Watts Intelliflow timer

Switching to a single 5W through-hole resistor seems promising. The 3.0.1 boards came back and still have just SMD footprints, but I tacked a 3.9kΩ 5W resistor across them and sort of hovering in the air over the board. The resistor gets pretty hot to the touch, but since it has air space around it, it appears to not be damaging anything else (or itself). In operation the unit switches back and forth between 4W and 0.75W, which seems to be able to successfully trick the Intelliflow into leaving the water on for 6 hours. The 3.1 boards with the TH footprint should arrive shortly and should be successful if my testing has any value.

Well, stress testing of the V3 design showed some flaws. The 9.1kΩ resistors just get too hot and damage the board. So that's no good. I'm going to try 10kΩ because I have them handy, but there's probably a balancing act that's going to be necessary between getting too hot and pulling enough current through the Intelliflow to trick it into staying on. The previous board had just under 3 watts on the DC side and it wasn't perfectly reliable. Going from 9.1 to 10 kΩ drops the power on the AC side from 4.75 watts to 4.33 watts. The next most logical step would be to try 11kΩ, which would be just under 4 watts.

The version 3 boards came back and there were a couple of errors that needed to be worked around, but in essence it works. I'm going to have a 3.0.1 board made that fixes the couple of mistakes (a couple of clearance issues and moving one of the outputs from the TPID pin to the TPIC pin, which is less sensitive to loading).

The v3 design uses two SSRs - one to hold the system power on for the duration of the sequence and another to switch the load on and off. The load is 3 9.1kΩ resistors in parallel, for a total of around 4.75 watts, plus another quarter watt or so for the LV side. The load is switched on and off at about a 50% duty cycle to try to fool the Intelliflow into thinking there's a busy washing machine there. But the goal of v3 has been reached - when the system is off the leakage of the power-hold SSR is practically zero.

The version 2 boards came back. This design has the SSR and button power system so that there is no standby load. One thing I discovered while the boards were being made is that the Tiny9 has a Vcc Level Monitor system (again, why they don't call it a brownout detector, I'll never understand). At first glance, it doesn't seem terribly helpful, as if you configure it for the "high" RESET voltage (VLM1H) is "typically" 1.6 volts, while the safe area curve for the chip bottoms out at 1.8 volts (the VLM1L threshold is even lower at 1.4 volts typical). Why they made this choice I can't understand. However, in practice it appears to work ok (at least for one example). When the timer runs out, the SSR is dropped and you can see the power LED fade out over the course of about a second as the power supply filter caps bleed out. If the VLM wasn't working, I'd expect the chip to reset as the voltage dropped, and of course the first thing the firmware does is turn the SSR back on.

So it appears that we don't actually need a hardware voltage watchdog chip separate from the VLM in the Tiny itself.

The upcoming v3 board makes another change, and will likely be the final version. In this one we've traded a second SSR for a cheaper open-frame DC power supply that's rated for only 1 watt. The second SSR switches a power load on the AC side. 3 10kΩ 2W resistors form a ~4.3 watt RMS load (at 120VAC). The DC load (both SSRs and the power LED) top out at about 600 mW (the controller load is negligible). The power supply is supposed to be about 60% efficient, so that's another watt of load, so the combined total should be plenty to fool the Intelliflow.

I've been studying design for off-line switching converter chips to roll my own supply with the goal of making it smaller. If you do go that way, you can also do away with the optoisolated components, since the design wouldn't be isolated at that point. But the issue with that is that in order to operate the triac in quadrants 2 and 3 you want the positive supply rail for the LVDC side to be anchored on one of the AC legs and for "ground" to be 5 volts lower than that. With that configuration, driving the gate low would result in a negative voltage relative to the T1 leg of the triac, which gives us the operating quadrants we want. The alternative, driving the gate positive means quadrant 1 and 4, which is a problem because making a triac sensitive in quadrant 4 is hard (and can result in increased dV/dT sensitivity, which is bad).

Another issue with the design is that putting the load on the DC side requires having a 3 watt power supply. Putting the load on the AC side just means that the DC supply only needs to do the equivalent of lighting 3 LEDs, which is less than half a watt (the load of the controller is minuscule).

So for a sort of middle ground, I've made a board version that uses an open frame AC-DC supply that's fairly small and rated for 1 watt. The load is on the AC side and is switched by a second SSR (the first is the power supply holdfast). For 120VAC, 3 15 kΩ resistors will draw just under 1 watt each. This winds up being just a bit less than when the load is on the DC side (because of inefficiencies in the DC supply), but should still be visible enough to the Intelliflow.

As before, the system will bootstrap with a pushbutton across the first relay's switch terminals. That will start power long enough for the controller to start up and activate the first relay, holding the power on for the duration of the timing cycle. The second relay will switch the load on and off to simulate washing machine activity. When time is up, the first relay will be turned off. As the DC supply voltage falls, the brownout detector will hold the controller in RESET until the power is too low to matter.

I took some measurements on the AC side using two different wall warts to see what the current flow would be with the load on and off. With the load off, the AC side sees about 100 mW. That's quite a lot, given that the theoretical DC load should be 800 µA or less at 5 volts (< 5 mW), but nobody ever said that wall warts were terribly efficient at low draw. With the load on, the AC side sees just over 5 watts of load. Given the theoretical DC load of about 2.7 watts or so, that's about 50% efficiency. Not... fantastic. The standby load will wind up being about a kW-hr per year. About the only way to make it any better would be an arrangement where the button would momentarily turn on AC power where it would be held on by an SSR for the timing duration. That's certainly possible, but it does complicate the design quite a bit (and raise the cost).

The board I have now does reliably start the intelliflow. Perhaps the solution is to cycle the load on and off in some pattern to keep the Intelliflow activated.

the downside with the current hardware is that the activation LED is in parallel with the load, so modulating the load will modulate that LED. If this works, another rev of the hardware will be necessary to have a separate GPIO pin just for that LED. 

EDIT: That seems to work. In fact, a duty cycle of about 50% seems to be about right. That's going to be good from the perspective of wasting power and from the board just getting hot from the dissipated energy.

At first blush, the first iteration of the board - with just under 5 watts of dissipation does make an Intelliflow turn on. You have to give it time to turn off, though. Even unplugging the 5 volt wall wart doesn't make the power turn off right away. I believe this is because newer Intelliflows have a 15 minute timeout before they turn off.

~5 watts of dissipation, however, makes for a great deal of heat (and, of course, wasted energy). I am going to try to remove some of the resistors to see if the unit remains reliable without getting so hot.

EDIT: Wow, it turns out two watts is enough. Who knew the Intelliflow was so sensitive? Well, now it looks like even 5 watts isn't reliable. Going to have to do more work.