Unusually High Output Offset Voltage in AD8337 LTSPICE Model

Hello govish87,

Currently, we are on a holiday break and we kindly request for your patience and understanding. We'll verify the offset voltage issue on dual supplies once we get back at the office.

Apologies for the inconvenience caused. I also agree that the spice model has an issue with single supply operation as I've already reported this internally in the past.

Best regards,
Paul

Hello govish87,

I've also used the SPICE web model of AD8337 and was able to get a result quite similar to Figure 74. (Vout = 4mVpp)
   

Please take note that on Figure 74, Vs = +/-2.5V was used. In addition, the ramp signal applied on the Vgain pin has a frequency of 1kHz. This could be the reason why you have obtained a different result from what is shown on Figure 74.


I hope this helps.

Best regards,
Paul

Hi Paul, 

Thanks for your response, and hope you had a good holiday season.
I'm still slightly confused, however. Here's an excerpt from the electrical characteristics section of the datasheet--note the Output Offset Voltage specified at +/- 25mV worst case at peak gain (though I understand that this measurement is also made with +/-2.5V supplies):

Does output offset performance change significantly across supply voltage variance? 

<EDIT>: To answer this question, I see this chart in the datasheet:

There does seem to be elevated offset voltage at increased supply, but still seemingly not on the order of the +50mV that I've been observing </EDIT>

Regardless, I associate Output Offset Voltage to be a DC characteristic in any case--why is it fair to measure output offset with an AC waveform? Is there a particular part of the SPICE model that may be responsible for this discrepancy between DC and AC offset performance, and is that representative of true device behavior?

Best,
Ishaan

Hello govish87,

- Yes, you are correct. Increasing the supply voltage will also increase the offset voltage.

- As checked also, it looks like the offset voltage of AD8337's SPICE model at +/-5V Supplies were not modeled. Apologies for the inconvenience. In real life, the typical offset voltage performance of the part at +/-5V should look like the one on Figure 15.

Regardless, I associate Output Offset Voltage to be a DC characteristic in any case--why is it fair to measure output offset with an AC waveform? Is there a particular part of the SPICE model that may be responsible for this discrepancy between DC and AC offset performance, and is that representative of true device behavior?

- Please correct me if I'm wrong. Are you referring to the offset voltage plots of Figure 15 and Figure 74? Please take note that the voltage at the VGAIN pin continuously changes, which effectively changes the gain and thus, the output offset voltage. The offset voltage specification from the electrical characteristics table were only taken at a specific gain setting (VGAIN = 0.7V, 24dB). The output offset voltage spec is a DC spec. 

Ex: Looking at the table below, the acceptable output offset voltage value for linear operation is from -25mV to +25mV, (+/-5mV, typically) when Vs = +/-2.5V
|

Now, if we take a look again at Figure 74, at around Vgain = 0.7V, we can get an offset voltage value of approximately -1.5mV. This value directly correlates with the Output Offset Voltage spec from the table above. -1.5mV is within the accepted output offset voltage value specified.


Best regards,
Paul

Hi Paul,

Thanks for the reply again. I'm curious about why I'm seeing different behavior of the output offset voltage in my DC operating point simulation than what you're seeing in the transient simulation--is there a characteristic of the AD8337 that leads to this discrepancy? I would expect a characteristic curve like the one shown in Figure 15, rather than the one I'm seeing in my LTSPICE charts. 

My hypothesis is that this is likely because operation at +/-5V split supply is not captured. Would it be possible to get a corrected SPICE model in the near future with device characteristics more completely modeled? I think this would help folks with issues similar to mine + the ones highlighted in the original post. 

EDIT:
I quickly verified, and with +/-2.5V supplies, the results for the DC operating point simulation are as expected; see below:

The final answer to my original question is *because operation of the AD8337 with +/-5V supplies is not accurately modeled*. It would still be desirable to get a corrected SPICE model in the near future with device characteristics more completely modeled. I think this would help folks with issues similar to mine + the ones highlighted in the original post.

Best,
Ishaan

Hello govish87 

Apologies for the inconvenience. As I mentioned on my last reply, the offset voltage of AD8337's SPICE Web model at +/-5V Supplies were not modeled. I'll report this internally for it to get fixed.

But rest assured that in real life, the typical offset voltage performance of the part at +/-5V should look like the one on Figure 15.

Regards,
Paul

Sounds good!

Thanks for all the insight and support--I really appreciate it. 

-Ishaan

Sounds good!

Thanks for all the insight and support--I really appreciate it. 

-Ishaan