OPA548: Can not bring output to GND

Hi Gudmundur,

You have to understand that the OPA548 is a high-voltage part. It is supposed to work in higher and dual supply up to +/-30V, which is also a reason why the output voltage swing is not rail-to-rail but 1V from the negative rail typically for small currents. The output swing is always defined relative to the supply voltage. Depending on the common mode, i.e. how the non-inverting input is biased, the signal might slam against one of the rails and the op amp saturates.

But more important is that you cannot ground the non-inverting input of a buffer. It is not a valid configuration.

The OPA548 might not be the best solution to operate in a low single supply while the output should swing all the way to ground.

If you are interested in more insights on input and output limitations, watch our precision lab course material that covers this topic.

What is the input/output voltage range of the application and the purpose? Depending on that, I would recommend another part and maybe another configuration.

Best,

Miriam

 

Thank you for the response Miriam.

Regarding my application:

My load is actually an inductor/electromagnet with an inductance in the range of 20 mH - 100 mH and a resistance of 9 Ohms. I want to linearly control the voltage across the inductor and I will probably need about 18V at the op-amp's output, might even want to go up to 27V if possible (so 2-3 amps). Since I only need current in one direction I was hoping to get away with a single supply-configuration. I will of course be adding protection diodes. I might change the configuration to a controlled-current configuration with a current sense resistor between the inductor and ground, and a snubber network on the output for stability (if that might change your recommendation).

Regarding my op-amp understanding and intuition-building:

I've watched the relevant Precision Lab videos. Input common-mode range of OPA548 extends at least 0.1V below negative rail so that shouldn't be the problem in this case, so I turn my attention to the output stage. Figure 13 of the datasheet seems to indicate that for almost-zero current the output should swing almost all the way to the negative rail, so it "feels" like things should work.

To elaborate a bit better on why I'm having a hard time comprehending why my circuit doesn't work: It's not like I'm asking the bottom-part of the output stage to sink any current from my load to pull the op-amp output pin low. The load already works to pull the op-amp output to ground, all I want from the op-amp is to stop pumping (up to) 100's of milliamps through the top portion of the output stage and into my load. Given that I'm within the common-mode range I don't see why the op-amp doesn't "see" that the inverting-input is too high and cut off the top-portion of the output stage to compensate.

I appreciate any advice, whether it relates to my application or my understanding.

Hi Gudmundur,

the precision lab tries to explain this issue on transistor level: you cannot cut off the top transistor; this is not how an op amp works. There will be always a voltage drop on the output. If you try to force the signal beyond this point, the op amp will saturate. If your main goal is to reach ground, you can use the LM7705 as a negative bias generator. This is feasible if you have a 3V-5V low impedance source available. If not, you could try to put a Schottky-diode in series with the op amp before the feedback. The result should be a very low voltage drop on the output.
What are your available voltage sources and what is the range of the current? Is your configuration still voltage-controlled?
I will simulate the Schottky-diode solution tomorrow to verify and to give you a number of the reduced offset voltage.

Best,
Miriam

I am currently using a bench power supply for prototyping. I was mainly hoping to move to a single supply configuration to reduce circuit complexity, so I will likely stick with a dual-supply system rather than adding components to generate negative voltages. Thank you for the suggestion though.

I am however interested in this Schottky trick you mention and look forward to see the simulation results.

Yes, I am still working with a voltage-controlled configuration.

I should probably also consider controlling the load with a simple power transistor rather than an op-amp, since I only need unidirectional current flow and a power transistor should allow me to cut off the current completely. (Using an op-amp came about since I originally needed to drive current in both directions through the load.)

I'm still trying to wrap my head around op-amps internal functions, so that I can ask less-stupid questions in the future :) The Precision Lab videos talk about why the op-amp can't drive the output all the way to the positive rail when supplying current, and why the op-amp can't drive the output to the negative rail when sinking current. I get where that comes from (Vsat+Vbe for biopolar output stages) but that doesn't explain (to me at least) why the op-amp doesn't allow the output to swing to the negative rail when the load is already pulling the output pin low.

This can't be purely due to bias current in the output stage because with no load the op-amp only draws 13 mA while putting a 1 Ohm resisitve load (R1 in schematic) puts more than 270 mA in to the amplifier and out the output pin.

Does the op-amp sense the current going through the bottom transistor and increase the output pin voltage, if needed, to maintain a minimum bias current through the bottom stage? This would explain why connecting the load to ground significantly increases current through the top transistor.

Thanks again for your help and comments.

Hi Gudmundur,

enclosed are the schematics showing a solution with a diode between the OPA548 and the feedback. The voltage drop at the diode results in a low voltage drop at the load.

Depending on the manufactured part, the offset can vary from -10mV to +10mV which is shown by Voff. The offset of the SPICE model is nullified by Vnull (2.8mV).

Next to these schematics are the not adjusted followers which slam against the rail with a grounded non-inverting input with a voltage drop of about 250mV at the load.

I also attached the TINA file which allows you to simulate the different configurations.

Is that a feasible solution for you?

3755.OPA548outputswing.TSC

Best,

Miriam