Upcoming FLM v3 - features, changes, points-of-interest

As presented in the status update and live at the grid:camp there is a new version of the Fluksometer visible on the horizon. (New Horizon?)
Even though I will just become an early adopter, there are some interesting differences to the current version 2 FLM notable right now:
Besides a more powerful processor there are "severe" changes in how the FLM detects current and voltage, enabling it to detect "all four quadrants of AC power flow" - this written, there are new current clamps and a special power supply provided with the FLMv3 essential to its operation.
With these hardware changes, there will also be a lot of software changes which I expect only very limited to be provided for the FLMv2 (even though handling seems quite similar, for example by interfacing sensors via the built-in MQTT broker).
An interesting feature thus taken from the source code are new MQTT topics reflecting the new features; beside the "well-known" /sensor/+/counter and /sensor/+/gauge there are distinct "flx" topics:

  • /device/+/flx/sar/+
  • /device/+/flx/sdadc/+
  • /device/+/flx/voltage/+
  • /device/+/flx/time

I would be eager to get to know some experiences of early adopters/beta testers to share here. This might help understanding to upgrade... (I will also work on a successor of local visualization on receiving my unit)

AttachmentSize
FLM3E_boxed.jpg671.35 KB
vk2him's picture

Looking forward to this development!

gebhardm's picture

Many thanks to Bart I yesterday received my FLM03E for in depth testing. See picture attached for the box content. Of course I instantly installed it in parallel to my FLM02; so you may take a glance at my dash for its readings and differences to FLM02 readings.
What "springs to the eye" (German idiom) immediately:

  • New, smaller current clamps with attached wires: Easier installation and less space occupied.
  • New power supply with 9V AC-out for instant voltage and phase shift measurement by the FLM: This means, of course, that you cannot combine devices on a common 12V rail as I do with my FLM02, RaspPi, network switch, and PV datalogger.
  • One more screw terminal, now seven instead of six.
  • No WiFi antenna plug and external antenna due to built in antenna.
  • Two network ports: WAN and LAN.
  • One more LED for second network port; LEDs located to the device right side instead of left on the FLM02.

Installation
If you are somehow familiar to this kind of tiny routers then the two network ports provide following behaviour: WAN is for the internet, and LAN is for the intranet connection. There is a DHCP server active on the LAN side, thus plugging a PC in here provides it an IP address; this did not change to the FLM02, so you get one in the IPv4 range 192.168.255.x.
Now you can access the FLM03E at its "gateway" address http://192.168.255.1 - on a browser the "usual" login page occurs: Log in by user 'root', password 'root' and you are in somehow the same outline as with the FLM02. On the now visible status page you get, as named, 'status' information.
Selecting the sensor tab, you may configure your sensors, seven in a row, firsth thress for current clamps, three for pulse input and one uart.
Due to the new power supply providing the capability to measure line voltage there is no voltage configuration for the current clamps, just name their maximum current - 50 in my sample.
Now the tricky part: Syncing with the Flukso site. I did it in the way that I connected the WAN port to my existing intranet, gaining internet access through the modem/main router - the WAN port gets an IP address from this, detectable by "nmap"; PC attached for configuration on the LAN port, as described before. "nmap" also shows open ports on the WAN side are 1883 (MQTT) and 8083 (some TCP-UDP service) - by default no access to the FLM03E's http port 80 from here...
So I initially got my FLM03E up and running; next steps are actual port configuration with respect to an "intranet device" and some more investigation on new capabilities.
New capabilities
As written in the thread topic, there are new MQTT topics; the FLM03 is very communicative on these leaving my connected RasPi with flmdisplay a bit breathless...
More to come.

jgysenbergs's picture

It seems the FLM3 is almost ready for shipment. Would be nice having a date or time scale and an indication of price.

bazzle's picture

Thanks for the headsup.
Does it show the "mains" voltage on the display/graph?

gebhardm's picture

For first assessment of the MQTT capabilities of the FLM03E, please refer to https://github.com/gebhardm/flm03/blob/master/MQTT.md - note that FLM02 and FLM03 are currently not directly compatible with this respect; it will require some more investigations before I am able to decide on how to adapt, for example my flmdisplay functionality.

gebhardm's picture

Next finding: To get more convenient access (my FLM03E is connected via cable on its WAN port to the LAN), the firewall has to be tweaked like for the FLM02. Now ip-port 80 with its web interface works without plugging a PC into the LAN port as well as logging in via SSH (ip-port 22); maybe reasonable to change the default password...
Now having a look at the built in visualization is just one SCP away ;-)

gebhardm's picture

For using one pulse meter on two FLMs in parallel you cannot just use two wires. To solve this issue, I made a "pulse doubler" feeding two optocoupler from one "open collector output". The idea should be clear from the schematics...
Now I can see my PV on both FLMs. A real tinkering weekend ;-)

gebhardm's picture

Interesting is the difference between clamp sensor settings "3p+n" and "1p".
For the FLM02 "1 phase" meant "each phase separate"; this is "not true" anymore for the FLM03. The FLM03 seems to always report on all clamps, so no "collection into one", when choosing "3p".
Here "1p" seems to be "measuring three different points on the same phase". In the configuration here you see a "shift" parameter that I assume to be the "phase shift", visible if running on a "real" three-phase environment. Corresponding to the detected phase shift then the readings look "strange" when compared to the FLM02 readings.
"3p+n" (we have a neutral return line) then shows "what is expected with respect to FLM02 comparison", even though readings differ in amount. What is more accurate would have to be measured with a Fluke.
Pulse detection works as expected, comparision shows practically no difference.
So I have to dig a bit deeper into the magic of "measuring AC" to really understand what is going on. To be continued...
Hello Roel and 3E, any experiences to share here?

ksofluk's picture

Thank you for sharing your experience with the new flukso - very interesting!

How accurate is this new way of measuring power usage (by measuring line voltage)? Is a pulse meter still required?

petur's picture

GEBHARDM what's the total power consumed by the new Flukso?

I am guessing the power adapter is a transformer so I'm a bit worried about the whole setup consuming more power.

OTOH I would very much like to switch to FLM03 because all the low power investments I have been doing last months do not show up with the current FLM02 because of the cos-fi of the PSU. (For example FLM02 thinks my new subwoofer consumes 10W in standby where it is 1W actually.)

Thanks for the info on FLM03 so far

gebhardm's picture

@ksofluk - Accuracy is, as I wrote still subject to investigation; at the moment it seems as if my configuration is not correct with respect to phase detection. If a pulse meter is still required: There are three analog inputs and three pulse inputs; if you measure a three-phase supply circuit by clamps, yes, then a pulse meter for PV is still required.
@petur: I measure an own consumption of the FLM03 with a direct power meter at 3W. With respect to the cos(phi), as I wrote, still subject to investigation...

petur's picture

Thanks, I can live with 3 watts ;)

gebhardm's picture

And yet another success - with not too much effort also the local visualizations run again; the corresponding patch is available in the corresponding flm03 branch at https://github.com/gebhardm/flmlocal/tree/flm03 - waiting for a round of refactoring ;-)

vk2him's picture

Great work Marcus !! Now to wait until v03 can be purchased in the store!

icarus75's picture

Port configuration will trigger the phase matching algorithm. So it's crucial to install all current clamps prior to configuring the ports via the local UI. A non-trivial resistive load should be present on each phase as well during the configuration step for successful matching. Selection of the electrical system (1P, 3P-N = 3x230V, 3P+N = 3x380V) will influence matching. 1P indicates you only have a single phase available in the dwelling, so phase shift can only be 0 or 180 degrees.

gebhardm's picture

Thanks Bart, as suspected ("3P+N" should match my configuration); would be a great feature if the determined phase shift then is presented also in the UI (by MQTT it is available already). The load issue is still subject to investigation - see around 06:00 - this is a coffee brewer and a toaster, so resistive per se.

icarus75's picture

@petur The FLM03 draws 150mA when fed 12V DC from my bench power supply. With an AC-AC adapter efficiency of 80%, total power consumption should hoover around 2.2W.

gebhardm's picture

Another tinkering weekend: To "demystify" the FLM03's way to measure currents I made a little app and graph display that may also lead you to the limits of MQTT processing: The FLM03 publishes its millisecond readings on topic /device/+/flx/current/+ - alone these readings make a Raspberry Pi "suffering", being incapable to display the readings in realtime. And IoT is actually about zillions of such readings...

jgysenbergs's picture

Impressive!

icarus75's picture

Any reason you can't directly subscribe to this topic via mqtt/ws from the browser and render the readings without intermediary Raspi?

gebhardm's picture

No reason; just not tried out (may be another option in the visualizations, indeed) - I use the RasPi (or my development Mac) here as an intermediate toolbox that I may break without harming the FLM (and of course get diverse and different brokers addressed with a simple Ctrl-o in nano).

gebhardm's picture

And nota bene: mqtt/ws works only when running against a dedicated, ws-enabled mosquitto instance (and I didn't ws-enable my mosquitto so far - all this extra work I'd like to avoid as being not part of the issue to address - which is still the proper calibration w/o firing up an oven).

gebhardm's picture

I'll try it out with mqttws31.js directly; maybe I just misunderstood some of its features...

gebhardm's picture

OK, Bart, you are right, it works also with MQTT over websockets (if you address the right port, which is 8083) - https://github.com/gebhardm/flm03/tree/master/MQTTrealtime as an alternative using the FLM03 native mqtt broker directly utilizing mqttws31.js.

gebhardm's picture

Rationale on using 'mqtt underneath ws' or 'mqtt over ws': For WAN access one argument to do server side MQTT handling is to avoid opening two ports for client access: 1080 and 8083 in my specific case; if I push MQTT messages through the http websocket then I have to just open (10)80 and my MQTT broker remains a closed shop...

petur's picture

@gebhardm maybe silly question but what about using a reverse proxy to reduce the number of open ports? Granted, that already requires websockets too...

gebhardm's picture

@petur:...simplicity of operations... (there are people that even regard a simple copy as "over their head")

Ryton's picture

Fantastic news!

Play the drums, blow the horn, pop the champagne!
The new, long-awaited FLM has arrived!