3V3 PIN to GND DC impedance

[sergiomarina]

ESP32 WROOM D0WD and D0WD-V3
I'm used to measure the DC impedance of the ESP32 before using it. The standard measures are 1,8-1,9KOhm and 4-500Ohm (depending on the positive tip of the tester.
I just received some ESP32 modules where the former impedance measures infinite (instead of 1,8-1,9KOhm)
Are these ESP32 modules defective?
I'm pretty sure they are defective, but I need a confirmation to be 100% sure and open a dispute towards the dealer.
Thank you.

[ESP_Angus]

Hi sergio,

By measuring DC impedance you mean that you're measuring resistance, is that right? Are you using a multimeter or some other type of tester?

This is probably not a meaningful way to characterize an active electronic part. The measurement is probably showing some interaction between the chip, module components, and the multimeter sensing voltage. However I don't think you can draw any conclusions from it.

Do the ESP32 modules you're testing not work when powered up normally? Can you measure the current draw into the 3V3 pin when you power it up, and/or the U0TXD pin output? This should give some indication of whether the module is fundamentally working.

[sergiomarina]

Hi, thank you for your quick answer.
Yes, "DC impedance" and "resistance" are the same.
I powered just one of those ESP32 and I found just 2-3mA the current consumption.
Of course no output on U0TXD.

I've taken the mentioned measures of the remaining ESP32s just scratching the plastic film of the stripe. I went through this precaution since I do want to wait for the dealer reply and then decide if it suites to open the ESP32s from the stripe. In particular the mattered stripe consists of 15 ESP32Es (chip D0WD revision 3).
At the first I've taken the first ESP32 of the stripe row and It had the correct V3V->GND resistances. I have soldered it onto the PCB and it is working flawlessly. Then I've tested an ESP in the middle of the stripe and I've found that only the ESP in the first position is performing properly (!)

As a matter of fact I have installed several dozens of ESP32 (D0WD and D0WD-V3) and I've always always found the resistance as per my previous post.
Even more, before powering the ESP32 my preventive test procedure is based on that measure, and I've never measured an infinite resistance.
Have you the chance to take the same measure? What result did you get? Please, let me know ...

[ESP_Sprite]

Anyone else doing this measurement is not going to give you any information, as an ESP32 is not a resistor and the measured resistance is dependent on e.g. the voltage the meter uses to measure, as well as random stuff like if any internal flipflops are in latchup or if the crystal oscillator starts doing its thing.

[sergiomarina]

Hi,
I agree that the multimeter exact value has a limited value, but I'm asking not regarding to the exact value but if infinite can be acceptable.
As a matter of fact:
- when the ESP32 does not work (please, see more detailed test result in my previous posts) my two multimeters measure infinite resistance (identical measure).
- vice versa, in the case multimeters measured a finite resistance (not important the exact value), ESP32s work flawlessly.
That is an evidence to me.
Have you the chance to take the same measure? What result did you get?
Please, let me know about the result of your measure.

Writing one week after.
desoldering the ESP32E from the pinboard I found the resistances between 3V3 and GND were in the expected range. Therefore I soldered wires instead of pinboard and the ESP worked 100% fine. So the problem was related to the pinboard rather than to the ESP.
Conclusion:
if a resistance (either one) between 3V3 and GND is infinite it does not means the ESP is defective (my apologies to have raised this flag). After the first poweron/test (performed successfully) the resistances (both directions) are as in the expected range: 1,8-1,9KOhm and 4-500Ohm. I made this test for 5 ESPs, and I've got the same result.

[sergiomarina]

Hi,
after two weeks I return to this subject since today I run into a disaster. Please, support!
Yesterday I just finished an ESP32 application for a meteo station and I started pre-production tests. Just this morning (after about 15 hours) the ESP32 got stacked. Looking at wireshark tracing the ESP did not recover after DEEP SLEEP cycle.
No reaction even after power down reset. No activities on WiFi and on serial port (so I couldn't monitor what was going on), still about 50ma current consumption. Exactly the same problem I encountered with the first ESP32 I tested as I reported in my first post.

Yesterday I prepared the test bench for the pre-production tests assembling a brand new ESP32 component and a pinboard. At the end of soldering, as usual, I measured the impedance 3V3 to GND impedance and I found not infinite values. I tested and I found about 90mA and 10mA current consumption in WiFi active and deep sleep modes. I started testing. After about 15 hours I found out that the server did not receive measured values.
After several power down reset I measured the 3V3 to GND impedance and I've found again infinite impedance, still the current consumption was about 50mA, but steady (meaning the ESP did not switch between ACTIVE and DEEP SLEEP status), furthermore the serial port was inactive toward the USB converter. So I dissoldered the ESP from the pinboard. After that I soldered the ESP with just wires. I measured the 3V3 to GND impedance and I found out it was again into normal range. I tested by monitoring current consumption (and WiFi activities) and I found out expected values between ACTIVE and DEEP SLEEP status. Therefore I decided to solder again the ESP together with the pinboard (the same pinboard since it was still in good conditions). After mounting I tested the impedances I they were as expected not infinite (just as a minor comment: meanwhile the ESP was cooling down the impedance changed from 1.7 to 1.45Ohm). I made a quick test: either WiFi and serial port run as expected. Now pre-production tests have been running again since 5 hours.
If the disaster happens again I am disappointed since it appears that these ESP32 are defective (as a matter of fact they are absolutely not reliable and useless for real IoT applications)
Have you any ideas, suggestions?

[WiFive]

There are millions of esp32 in the world and not many complaints of this nature. It could be either defective hardware or something you are doing wrong. So either try a different batch, try to reproduce the problem without soldering (small batch burn fixture), or send suspect samples to espressif for analysis.

[ESP_Angus]

Hi sergio,

If I understand correctly, you had an ESP32-WROOM module on your application PCB that was not working. You removed it, tested it in isolation and it worked. Then you resoldered it to the application PCB, and it worked again then. With all due respect, I don't see how you can conclude that the ESP32 module is faulty and it's not an issue with the board design or the assembly/soldering - as these are the variables that changed when the fault was resolved.

Are you able to share a schematic and a photo of your hardware, please?

If you haven't already, suggest downloading and carefully reading through the ESP32 Hardware Design Guidelines (some of these don't apply for designs using modules, but some of them do.)

EDIT: Also suggest checking your schematic has the key elements shown in the Peripheral Schematics section of the ESP32-WROOM datasheet, in particular a pullup resistor on the EN pin and the necessary capacitors on the VDD/GND connection.

[sergiomarina]

Hi Angus. I'm reading your post just now. Sorry for that but I did not know how to activate the alert feature (if any). I agree that the root cause search is not simple. I think is a defect of the ESP32 since the infinite impedance is not possible to be due to a bad weld. I would say that a bad weld could lead to a low impedance (and not infinite). But even most clear is that I measured touching directly the ESP32 pins with multimeter tips.
I have new findings from yesterday session. A ESP32E mounted on a pinboard active to collect an analog signal since beginning of May stopped to send sample to the server. Since it happens almost every week (quite bad but not a disaster and it could be due to several causes) I have implemented a mate ESP32 that resets it by powering it down. But after several power down resets I realized I faced again the issue again (ESP32 completely stacked, even the serial port was dead). So I did some testing to collect as many info I could. I have warmed up to 100degC (tested the impedance) and to 200 (tested the impedance). No changes. So I desoldered heating a little more. Still infinite impedance. Btw you are right this measure depends on the multimeter. Recently I've bought another multimeter and it measures infinite in either cases, so not usable for this test. Next test I gently touched the two pads 3V3 and GND (only those ones) with welder tips so to melt the tin. Just a quick pass! I tested with the multimeter and the impedance was again about 1400Ohm (anyhow NOT infinite). I've re-welded (re-using the same pinboard) and it worked just fine. Right now that ESP is again on the field. What happened in that 2" timeframe window? Very short window, I would say negligible. What happened? impossible to say to me!
These are low quality (if not defective) ESP32 components. The most likely conclusion I see.
In a previous purchase I bought 4 ESP32D WROOM from another dealer and they reliably work since 1 year. So I have not a large statistic, but it is 100% negative.
Sorry I have not taken picture to that ESP32 but I have other ones similar that can give you an exact idea of the SUT.