LMG1020: How to improve rise time (Tr) if connected to a GaNFet (EPC2001C)

Hi Tianlei,

Thanks for reaching out about lmg1020.

Thanks for sharing debug info.

If the issue is happening on both eval boards then the laser load could be the cause for long tr.

What laser are you using? How long are the laser leads? They should be as short as possible for a low inductance power loop. 1inch is already very long. 20nH of power loop inductance is gained. And 1nH is the goal for Lidar.

This can lead to long fall time due to parastics source inductance preventing the GaN from turning off.

The long rise time can be due to similar affects.

Can you achieve 1ns tr with an SMD laser diode?

Thanks,

Hi Jeff,

Thank you for you replay.

I use a 1W VCSEL which package is 3mmx3mm. 

The laser leads are very short. When we desined our driver board, we plasement laser, discharge capacitor, EPC2001C and LMG1020 as close as possible. 

For the EPC9126 evaluation board, I think it should also follow the design rules. So both board should have short leads. 

Now, I am not focusing on laser's rise / fall time. I am focusing on LMG1020's rise /fall time when it has a load. 

And again, I can get a 400ps rise time if LMG1020 dose not have a load but LMG1020's rise time changes to 3ns when soldered a GaNFet (EPC2001C).

I also did the following experiments but won't change anything. I think what I am talking the long rise time is caused by a GaNFet not a laser or circuits layout after GaNFet . 

1. With or without laser.

2. Turn on or off laser changing power supply. 

Hi Tianlei,

Thanks for the detailed update.

Can you confirm the gate resistance from the 1020 output to the Gan gate?

3ns rise time could mean a 10ohm gate resistor.

0ohms can be used to achieve 400ps tr as long oscillations and thermals are ok.

Thanks,

Hi Jeffrey,

Thank you for your reply.

If there is no GaNFet out of LMG1020, the rise time I measured is 400ps, in spite of 0ohm, 2ohm, or 10ohm.

3ns rise time is measured the test point afer 0ohm or 2ohm gate resistance when there is a GaNFet(EPC2001C) after gate resistances. 

My point is GaNFet influences LMG1020's rise time, the reason may be GaNFet's capacitor or LMG1020 drive current is too small. 

Tianlei

Hi Tianlei,

Thanks for the update,

Can you confirm the GaN and LMG1020 is not damaged?

Does the same thing happen with a new GaN and 1020 soldered?

Can you share your schematic and layout so I can see what is happening from the 1020 output to the GaN gate?

Thanks,

Hi Jeff,

Yes, I confirm the GaN and LMG1020 is not damaged. 

I tested several my boards and one EPC's evaluation board: EPC9126 and got same results. 

I attached EPC9126's documents. 

May I have your email and send you our design file separately?

Thanks. EPC9126xx_qsg-rev3.pdf