Dear all,
This is an urgent request. Our manufacturing process is stopped.
We are Lectra company and produce computer boards in large CNC systems (turnover = 200M€/year). 400 boards / year for 8 years and a design never changed.
We use LM2678 to generate several power voltages from a main 24VDC.
We manufacture this board since 2006. We didn't encounter any LM2678 problem till several recent weeks.
We have two subcontrators which manufacture each half the overall quantity. The issue is observed only at one.
The issue is: the 3.3V voltage changes according to the board activity. We observe a drop about 0.2 to 0.6 V. Our design was calculated with Wenbench in 2004 for 3A. The typical current is 1.8A. I attached the power design CAD file.
A important think: we observe the issue only at one subcontrator N°2.
At the subcontrator N°1, the part marking is : KM26ACE3. The parts run well.
At the second subcontractor N°2, the LM2678 parts don't run well and their marking is : 35ACFSUG3, 38A59NUG3, 38AJ9NUG3.
We tested former components on these boards and they run well. Their marking is JM31AC. We tested good parts from the subcontractor N°1 and they run well on boards manufactured by the subcontractor N°2.
A wrong LM2678 doesn't run well on boards manufactured by both subcontractors.
So the issue follows the LM2678 and doesn't depend on the board and the subcontractor.
On the wrong boards, with the good parts, we tested up to 3.5A by connecting 2 boards on the same 3.3V (only on regulator drives the 2 boards obviuosly). No problem observed. So the design seems to be good for us and the current margin is important.
The design uses a ground plane for the GND and large leads. We use LESR components.
We think we are facing to a process derating on the LM2678 components (no problem during several years and a recent issue at only one subcontractor).
On the wrong parts, we tried to remove the Cboost capacitor on one board. And we observed the LM2678 seems to run well (voltage equal to 3.3V exactly, whatever the output current) ?
Is this capacitor mandatory ? We mount the regulator on a large dissipator.
This morning, we are going to test an RC (10nf + 10ohms) on the Shottky diode. I am waiting for the result.
Can you analyse the parts markings ? Do you thing we are facing to a wrong components batch ?
Thanks for your very quick reply.
Lérisson Jean-Pierre.
jp.lerisson@lectra.com (it's better) or jp.lerisson@gmail.com
www.lectra.com (France)
