OPA548 - noise & current limit

Hello Thomas,

I would not expect the OPA548 current limit action would create the excessive noise you described; therefore, I expect it is something related to the application. It is easy to pick up 10mV of noise from the environment when using a breadboard, but 1V is unexpectedly high. Not having a final PC board set-up can lead to unexpected behaviors so that may be a factor. Unnecessary lead and board inductances and a questionable thermal environment can lead to problems. Here are some questions about your application:

• Have you observed the your power supply voltages with an oscilloscope as the OPA548 output goes in and out of current limit, both sinking and sourcing current?
• How much current is being delivered to the load when operating and where are the current limits set?
• What sort of thermal management is in place?
• What is the load type being placed on the OPA548 output?

A schematic of the application would be helpful and information about exactly how you are applying the OPA548 would be helpful too.

Regards, Thomas

PA-Linear Applications Engineering

Hello Thomas,

first I have to thank for that quick response!

Unfortunately I haven't any shematic. But as I said, it's a normal non inverting OPA. I used capacitors as in figure 1 of the opa548 datasheet (R2=5k1, R1=10k). I also tried the output stage compensation (figure 14) that is recomended if current limit is used with no big effect.

The application will be a simulator for a battery cell, with the possibility to simulate charge and discharge. In the end application the opa is powered by two isolated DC/DC converters (8V, 3V). It should be very precise, at least 10mV or better. Unfortunately I can't make a PCB while this first developing phase...think I have to learn much more about analog design because of noise etc... ;).

Have you observed the your power supply voltages with an oscilloscope as the OPA548 output goes in and out of current limit, both sinking and sourcing current?

No.

How much current is being delivered to the load when operating and where are the current limits set?

I've set the limit to about 450mA and increased the input with the pot till limiting was reached.

What sort of thermal management is in place?

D2PAK soldered on a 5 square inches single sided copper area/PCB with a soldered heat sink (should be about 10°C/W together). Don't know the exact voltage but I kept it under the limit where thermal shutdown starts (about 3V I think).

What is the load type being placed on the OPA548 output?

At the moment it's a adjustable PSU, which can source and sink current (Hameg HM8143).

Regards,

Thomas

Hello Thomas,

I've been giving what you are observing with the OPA548 behavior more thought based on your answers. The OPA548 has been around since the mid-90's and there haven't been any reports about the behavior you describe that I have found, so I do think we need to concentrate on the application.

I don't find anything wrong about the way you have applied the OPA548 from your descriptions of the applications circuit. I found and viewed the Hameg HM8143 product description and specifications, and it appears it would be a good sink/source load for the OPA548. The OPA548 current limit function is fast and anything that would cause the function to trigger could appear as a noise. Therefore, I think is imperative that an O-scope image be obtained of the output when sinking and sourcing, the input drive signal and the supply lines. Without knowing the exact characteristics of the noise we can't be sure what we are really up against. Such information may help us get to a determination more quickly.

Regards, Thomas

PA-Linear Applications Engineering

Thomas,

Are you using the Hameg's load SENSE capability in your OPA548 application?

Regards, Thomas

PA-Linear Applications Engineering

I have to confirm if it is the posted PSU type when I am in the company again, just identified the type from the Hameg website. But I haven't connected sense wires, no.

Because I'm still student I'm only working one day a week. I will try to take some pictures this or next week and post again.

Many thanks again and have a nice weekend,

Thomas

Hello Thomas,

here ar two scope pictures of the OPA548 in current limit, one while sourcing and one while sinking current.

Best regards,

Thomas

5428.OPA548_oscillation.zip

Hello Thomas,

The images of the waveforms you supplied do suggest the OPA548 is oscillating in the application. The horizontal divisions appear to be set at 2us/div and each repeating cycle has a total time a bit less than 2us. Therefore, the oscillation has a frequency in excess of 500kHz. If you have a look at the OPA548 Typical Curves, Open-loop Gain and Phase vs. Frequency, on datasheet page 4, you will see that the open-loop gain (Aol) crosses unity gain between 500kHz and 1MHz. Oscillation occurs when the phase margin has been reduced to the point where the feedback becomes positive and this often occurs as the result of the load applied to the amplifier. The degradation in phase margin is almost always associated with a capacitive load. The capacitance, in conjunction with the output resistance, adds a pole into the feedback loop and that is where the phase shift comes about. You may need to investigate the active load impedance characteristics. It would be helpful if we had the exact schematic, exact load, and PC board layout information.

Additionally, the operational-amplifier is not intended to be operated at the current limit point where the gain-phase characteristics can be different than they are in the linear region of operation. The current limit is intended to be a fail-safe limit and not a precision limiting function.

Regards, Thomas

PA-Linear Applications Engineering

Hello Thomas,

you are talking about the shematic of the Hameg PSU? I think, I will first try to use some other kind of source/load at the output. So I will see if the Hameg is the point.

The application is not intended to use the current limit all the time, in the application it's also a safety feature.

With the output compensation described in the datasheet it isn't possible to prevent oscillating with capacitive loads?

Regards, Thomas

Hi Thomas,

Well, the Hameg PSU represents a load when the OPA548 is sourcing current, and a current source when the OPA548 is sinking current. The PSU represents a different impedance to the OPA548 that depends on the operating point. Conversely, the OPA548 plays the same sort of role if we look at the PSU as sourcing and sinking current. There are appears to be an operating point where the load that the PSU presents to the OPA548 looks resistive and has a capacitive component as well. The equivalent load resistance is governed by Ohm's law, but the load capacitance portion may not be readily evident. It may be a capacitor connected internal to the PSU output.

The snubber circuit shown in datasheet Figure 14 may be correct for some loads but not all. As the datasheet states in the Output Stage Compensation section, "Some variations in circuit value may be required with certain loads." Therefore, it may be necessary to experiment with different compensation values, or methods, when the Hameg PSU is applied to the output, to assure stability under all load conditions.

Regards, Thomas

PA-Linear Applications Engineering