Hacking a Grundfos Comfort AutoAdapt recirculating hot water pump for ESPHome

[ncarney (Neal)]

Another thanks to the OP for giving me the courage to hack my Grundfos. TL;DR: I removed the front control board to make it “dumb”, which allowed me to control it with Home Assistant via a relay.

Details: I have a Grundfos ALPHA 15-55 SUC HWR-D Comfort. It is designed for on-demand recirculation which pairs nicely with my tankless heater. It uses remote Bluetooth buttons to initiate cycling the pump on for 5 minutes. Unfortunately, Grundfos did a HORRIBLE job with the Bluetooth implementation. First failure is that the remote devices regularly need to be re-paired (when they actually work). Second failure is that I had to purchase their expensive Bluetooth extender in order to get them to work, even with remotes that were not far away from the pump. Third and fatal failure is the Bluetooth remotes all seem to experience some kind of hardware failure. I have three remotes and they are all dead. There are many reports of others on the web with the same issue. I took one apart and hooked up to power supply set at 3.3V. I can see that the remote powers up and an LED flashes, but one or more components on the PCB are dead. It’s all SMD and I didn’t have the patience to try to trace everything. It appears Grundfos must know about the issue, because from what I can tell they no longer make a similar model and you can no longer purchase the Bluetooth remote as a spare part (one vendor is selling for $500 each!).

Unfortunately, before reading this thread, it seemed it was not possible to put the pump into “always on” mode. If Grundfos is reading, you should put an override switch on all your “intelligent” devices to fallback into always on mode for external control.

Encouraged by this thread, I removed the front panel to expose the top two boards. I removed the topmost control board (which has the LEDs and the button switch), and then powered up the pump. Sure enough, the pump ran non-stop. Success! Thanks to the OP I did not even have to test any of the power connections to determine the logic. I quickly plugged into a power-monitoring relay I was already using to monitor whether the pump was actually on, modified my Home Assistant automation, and now have full control.

I’m actually glad the Bluetooth remotes died. I always desired to control via Home Assistant rather than dedicated remote. Thanks to this thread, it’s now possible.

Front cover of pump installed:

Cover removed with second board exposed:

Top control board that was removed from cover:

Backside of cover after removal, with board removed. You can see there are three tabs in 12, 4, and 8 o’clock positions. Just used a small screwdriver to pry up the cover.

[pdw]

Interesting! Internally, it’s quite different to mine, and the connector to the front board has 12 pins rather than 6, so there’s obviously a bit more going on, but it sounds like the motor control signal works the same. It’s also interesting to me that their design uses a separate buck regulator for the control board rather than piggy backing off the linear regulator on the 2nd PCB – I had to do the same thing due to the higher power of wifi/bluetooth capable microcontroller.

Are you able to see what microcontroller they’re using on the control board?

[ncarney (Neal)]

Unfortunately, my pump is in a tight space and I had to remove it from the water system in order to work on it. This meant I had a short window of time before my spouse would be upset with me for no hot water. So I went about my business very quickly and did not do any investigating.

Indeed, I am puzzled by my 12 pin connector vs your 6 pin. My pump doesn’t seem to do anything more remarkable than yours. There are two variants of my pump: my variant uses the Bluetooth for remote control and there is another that uses the Bluetooth to get temperature data from a remote temperature sensor. I understand your sensor is located within the pump housing.

I did not look to see what microcontroller they’re using…the picture I took is the closest I will get until I take it apart next time. At some point I may indeed fabricate a board like yours using ESP32 simply so that it is a self-contained solution. I noticed the buck regulator on the control board and certainly will use the same approach if I undertake a board.

[ncarney (Neal)]

Shortly after writing my previous post I realized I must be losing my marbles. Of course I removed the control board and can easily inspect it. The markings on the microcontroller are a little inconclusive so I enlisted the help of ChatGPT. It thinks it is a custom TI chip similar to the CC2640. The standard CC2640 is a QFN package whereas the chip shown is a QFP. Below is a close-up of the chip. I also included a close-up of the bluetooth antenna trace at the top of the board above the buck converter.

[ncarney (Neal)]

…also, the 12-pin connector appears to connect to only 6 traces, so not sure why they changed from 6-pin. Maybe a different variant requires more?

[pdw]

@ncarney

Of course I removed the control board and can easily inspect it.

LOL

@ncarney

It thinks it is a custom TI chip similar to the CC2640. The standard CC2640 is a QFN package whereas the chip shown is a QFP

I think it’s half right That chip looks like a 48 pin QFN / VQFN, which is one of the standard packages listed on that link.

I’m guessing the 16 pin QFN in the antenna path is an amplifier. I’m wondering what the 8 pin IC is at the bottom. I’m guessing 3 of the 4 6-pin ones scattered around the board are transistors driving the LEDs. Not sure about the one nearest the connector.

[Brad_Ford (Wombo1)]

And my board has a 14-pin connector… Grundfos Alpha2 25-50N

[ncarney (Neal)]

Looks like there is a wide variety with Grundfos. At first glance your top board appears to be a single layer I/O board rather than a control board. The side shown just contains the LEDs and button switches, and I don’t see any vias so I’m guessing the back side is empty? If so I’m guessing the pump still runs in default mode (auto adapt) even with that board removed.

I’m a novice so hopefully @pdw will chime in. It could be that the second board is your control board and the big IC at the top of that board drives all the LEDs.

[pdw]

Yeah, looks like a single layer LED/switch board. The 0 ohm resistors are jumpers to allow traces to cross in the absence of a second layer. It’s multiplexed as there are a lot more LEDs and switches than there are pins on the connector.

It looks like there’s a single microcontroller on the second board (square QFN chip to the right of the connector). I’m not sure about the big chip at the top, but I’d guess it’s a motor controller of some kind. It looks like the PSU is on this same board. There’s a bridge rectifier on the right, and it looks like the transformer is on the back at the bottom.

This one would be very hard to mod

[quadflight]

So all you had to do was remove the top PCB and it’s always on, controlled by an outlet?

[ncarney (Neal)]

Yes, that’s correct. Very surprised at how it easy it was and very happy with how well it works now. The Bluetooth was very unreliable.

[reddtekk (Jim)]

Many thanks to @pdw for work on this and for everyone else’s contributions as well!

Going back to the Alpha2 25-50N that @Brad_Ford has (I have a few of them as well), would it not be feasible to basically piggyback that board with an ESP32 and hijack it? One could use the status LEDs as feedback and simulate button presses. Couldn’t start and stop it using this method, but could change the mode of operation and the speed/pressure.

A bit of a low priority for me at the moment, but it would be nice to be able to add some smarts to the older models. Perhaps I’ll take this on once I have some higher priority things taken care of.

[pdw]

It’d be a fun challenge It’s multiplexed so spotting which LEDs are on isn’t straightforward.

[reddtekk (Jim)]

That adds a challenge I hadn’t considered…but again a longer term low priority thing for me.