Author Topic: ESP8266 unstable when using ceramic cap, but stable with electrolytic  (Read 3952 times)

0 Members and 1 Guest are viewing this topic.

Offline synopticaTopic starter

  • Contributor
  • Posts: 41
  • Country: au
Hi brains trust!

I'm working on an ultra-low power ESP8266-based sensor network for my home.   I'm trying to keep them as simple as possible, and run them off either a single 18650 cell, or 3x AAA/AA cells (got a ton lying around!)

I'm using an HT7333 regulator for their low dropout and quiescent current (datasheet: http://www.angeladvance.com/HT73xx.pdf)

With no caps, this regulator cannot provide stable power to the ESP - it regularly crashes.  If I throw a 22uF electrolytic across GND/VOUT, it's rock solid, but will drain the batteries over the course of a couple of weeks.  I grabbed a reel of 22uF multilayer ceramic SMD caps, but I'm getting similar instability to when no cap is used.  I specifically chose the ceramics for their low-leakage properties.

What am I failing to consider here?  I realise that the ceramics are non-polarised - would such an application require polarised caps?

Thanks in advance!
 

Offline wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv
Re: ESP8266 unstable when using ceramic cap, but stable with electrolytic
« Reply #1 on: January 02, 2019, 11:47:44 pm »
Many LDOs are unstable with ceramic capacitors due to their low ESR. Many datasheets don't mention this. If you see electrolytic/tantalum output cap in example in datasheet, you should assume LDO is not suited to work with MLCC. Those which are stable with MLCC usually say it explicitly in datasheet or at least provide suitable ESR range.
http://www.ti.com/lit/an/slva115/slva115.pdf
« Last Edit: January 02, 2019, 11:59:02 pm by wraper »
 
The following users thanked this post: synoptica, Mr. Scram

Offline wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv
Re: ESP8266 unstable when using ceramic cap, but stable with electrolytic
« Reply #2 on: January 02, 2019, 11:55:07 pm »
Quote
If I throw a 22uF electrolytic across GND/VOUT, it's rock solid, but will drain the batteries over the course of a couple of weeks.
It has nothing to do with capacitor. 4uA quiescent current of HT73xx is higher than leakage current of non faulty electrolytic capacitor. ESP8266 is dominating in current consumption unless it's in Power OFF mode.

 
The following users thanked this post: synoptica

Offline synopticaTopic starter

  • Contributor
  • Posts: 41
  • Country: au
Re: ESP8266 unstable when using ceramic cap, but stable with electrolytic
« Reply #3 on: January 03, 2019, 12:16:23 am »
Quote
If I throw a 22uF electrolytic across GND/VOUT, it's rock solid, but will drain the batteries over the course of a couple of weeks.
It has nothing to do with capacitor. 4uA quiescent current of HT73xx is higher than leakage current of non faulty electrolytic capacitor. ESP8266 is dominating in current consumption unless it's in Power OFF mode.



That's interesting...  It's definitely going into deep-sleep correctly, but you're right - thinking back now, when I was only getting about a week or so, I was still using an AMS1117.

That being said, I though I read that an electrolytic would drain somwhere in the low-milliamps range. Maybe it's time to shell for a uCurrent Gold ;)

Thanks for the advice on the MLCC caps though - it's good to know!

What would be your recommendation in this instance - a low-leakage electrolytic, or tantalum?
« Last Edit: January 03, 2019, 12:19:41 am by synoptica »
 

Offline wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv
Re: ESP8266 unstable when using ceramic cap, but stable with electrolytic
« Reply #4 on: January 03, 2019, 12:32:42 am »
Small electrolytic caps usually draw nanoamps after they stay some time under voltage. Just open electrolytic cap datasheet, it's usually specified 3-4 uA maximum at rated voltage (and will drop at lower voltage). AMS1117 have around 5mA quiescent current.
« Last Edit: January 03, 2019, 12:37:36 am by wraper »
 
The following users thanked this post: exe

Offline Mr. Scram

  • Super Contributor
  • ***
  • Posts: 9821
  • Country: 00
  • Display aficionado
Re: ESP8266 unstable when using ceramic cap, but stable with electrolytic
« Reply #5 on: January 03, 2019, 12:33:29 am »
Many LDOs are unstable with ceramic capacitors due to their low ESR. Many datasheets don't mention this. If you see electrolytic/tantalum output cap in example in datasheet, you should assume LDO is not suited to work with MLCC. Those which are stable with MLCC usually say it explicitly in datasheet or at least provide suitable ESR range.
http://www.ti.com/lit/an/slva115/slva115.pdf
How about supplying the ESR artificially? I understand that the loop area may become undesirably large.
 

Offline boB

  • Frequent Contributor
  • **
  • Posts: 341
  • Country: us
    • my work www
Re: ESP8266 unstable when using ceramic cap, but stable with electrolytic
« Reply #6 on: January 03, 2019, 12:44:40 am »

Quote
How about supplying the ESR artificially? I understand that the loop area may become undesirably large.

Yes, this is done a lot.  We need a SMT resistor just for this as 1.0 Ohms is more than necessary usually.
2 in parallel works.

But the LDOs I have used like larger electrolytics just fine and the data sheets for the regulators state this in their graphs.

K7IQ
 

Offline Mr. Scram

  • Super Contributor
  • ***
  • Posts: 9821
  • Country: 00
  • Display aficionado
Re: ESP8266 unstable when using ceramic cap, but stable with electrolytic
« Reply #7 on: January 03, 2019, 01:10:40 am »
Yes, this is done a lot.  We need a SMT resistor just for this as 1.0 Ohms is more than necessary usually.
2 in parallel works.

But the LDOs I have used like larger electrolytics just fine and the data sheets for the regulators state this in their graphs.
Thanks. I was thinking of a situation where electrolytics aren't desirable due to their other properties, like inferior longevity.
 

Offline synopticaTopic starter

  • Contributor
  • Posts: 41
  • Country: au
Re: ESP8266 unstable when using ceramic cap, but stable with electrolytic
« Reply #8 on: January 03, 2019, 06:31:41 am »
Thanks again to all who commented; I've picked up a couple of tantalums and low-leakage electrolytics - I'll otherwise configure the sensors identically and will see how we go!
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: ESP8266 unstable when using ceramic cap, but stable with electrolytic
« Reply #9 on: January 03, 2019, 02:06:22 pm »
Big fail LDO -- no design win for me!

No performance curves
No stability region
No transient response plot
No note on capacitors value or ESR
No Iq vs. Vin plot (honorable mention: headline text says device is "CMOS", so Iq is probably pretty stable and low)

If I were forced to use it, I would very carefully test these parameters, characterizing devices in my lab.  I would similarly demand incoming component inspection (sampling would probably be okay) or at least in-circuit testing, to verify that production devices match expectations.  These are onerous demands, that will noticeably increase the production cost of the final product.

Or if not in a production setting -- why cost yourself $50 of frustration buying a $0.10 cheaper regulator?  Treat yourself, you're worth it. ;D

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 
The following users thanked this post: exe

Offline blackbrayn

  • Contributor
  • Posts: 20
  • Country: ro
Re: ESP8266 unstable when using ceramic cap, but stable with electrolytic
« Reply #10 on: January 10, 2019, 01:39:04 am »
Had similar problems with a esp32 , check the oscilloscope captures from the post , the capacitor is helping the initial wake up in your case i suspect , it did not help in my case and i didn't have time to continue investigating for now.

https://github.com/espressif/arduino-esp32/issues/796

Sent from my ANE-LX1 using Tapatalk

 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf