솔루션 찾기 기술 지원

Difference in Transformer designs between PIxls and PIExpert.

Posted by: jkaramchandani… on

Dear sir/ madam,

I have generated one LNK625PG based designs in PIxls and PIExpert. I have following queries regarding the generated designs which I have attached along with this mail.

1. Why does PIExper and PIXls (spreadsheet) show bit different transformer designs. While primary and secondary winding configurations are almost same, bias and feedback winding no. of turns are different in both?

2. Why doesn't PIExpert shows integral no. of winding layers, instead it shows 0.68, 0.85, etc? Is there any way to make no of winding layer a integral no. in PIExpert or other way whether transformer vendors can wind in transformer in 0.65, 0.85 layer, etc?

3. It seems that PIExpert gives us all required design details but I am only confused with transformer design i.e which one to follow PIExpert or PIxls.

4. Is there any way we can map exact spreadsheet transformer details into PIExpert and obtain the schematic and other details?

5. Also feedback resistors values are different for both transformers in PIxls and PIExpert, pl advice which one to follow?

Thanks,
jagdish

Files

댓글

Submitted by PI-Wesley on 05/11/2012

Hi


1. This is due to the fact that PIExpert assumes that both worst case (low) current limit and worst case (low) switching frequency can happen at the same time, whereas the PIXLS software use the I2F parameter of the datasheet. This means that PIExpert is a bit too conservative in the design, and the PIXLS does it in a more optimal way.

2. The non-integral number just helps to see how much the last layer is full. Usually (especially if shields are present) you should try to fill the layers as much as possible. However this may not be always possible due to the transformer being full already. It's a trade off really, and the precise numbers help the designer make a choice.

3. I would follow PIXLS in the specific case of LNK-CV. We are working to correct the PIExpert version on the next patch

4. You can do a manual design, but you won't be able to fix the inductor value. What this means right now for PIExpert, is that because of the the slightly bigger inductance, you will notice that the output current ripple is a bit lower than on your pixls design. Other than this no other critical values should be different.

5. That depends on which transformer you choose. Keep in mind that you may have to fine tune them in a real prototype.

Submitted by jkaramchandani… on 05/12/2012

Hello sir,

Thanks for your response and clarifications. I would like to ask general questions,

1. How reliably PIExpert provides the design? Can I go ahead with PCB based on PIExpert design that I have attached? I shall appreciate if you could just have look at that and advice us?

2. I feel bit comfortable with PIExpert since it provides all details right from schematic, layout and BOM. So for other future design, if I use PIExpert, then how close to good design does PIExpert provides the solution?

3. What are the recommendations that you will provide while working with PIExpert for any family of device (not just Link-VC)?

I shall highly appreciate if you could pl reply to my above comments.

Looking forward to your reply.

Jagdish Karamchandani
Technical Lead (R&D)
Cimcon Software India Pvt. Ltd.
+91-079-26578639
Fax: +91-079-26578659

Submitted by PI-Wesley on 05/14/2012

Hi


1. PIExpert is very reliable in terms of getting you a working design, however it likely won't provide you best in class performance, also because the ideal performance changes a lot with the application (industrial, consumer, high-end, low-end, etc). Some examples of optimizations should be done on the bench include:


a. Primary clamp circuit. We tend to be conservative in the clamp selection, in order to be sure that it will correctly operate even if the transformer parasitics are worse than expected by the software. The clamp selection itself is usually a compromise between standby efficiency, full load efficiency, EMI, and price. This needs to be checked by the design engineer.


b. EMI components. It is extremely difficult to predict in advance the EMI performance of the system, as it is dependent on layout, loads to be connected, case, applicable standards for individual cases vary, and there is a number of other factors. We base these values on experience, but once again for optimization, prototype testing is essential


c. currently no surge components are include (like a MOV).


2. PIExpert will give you a good head start, but as mentioned on point 1, a good design is highly dependent on the objectives of an individual project (space constraints, price, performance).


3. I always advise you to take a look at the application note as it goes into a lot of detail on the different trade offs you can do (you can find the app notes here). We also have an extensive library of existing designs here that can help you a lot in terms of performance analysis.


I hope this helps you further.

Submitted by jkaramchandani… on 05/14/2012

Hello Sir,

Thank you very much for your answers, your responses clarified major doubts we have.

Jagdish Karamchandani
Technical Lead(R&D)
Cimcon Software India Pvt. Ltd.
+91-079-26578639
Fax: +91-079-26578659