After sending a design to a fab house, I always take a deep breath, and start looking at my design from the relaxed position that I'll just have to wait for my boards to arrive. Now is the time to do some administrative chores such as making a partlist, look at assembly sequence, read through that part of the datasheet that I skipped earlier because I was in a rush... hey.... what's that?
I found some new insights that might be interesting to look at when testing my design:
- I was looking for the type of energy harvesting transformer I'd need; I had seen that for higher input voltages, not a 1:100, but 1:50 could give the best performance, with a bit of tuning. From the datasheet I gather that it doesn't matter that much; it mostly deteriorates the startup time, because you'll need a higher input voltage. The fact that the converter will run sub-optimal after that doesn't change much for my application. During cooking, the excess energy can be dumped in VSTORE (see below)
- The designation LTC3108-1 is not a typo, the LTC3108-1 is a different part from the 'normal' LTC3108. That actually caused my heart rate to rise..... My first thoughts were: what if I've selected the wrong part, that starts at 2V instead of 20mV input voltage? What if the 'good' part has a different footprint? Arrggghhh...
Luckily the LTC3108 is not that much different from the LTC3108-1. The biggest difference is that the output voltages can be programmed to different values. Other than that, I haven't found any significant difference.... Phew....
- Talking about voltages, that LTC3108-1 does have an attractive feature to make a 3V power supply. The datasheet of the LTC3108 mentions that you could run a microcontroller of the VLDO pin, which generates a 2.2V output. That made me think. Why wouldn't I use that pin? 2.2V looked very low to me, but the EFM32GG940 is happily running at a measly 1.85V, according to the datasheet. The LCD has its own boost converter, which may run up to 3.6V, so no problem there, unless the boost converter takes more energy than I would gain from lowering the supply voltage from 3.3 to 2.2V.
The only device that 'needs' 3V is the buzzer. In a respin of the design it might prove to be more efficient to go to 2.2V, and have a separate switched capacitor boost converter provide the 100ms of sound that I'd like to produce from the buzzer ('eggs are ready')
- Now that I'm looking at the buzzer: this is the current consumption over voltage:
The datasheet says it only works from 3V, which is the first thing I'm going to test... The curren tuse is something like 2mA at 3V. I decided to use the Vout2 of the LTC3108 to drive the buzzer, I thought it would be OK to have some current limited power switch handle this instead of a GPIO of the EFM32. If I'd like to use 3V, 2mA for 100ms, that would cost me 600μJ (microJoule) per beep. I'd prefer that energy to be taken from the Vstore capacitor. The Vstore capacitor charges to 5.25V while input power is abundant enough to keep the switching output (Vout and Vout2) of the LTC3108(-1) at the programmed level.
The amount of energy stored on a capacitor is 1/2CV2. If I want to get three beeps of 600μJ each, and don't want the capacitor voltage to drop below 3V, I have a voltage drop of 5-3=2V, and I need 1.8mJ (milliJoule). Capacitance needed is: 1.8x10-3*2/(2)2= 900μF. Whoops, that's more than what I accounted space for, and this calculation doesn't take into account losses..... Luckily I can patch in a electrolytic capacitor.
Now what if the buzzer would run off 2V? Even if the current would stay the same (it doesn't, it would lower), the amount of capacitance needed would drop drastically because the voltage is squared in the formula above: 1.8x10-3*2/(3)2= 399μF. That would even be possible with the ceramics I had originally planned!
- Once again, "It's always the fine print that trips you up [source: DAB]", and to get the most out of energy harvesting: read the datasheets, and make system-level design decisions. I'm going further in reading datasheets, todays' exercise has given me a lot of inspiration to do some experiments with the hardware....