Bulk test of nodeMCU's

Moderators: grovkillen, Stuntteam, TD-er

Post Reply
Message
Author
Shardan
Normal user
Posts: 1112
Joined: 03 Sep 2016, 23:27
Location: Bielefeld / Germany

Bulk test of nodeMCU's

#1 Post by Shardan » 04 May 2017, 20:02

Hello all,

for testing a fix i've used several nodeMCU's.
This lead to a bulk quality test of these boards as i run in several problems.

I've used two different versions, the older one with CP2102 as USB-TTL converter (a white back label says "AMICA" for brand) and a newer version with a CH340G, labeled with "LoLin".
You can easily differ between the used USB-TTL chips;
.
USB2TTL_Diff.jpg
USB2TTL_Diff.jpg (78.85 KiB) Viewed 3427 times
.
For short: The results were disappointing.
From 22 new nodeMCU's i tested six were broken!

The errors:
Two of the "AMICA" nodeMCU's run into crashes from booting on.
I tried to clean up the flash with a "zero" binary and flash ESPEasy again, but no better result.
This might be caused by a faulty flash, but i'm not sure about that.

The "LoLin" nodeMCU's gave an even worse result.
4 out of 10 tested MCU's did not work.
2 did not show any reaction, no "flash" of the blue LED on power up, no answer on serial. ESPlorer did not even detect the COM-Port.
One showed the COM-Port but it was not answering on ESPlorer and could not be flashed
One seemed to be OK first, i flashed, blue LED flickered while flashing. After restart the original AI-Thinker firmware showed up.

Further examination showed that these MCU's were mechanically damaged.
Look at this picture, a gross example:
.
DSCF0137.JPG
DSCF0137.JPG (65.95 KiB) Viewed 3427 times
.
Did you notice the dent on the ESP-12E? Obviously these parts have undergone rough handling.
I'm not able to decide if that happened on transport, on manufacturing or in store.

I examinated the other broken MCU's and found similiar traces. Things like broken connections in the PCB,
cut off soldering points and such. Some of these only visible with a strong magnifying lens or a microscope
or by checking connection by connection with a multimeter.

I've ordered some LoLin MCU's again to recheck and make sure if this was a problem of rough transport or if
they handle these parts forcefully on manufacturing. I'll report back.

Meanwhile i've quick-checked my stock of WeMos D1 mini, they all at least boot up without issues.
A long term testing will follow.

A bunch of nodeMCU's is running on permanent testing atm. It seems they are running fine atm (Uptime around 1200 minutes now).
.
DSCF0130.JPG
DSCF0130.JPG (81.23 KiB) Viewed 3427 times
.
So what are the consequences?

For future checks & tests for now I'll rely more on the WeMos boards I got
(Aliexpress - WeMos Shop. Be aware there are remakes from other manufacturers, I did not test these!)
And for sure it obviously pays to try another nodeMCU if something does not work or shows strange behaviour.


Regards
Shardan
Regards
Shardan

User avatar
toffel969
Normal user
Posts: 469
Joined: 03 Jan 2017, 10:58
Location: Germany

Re: Bulk test of nodeMCU's

#2 Post by toffel969 » 04 May 2017, 21:23

Shardan wrote:
04 May 2017, 20:02
Hello all,

for testing a fix i've used several nodeMCU's.
This lead to a bulk quality test of these boards as i run in several problems.

I've used two different versions, the older one with CP2102 as USB-TTL converter (a white back label says "AMICA" for brand) and a newer version with a CH340G, labeled with "LoLin".
You can easily differ between the used USB-TTL chips;
.
USB2TTL_Diff.jpg
.
For short: The results were disappointing.
From 22 new nodeMCU's i tested six were broken!

The errors:
Two of the "AMICA" nodeMCU's run into crashes from booting on.
I tried to clean up the flash with a "zero" binary and flash ESPEasy again, but no better result.
This might be caused by a faulty flash, but i'm not sure about that.

The "LoLin" nodeMCU's gave an even worse result.
4 out of 10 tested MCU's did not work.
2 did not show any reaction, no "flash" of the blue LED on power up, no answer on serial. ESPlorer did not even detect the COM-Port.
One showed the COM-Port but it was not answering on ESPlorer and could not be flashed
One seemed to be OK first, i flashed, blue LED flickered while flashing. After restart the original AI-Thinker firmware showed up.

Further examination showed that these MCU's were mechanically damaged.
Look at this picture, a gross example:
.
DSCF0137.JPG
.
Did you notice the dent on the ESP-12E? Obviously these parts have undergone rough handling.
I'm not able to decide if that happened on transport, on manufacturing or in store.

I examinated the other broken MCU's and found similiar traces. Things like broken connections in the PCB,
cut off soldering points and such. Some of these only visible with a strong magnifying lens or a microscope
or by checking connection by connection with a multimeter.

I've ordered some LoLin MCU's again to recheck and make sure if this was a problem of rough transport or if
they handle these parts forcefully on manufacturing. I'll report back.

Meanwhile i've quick-checked my stock of WeMos D1 mini, they all at least boot up without issues.
A long term testing will follow.

A bunch of nodeMCU's is running on permanent testing atm. It seems they are running fine atm (Uptime around 1200 minutes now).
.
DSCF0130.JPG
.
So what are the consequences?

For future checks & tests for now I'll rely more on the WeMos boards I got
(Aliexpress - WeMos Shop. Be aware there are remakes from other manufacturers, I did not test these!)
And for sure it obviously pays to try another nodeMCU if something does not work or shows strange behaviour.


Regards
Shardan
Shame, what a disappointment. Sorry to hear that. All nodemcu (8 in total) I bought were ok, though I bought them in german shops (www.ex-store.de and www.exp-tech.de) . reliability at a certain cost...

Regards Sebastian
Domoticz on Raspi 2 -- 14 ESP units (hacked Sonoff,NodeMCUs, Wemos, self-built units) running with RC140- Mega 2.0.0 dev8

User avatar
lucaberta
Normal user
Posts: 59
Joined: 09 May 2016, 11:26
Location: Lausanne, Switzerland
Contact:

Re: Bulk test of nodeMCU's

#3 Post by lucaberta » 05 May 2017, 00:31

I too had hardware issues with a couple of NodeMCU v1.0 boards. Really bad soldering leaving open lines instead of good connections. I had to resort to some jumper wire to close the connections as soldering would have been too difficult.

Never had a problem with any of the Wemos devices, There are fewer GPIOs exposed, but the build quality is so much better, and they are smaller in size. I also love the choice of pin headers provided by Wemos with each board and shield. Those are the boards I use mostly these days. Too bad there is no flash button, though!

Andreas Spiess reviewed the new Wemos D1 mini and D1 light just today, his video appeared on my YouTube subscriptions a few hours ago, worth watching:

https://youtu.be/JslhaM5FHZI

User avatar
toffel969
Normal user
Posts: 469
Joined: 03 Jan 2017, 10:58
Location: Germany

Re: Bulk test of nodeMCU's

#4 Post by toffel969 » 05 May 2017, 11:10

lucaberta wrote:
05 May 2017, 00:31
I too had hardware issues with a couple of NodeMCU v1.0 boards. Really bad soldering leaving open lines instead of good connections. I had to resort to some jumper wire to close the connections as soldering would have been too difficult.

Never had a problem with any of the Wemos devices, There are fewer GPIOs exposed, but the build quality is so much better, and they are smaller in size. I also love the choice of pin headers provided by Wemos with each board and shield. Those are the boards I use mostly these days. Too bad there is no flash button, though!

Andreas Spiess reviewed the new Wemos D1 mini and D1 light just today, his video appeared on my YouTube subscriptions a few hours ago, worth watching:

https://youtu.be/JslhaM5FHZI


In comparison I would say:

NodeMCU advantages
- bigger 3.3 V power regulator with, enables you to drive many sensors/actuators without secondary power regulator
- more 3.3V and GND pins allow for direct connection of sensors with dupont wires: great for prototyping
- probably very good as your first ESP device

Wemos D1 mini advantages
- smaller form factor
- D1 Mini pro version has external antenna connector and at least 4Mb flash --> great if WiFi coverage is not very good (note the necessity to shift the 0-Ohm link)
- Better built quality.

I also somewhat prefer the Wemos D1 pro mini.
Domoticz on Raspi 2 -- 14 ESP units (hacked Sonoff,NodeMCUs, Wemos, self-built units) running with RC140- Mega 2.0.0 dev8

Shardan
Normal user
Posts: 1112
Joined: 03 Sep 2016, 23:27
Location: Bielefeld / Germany

Re: Bulk test of nodeMCU's

#5 Post by Shardan » 05 May 2017, 14:25

toffel969 wrote:
04 May 2017, 21:23
Shame, what a disappointment. Sorry to hear that. All nodemcu (8 in total) I bought were ok, though I bought them in german shops (www.ex-store.de and www.exp-tech.de) . reliability at a certain cost...

Regards Sebastian
Hello Sebastian,
I've checked through the Shops you mentioned. With Exp-Tech i'm sure that are the same i use ("AMICA" with CP2102 USB-Chip).
With the other shop they don't show the backside so i can't say for sure, but i assume they are same too.
From 18 tested of those 2 were broken, for what reasons ever. Not nice, but i can live with that as i pay less then half the price.
But these are version 2-NodeMCU's and they are getting more and more rare on the China market.

The other 10 I tested ("LoLin")were version 3 with a CH340G USB chip. and I found much more broken ones.
As far as i remember the broken ones came in one lot, but I'm not 100% sure about that. They might have
get mixed up over some time.
So it might be a "Lemon Produktion" or they have been damaged on transport. As said i can't decide that atm.

As soon as the new lot arrives - may take some weeks - i'll see if that is such a poor quality too.

Regards
Shardan
Regards
Shardan

Shardan
Normal user
Posts: 1112
Joined: 03 Sep 2016, 23:27
Location: Bielefeld / Germany

Re: Bulk test of nodeMCU's

#6 Post by Shardan » 05 May 2017, 14:44

toffel969 wrote:
05 May 2017, 11:10
In comparison I would say:

NodeMCU advantages
- bigger 3.3 V power regulator with, enables you to drive many sensors/actuators without secondary power regulator
- more 3.3V and GND pins allow for direct connection of sensors with dupont wires: great for prototyping
- probably very good as your first ESP device

Wemos D1 mini advantages
- smaller form factor
- D1 Mini pro version has external antenna connector and at least 4Mb flash --> great if WiFi coverage is not very good (note the necessity to shift the 0-Ohm link)
- Better built quality.

I also somewhat prefer the Wemos D1 pro mini.
In fact this were the reasons why I've chosen the nodeMCU's. More GPIO and a somewhat "fat" regulator and many pins and led out GPIO's.
It's the perfect starter device for some breadboard testing indeed.
(The (un-)reliability of breadboards is on another note.... ;) )

The WeMos D1 Mini has it's advantages and disadvantages too.
It is possibly the best choice if you have a well defined project and want a quick solution.
Switching an air conditioner or humidifier are examples for that.
Take a WeMOS mini, a relay shield and a sensor shield and you're go.
No doubt the genuine WeMos D1 are better quality compared to nodeMCU's.
Again, meanwhile there are remakes of these boards on the market, not tested yet.

A disadvantage is the fixed layout. Combinations are limited, a second relay shield - not possible.
The PCB layout defines the GPIO for the relay shield so two relay shields will have same GPIO.
Yep, one may use a relay shield and a standard relay board of course...but I feel this breaking
the idea a bit.

As toffel said, the small 3,3V-regulator on the WeMos puts up some further limits.
If you stay with the standard WeMos shields it's ok. If you use it as a
hardware development base it gets limiting.


Regards
Shardan
Regards
Shardan

krikk
Normal user
Posts: 115
Joined: 28 Feb 2017, 07:57
Location: Austria
Contact:

Re: Bulk test of nodeMCU's

#7 Post by krikk » 05 May 2017, 20:30

wemos d1 r2 is also very good for started, but seems like wemos does no longer sell them:

https://www.aliexpress.com/item/WeMos-D ... 65666.html

User avatar
toffel969
Normal user
Posts: 469
Joined: 03 Jan 2017, 10:58
Location: Germany

Re: Bulk test of nodeMCU's

#8 Post by toffel969 » 06 May 2017, 12:55

krikk wrote:
05 May 2017, 20:30
wemos d1 r2 is also very good for started, but seems like wemos does no longer sell them:

https://www.aliexpress.com/item/WeMos-D ... 65666.html
Yes true, it also has remarkable V-Inputrange and provides 5V strong enough to drive relays etc. If size doesn't matter (its the same form factor as arduino UNO) ;-) then just the R2 is also a very good option.
Domoticz on Raspi 2 -- 14 ESP units (hacked Sonoff,NodeMCUs, Wemos, self-built units) running with RC140- Mega 2.0.0 dev8

costo
Normal user
Posts: 490
Joined: 21 Nov 2015, 15:03

Re: Bulk test of nodeMCU's

#9 Post by costo » 07 May 2017, 16:47

Shardan wrote:
04 May 2017, 20:02
.............
I've used two different versions, the older one with CP2102 as USB-TTL converter (a white back label says "AMICA" for brand) and a newer version with a CH340G, labeled with "LoLin".
You can easily differ between the used USB-TTL chips;
........
I have used the Amica type for some time , the type you call the ´older´ one. I ordered that one because I liked the feature of being smaller. On a breadboard you can plug dupont cables next to the headers. The Lolin one is too wide for that.
Mine worked fine, it was well packaged and had no visible damage at all. After I flashed it maybe 50 to 100 times then it started acting weird. Rebooting with no cause and it got worse every time I flashed again. At first I suspected the power circuit along the AMS1117. I soldered some extra capacitors of 100nF on each side and then it behaved well again for a few days and flashes. Then started to spontaneous reboot again.
I ordered a new one with the CP2102, this time it was nameless, no Amica on it. Up to now it works fine but I did not reflash it that many times.

I suspect the ESPmodule, particularly the flash chip. The ESP modules, although marked with AI thinker, are probably manufactered by different factories. Cheaper ESPboards may use the cheapest flash chips from the chinese market which are famous for their feature of only to be flashed maybe a 100 times after which they are corrupt.

None of my Wemos D1 modules give me troubles so maybe the quality of the ESP module is better.

Shardan
Normal user
Posts: 1112
Joined: 03 Sep 2016, 23:27
Location: Bielefeld / Germany

Re: Bulk test of nodeMCU's

#10 Post by Shardan » 07 May 2017, 17:43

Hello costo,

the "Amica" ones are somewhat reliable here too.
As i buy them from China for a really low price it's acceptable to me if one out of ten is defunct.

I bought two lots of five of the "LoLin" and sorted out half of them - that is definitely not acceptable.
As i stored them together in one drawer I'm not sure but it might be one lot that is broken.
As i know the packaging (bubble wrap and enevelope) it might be a transport problem.
A heavy packet fallen on the envelope and that's it. Who knows?

The "AI Thinker" is not a manufacturer. It's the firmware that the ESP's are delivered with from Expressif.
Most manufacturers put it on the ESP. Remember that the ESP's you usually get on the market are not
made by Expressiv, they just use the ESP8266 from Expressif. Expressiv delivers the "Wroom-01 / 02" ESP but
it is not very widespread.

I had downloaded the datasheet for the 1MB flash used in sonoff basic and some ESP-01 for a reason
these days. They say 100.000 cycles per bit. Anyways, that's a datasheet from the original manufacturer, Xircom.
Who knows were some ESP manufacturers buy their flash chips. i think the first priority says "cheap".... ;)


Regards
Shardan
Regards
Shardan

costo
Normal user
Posts: 490
Joined: 21 Nov 2015, 15:03

Re: Bulk test of nodeMCU's

#11 Post by costo » 07 May 2017, 19:04

Shardan wrote:
07 May 2017, 17:43

Who knows were some ESP manufacturers buy their flash chips. i think the first priority says "cheap".... ;)
Yeah right, some of them must be so evil that they use components that were rejects in the manufacturing proces. But you never know in advance which seller sells bad stuff. So I usually order my stuff in single quantities from different sellers.
Some sellers on Ebay/Aliexpress just sell rejected garbage very cheap and hope the buyer will not complain (in time). Or they offer a small amount of money back for the inconvenience.

I said before: ¨None of my Wemos D1 modules give me troubles so maybe the quality of the ESP module is better.¨
but I forgot to mention that my first Wemos had the CH340 chip mounted 180 degrees wrong, pin 1 was soldered where pin 8 was supposed to be. So definitly no control from the manufacturer. Wemos can be bad too.

Post Reply

Who is online

Users browsing this forum: Google [Bot] and 18 guests