NODO OTGW board COM port continuous connect/disconnect

[Kermit]

Recently replaced my Vokera BeSmart OT wireless box + stat with a WundaSmart Hubswitch. They seem to play nice together, but when I attached the NODO OTGW board (V2.11, now with V6.6 FW on)), things went strange. The gateway (with WiFI D1 mini at the time) had been used frequently on the BeSmart OT link with no probs, so pretty happy it was basically functional about a year ago. However, I think it might be damaged, as I have had some problems getting working with the new HubSwitch.

With its WiFi configuration, it looks like connecting it to the OT bus was causing the supplies to brown out.
Removing the D1 WiFi module and attempting to connect a PC over serial caused the COM port to connect/disconnect continuously

Loading the diagnostic.hex file and connecting therm to boiler and running test 4 (delay checks) caused the COM port to disconnect.

Finally pulled the PIC out and went through the trouble shooting. All OK apart from the measuring pin 18 while floating pin 3 or connecting it to ground with the OT terminals connected together. Pin 18 was a few mV with pin 3 floating.
Connecting pin 3 to ground didn't make it go to 3V3 (maybe 100mV). It also caused the supplies to brown out and the USB fuse to drop lots of volts and get hot. Something it consuming a lot of current!
Without the OT loopback, connecting pin 3 to ground causes no issues.

However, without a schematic of the NODO board, it is a bit tricky to figure out what could be broken. Any suggestions for further checks?
Is the schematic available anywhere?

[hvxl]

Your test shows there's a problem in the thermostat or boiler interface of the OTGW. These interfaces of the Nodo shop OTGW are still the same as the original design. So you can refer to the schematic on the otgw web site. Except for the diodes, even the component numbers match. On the Nodo board D1 through D4 form the rectifier bridge, D5 is 4v3, D6 is 4v7, and D7 is the 15v zener. A couple of resistors have a different value because of the change from 5V to 3.3V: R8 is 820 and R9 is 3k3.

Just to confirm that I understand correctly: You measured the current on the Therm connector with pin 3 open and again with pin 3 connected to ground. In both cases you found the expected current values? This would mean that the current limiting functionality is working correctly. Then I don't understand how looping the Therm interface to the Boiler interface can result in high current being drawn. It should always be limited to around 20mA, even if you completely short out the Therm interface. The Boiler interface is electrically isolated from everything else. So short-circuiting its input is the worst thing it can do.

Another mystery is how the OTGW was damaged in the first place. In the WundaSmart Hubswitch manual I see that it is powered from the mains. I would expect the opentherm interface of the Hubswitch to be isolated from the mains. But that would be a possible way to cause large currents to occur when the OTGW is grounded via the Hubswitch on its Therm interface and via a computer connected to the USB interface. Can you check that there is no electrical connection between the opentherm interface of the Hubswitch and its mains terminals? It would be a very bad design if there is.

[Kermit]

Thanks for the quick response, and in particular for the info about the component numbers being the same- I checked the diodes and gave up when I found they were different. If the transistors have the same numbers, that makes it a lot easier to trace and debug.
Sounds like the best bet is to remove the PIC and repeat the loopback test (briefly) to figure out where the current is going.

And I confirm that without the loopback between the boiler and therm connection, connecting pin 3 to ground and a DMM in current mode across the therm connections resulted in the a current of IIRC 19mA.
Repeating the Pin3 to ground with the loopback in place gave me the high current and brownout case. As you say, it is strange that the loopback (which at worst is a dead short), could draw something other than 19mA

As to how it happened in the first case- it is possible the board was mishandled, though I cant see any physical damage.
I check the voltage of the OT terminals of the WundaSmart against the ground (DC and AC) of the OTGW when they are disconnected and see a few mV. There is high resistance between them. I too wondered if there might be some mains level offset between then, but it looks OK.
The voltage across the WundaSmart OT terminals is again a couple of mV.
The boiler has about 24V OC on its terminals and no significant voltage between the OTGW and either terminals.

I was surprised that when the WondaSmart and boiler are connected (no OTGW) I see about 6V across the OT terminals. Is that expected?
They do appear to be working OK!

[hvxl]

I was surprised that when the WondaSmart and boiler are connected (no OTGW) I see about 6V across the OT terminals. Is that expected?

Yes. When the opentherm line is idle (between messages, no smart power), it is at low voltage (< 8V) and low current (5..9 mA).

[Kermit]

I should have checked the OT spec!
Anyway, I'll have time later in the week to do a bit more debugging. Im guessing one of the 3 rails is getting pulled down, so finding which one will give me a clue as to what is going wrong. I can trace the expected voltages around the interface, less so around the power supplies.
Looks like the 24V boost is built around U1 (SDB628 family) and the 3V3 linear reg is U4, so I can have a poke around there to see what is dying.
Is there any testpoints I should check?

[Kermit]

Looks like the culprit is the polyfuse. When I loop the therm and boiler contact together and connect pin 3 to ground, the USB current is 130mA, and the fuse gets annoyed and drops about 2 volts, so the supplies brown out.
Shorting it out and repeating shows normal operation (pin 18 reaches about 2.9V). The OT line current increases from 6mA to about 20mA. OT voltage stays about 17V.
The 3.3V LDO output seems to be stable at this time, but the boost converter drops about 20mV, so Im guessing its a load on the 24V rail.

Is 130mA a bit big, or expected? Its 650mW, but thats 20mA from the 24V converter (well within spec) and at least 17V x 20mA = 340mW is just the OT bus power.
The diagnostics build test 4 (loopback) ran OK with the polyfuse shorted, but browned out the supplies before.

Anyway, I'll try to track down some replacement fuses and give it a try. Looks like it is a Bourn MF-PSMF020X (Vmax 9V, Itrip 500mA, 0805)- is it possible to confirm?

[tjfs]

I'm sorry to hear this. Can you read the marking on top of the polyfuse and do you remember when you purchased the gateway?

Tim

[Kermit]

The polyfuse has just the number 4 on top, which matches the markings on the Bourn PTC MF-PSMF020X-2, rated for 9V operating at 200mA, tripping at 500mA.
Bought the board assembled from ebay from some mysterious person in Milton, Cambridge in January this year (thought it was older than that!)

Could I check if 130mA from the USB when its connected to the boiler and thermostat is the expected current?

And happy to report the board is back in action with the slightly dodgy expedient of bridging the PTC. The USB power supply does have current limiting, so that should suffice for the moment.

[tjfs]

We did have an issue where a small number of boards were fitted with the incorrect fuse but I didn't realise that any 2 11 boards were involved.

It's fine to run the board without the fuse in circuit.

The Bourns datasheet shows the marking for the 500mA fuse is 4, and the 200mA fuse is 2. I'm unsure why your fuse is failing at lower current.

I'm glad you fixed the problem with your gateway.

Tim