Domoticz

I am working on getting the EPS8266 based P1-wifi gateway (see romix.macuser.nl to draw its power from the Smart meter, rather than from a separate micro usb power adapter.

The DSMR 4.2 specifications rate the 5v power supply (pins 1/6 of the P1 port) at 100mA. The ESP8266 most of the time draws between 70 and 80 mA. However, during startup there are noticeable spikes above 100mA (even my multi meter shows them) and occasionally (during transmit) it also spikes above 100 mA.

A 4400 uF capacitor can be used to take care of the transmission spikes, but won't keep the power up below 100mA.
Sending a P1 datagram to Domoticz takes about 6 mSec, not an awful lot of time. The 4400 uF capacitor should be able to provide a sufficient buffer.

Does anyone have an idea how to keep the ESP8266 (Wemos D1 mini board in my case) within specs?

Alternatively, does anyone have more information about the P1 power supply in the Smartmeter (Kaifa, for instance). Is it really limited at 100mA, or can it deliver more? Is it fused? If so, will the fuse be permanently blown, or will it reset?

Ronald

For specifications of the P1 port see: http://files.domoticaforum.eu/uploads/S ... l%20P1.pdf

See chapter 4.2 Power supply
The power supply for the P1 port shall be able to withstand short circuits. Manipulation of the
power supply lines shall never influence any other part of the meter,
When no device is connected through the P1 port, the power consumption of the P1 circuitry
shall not be included in the register values. When a device is connected to the P1 port, the
power consumption of the P1 circuitry shall be included in the register values. The P1 port
will function and supply power independent of the state of the E breaker.
The power supply will supply a maximum current of 100 mA at 5 Volt. Overload protection
shall be implemented as a current limiting mechanism.

Yes, I have read the documentation. This leaves many questions open though.
What kind of overload protection? Glass fuse? A self-recovering electronic fuse?
How should we understand 'able to withstand short circuits'?

Currently, I am investigating whether installing a Fairchild FPF2004 load switch and a large capacitor (4700uF) would do the trick.
https://www.fairchildsemi.com/datasheets/FP/FPF2000.pdf

Will keep you posted

I have experimented a bit with an FPF 2000 on my end, combined with a large cap (up to 4700µF 16V). When fed with 5v, the voltage drops to about 3.1V at the 5V terminal at my Wemos D1 mini, the 3v3 regulator outputs the same (slightly above 3V). This seems to be insufficient for the circuit to work properly. The blue ESP led does flash once during startup, but I can't connect to it through wifi, nor does the led flasher I included in main loop give any sign of life. In other words, I'm pretty sure the module does not work at 3v. Spikes on my (simple) scope are mostly gone with tha 4700µF capacitor. If removed, the image is quite lively and the blue led flashes every few seconds (signalling restarts).

What puzzles me is the large voltage drop over the FPF 2000. Unloaded the drop is almost .7 V already.

Several options.
1. Given the load switch in the P1 port, incorporating a load switch at the consumer end (our little ESP gizmo's), might be redundant and actually doing more harm than good. Giving it a try with just a large Cap to buffer the spikes might still do the trick. I am fairly hesitant to give it a try though.

2. incorporating a proper load switch/buffer might require more than just the FPFP2000 and some caps. There is an interesting Fairchild application note (AN-8019, entitled Reliable USB Modem design using the combination of an integrated load switch and a buck converter) that discusses a problem similar to ours: how to handle 2A peak currents in GSM/GPRS setups. The buck converter handles the voltage drops caused by the load switch's actions. The scheme is not useful as is for our problem because it outputs 3.6V and is based on different characteristics of both input and peak load.

One drawback of the fpf based solutions is that these chips are really tiny (5 pin space saving sc-70 package. Well space saving indeed, but a bit of a nuisance to solder).

An update. i have a working prototype with a small LIR2540 rechargeable battery as buffer for the power spikes. It is charged by an MCP73831 li-ion charger chip. Works like a charm.
Am designing new PCB, with LIR2032 instead of the bigger LIR2540. The MCP means moving to smt components. production will be a challenge. Also throwing in a SN74AHC1G04 inverter instead of the BC547b. Updates will be posted on http://esp8266thingies.nl soon.