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.

OPA355 ckt with asymmetric overshoot

Other Parts Discussed in Thread: OPA355, LMH6629

I am stumped.   I have an OPA355 ckt configued with gain=+2. Rf=499ohms, Vs=+/-2.5V.  I have done everything I can think of to minimize stray capacitance.  The amp has plenty of decoupling capacitance.

The Problem:
When I apply a 0.5V, 10Mhz square wave to the input, the output has a large overshoot (more than 50%); but only on the falling edge.  What could cause this asymmetry?  

As an experiment I changed the gain to +1.  With a 499 ohm feedback resistor the overshoot remains but it disappears if I short across Rf.

  • Perhaps the asymmetry is in the amps output?

    If I place 5pf cap across the 499ohm Rf the overshoot is not noticeable; but this limits the theoretical frequency response to less than 100MHz which is not good enough.

    I believe I have a good layout; but maybe I have underestimated something.  

    The symptoms are similar to excess capacitance on the inverting input. But there is no way I have 5pf of stray capacitance on the inverting input node.  

    Is there an eval board for the OPA355 or at least documentation of the layout that was used to take the measurements in the datasheet? I couldn't find anything online.

    I have inserted my layout; maybe there is something I am missing?

     

  • Hello Paul,

    I'm not very familiar with OPA355 and the possible overshoot causes, so I won't make any comments. However, you should be able to use the SOT23 EVAL from other high speed amplifiers (e.g. LMH6629, etc.)  for this device. The Enable pin may not be routed on the board for SOT23-5 devices but I'm sure you don't need that for your purposes. This board is for the LMH6629:

    http://www.ti.com/tool/lmh730316

    I recommend you specify your load, input voltage, conditions in more detail and may be even attach a scope photo.

    Regards,

    Hooman

  • Hi Paul,

    I take back my comment about the Enable pin. This board I mentioned has that pin routed properly (so no issues).

    Does it help if you skew your OPA355 supply voltages (i.e. V+ = 4V, V-= -1V for 5V total)? With +/-2.5V supplies and a 0.5V input, you are approaching the upper Vin CM voltage range of V+ - 1.5V (=2.5- 1.5= 1V). So, that's why I asked about skewed supplies to see if that affects anything?

    What about your load? Is it DC coupled and returned to ground or some other scheme? Is there any load at all? Are you putting your scope probe tip right on the OPA355 output pin for measuring? What if you isolated the tip or used a 50ohm series load and terminate the scope in 50ohm (RL=100ohm).

    Is your input source coming from a 50ohm lab Generator properly terminated in 50ohm near the input of the amplifier?

     

    Regards,
    Hooman

     

    Regards,

    Hooman

  •  "Does it help if you skew your OPA355 supply voltages"

    I can not try that easily.  I was able to shift the input negative by 0.25V without any effect.  Besides the datasheet shows a pulse response under the same conditions as mine but with a 1Vpp input versus my 0.5Vpp. 

    "What about your load?"

    All of my measurements of the output are isolated with a 75 ohm resistor. I have tried both no load and 150 ohm DC coupled to ground - with no change.  One time I did put the probe right on the output - also without effect.

    "Is your input source coming from a 50ohm lab Generator properly terminated in 50ohm near the input of the amplifier?"

    It is a 75 ohm source; 75 ohm cable , and 75 ohm terminator - besides I have verified with the scope that the input to the amplifier does not show this overshoot.  I don't have a general purpose lab generator that can operate at these frequencies so I am using a purpose built ckt.

  • Hi Paul,

    Couple of comments:

    1. Supply Skew: For the effect I was noting, you'd need to raise V+ (not V-) to allow more input headroom. I would not bet that the scope photo in the datasheet is with input close to V+ (low headroom, as you are). The input CM in that image is not specified.

    2. Scope setup: I would make sure you capture some images (with no scope probe, 50ohm terminated scope and 50ohm in series with output) to compare.

     

    Regards,

    Hooman

  • "For the effect I was noting, you'd need to raise V+ (not V-)"

    I didn't say I raised V-; I said I dropped the input voltage toward V-.

    As per suggestion I moved GND and therefore the midpoint of the signal to  +Vs - 3.4V and there was no effect on the overshoot.

     I think it is very reasonable to assume that the center of the pulse waveform shown in the datasheet is GND (Vs/2); but you are right, it is not specified.  But also my original input signal was not really close to the edge of the common mode input range; it maxed at +V - 2.25V. 

    "I would make sure you capture some images (with no scope probe, 50ohm terminated scope and 50ohm in series with output) to compare."

    What would that show? My probe is calibrated and the same probe does not show overshoot on the input.  If the probe was bandwidth limiting the signal I would see less overshoot.  Attaching a coax cable to the circuit is not a trivial task.

  • Hi Paul,

    I understand the points you've raised and I agree with you. The scope return (ground lead) has fooled me with overshoot many times in the past, and that's why I prefer the coax to 50ohm scope termination for this kind of testing. But looks like you feel confident that's not the cause.

    Removing ground plane around / below the inverting input is also something to look at (although your layout does seem to be cleared off). Touching the scope tip to device pins would also tell you if you any of them is sensitive to capacitance and that could be a way to look for causes. If you had another device (different part number altogether but with similar speed), with similar pinout to drop-in, that could tell you more about your layout.

    As I said earlier, I'm not very familiar with this part's behavior.

    I'm hoping others more familiar in this forum would chime in.

     

    Regards,

    Hooman

  • Hi Paul,

    Ok. I'll throw in my last 2-cents:

    1. If you configure the amplifier as inverting instead and the overshoot disappears, then this points to OPA355 not being happy with its Common Mode voltage riding a fast edge. However, I know that this would take board surgery to do.

    2. Along the same line as #1 above, can you slow down the input edge rage / rise and fall time to see if you can affect things? An RC LPF on the input edge may tell you something. Some devices do not behave as well with super-fast input edges.

    3. Have you tried the datasheet Av=+2 feedback resistor condition of 604ohm (instead of 499ohm) to make sure that's not any cause (but I know if is very unlikely that it is)?

    Frequency response testing (if it were available) could also help to make sure your step response probing is not an issue. But I know that it may not be easy on your board. Using the EVAL board may be?

     

    Regards,

    Hooman

  • Hooman, Thanks for taking so much time with my problem.

    "1....configure the amplifier as inverting..."

    The datasheet indicates this should not be necessary and would not work in my application so the point seems moot.

    "2...slow down the input edge..."

    I may try this but... the amp is supposed to have 450MHz bandwidth which equates to a 0.8ns rise time and again my application requires this tolerance.

    "The scope return (ground lead) has fooled with overshoot many times ..."

    I am using a low inductance ground spring attachment on the probe tip.

  • Hello Paul,

     

    I have read this thread and I do not have any specific guidance either.  Can  you tell me the make and model of the prope you are using? 

    You were probing at the mid point of a 150 Ohm load correct?

    What is the destination for the signal from the OPA355?  Have you only measure the OPA355 with a probe?  Have you tested the full system without a probe?

     

    Regards,
    Loren

  • This is definitely NOT an issue with the probe or the scope.  With nothing changed except the value of the feedback resistor (499 ohm vs 0 ohm) I can make the overshoot appear or disappear.  

    I suppose it is possible that 75 ohms is not enough resistance to isolate the probe capacitance from the output -- but if that were the problem then reducing the gain would make the amp less stable not more stable.

    Perhaps if you explained your thinking regarding possible probe related issues? 

     

    "You were probing at the mid point of a 150 Ohm load correct?"

    I usually probe on the far side of a 75 ohm resistor from the output.  I have experimented with an additional 75 ohm resistor from there to gnd (150 ohm load) and also with no resistor to gnd (i.e. no load).  

    "Have you tested the full system without a probe?"

    I am not sure I know what you mean; but, I am building the signal chain one stage at a time and this is the first stage. I don't think adding more untested circuits would help clarify the issue.

  • Do you know what the capacitance is of the probe you are using?  Yes, I do suspect the probe could be causing measurement issues.  When we use probes in the lab we usually use high bandwidth FET probes with 1pF or less of capacitance.  For pulse response measurements we do not use probes at all. 

    What else is attached to the amplifier output beyond the screen shot of your layou?  Can you easily disconnect it?