ESP8266 Community Forum

Rumor has it that the ESP8266 will consume up to 300mA.

I'm measuring about 1A in bursts of about 1ms in my application right now. A little guesswork is required to make the measurements, so the results might be off by say up to 50%, but the order of magnitude is significantly different from the 300mA that is normally quoted.

Although I can't see exactly what happens, I think this happens when a packet is transmitted.

The image currently on my scope has a 800us linear droop of the 3V line, followed by a "recovery" in almost 200 microseconds. A probe on the 0.4ohm current-shunt shows that about 800mA of current is being consumed during both the droop and the recovery period. (and that current stops immediately as the recovery completes).

I think my MCP1802 is in current-limit mode during the whole droop-recovery period. At about 800mA, while the datasheet mentions something like 380mA typical. Weird.

From the fact that the recovery is about 4 times faster than the droop, 800mA is about 4 times more than the "deficit" that occurs during the droop. So the total current during the droop is about 0.8 + 0.2 = 1.0A.

My question: Can anybody confirm that the ESP will draw about 1A from the 3.3V line in bursts when transmitting?

rew wrote:Rumor has it that the ESP8266 will consume up to 300mA.

I'm measuring about 1A in bursts of about 1ms in my application right now. A little guesswork is required to make the measurements, so the results might be off by say up to 50%, but the order of magnitude is significantly different from the 300mA that is normally quoted.

Although I can't see exactly what happens, I think this happens when a packet is transmitted.

The image currently on my scope has a 800us linear droop of the 3V line, followed by a "recovery" in almost 200 microseconds. A probe on the 0.4ohm current-shunt shows that about 800mA of current is being consumed during both the droop and the recovery period. (and that current stops immediately as the recovery completes).

I think my MCP1802 is in current-limit mode during the whole droop-recovery period. At about 800mA, while the datasheet mentions something like 380mA typical. Weird.

From the fact that the recovery is about 4 times faster than the droop, 800mA is about 4 times more than the "deficit" that occurs during the droop. So the total current during the droop is about 0.8 + 0.2 = 1.0A.

My question: Can anybody confirm that the ESP will draw about 1A from the 3.3V line in bursts when transmitting?

I have no idea how are you exactly doing the mesurements out there but it looks out of scale somehow. Maybe I'm wrong but migh worth to explain in detail.

I have constantly measured avg value of 70mA with bust up to 120-140mA. This is for ESP07 modules only in client mode. For some reasons ESP01 goes around 180mA. But 1 Amp...it's from another league.

As soon as I have some free time will finish and publish some extensive data about power consumtion. For now been busy with tests with a new toy, MPSM v2 Board

Here is my scope screenshot.

The yellow trace is the current consumed. Down is more current. the gridline above the horizontal part is about "zero" current, so there is a little "CPU-is-running" consumption. The series shunt resistance that I used is 0.4 ohms, so the 200mV/div translates to 500mA/div.

The blue trace is the 3.0V voltage in my system.

To analyse this, first you need to inspect the "recovery" phase of the 3.0V powersupply line. As you can see, the line recovers at 400mV/100us. During the recovery, the current is about 900mA, and after the recovery the current drops by about 800mA. So the rest of the system is using about 100mA (or my initial calibration of the gridline has drifted a bit). Anyway, it is clear that this 800mA is being used to recharge the capacitors on the 3V line.

With I = C dV/dt we can transform this to: C = I dt/dV = .8 . 100us / 400mV = 0.2mF = 200uF. This is about right: I Have two 100uF capacitors on the 3V line.

I double checked the shunt resistor: 200mV at 0.5A -> 0.4 Ohms.

Of course "average" values are much less. The thing is that I'm worried that the negative spikes on my 3V line might crash the CPU. Two or three times so far the system has crashed. (The system is meant to be low-power, so the CPU goes into deep-sleep most of the time. So when it wakes up, it goes through the boot sequence). So when I come back, I see it has booted, but not started my user_main function.... Actually, that's weird, as the transmit spikes will not be happening at that point in time.

Anyway, These dips on the 3V line are "not good". Until they are fixed, I'm not betting on the system staying up. To design the power supply system for this, I need to know the specifications: How much is the maximum current that the ESP will draw. These measurements seem to indicate: "more than 800mA". Once that is confirmed I can design something that can deliver "at least 800mA"....

rew wrote:Here is my scope screenshot.

dsc05768_small.jpg

The yellow trace is the current consumed. Down is more current. the gridline above the horizontal part is about "zero" current, so there is a little "CPU-is-running" consumption. The series shunt resistance that I used is 0.4 ohms, so the 200mV/div translates to 500mA/div.

The blue trace is the 3.0V voltage in my system.

To analyse this, first you need to inspect the "recovery" phase of the 3.0V powersupply line. As you can see, the line recovers at 400mV/100us. During the recovery, the current is about 900mA, and after the recovery the current drops by about 800mA. So the rest of the system is using about 100mA (or my initial calibration of the gridline has drifted a bit). Anyway, it is clear that this 800mA is being used to recharge the capacitors on the 3V line.

With I = C dV/dt we can transform this to: C = I dt/dV = .8 . 100us / 400mV = 0.2mF = 200uF. This is about right: I Have two 100uF capacitors on the 3V line.

I double checked the shunt resistor: 200mV at 0.5A -> 0.4 Ohms.

Of course "average" values are much less. The thing is that I'm worried that the negative spikes on my 3V line might crash the CPU. Two or three times so far the system has crashed. (The system is meant to be low-power, so the CPU goes into deep-sleep most of the time. So when it wakes up, it goes through the boot sequence). So when I come back, I see it has booted, but not started my user_main function.... Actually, that's weird, as the transmit spikes will not be happening at that point in time.

Anyway, These dips on the 3V line are "not good". Until they are fixed, I'm not betting on the system staying up. To design the power supply system for this, I need to know the specifications: How much is the maximum current that the ESP will draw. These measurements seem to indicate: "more than 800mA". Once that is confirmed I can design something that can deliver "at least 800mA"....

With all respect, for a decent current measurement I think is better to use a proper Amp Meter or multimeter. The values I was talking about were obtained using a HP 34401A and a Brymen 869s, both tested and calibrated.

Take a look at my last MPSM Board project power supply, it has a dead simple power stage LDO and powered up by just a Li-ion Battery Cell . Never ever seen it over 120mA in more than a week of running.