Laboratory power supply using OPA549

[satanpit]

Hi, everyone, I am a hobbyist and don't have enough experience developing circuits and need help. I have two power opamp OPA549 and I decided to make a power supply using that amplifier.

I made a scheme using an external opamp for the voltage/current control loop. I know that I can use internal current limiting, but they have bad linearity.

My question is it a good way to use OPA549 like that and what pros and cons of this construction?
Maybe I can use only one opamp for the current control loop.

The scheme in the attachment

[blackdog]

Hi,

A very important lesson is this when developing LAB Power Supply: Phase Margin and Gain Margin.
Use google to search for articles on this.

The schematic of your setup is never going to work properly.
The reason is that you will never get the phase margin and gain margin right.

You have included in the main amplifying element and that is your OPA549 two more amplifying elements.
That is your LT6105 and an OPA2182 with also a low pass filter R6 and C6!
This is really never going to work.

If you really want to experiment with the OPA549 look at figure 10 and figure 11 in the datasheet.

Another important point if you are going to use Power Opamps, they are basically meant for AC signals and I mean the dissipation in the power transistors in the OPA549 and other Power Opamps.
So if you are going to dissipate a lot of single-ended, look carefully at the maximum allowed for the opamp you are using.
If it is not in the datasheet, send an e-mail to the manufacturer about it.

Translated with DeepL.com (free version)

Kind regards,
Bram

[satanpit]

@blackdog thank you for your reply.

I saw the datasheet of OPA549 and Figure 10 and Figure 11 but in this case, I must use the internal current limit circuit which is not linear.

I did some base tests on the breadboard and it worked but maybe you are right and it will not be stable or I will have other problems.

[David Hess]

My question is it a good way to use OPA549 like that and what pros and cons of this construction?

The advantage is that the power operational amplifier should have built in thermal protection and safe operating area protection for its output transistors.  The class-AB output should also be able to source and sink current.

Maybe I can use only one opamp for the current control loop.

Move the current loop's error amplifier to the output voltage and make the comparison directly without the instrumentation amplifier.

[satanpit]

David, Thank you for the reply,

The advantage is that the power operational amplifier should have built in thermal protection and safe operating area protection for its output transistors.  The class-AB output should also be able to source and sink current.

Yes, I agree with you in these cases, but I have never seen circuits like that, and I think maybe it's the wrong way.

Move the current loop's error amplifier to the output voltage and make the comparison directly without the instrumentation amplifier.

I am not sure that I understand you.
How I can make the comparison directly without the instrumentation amplifier?
Could you please show some examples?

[David Hess]

The advantage is that the power operational amplifier should have built in thermal protection and safe operating area protection for its output transistors.  The class-AB output should also be able to source and sink current.

Yes, I agree with you in these cases, but I have never seen circuits like that, and I think maybe it's the wrong way.

There is nothing wrong with it, but it will usually cost more than a discrete transistor implementation.

Move the current loop's error amplifier to the output voltage and make the comparison directly without the instrumentation amplifier.

I am not sure that I understand you.
How I can make the comparison directly without the instrumentation amplifier?
Could you please show some examples?

The idea is that instead of using an instrumentation amplifier to create a ground referenced signal representing the current and making the comparison at ground, the reference voltage is moved to follow the output so the comparison can be made referenced to the output instead of ground.

In the example below, R55 is the current shunt and U70 is the error amplifier for the current control.  The reference voltage is generated across R65 in parallel with R66 by constant current source Q60.  Since the bottom of parallel connected R65 and R66 is attached to the bottom of the current shunt, which is also the output, the reference voltage follows the output voltage and U70 only sees the difference between the voltage across the current shunt and the reference voltage from adjustable divider R65.

[Berni]

This is a perfectly valid way of doing it. It just tends to be more expensive in parts cost.

I built a similar kind of power supply for the use of simulating power brownout and automotive cranking waveforms. So it is essentially just a DAC that is amplified up by a high power opamp. Using an opamp there meant it saved me a lot of design effort for a one off project (so time was more important that parts cost) and i needed the push pull ability on the output since some of the waveforms it was required to reproduce are pretty fast.

The one thing to watch out for when using this in a PSU is to make sure the output does not become unstable once a lot of capacitance is placed on the output. So you might want to add a sizable ferrite and capacitor to the output. This hides the load from the opamp at high frequency, presenting a consistent impedance, that you can then tune your regulation loop response to work well with. In general having a sizable capacitor directly at the output of the PSU is a good ting since it cleans up noise and improves regulation during sharp current transients of the load.

Also keep in mind that having push pull capability is 'weird' to have in a PSU. Yes there are some PSUs out there that do it, but most are push only. Also your current limit will only work for push operation in your schematic (Not necessarily wrong, i have a commercial PSU that does the same, just an unusual feature)

[satanpit]

The idea is that instead of using an instrumentation amplifier to create a ground referenced signal representing the current and making the comparison at ground, the reference voltage is moved to follow the output so the comparison can be made referenced to the output instead of ground.

In the example below, R55 is the current shunt and U70 is the error amplifier for the current control.  The reference voltage is generated across R65 in parallel with R66 by constant current source Q60.  Since the bottom of parallel connected R65 and R66 is attached to the bottom of the current shunt, which is also the output, the reference voltage follows the output voltage and U70 only sees the difference between the voltage across the current shunt and the reference voltage from adjustable divider R65.

Thank you, David.
Exciting idea, I have never seen it before.
But I am not sure that this way is better than using an instrumental amplifier

Also, thank you, Berni for the reply.
I have some clarifications. My idea is to make a PSU with digital control.
I decide to use DAC for set voltage and current, and INA239 for monitoring.
Maybe I confused you using VSS, it is only for making zero on output, I plan to use only -3V for VSS and +30V for VDD

About ferrite I think about that because when I tested it on a breadboard I had some unstable state with capacitance load.
But I am not sure that a big capacitor in output is a good idea.
It will have negative consequences like slow reaction.
I think if I use OPAMP it is not necessary to have any capacitors in the output

[David Hess]

The idea is that instead of using an instrumentation amplifier to create a ground referenced signal representing the current and making the comparison at ground, the reference voltage is moved to follow the output so the comparison can be made referenced to the output instead of ground.

In the example below, R55 is the current shunt and U70 is the error amplifier for the current control.  The reference voltage is generated across R65 in parallel with R66 by constant current source Q60.  Since the bottom of parallel connected R65 and R66 is attached to the bottom of the current shunt, which is also the output, the reference voltage follows the output voltage and U70 only sees the difference between the voltage across the current shunt and the reference voltage from adjustable divider R65.

Thank you, David.
Exciting idea, I have never seen it before.
But I am not sure that this way is better than using an instrumental amplifier

Configuring the error amplifier to operate at the output common mode voltage directly without a level shift has advantages:

1. The cost of the instrumentation amplifier is removed.

2. The error from the limited common mode rejection of the instrumentation amplifier is removed.  There are no resistor matching requirements for high common mode rejection.  Operational amplifiers also have limited common mode rejection but it is usually higher than the common mode rejection of an instrumentation amplifier and it requires no special parts to achieve it.

3. Delay inside the feedback loop from the instrumentation amplifier is removed increasing performance and simplifying frequency compensation.

4. The compromise between speed and precision of the instrumentation amplifier is removed.

Disadvantages include:

1. Now the reference voltage needs to be level shifted to the output voltage.  In my example above this is done with a current source controlled by the reference.  If greater precision was required, then this would require an operational amplifier, but the demands on this operational amplifier are not great and the speed versus precision tradeoff is reduced greatly.

2. The current source or sink for the above must have compliance above the highest output voltage, or compliance lower than the lower output voltage which would require a negative supply.  The example I works on the high side although a negative supply was available.

3. The error amplifier must have a common mode input range which includes the full output range.  In my example above, there is a negative supply available so that the error amplifier for current mode can operate down to zero volts.

Williams/HP used a similar design but included a low voltage floating positive and negative supply for the control circuits.  This low voltage supply used the output voltage as common, which also allowed low voltage control circuits to work with high output voltages without level shifts.  This also significantly boosts the common mode rejection of the error amplifiers.

[satanpit]

Thank you for the details.
I will try to make it on the breadboard