OPA244 in PSU

[GigaJoe]

Hello everyone.

I have 4 laptops PS, 20V 2A.  Willing to mage 4 independent channels 1-15V, 1A Max; Finally I prototype just one PS. a large multiply LC filers, Ground wire, 2-sense etc..  Finally i got what I want , around 50 microVolts change when load changing 0 to 1A. And not able to detect a noise from the laptop PS. ( scope 1mv/div shows nothing .... ) OK ...
Schematics are basic, In my existing I use OP07 , it connected as a single supply and has 0.7 Volt bottom line, so PSU produce around 1.6V ...

QUESTION: If I will use OPA244 instead OP07. will my minimal output decreased ? I would be happy to have 0.5 V output.  Or advise on any reasonable precision opamp, with single supply and correctly operating with millivolts above the ground.

[ The schematic a basic, simplified for simulator, R1 - just load for VR ( otherwise don't work) ]


2 question, not important but nice to have some idea:
to advise a current limiter as a trigger. Planning to shunt pin-3 on opamp to drop the voltage. And button to recover. And need to thing to link triggering signal to other PS, and power sequencing, keeping the ground independent  ( relay, or some opto things ... )
 

[David Hess]

Schematics are basic, In my existing I use OP07 , it connected as a single supply and has 0.7 Volt bottom line, so PSU produce around 1.6V ...

The OP07 is limited by its input common mode voltage range which does not extend to ground.

QUESTION: If I will use OPA244 instead OP07. will my minimal output decreased ? I would be happy to have 0.5 V output.  Or advise on any reasonable precision opamp, with single supply and correctly operating with millivolts above the ground.

Any single supply operational amplifier, which means one with an input common mode voltage range that includes its negative supply, should be able to provide a zero volt output in this circuit.  The OPA244 is one such.

[Kleinstein]

The 7818 need local caps at its input and output to make it stable. However there is no real need for this regulator. If needed at all a LC filter will be likely more effective.

The circuit might oscillate with a capacitive load at the output. Usually it needs an extra local feedback around the OP. It would be kind of a lucky find if the speed of the OP is just slow enough for the output stage.

It would be a good idea to at least a shunt for current measurement, so one can display the actual current without having an extra drop and maybe add a current limit later. The shunt might interact with the voltage regulation and it is thus important to plan with it from the beginning.

[GigaJoe]

Yes, Thanks,
The real implementation has a lot caps , for 18V stab in amd out, and caps for V+ opamp , caps for voltage ref V+ , and some output ...
Without stab I see some noise , actually, it a low power TO92 - stab, from ebay like a $1 for 20

I did order OPA197 - it seems pretty good for a common use, and has a single supply option.

[TheUnnamedNewbie]

Yes, Thanks,
The real implementation has a lot caps , for 18V stab in amd out, and caps for V+ opamp , caps for voltage ref V+ , and some output ...
Without stab I see some noise , actually, it a low power TO92 - stab, from ebay like a $1 for 20

I did order OPA197 - it seems pretty good for a common use, and has a single supply option.

Don't overdo the capacitance on the output, it can just end up slowing down your circuit further instead (and make it unstable)

[Zero999]

What purpose does the LM7818 serve?

The op-amp and voltage reference can be powered directly from 20V and it would still work. Reducing the power supply voltage to the op-amp will only increase the drop-out voltage, because Q1 has no voltage gain at all.

[GigaJoe]

the source is 20V laptop PSU under the load over 1V p-p 1XX Khz output,  VR give an additional filtering as opamp has much lower noise rejection in a single mode supply.
Just an idea, that I assume, but I focused on very precision voltage output.

[IanMacdonald]

The CA3140 can operate with both inputs near ground, and is probably the most suitable error amp. 

http://www.intersil.com/en/products/amplifiers-and-buffers/all-amplifiers/amplifiers/CA3140.html

The other possible issue is that the burden voltage of a darlington and the opamp combined is going to lose maybe 3v of max voltage.  Maybe no problem if you have 5v to spare, but it's not an ideal topology.

[David Hess]

the source is 20V laptop PSU under the load over 1V p-p 1XX Khz output,  VR give an additional filtering as opamp has much lower noise rejection in a single mode supply.
Just an idea, that I assume, but I focused on very precision voltage output.

The schematic below shows two important requirements for precision:

1. The output divider, reference, and ground return need to use a single point ground; otherwise the ground return current will modulate the output voltage.  The same constraint applies to the top of the voltage divider.

2. The power dissipation of the error amplifier needs to be constant to limit temperature changes which would lower precision and the lower the power the better.  (1) This means unloading the output with a current buffer; an emitter or source follower works well for this.

The CA3140 can operate with both inputs near ground, and is probably the most suitable error amp. 

The CA3140 is a good choice except that its FET input yields higher input noise and lower precision.  The only reason it would be better here than the OPA244 is that the OPA244 is a micropower part but see below about precision in connection with the LT1021D-5 reference.  Note that the speed of the CA3140 will require extra frequency compensation.

In the past I have used the OP-07 (inexpensive!), LT1001/LT1097, and OP-27/LT1007 for designs like these but of course they will not work here because a single supply part is needed.  The various single supply replacements for the OP-07 like the LT1006 (2) and OP-90 can work but are higher noise because they are lower power.

How much precision is really needed in this case?  The LT1021D-5 reference is specified for 15uV/C typical and 100uV/C maximum drift and a noise of 75nV/sqrtHz so all of these operational amplifiers are overkill except maybe for power supply rejection ratio if the amplifier's supply is unregulated.  Filtering the output of the reference would lower output noise and would be a good reason to use a low input bias current operational amplifier like the CA3140.

(1) Thermal feedback from the output stage to the input stage of the operational amplifier limits open loop gain (!) thereby limiting precision.  How weird is that?

(2) The LT1006 is the single version of the dual LT1013 and quad LT1014 which are improved LM358/LM324 replacements.

[GigaJoe]

Thank you David,

I had to realize a common ground point connection, during my practice on prototype. So it will 4 terminals in the end. An actual result over around 20-40 uV when load changed from 0 to 0.5 A. But when i had circuit without variable resistors, connected to V-ref.? just straight to opamp, V-out was 15V and hadn't change at all (0 uV) at any load. - That puzzle for me. 

But even using 2 multi-turn resistor 10K and 100 it difficult to tune with such a precision. so fractions of millivolt that would be reasonably enough. But I like to have it stable for long period, let say a week or so, without much floating. Output tran - dargington.


My second point is Current protection circuit , adding a trivial one may disrupt voltage schematics. So I'm trying to figure-out something like trigger base, set before VR - no luck yet ... Any help - so appreciated.

[David Hess]

I had to realize a common ground point connection, during my practice on prototype. So it will 4 terminals in the end. An actual result over around 20-40 uV when load changed from 0 to 0.5 A. But when i had circuit without variable resistors, connected to V-ref.? just straight to opamp, V-out was 15V and hadn't change at all (0 uV) at any load. - That puzzle for me.

With a 4 terminal connection using the circuit example I posted and a precision operational amplifier, I got load regulation of less than 1uV from 0 to 1 amp.

But even using 2 multi-turn resistor 10K and 100 it difficult to tune with such a precision. so fractions of millivolt that would be reasonably enough. But I like to have it stable for long period, let say a week or so, without much floating. Output tran - dargington.

I like using separate coarse and fine potentiometers however the stability of the coarse potentiometer is going to be a problem.  The best option there is to use a multi-turn wirewound for the coarse potentiometer while keeping the single turn fine potentiometer.

My second point is Current protection circuit , adding a trivial one may disrupt voltage schematics. So I'm trying to figure-out something like trigger base, set before VR - no luck yet ... Any help - so appreciated.

You could replace the power transistors with a 3 terminal integrated voltage regulator which provides its own current limit and thermal protection as shown in the schematic example I posted although this will raise the minimum output voltage to 1.25 volts if you use an LM317 and an amplifier which can pull its output all the way to ground without the buffer transistor.  The LM317's adjustment pin current is microamps so no buffer transistor is needed but it requires a minimum output load of like 5 milliamps.