0
\$\begingroup\$

In a product we designed, the LAN drive circuit(MII mode) is composed of LPC1768 and DP83848CVV. The function is very normal. There are no problems with factory testing. After using it in the hands of customers for 2-3 months, the product does not work one after another. Not even the computer. Analyzing the returned products, it was found that the RESET PIN of the DP83848C was short-circuited to the ground, and the LPC1768 could not be started. The RESET foot of DP83848 is directly connected to the RESET PIN of LPC1768 and then grounded by a 0.1uF capacitor.

on damaged PCB, if i desolder DP83848, LPC1768 works fine.

I have attached the Schematic PDF for reference.

We also have applied ESD protection as shown in LINK , we applied Basic ESD Protection, from PDF link , we used SP3002 for same. But that also didnt work. Now we are deadlock and we have to resolve this, we would like to know if anymore data is required from my end .we would like to receive some guidance to resolve this issue

\$\endgroup\$
2
  • \$\begingroup\$ What type of capacitor is C31 on the output of the REG1117? If it is ceramic, then you risk significant overshoot (or worse) at startup. Most 1117 variants have a minimum output ESR requirement for the output capacitor. \$\endgroup\$ Feb 27, 2020 at 10:48
  • \$\begingroup\$ Hi Peter, Thanks for your valuable time and response, yes it is ceramic (C31). \$\endgroup\$
    – ht_admin
    Feb 27, 2020 at 12:01

1 Answer 1

0
\$\begingroup\$

The DP83848 absolute maximum for input voltage is 4.2 + 0.5V = 4.7 V.

The LPC1768 inputs are 5V-tolerant, and can withstand up to 5.5 V.

I think your problem is located in the power supply design. You use two pi-filters, and this type of filters can be extremely dangerous on specific power level rise times, with the circuit oscillating up to 2x supplied amplitude, thus killing your electronics.

Why 2-3 months? I would say coincidence, but looks like something in the reset pin input circuit being constantly worn out (e.g. protection diode), and finally failing with short circuit. Check all other pins to see if there're any other with the same symptoms.

Perform a simple test: connect scope to the 3.3 V power rail and see how power voltage behaves when you turn the system on, and turn it off - using your lab power supply and the one(s) used in the field. It would be great if you will see clear exponential increase up to 3.3V, but it may also happen that power fluctuates between 2 and 5 volts for several milliseconds.

\$\endgroup\$
5
  • \$\begingroup\$ Hi Anonymous, Thanks for your valuable time and response, The L6 and L2 are actually Ferrite bead but they are not implemented, instead 0E is used instead of L6 and L2. \$\endgroup\$
    – ht_admin
    Feb 27, 2020 at 12:02
  • \$\begingroup\$ Did you perform the test asked for? \$\endgroup\$
    – Anonymous
    Feb 27, 2020 at 12:18
  • 1
    \$\begingroup\$ Hi Anonymous, on the on used in field Power supply, there is not fluctuation on power up or power down, i am trying to arrange Snapshot, and also test on lab power supply. Will update on that. \$\endgroup\$
    – ht_admin
    Feb 27, 2020 at 12:46
  • \$\begingroup\$ Hi Anonymous, I have addred images for condition : test: connect scope to the 3.3 V power rail and see how power voltage behaves when you turn the system on, and turn it off - the one(s) used in the field Images brief: IMG_20200228_185117835 -> on 5Vdc point, Without LAN or Serial (RS232) IMG_20200228_185934849 & IMG_20200228_190145025 -> full load with EThernet and Serial input ON. IMG_20200228_185546190 -> on 3.3VDc test point, without LAN or Serial \$\endgroup\$
    – ht_admin
    Feb 28, 2020 at 13:43
  • \$\begingroup\$ Good news in overall. Now set scope to 1 us (or even less, try different micro or nano-timing), and make it start sampling on power on trigger (rising edge). Thus you will also ensure that there're no transient issues at the "microlevel" which are not seen (or not sampled) on "macro" 1-second level. \$\endgroup\$
    – Anonymous
    Feb 28, 2020 at 14:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.