ESP8266 Community Forum

- Tue Apr 21, 2015 1:49 pm #15223 See viewtopic.php?p=10530#p10530 and further posts.

Summary: ESP-01 goes up to ~325 mA during peaks.

- Thu Apr 23, 2015 10:43 am #15389

I know, probably you will tell me that Ohm Law is the same but I consider HP34401A a more appropriate device than a Oscilloscope to properly measure current requirements and validate some data.

Instead of 100uF, I have 200uF near the ESP chip. And I see the high current for about 1ms. During that time, the voltage will sag depending on the "short term current limit" of the LDO. Because my application is sensitive to "static current", I cannot afford an LM1117 like LDO that draws about 5mA without at the output. Your LM1117 will supply the 1A burst from the input capacitors. The scope shows that the burst/current peak lasts for about 1ms. This is something completely different from ns length peak. In fact the factor-million in the time means that we're talking about a factor of a million in the amount of energy involved. So a 10A 1ns peak requires 100 thousand times less capacitor than a 1A 1ms peak. So where IMHO the 200uF barely works for the 1A 1ms peak from the ESP, 100nF will work perfectly for your 10A 1ns peak.

- Thu Apr 23, 2015 11:06 am #15392

See viewtopic.php?p=10530#p10530 and further posts.

Summary: ESP-01 goes up to ~325 mA during peaks.

Thanks for the link. But that measurement was done with the scope not set correctly. The peak could very well be much larger than 325 mA measured there.

An LDO is very capable of reacting to input changes at the 10 us scale. The plot in the link shows a timescale on the order of seconds (10s for the whole plot) so it is very likely that the individual points are averaged over much longer than the microseconds we're interested in.

I have a battery as a powersource. So it should not be a problem to draw 1A for short bursts. Similarly, I would prefer to have say 4.7 uF of capacitor being enough. This capacitor should "last" the first 10 microseconds before the LDO has a chance to adjust. But according to my measurements the MCP1802 goes into current-limit mode, leaving the capacitors to provide the remaining current. The sag on the VCC line down to about 2V is tolerated by the ESP. Neat, but not proper circuit design.

I HAVE seen my ESP8266 crash, so unless I'm able to keep VCC above 2.8V I'm considering these current-spikes and the voltage-sags as a probable cause.

In order to design a proper circuit I would need to know if the 1A is "about right". So far everybody keeps saying "peaks of about 300mA" with IMHO improper measurements.

- Fri Apr 24, 2015 1:43 am #15456

See viewtopic.php?p=10530#p10530 and further posts.

Summary: ESP-01 goes up to ~325 mA during peaks.

Thanks for the link. But that measurement was done with the scope not set correctly. The peak could very well be much larger than 325 mA measured there.

An LDO is very capable of reacting to input changes at the 10 us scale. The plot in the link shows a timescale on the order of seconds (10s for the whole plot) so it is very likely that the individual points are averaged over much longer than the microseconds we're interested in.

I have a battery as a powersource. So it should not be a problem to draw 1A for short bursts. Similarly, I would prefer to have say 4.7 uF of capacitor being enough. This capacitor should "last" the first 10 microseconds before the LDO has a chance to adjust. But according to my measurements the MCP1802 goes into current-limit mode, leaving the capacitors to provide the remaining current. The sag on the VCC line down to about 2V is tolerated by the ESP. Neat, but not proper circuit design.

I HAVE seen my ESP8266 crash, so unless I'm able to keep VCC above 2.8V I'm considering these current-spikes and the voltage-sags as a probable cause.

In order to design a proper circuit I would need to know if the 1A is "about right". So far everybody keeps saying "peaks of about 300mA" with IMHO improper measurements.

I don't understand exactly what are you considering improper measurements but meanwhile are you sure that you don't have some nasty LDO oscillations somehow? Check input and output caps, ESR, traces.
From what you describe might be a reason...
Do you have the proper load on output? Try also a test with MCP1702, see what's happening, if you think you have your reasons to avoid 1117. You might be amased by the results.

ESP8266 MPDMv4 AC MAINS Dimmer Home: http://esp8266-projects.com/2016/04/mpdmv4-mains-dimmerswitch-html/
Available on Tindie: https://www.tindie.com/products/nEXT_EVO1/universal-ac-mains-dimmer-mpdmv41/