Thanks for your model. We are testing the EN63A0 and EN6337 model vs your Excel file and give you our feedback.
Regarding the EN63A0, in the datasheet it is writing that there is a type IV control loop. Could you explain me what is a type IV control loop. I know type I, II and III but never heard type IV.
Perhaps if you have a link (not found on Google).
Please find in attachment a little report based on your Excel tool and Pspice model for EN63A0 and EN6337.
We can see some differences.
For EN63A0, correlation seems OK up to 100 KHz but not at higher frequencies so PM seems OK but not GM.
For EN6337, correlation seems OK after 100 KHz, but not at lower frequencies.
We would like make our stability analysis based on Spice simulation rather than on the Excel file because we have a lot of parameters like different kind of capacitors... which couln't tough into account into Excel.
Thanks for your feedback
Hello @MBaqu ,
Thanks a lot for sharing the correlation report.
As i can notice from the report there is no big difference between the Excel tool and the PSpice model.
The little difference between the two tools that the spice model has more model complexity which its very hard for us to model that models accurately.
We managed to improve that, by applying some model parameters adjustment when you click the compensate button which will help to cover all the circuit variation on the actual silicon due to the process variation.
Using the pspice model should be okay and its results are accurate enough, but i would recommend you to check your final results with the excel tool even if you didn't model your circuit accurately inside the Excel tool, which will be a good reference for you to audit the pspice model results.
I just add a summarize in the report :
About the spice model, we will continu to correlate with our measurements, I could be able to provide our report to you but not on the public forum if you are interested.
I have again two questions :
- Regarding correlation, we have just the EN6337 model in no load which seems a little inaccurate, is it possible to improve it or no (if it is time consuming, we can try to help you on this process, we have some spice guys here (but again not in public forum)?
- Could you explain me the type IV compensation network, I know type I, II and III but not IV.
Hello @MBaqu ,
Sure, i am interested on your report, you can share it with me through a private message on Forum, i will send you and message so you can replay from there.
Regarding the EN6337 AC model, it is actually support Loading condition but you need to update the ILOAD parameter to the required load only as shown below.
Please note that when you change that parameters it is changing some internal components value that will be effected by the loading, so please don't remove or rename that parameter from your circuit.
But regarding the external current source, it will not be effected by that ILOAD parameter where under the AC analysis the current source represented as an open circuit, this is why if you measure the current it will be 0A.
You can replace that current source by a resistive load as shown below and use the ILOAD parameter to change the loading.
I added 0.001A to the calculation to avoid any effect when the iload is 0A.
Regarding type4 compensator, unfortunately i can't share any details about it where it is confidential information, but i could say it is very close to type3 but with some modification to improve the loop stability at higher bandwidth.
We made new simulations with all parameters of the EN6337 spice model modified, we missed that in our previous simulations indeed.
Please find a new little report on the EN6337. Trend of curves are correct but phase margin isn't really the same of Excel file (gap of more than 10°).
Hello @MBaqu ,
Thanks a lot for sharing your updated report.
i understand there is a 10Deg between the simulation and the Excel tool, but there is no issue on that where in simulation this will gives a worst condition than with the Excel tool, the tool has an internal algorithm that will do another circuit variation when you click on Compensate button and the tool will work to simulate that different conditions which will be close to what happens in real world.
The bode plot that the tool shows is the nominal condition without any circuit variation, so you will need to click compensate button to make sure that the system is stable.
If you would like i could help , if you could share the input/output conditions, the output caps details and if there is any second stage LC filter after the local output caps and i will help to suggest a suitable feedback components.
Regarding our correlation between measurements and simulation, I have again one question:
- What is the maximum frequency the spice model support for accurate simulation? Is there is some limitation close to the switching frequency of the component for exemple?