the regulated voltages and the current capabilities of the transistors, and after some hours playing at Proteus, i came to this topology:
4.png

Here some things i have observed:
- Without any load and with the opamps disconnected, the voltages are as expected.
- Adding two 1K pots, one between the virtual GDN and V+ and other between GND and V- for simulating loads, the voltage at the rails change a lot if they are unbalanced, so this circuit cant stand for too much components on a single side of the dual supply
- If i connect the opamps, the V+ rail goes to over 700V (see image 5 attached)
- If i connect an axiliary load (arbitrary 500 ohms) to V+ and V-, the voltages come back to normal ranges (see image 6 attached)

As i am just a student and not have engineer degree, my question is, can i use this topology as a dual supply?
This circuit have some impedance mismatch that will cause troubles if i use it? (I have ordered some components to make some tests, but they not arrived yet, so i cant make real world tests right now)
Anyone recomends some changes on this circuit or other options to achieve the dual supply from this single 24v imput?

My goal is to power 5 opamps and have some other components (like voltage divisors between V+ and GND for exemple), and the overall current consumption (other than the auxiliary 500 ohms load) will be no more than 100mA.

Hope i can help other people with this idea too :)
 

Attachments

crutschow

Joined Mar 14, 2008
34,283
I hate to be a wet blanket but, as you observed, the serious flaw with your circuit is that it requires the plus and minus supply output currents be exactly equal, as there's no path for any difference current (such as any auxiliary loads) back to the V1 supply.
You show such current going into the ground (symbol) but there's no path from that to V1.

What you need is a virtual ground or "rail-splitter" circuit to absorb any difference current between the plus and minus outputs so that it can return to V1.

Below is the LTspice simulation of a such a circuit using the common LM317 and LM337 voltage regulators. along with a TL431 reference voltage to generate the needed circuit bias voltage.
The regulators take any unbalanced ground current and route it back to V1
For example if the plus output current is higher than the minus output current, then U1 will turn on and route it back to the bottom of V1.
Similarly U2 will route any excess minus current back to the top side of V1.

The simulation shows the output being evenly split as the input voltage goes from 10V (±5V out) to 24V (±12.5V out).
The R+load (green trace) resistance changes with simulation time to show the the voltages remain evenly split even though the relative load currents change.

1580515159702.png
 
Last edited:

Thread Starter

leogomide

Joined Dec 19, 2019
8
@crutschow you just have made my day! <3
BIG thanks for this circuit, is exactly what i needed

1580522728658.png
Following this other exemple from the link, i have some other questions about it:
- This exemple will work with the TL431?
- Are the diodes D1-D4 and resistors R3-R4 really needed?

About the below example:
- The circuit of the mosfet for reverse polarity protection is good this way?
- The use of C5 will improve significantly the rail stability, or its not very important, since the output current will be very low?
- I need to change R42-R43 if i use an 12V input, for +6/-6V? On proteus simulations this values are good for 12V input too
- You recomend any changes on this circuit?

1580525692165.png

And again, HUGE thanks for your support :D
 

Thread Starter

leogomide

Joined Dec 19, 2019
8
Just one last question, can i use the 1.5A version of LM317 together with the 100mA version of the LM337?

Have a nice day :)
 

dendad

Joined Feb 20, 2016
4,451
As the circuits are virtually in series, the currents drawn need to be the same.
You have an option to change to a split supply, or use a inverting buck regulator for the -Ve supply and a buck for the +Ve supply. The regulators need to be independent.
 

crutschow

Joined Mar 14, 2008
34,283
Following this other exemple from the link, i have some other questions about it:
- This exemple will work with the TL431?
- Are the diodes D1-D4 and resistors R3-R4 really needed?
Yes, it will work with the TL431 as it has the same reference voltage.

D1-D2 are likely to prevent any damagning currents into the regulators during startup.
Not sure what D3-D4 are for but there's no harm in including them.

R3 and R4 are to minimize any current due to the tolerances of the TL431 and regulator reference voltages, when there is otherwise no unbalanced current, and should be included
- The circuit of the mosfet for reverse polarity protection is good this way?
- The use of C5 will improve significantly the rail stability, or its not very important, since the output current will be very low?
- I need to change R42-R43 if i use an 12V input, for +6/-6V? On proteus simulations this values are good for 12V input too
- You recomend any changes on this circuit?
Yes, that is a low-voltage drop way to guard against reverse polarity.

Stability is independent of the current draw so C5 is should always be used.

As long as the minimum current through the TL431 is at least a mA, you shouldn't have to change the R42-R43 resistor values.
can i use the 1.5A version of LM317 together with the 100mA version of the LM337?
Only if the plus output current is never more than 100mA greater than the minus output current.
The regulators need to be independent.
What's "independent".
They can both share the same common.
 

Thread Starter

leogomide

Joined Dec 19, 2019
8
So this is the final circuit that i choose to use:

1580589657565.png

Just the final considerations:
- Any problems to replace the 1N4002 with the 1N4007?
- Any problems use ceramic capacitors with lower value on C1 and C5? The original have polarized caps and a very bigger values

About the D3 and D4 in the below circuit, they are clamping diodes?
Lets say that because an ESD strike, the V- line has a bigger voltage than GND, so the diode will conduct the current to GND, right? Same if V+ has lower voltage than GND
However, i think that are better ways to do this ESD protection with cheap parts, an exemple it to use two SMAJ15CA (Datashhet) between V+/GND and V-/GND, thats correct?
1580589815347.png
 

crutschow

Joined Mar 14, 2008
34,283
Any problems to replace the 1N4002 with the 1N4007?
No.
You can almost always replace a lower voltage rated diode with a higher voltage one of the same series.
Any problems use ceramic capacitors with lower value on C1 and C5?
No.
About the D3 and D4 in the below circuit, they are clamping diodes?
Yes.
I believe they are primarily to protect the circuit if a reverse voltage (such as from a battery) is reverse connected across the outputs.

You could use the SMAJ15CA in place of the diodes, but It's not clear that will provide much additional protection in this case.
 

Thread Starter

leogomide

Joined Dec 19, 2019
8
In this case, its recommended to tie the GND of the circuit to earth (PE)?
Im plannig to use this circuit in an industrial environment
 

crutschow

Joined Mar 14, 2008
34,283
In this case, its recommended to tie the GND of the circuit to earth (PE)?
Im plannig to use this circuit in an industrial environment
The GND will go to the common of the circuit it is powering, but whether that goes to earth ground depends upon the circuit function, and the electrical environment.

Typically a particular industrial environment will have rules for electrical equipment grounding that must be followed.
 
Top