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.

OPA344: What is the cause and what is the solution for this strange conditioner?

Part Number: OPA344
Other Parts Discussed in Thread: LM2576

Dear All,

Please see the bellow schematic:

I experimented the above circuit giving a 50 Hz triangular wave with 3.3V peak to peak voltage and  1.65V of offset voltage. The out put is as follow:

I found that if I give the input a constant 2.6V the out put is as follow:

If i give lower than 2.6V or higher than it at the input, the oscillation removes. Means the above oscillation only exits in 2.6 V of input voltage.

Can you please explain me the probable causes of the strange behavior? And please give me the solution to the problem?

Regards, Ras

  • Former Member
    0 Former Member

    Hello Ras,

    I'm wondering, is the first shot of the output conforming with what we expect to see?  I ran a quick AC sweep in TINA and it looks like we'd expect a clean unity gain.  Also, I checked the circuit's stability and there is 55 degrees of phase margin.  So given this information, I'm not sure if the first output shot makes sense.  Is each mark in the scope shot 1V or 100mV?

    Also, is there any type of loading going on at the output?

    Have also included a copy of the TINA circuit I'm using for your reference.Ras_Sharif.TSC

    Regards,

    Daniel

  • Hello Daniel,

    Thank you for the reply.

    The complete circuit is as follow ( 40pF is the ADC input capacitance ):

    The volt per division of the scope is 1V.

    I wondered as you. Probably the OPA344 has burn out.

    Regards,

  • Former Member
    0 Former Member in reply to Ras Sharif50

    Hi Ras,

    I ran another AC sweep and stability analysis with the loading shown here and did not see anything that might explain the behavior in the scope shots, although I did not expect to see anything.  The behavior is pretty strange and the only thing that stands out to me is that the odd behavior seems to all occur about one diode drop away from the V+ rail.

    Do you have another unit that you could swap in for testing?  That seems like a fairly likely scenario.

    If that doesn't work, we can take some next steps from there.

    Regards,

    Daniel

  • Hi Daniel,

    The behavior in the first waveform implies a hysteresis comparator.

    I try to test some probable conditions.

    Regards, Ras

  • Hi Ras,

    just take a fresh OPA344 and try again.

    Kai

  • Dear Kai,

    Thank you for the point.

    I tested another OPA344 circuit. I get same result.

    I encountered an abnormal response form an opamp for the first time.

    Is it probable that the supply contain a certain frequency which amplified when the input voltage reaches 2.6V!?

    Please, if possible, test the circuit, although I think the problem refer to my circuit?

    Regards, Ras

  • Hi,

    I tested some conditions.

    The circuit oscillates at 200Hz even if the input is zero.

    When the input is around 2.6V the oscillation amplifies and become disturbing.

    Regards,

  • Former Member
    0 Former Member in reply to Ras Sharif50

    Ras,

    You could try taking shots of the input or even run an FFT on the supply.

    However, I think we'll try building it here on our end.  That might be the best way forward, but we don't have the part in right now.  So, we won't be able to get back to you until next week.  Is that ok?

    Regards,

    Daniel

  • Hi Daniel,

    I will place the input and FFT of the supply here, tomorrow.

    Next week is ok. I should to solve the problem.

    Regards,

  • Hi Ras,

    can you show scope plots of the input signal and the supply voltage of OPA344?

    Is the supply voltage of OPA344 properly decoupled?

    Do you use a breadboard?

    Kai

  • Hello,

    Thank you for the replies and excuse me for the delay.

    Please see waveforms and ffts.

    - The supply voltage:

    - The supply fft:

    - Another supply fft:

    - input signal:

    The OPA344 is in a pcb. I provide each OPA344 with a separate low esr 10nF capacitor closed to the opamp.

    Regards,

  • Hi Ras,

    there's huge noise on the supply voltage! And the noise packets seem to occur synchroneously to the input signal when reaching the supply rails.

    You should use a much better source for the supply voltage (stiffer and less source impedance) and/or use a much bigger decoupling capacitance. I would recommend a 100µF electrolytic cap in parallel to 100...470nF/X7R for a test.

    Where do you have the 3.3V supply voltage from? Is there more circuitry on the PCB which runs with this 3.3V supply?

    Kai

  • Hi Kai,

    Thank you for the reply.

    This is the supply circuit:

    Noise packets which are stable in the supply voltage travels through the input voltage and do not stand at the top o f the signal.

    There are other circuits; Some other opamps and a mcu. Although the digital circuit is not mixed in the analog circuit.

    Can we expect the OPA344 behaves similar to hysteresis comparator if the supply has such noise?

    I add some electrolytic capacitor but there is no any improvement.

    Regards

  • Hi Ras,

    the supply voltage of an OPAmp must be stable, clean and noisefree. But for some reason your 3.3V supply voltage isn't clean at all! See how huge the noise is. Your input signal is even exceeding the supply voltage, when the noise makes the supply voltage fall down to 3V. This is not acceptable at all. You must improve the 3.3V supply voltage.

    I guess that the LM2576 generates a big switching ripple at the +5V output and that this ripple punches through the LF33 due to a poor input ripple rejection at higher frequencies. Figure 23 of datasheet of LM2576 shows how to add an optional output ripple filter. You should add this recommended post ripple filter.

    Alternatively, for a test, you could supply the LF33 with a different +5V supply voltage, which is free from ripple.

    Can you show scope plots from the +5V supply as well? I guess that we will see lots of ripple...

    Kai

  • Dear Kai,

    I probe the supplies and saw the huge noise supplied from an external switching adapter to the LM2576 and propagates to the supply chain.

    I eliminated the noise by changing the main supply. Nevertheless the opamp behave as before.

    Regards,

  • Dear Kai,

    Please consider my following found as a point ( and not the solution to the problem ).

    I scaled down the input from [0 3.3] to [0 2.4] and increase the gain form 1 to 1.375 and in this configuration there is no any problem.

    Regards,

  • Hi Ras,

    can you show a scope plot of the +5V supply voltage at the input of LF33?

    Have you mounted the post ripple filter behind C71? And remove ZNR1 for a test. The zener diode could become conducting during normal operation due to manufacturing tolerances of zener voltage.

    Another idea is to run the OPA344 completely separated form the rest circuitry with an own supply voltage. Disconnect everything from the OPA344 section which isn't needed for its operation. If the problem still exists (even with a fresh OPA344) then it might have to do with the OPA344 itself, maybe caused by the used low pass filter topology in combination with rail-to-rail operation. Then, reducing the input sihnal and increasing the gain could be a remedy.  

    Kai

  • Former Member
    0 Former Member in reply to Ras Sharif50

    Thank you for all the input Kai!

    Hello Ras,

    I have several comments to make.  First, I think it is pretty clear that the supply is rather noisy.  I would have two concerns from this issue.  First, the amplifier's performance could be affected.  This is especially the case if the noise coincides with the input approaching the supply rails.  As Kai has alluded to, having the input exceed the supply could definitely lead to erratic behavior.  It's difficult to predict what that behavior would look like, but having the amplifier rail unexpectedly is possible.  Another thing to consider is that when you overdrive the input of the amplifier, you are also overdriving the output of the amp.  When this occurs, the amplifier's output will not respond properly for a certain period of time as it recovers from the overvoltage condition.  This is referred to as "overload recovery time" (see this post).  It may explain the "hysteresis" type behavior that is being experienced.

    I would also like to point out that if your analog supply has this much noise, then everything downstream will feel the effects.  This can be especially problematic if you plan on using the voltage to create a reference for the ADC.  Putting decoupling capacitors on the amplifier is good, but will not be enough on its own.  The signal out of the buck converter needs to be clean.  Kai's suggestion to add the optional filter may be valuable to you.  Another thing you can do is open another thread here on e2e, but for the buck converter.  Try to see if someone on that team can make suggestions regarding your output filtering.  It may even be a layout issue.  The LM2576 also has plenty of information on the datasheet for "best practices," which you may find useful.

    Scaling down the input and increasing the gain may be working because it then ensures that the amplifier's input does not exceed the supply voltage, even with noise on the supply.  I'm not sure how cleanly you can test all these things if you are using a PCB.  I have put in a part request for the OPA344.  The samples should arrive soon and then I will try to mimic your use of the part with clean supplies.

    Regards,

    Daniel

  • Hi Kai,

    I have the plots of the 3.3V supply now as following. The 5V at the input of LF33 was completely similar.

    I couldn't place a post filter on the pcb for now but i have removed the zener diode when I explain the results in my previous posts.

    I will test the OPA344 in a completely separated circuit in the first possible time.

    Regards,

  • Dear Daniel,

    As I tried in previous posts I eliminated such huge noise by replacing the main supply to the LM2576 and finally the supply is fairly clean.

    I should to remind the input to the opamp didn't override anymore. In my second test which was a constant 2.6V voltage at input the problem rises to exits.

    I didn't see any problem in my other OPA344 opamps in the circuit. It is probable that the reference voltage had some noise but certainly there is now such huge noise now.

    Thank you for doings to test the circuit. This is so valuable for me hope helping to solve the problem.

    Regards,

  • Former Member
    0 Former Member in reply to Ras Sharif50

    Hello Ras,

    Sorry for the delay in my response and thank you for the update.  Could you please summarize again the results of your testing with the now clean voltage supply?  I did not understand the portion where you state, "I should to remind the input to the opamp didn't override anymore. In my second test which was a constant 2.6V voltage at input the problem rises to exits."

    I am certainly happy to help by testing the circuit myself.  I am currently waiting for the parts to arrive and they should do so soon.  However, there is one piece of bad news.

    As you may have read in the American press, one of our campuses was recently damaged by a tornado and is closed.  Unfortunately, this is where I was hoping to conduct the testing for your circuit.  Consequently, there may be some delay before I can get some results for you.  It may take until later this week and possibly even early next week.  Please know, however, that this issue is a priority for me and I will do what I can to get some results as soon as possible.

    Regards,

    Daniel

  • Hi Daniel,

    I hope you all are fine?

    Kai

  • Hi Daniel,

    The problem persists with clean supply.

    I had conducted two test; The first was with a triangular at input of opamp. In triangular test, the output is as that you had saw. The second was with a constant voltage at input. Only in around 2.6V the output oscillates. With voltages lower than 2.55 and higher than 2.65 the problem goes away.

    I wish people in that territory return to the normal routine of life as soon as possible. There is no problem for the delay has arisen to the tests.

    I appreciate for your time and obligation.

    Regards, Ras

  • Former Member
    0 Former Member in reply to Ras Sharif50

    Hello Kai and Ras,

    Thank you for your concern.  Fortunately, the building was nearly empty at the time and no one was seriously injured.  However, the damage is significant.

    I think I now understand the results of the further testing.  I will try to reproduce the circuit as soon as I can.  I may also get the opportunity to speak with a colleague about the matter tomorrow.

    I will keep you informed and reply as soon as I have an update.

    Regards,

    Daniel

  • Former Member
    0 Former Member in reply to Former Member

    Hello Ras,

    Unfortunately, it now looks as though I will not have access to my lab until early next week at the earliest.  I'm also not sure if the site will be ready to receive samples.

    Due to this delay, I am going to ask another of my more experienced colleagues to take a look at the issue.  Hopefully, he can share some thoughts while we are waiting to be able to perform a physical test.

    I will continue to provide updates as I can.

    Regards,

    Daniel

  • Hello Ras,

    The OPA344 is rail to rail input. The data states (on top of page 10 in data sheet) that the input stage switches between the NMOS and PMOS input pairs near V+ - 1.3V and the observed instability is near  V+ - 0.6V. On both sides of the switch the input offset voltage can change quickly which can make the effective gain of the "unity gain" setup not be a constant 1.00. It is also a possibility that the upper range is not working on this device under test.  What percentage of samples show an instability?

    As a test, try removing the 22nF feedback capacitor and check for instability or any other issue.

    Your schematic does not show the output load. Where does the output go?

  • Former Member
    0 Former Member in reply to Ron Michallick

    Ron,

    The output goes to an ADC with an input capacitance of 40pF.

    Regards,

    Daniel

  • Thank you for your doings Daniel,

    Hello Ron,

    Please know the problem exists in a basic buffer circuit without load at output. The spoiled triangular wave ( as in first post ) outputs under rail to rail input condition.

    I found under a certain constant input voltage around 2.6V the output starts to fluctuate by itself.

    I am studying the datasheet about the transition voltage you noticed. In which condition the problem you pointed rises? Which configuration does prevent the operation?

    Regards, Ras

  • Hi Ras,

    I would build a test circuit only with the OPA344. Drop all the rest of the circuitry. Also, use a fresh OPA344 for the test and a clean, stable and noisefree supply voltage!! Run the OPA344 as a unity-gain follower as shown in figure 5 of datasheet. Also mount an isolation resistor of let's say 100...220R directly to the output as shown in figure 5 and mount a proper power supply decoupling capacitor of at least 100nF/X7R directly across the supply pins of OPA344.

    Now apply a sine wave of 50Hz to the input and observe the output voltage of OPA344 (behind the isolation resistor!) with the scope. Then, apply your 50Hz triangular signal to the input of test circuit and observe the output signal again.

    If everything is ok up to this moment, modify the test circuit and check the active low pass filter from your original post.

    Divide and conquer!

    Kai

  • Hello Kai,

    It is Exactly the thing that i have done in my last tests.

    Thank you for the efforts you make.

    Ras,

  • Former Member
    0 Former Member in reply to Ras Sharif50

    Hello Ras,

    I wanted to provide you with an update on my attempt to independently recreate your results.

    I have, within the last hour or so, been re-granted lab access.  I am waiting to find out what equipment is damaged and what is usable, which should happen soon.

    The OPA344 samples I ordered a week and a half ago now seem to have been re-routed to a different location.  I am attempting to track them down.

    If all goes well, I hope to be able to go into the lab and recreate your test by the end of the day tomorrow here in Dallas.

    I regret that this issue could not have been handled more quickly, but the situation was rather out of my control.  Thank you for being patient and please know that I am doing what I can to get this done quickly.

    Regards,

    Daniel

  • Hello Daniel,

    I am very grateful for your attempt and attention specially in your now situation.

    I temporarily solve the problem with limiting the input to the opamp.

    But I should to understand the cause and solution for the problem raised since I should to scale the input of opamp to full rail to rail.

    Regards, Ras

  • Former Member
    0 Former Member in reply to Ras Sharif50

    Hello Ras,

    Good news: The OPA344 samples have arrived and it looks like I will have all the equipment I need to build your setup and test it.  Unfortunately, I was not able to get the setup built and run today.  But, I expect to have results tomorrow.  Hopefully, this will help us to determine why the problem goes away when you limit the input, but appears during rail-to-rail operation.

    As a final question out of curiosity, have you tried changing the input frequency of the signal by one decade in each direction?  Say, from 5Hz to 500Hz?  It should not change anything.  But, it is something that one of my colleagues mentioned when I was discussing the issue with him today.

    I will have an update for you tomorrow.

    Regards,

    Daniel

  • Hello Daniel,

    Thank you for good news.

    As I remember, yes, I changed the frequency and the problem persist. For now, it is not possible for me to check again. If it is necessary I can do it in the first possible day?

    Regards,

  • Former Member
    0 Former Member in reply to Ras Sharif50

    Hello Ras,

    Thank you for your patience.  I was able to run some tests on your circuit today.  I used the OPA344 in the configuration shown in your original image.  In other words, I did not place an output load to mimic the ADC input.  First, I put in a 50Hz triangle wave at 3.3Vpp with a 1.65V offset.  Everything looked normal as the output showed little to no attenuation and only slight phase delay.  This was as expected.

    Next, I input a DC voltage at 2.6V.  The output was a clean DC voltage with no strange oscillation.  Then, I moved the DC input voltage from 2.4V to 2.8V in increments of 20mV.  Still, I saw nothing strange at the output.

    Below are my scope shots.  Time divisions are 4ms/division and voltage divisions are 1V/division.  Both channel 1 and channel 2 have the same 0V offset.  Channel 1 in yellow is the input.  Channel 2 in blue is the output.

    It looks as though my setup is behaving as expected, while your setup is not.  Is it correct that you have tried running this test while disconnected from the ADC?  The only thing I can think of that might be different for our tests is that I have a separate power supply and no load.

    Please let me know your thoughts.

    Regards,

    Daniel

  • Hello Daniel,

    So Great tests and plots. Thank you for the valuable effort and time.

    Now I am certain the problem relates to the pcb or other component in the circuit.

    I should to check all of the probable causes in the circuit.

    These tests are postponed to a possible time in the future for me.

    I will put the results here step by step. This may take time.

    I would like to appreciate you for contribution.

    Regards, Ras

  • Former Member
    0 Former Member in reply to Ras Sharif50

    Hello Ras,

    Thank you for your patience and you are welcome.

    If you need any other help, please feel free to ask more questions on e2e.

    You may wish to start a new forum post if it is not directly related as this thread is becoming quite long.

    Best wishes,

    Daniel

  • Hi Ras,

    what scope probe are you using when measuring the output voltage of OPAmp? Is it a 1:10 or 1:1 scope probe? How much is its capacitance?

    Have you mounted the isolation resistor I recommended in this post?

    Can you show a photo of your setup?

    Kai

  • Hello Kai,

    Thanks for your help.

    The scope probe is 10:1 with maximum 12 pF input capacitance.

    I have installed a 100 ohm resistor at output.

    Please look at my prototype:

    I can't follow my tests to find the probable causes. I want to add the post ripple filter after LM2576 and change the ldo.

    Regards, Ras