Power Supply

Status
Not open for further replies.
Ordinary opamps have a max supply voltage of 36V.
The MC34071, TLE2141 and a few others have a max supply voltage of 44V.
The OPA445AP has a max supply voltage of 90V.

An amplified 5.6V zener diode is buffered by the opamp so it is very stable.
Here is the reference voltage from a power supply project on another website:
 

Attachments

  • amplified zener diode.PNG
    3.8 KB · Views: 141

The only advantage of my diagram is that for a fixed voltage output it works fine. If you have and variable output voltage then you always have to adjust the current sense pot. This circuit is for a 12V lead acid (gel-cell) charger, so the output voltage is set to the charge voltage and then the current sense is adjusted. When the charge current drops to a set level the LED is turned on to indicate the battery is fully charged.

A differential amplifier is one way to go with a variable supply.
 
Ah okay then, thank you for the explination.

I'll get re-designing the circuit in an hour or so to get rid of the dual regulators. The next revision should be up tonight.
 
I think the unregulated supply will be +33.3V at full current. You don't want the outputs of IC5 and IC1 anywhwere near saturation which is about +31.3V.

But the output of IC5 is trying to be +31.9V. Turn down R25.

With the output of IC1 at +30V then I think the minimum output from the project is about +27V if C1 has a much higher value.
 
Okay then, so I need a larger C1 and I need to set a max for the PSU. I might put R25 as a 5k preset so it can be set after the build. This would allow for a larger input voltage to be used in order to get to 30V output, if required.

What size C1 would you suggest ? I found a 63V 10,000UF cap for £5.76 in CPC, would that do ? I don't want to end up with a design which will cost hundreds lol.

What is the formula for the amplified zener diode circuit ? I only managed to get that output through lots of trial and error in Livewire. :x

Thanks.
 
An opamp tries to keep its input voltages the same.
Then R27 will have the zener diode's 5.6V across it. Use Ohm's Law to determine its current and use Ohm's Law again to calculate R25 with that current.

I have a graph for 120Hz ripple voltage. 10,000uF at 3A will have a ripple voltage of about 1.5V p-p and 100Hz will have 1.8V p-p.
 

Attachments

  • Full-wave rectifier ripple voltage.PNG
    32.7 KB · Views: 149
Ah thank you, I understand now. I presume that R26 doesn't really matter and that it's jsut to provide a bias current for the zener ? I just picked a value which would give about 10mA through the zener.

Looking at the graph I guess there isn't much point in this case for picking a larger value that 10,000uF.

Thank you very much for everything.
 
I've had a thought about linear power supplies in general. Instead of using a standard inverting op-amp as the error amplifier, why not use an Integrator. As the voltage drops, the base current of the transistor will keep rising untill the voltage drop is nil.

Thoughts ?

Edit: It works perfectly in Livewire. Maybe a low pass on the output of the op-amp would be nice to stop random frequencys flowing round the circuit. Pic:
https://yngndrw.hostilezone.net/uploads/Integrated%20PSU.jpg
 
Last edited:
A power supply must be quick enough to see the output voltage drop a little when it is loaded or rise a little when it is unloaded then correct the error as soon as possible before it gets very bad. An integrator will slow it down too much.

It is called Transient Response.
 
Ah I see, how about using a mixture of a standard error amplifier and an integrator ?

The integrator would be setup to be as fast as possible with a low pass on the output to try and stop any oscillation, the standard amp would be setup as normal.
Both amps would then be fed into a summing amp, which would be biased towards the standard amp.

If the standard amp got 3/4 of the bias and the integrator got 1/4 of the bias, the maximum Transient Response would be 125% of the time that it would normally be. Slightly slower but that's in a worse case scenario. Surly that would be acceptable ?

Of course it means that the integrator needs 4 times the voltage swing in order to be 100% useful, again in a worse case scenario.

Am I looking at this wrong ?
 
You need to analyse the speed and phase shift of the opamp, the driver transistor, the output transistor and the output filter capacitor to see the transient response or oscillation, then add series and shunt capacitors in the negative feedback loop to make it quick without ringing or oscillation..
 
So you think it would be worthwhile if the time was spent to analyse and optimse it ? I'm just wondering why a modified version of this method hasn't been used before - In controller IC's I mean, because I'm yet to find a PSU whose output doesn't change from a large range of loads.
 
The error amplifier in a power supply has an extremely high voltage gain so it reduces the output voltage change from no load to full load to a very small amount. The wiring to the load has a higher voltage drop than the power supply regulator.

The error amplifier and the high current output transistors are slow to react so there is a transient at the output when the load current is quickly changed. Good design minimises the transient. Poor design causes oscillation.
 
Ah see, so the best ways to reduce voltage drop from high loads is:
1) Use a higher gain error amplifier.
2) Connect the voltage feedback as close to the load as possible.

Is this correct ? What other methods could be used to help ? (Not taling about rapidly changing loads here, I mean the different output voltages when a static high resistance load is used and a static low resistance laod is used.)
 
A 7805 regulator has a typical output voltage drop of only 15mV when its load changes from 5mA to 1.5A. That is pretty good voltage regulation.
 
Although its output impedance goes up with frequency, it's effectively like an inductor connected in series with the output.
 
Ah fair enough. How about with higher currents. Let's pick say .. 0 - 10A.

I know it would be too inefficent and should be done through switching, but I'm interested.
 
Status
Not open for further replies.

Similar threads

T
  • Question
Power problems
Replies
10
Views
810
rjenkinsgb
  • Question
Bench Power Supply
2
Replies
20
Views
5K
Tony Stewart
Handy Bench Power Supply
Replies
8
Views
3K
schmitt trigger
S
A Simple Power Supply Project
Replies
1
Views
2K
Tony Stewart
  • Question
Powering RPI5 8gb from 5v supply
Replies
3
Views
2K
aliarifat794
A
Menu
Log in
Register
Cookies are required to use this site. You must accept them to continue using the site. Learn more…