Q. What are the target applications for the LTC3108 and LTC3109?
A. As discussed in the datasheet, these parts target applications that require a low average power (in the milliwatt range or less). This could be a micro-controller that is running continuously, or a wireless sensor/transmitter that is operating at a low duty cycle. The LTC3108 and LTC3109 were designed to be powered by any source that can provide a minimum input voltage of about 25mV, with a low source resistance (ideally 3Ω or less.). The LTC3109 can operate equally well from input voltages of either polarity (or a low frequency AC input).
Q. How much temperature differential do I need to start-up the LTC3108 & LTC3109?
A. With a 40mm square TEG and proper heatsinking, the LTC3108 & LTC3109 can startup and regulate VLDO & VOUT (no-load) with a dT as small as 1°K or less. Typically however, with smaller TEGs and heatsinks, a temperature differential of about 5°K is required.
Q. How much power can I get out of the LTC3108 or LTC3109?
A. This depends on the input voltage and the source resistance (i.e. which TEG you use, how much heatsinking is provided, and how much dT is available). There are curves in the datasheets showing the output current capability for different input voltages and different turns ratios. In general, the average output power capability is limited to about 15mW maximum, even at high VIN. At the other extreme, for very low VIN, the average output power may be less than 50µW.
Q. So my customer can’t even use the LTC3108 or LTC3109 to power an LED?
A. No – probably not continuously. The average output power required is too high (we must still obey the laws of physics). However it can pulse an LED very nicely!
Q. What if my customer needs more output power than the LTC3108 can provide?
A. The LTC3109 can be configured for uni-polar operation with a single transformer. In this configuration it will provide more than double the current of the LTC3108. The input resistance in this configuration is 1 Ohm, so it’s a better load match to low resistance voltage sources of 1 Ohm or less.
Q. What’s the difference between a TEC (thermoelectric cooler) and a TEG (thermoelectric generator)?
A. Usually nothing. In some cases, when making a thermoelectric device for power generation, manufacturers will use a higher temperature solder to allow for higher operating temperatures and higher output power. (When I use the term TEG in the datasheet, I am simply referring to the fact that we are using a Peltier module to generate power, not to cool.)
Q. How do I choose a TEG for a given application?
A. This will depend on the average power required by the application, and the dT available across the TEG. The output power of a typical TEG is around 90µW/°K – cm2, assuming proper heatsinking. Taking into account conversion losses, figure about 25µW/°K – cm2 maximum at the LTC3108 output. For example, a 40mm x 40mm TEG (with <2.5Ω source resistance) with a dT of 10°K across it will produce about 4mW of maximum output power from the LTC3108. Note that maintaining a dT of 10°K will require very good heat transfer with such a large TEG.
Q. TECs don’t seem to be specified for power generation purposes, so how do I know which ones will work best for harvesting?
A. Manufacturers specify the maximum voltage and maximum current rating of the TEC (when used for cooling purposes). For the highest output power in an energy harvesting application, choose the TEC with the highest voltage rating and the highest current rating. This will generally provide the highest output voltage for a given dT, and the lowest source resistance to allow for the most power transfer to the load. HOWEVER, the larger the TEG is, the lower its thermal resistance is, meaning that it will require more heatsinking on the cool side of the device to maintain a dT across it. Therefore, larger TEGs are only beneficial if you use a larger heatsink. Applications that don’t need as much power will perform well using a small TEG (say 15mm on a side) and a small heatsink.
Q. What kind of open-circuit output voltage does a TEG produce?
A. This depends on many factors (manufacturer, model number and absolute temperature), but as a rule of thumb, the open-circuit voltage is about N*0.25mV per degree Kelvin across the device, where N is the number of couples (which is published in the TEG datasheet).
Q. What about the new thin-film TEGs from Micropelt and Nextreme? Do the LTC3108 & LTC3109 work well with them?
A. These new thin-film devices are very small and thin, and are generally designed to provide a higher output voltage (volts) at a low output current. In the case of the Micropelt device, the source resistance is very high (over 300 Ohms), and is therefore a very bad match to the LTC3108 or LTC3109 input. The Nextreme device has a much lower resistance and may work well, but requires a much higher dT (typically 50 – 70°K). The LTC3108 and LTC3109 were designed to work optimally with “traditional” TEGs (which usually have a source resistance between 0.5 and 5 Ohms) that are widely available from many manufacturers. The LTC3105, 250mV Boost converter with MPPC is a better match for higher resistance sources like the Micropelt device, as the input resistance of the converter can be programmed to match the source resistance.
Q. Isn’t the small size of these thin-film devices a big advantage?
A. Perhaps, but remember that any of these devices will require heatsinking to the ambient, and attachment to some source of heat (or cool), so in the end, the solution tends to be rather large, regardless of the size of the TEG.
Q. I understand that the load resistance should match the TEG source resistance for maximum power transfer. How do I do that?
A. You can approximate the TEG resistance by dividing the maximum voltage rating by the maximum current rating. Choose a module with a resistance of 2.5 Ohms or less for optimal power transfer. The input resistance of the LTC3108 harvester circuit is shown in the datasheet curves, but is generally >= 2.5 Ohms. The input resistance of the LTC3109 is typically 5.5Ω in autopolarity configuration. However, even if the matching is not perfect, it is preferable to use the lowest RS TEG available to maximize power capability. The caveat to this is that lower resistance TEGs are usually large and also have a lower thermal resistance. Therefore, they will also require a larger heatsink to maintain the desired dT across the device.
Q. You keep talking about the heatsinking. Why is this so important and how do I size the heatsink?
A. A TEG can only produce output power if there is a temperature drop across it, which requires heat flow through it. Therefore a heatsink is generally required on one side of the device, to maintain a dT across it. Otherwise the whole TEG will just reach the same temperature as the hot or cold side, and no power will be produced. (An exception to this would be if the TEG were sandwiched between two surfaces at different temperatures which each had a large thermal mass and low thermal resistance, such as a hot and cold water pipe. In this case of course, no added heatsink would be required.) For optimal power output, the heatsink should have a thermal resistance as low or lower than that of the TEG. Since the typical TEG has a thermal resistance of around 1-10°C/W (depending on surface area and thickness), this requires a reasonably good heatsink. Larger TEGs will have lower thermal resistance, and therefore require larger heatsinks. The heatsink directly impacts the dT, which directly affects the output power capability.
Q. How much do TEGs cost? I’ve heard they are expensive.
A. Getting TEC or TEG volume pricing seems to be a challenge. In single-piece quantities, the US manufacturer’s and distributor’s prices usually range from $15 to $30 apiece. However, don’t panic - you can buy a single 40mm TEG on Ebay for less than $5 including shipping from China, so volume pricing should be less than the cost of our IC. Some manufacturers are now offering small TEGs for about $1 in high volume.
Q. Where can my customers buy a TEC or TEG, and what part numbers are recommended?
A. Some of the big TEC/TEG manufacturers and distributors are: Tellurex, Marlow, CUI, Ferrotec, and Laird. The specific part number that will be required or work the best is application dependent.
Q. How do I size the CSTORE and COUT capacitors?
A. This is explained, with two detailed examples, in the LTC3108 and LTC3109 datasheets. Just remember that any pulse load current must come from the VOUT capacitor. The Store capacitor is only for providing a small current to recharge VOUT at times when the input voltage source is lost. For this reason, the storage element (battery or capacitor) does not need to be low ESR, and can in fact be hundreds of Ohms.
Q. Can I use Li-ion or the new thin-film lithium rechargeable batteries for the storage element?
A. Yes, but you will need other external circuitry (the LTC4070) to protect the battery from over-charge and over-discharge. Be very careful if using the new thin-film batteries from Infinite Power Solutions or Cymbet, as they have very little capacity. A special version of the protection circuit is required that draws NO current when the battery reaches UV. Otherwise the battery can become over-discharged and damaged if left uncharged for a day.
Q. What is the conversion efficiency of the LTC3108 & LTC3109? (and why is it so low?)
A. The datasheet has curves showing the efficiency for different input voltages and turns ratios. In general the efficiency from VIN to VOUT is only 20-40% (note that this does not include losses in the TEG itself due to source resistance). Remember – you are stepping the voltage up by a factor of 100! There are many sources of power loss (gate drive, transformer, rectifier), especially when the total power may only be 100’s of microwatts. However, the major source of power loss is usually the NMOS Rdson. At very low VIN, even the 6-7µA quiescent current of the IC becomes a significant loss.
Q. How do I choose the optimal step-up transformer ratio for an application?
A. The datasheet provides curves of IOUT vs VIN for different turns ratios to help in optimizing the transformer selection for a given application. However, you can use these simple rules to select the ratio:
- If the application must startup at the lowest possible voltage (down to 20mV), use the 1:100 ratio
- If the application must startup at the lowest possible voltage (down to 20mV), use the 1:100 ratio
- If the minimum input voltage (under load) is at least 150mV, use the 1:20 ratio
Q. Are there other choices available for the step-up transformer besides those from Coilcraft?
A. Yes, Wurth now makes transformers that are essentially identical to the Coilcraft parts.
Q. Why is the maximum input voltage rating so low? (2V at SW, or VIN*Ratio < 50)
A. There are a number of reasons. First of all, we are trying to avoid very high secondary winding voltages for safety reasons and to prevent exceeding the insulation rating of the transformer windings (conservatively rated for 100V), or the rating of the C1 and C2 capacitors. Secondly, we need to limit the maximum voltage seen from drain-gate of the NMOS devices when they are off (VGATE can equal -6V). Finally, if VIN is 300mV or more, you should use the LTC3105, and if it is 800mV or more, choose the LTC3525 or the LTC3526L.
Q. Do we have any patents on the LTC3108 or LTC3109?
A. We don’t on the LTC3108. The basic resonant architecture it uses is covered by numerous patents dating back over 20 years, and is now public domain. We do have a patent pending for the autopolarity architecture of the LTC3109.
Q. What about solar harvesting applications with the LTC3108?
A. In general, you should look at the LTC3105 first for these applications. The LTC3108 requires a minimum input current of a few milliamps (at the converter input) just to startup, and may not be a good load match for a PV cell. Therefore, small solar cells that have a short circuit current of less than a few milliamps will not work with the LTC3108 (or LTC3109).