This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

OPA1642: Unexpected behavior in single supply circuit

Part Number: OPA1642
Other Parts Discussed in Thread: TL072, LM833, , OPA1656, OPA1655, OPA2156

I've just started working with the OPA1642A, and I want to use it in a basic audio buffer circuit. I've done this with older parts for years (LM833, TL072, etc.). I'm seeing something unexpected with regard to the bias voltage. With no signal input, Vmid measures really close to V+/2. However, as I give it an input signal (1kHz sine), the bias voltage drops, presumably due to current flowing into the op-amp. With a 9V supply and a 6V p-p input signal, I measure a 1.4V voltage drop across R4, which reduces the bias voltage at the positive input.

This is unexpected, I don't get this with other op-amps. There's some voltage drop, but an order of magnitude less, and it doesn't vary so wildly with input signal level. Can someone help me understand what's going on here?

A few more notes:

- I'm trying to maximize the headroom I can get out of a 9V supply, that's why I'm using a rail to rail opamp. Because the bias voltage drops, the bottom of the waveform clips early compared to what I would expect.

- I need the 2M resistors in there, because I need a 1Meg input impedance.

Thanks in advance!

  • Hey Ron, 

    I am currently looking into this and will respond by end of business tomorrow. 

    Best Regards, 

    Chris Featherstone

  • Hey Ron, 

    I think you may be running into an input common mode limitation on the device. The linear input common mode range of the OPA1642 is 3.5 V from the supply rail. What is the common mode voltage that the input signal is centered around measured at the non-inverting terminal?

    Best Regards, 

    Chris Featherstone

  • Hey Ron, 

    Just to follow up, I assume the input common mode voltage is 3.1 V under the condition you mentioned where the drop occurs at R4. Does this mean the scope capture is showing your input signal centered around 3.1V under this condition? The input bias current is 20 pA max at room temp so it seems unlikely that the input has a lot of current flow through it at DC. The most error due to input bias current across the 2M resistor at the non-inverting terminal I would expect is 40 uV. However the input bias current will increase as shown below at high common modes i.e your input signal level entering the non-linear range. Under the condition you are seeing this issue try lowering the input signal level as experiment to see if it starts to go away and aligns with the linear common mode range. This can help us rule that out as the culprit if the issue persists. 

    Best Regards,

    Chris Featherstone

  • You're correct, it is centered right around 3.1V when the input signal level is high. It looks like I may have selected the wrong part for the job. I'm mostly a digital guy, so I didn't even think to check that parameter on the data sheet.

    Do you have a recommendation for a different opamp that might be better suited? I'm looking for good overall audio performance, handling clipping without flipping phase, and maximizing headroom with a 9V single supply. Ideally one that's available in a VSSOP-8 package, but that's not required.

    Thanks!

  • Hey Ron,

    No problem. I will take a look at the previous devices you mentioned for a comparison and try to narrow down some good suggestions for you.

    What is the intended application that this amplifier will be used for. Will it be used to drive a speaker or headphones, long cabling or anything? What is the source of the signal? Microphone, guitar, DAC etc? The reason I ask is because it can help us provide better guidance with a general idea of the application.

    We do have possibilities such as the OPA1655/OPA1656 that have slightly less common mode voltage range relative to the LM833 you mentioned. 

      

    Input and output Rail to rail options are available such as the OPA2156. The input rail to rail is achieved by using a two stage input, PMOS and NMOS. There will be some crossover distortion if you cross the PMOS and NMOS region. This is a tradeoff to achieve rail to rail capabilities without using a charge pump.  The Iq here is higher but is a possible option. Below is two figures showing the input offset voltage across common mode voltage. You can see the PMOS to NMOS transition region. 

    Best Regards, 

    Chris Featherstone

  • It's intended for a guitar input, hence the high input impedance. This is part of a mixer circuit. The rest of the circuit is currently not populated so I can test this section in isolation. Its output feeds one input of an inverting summing amp with 10k input impedance. There's also an output jack that allows direct access to the buffered output, and it's intended to go to an amplifier or other high impedance input (typically several hundred K to 1M)

  • Hey Ron, 

    Thanks for the information. For a high impedance source like the guitar it makes sense to choose a JFET input op amp. The JFET is high impedance, has low bias current and is low noise. JFET inputs sit nicely between bipolar and cmos inputs with the best of both worlds of high input impedance and low noise. It doesn't sound like we need to worry about output drive capability since the output is feeding a high impedance loads. The main concern that you highlighted is input common mode range relative to the supply. For guitar it is common in setups to have 9V and 12V. Is it possible you could increase the supply to 12V such that you would increase the common mode range of the amplifier and stay inside the linear range? If so you could still use the OPA1642 or even the OPA1656. Is the supply limited to 9V for this application?

    Best Regards, 

    Chris Featherstone

  • 9V is the de facto standard for guitar related equipment, and it's very common to use a 3rd party multi-output DC supply to power their gear instead of whatever wall wart is included with the product. Typically those supplies have 9V outputs (some will have 12V or higher outputs in limited numbers). So, maximizing performance at 9V is still a goal, but we'll certainly recommend using 12V supplies. Even at 9V, the 6V p-p headroom is sufficient for most cases, but we try to accommodate unusually hot signals as much as possible. Sometimes, I'll go the route of using a DC-DC converter to generate a bipolar supply, but that opens up the whole FCC testing issue. As it is, this product has no oscillators or clocks.

    As I understand it, switching to OPA1656 may get me another volt of headroom. Is there any reason I shouldn't give it a try? Any tradeoffs I should be aware of?

  • Hey Ron, 

    Yeah, the OPA1656 will provide an additional 1.25V of headroom going to the positive rail. It can go to the negative rail but not below for the linear range. 

    OPA1656

    OPA1642

    I put together a table below showing devices that have high input impedance and some key parameters for comparison. The OPA1656 has higher Iq but also higher gain bandwidth and very low THD+N. The package types and pricing are also shown in the attached excel table.

    Audio op amps-parametrics-20230616123008.xlsx

    We also have discrete JFETs now that can be paired with any op-amp to provide high input impedance and low noise with flexible biasing. This would be in gain however. I wanted to highlight them as a consideration for other use cases for guitar inputs. 

    https://www.ti.com/lit/an/slpa018/slpa018.pdf?ts=1686945129871&ref_url=https%253A%252F%252Fwww.ti.com%252Fproduct%252FJFE150

    Best Regards, 

    Chris Featherstone

  • Thanks very much! I think it's worth putting together a prototype with the OPA1656 and see how it goes. The discrete JFET input stage also seems really interesting, I'll have to experiment with that as well.

    I really appreciate all of the advice, thanks so much!

  • Ron, 

    Anytime! Reach back out to us if you have anymore questions. We have a lot of audio enthusiasts here and are happy to help. Quite a few of us play instruments as well. 

    Best Regards, 

    Chris Featherstone