January 8, 2006
about 2 months ago
The 73.2k resistor on Isense output pin of INA169 is not a source of series impedence… it is a shunt to GND that converts the INA169’s output (which is actually current) into an analog voltage signal. This is described in the spec sheet for INA169. It’s sometimes required for ADCs to use an amplifier pre-conditioner, which amounts to an Op-Amp with output tied to input reference… it provides extra muscle to ensure analog input is passed into the ADC with minimal attenuation. I’d suggest it as a very good idea for all products with ADC, such as hobby autopilots, to use ADC amplifiers between outside world and the ADC. ADC amplifiers are available as multi-channel ICs in small packages. If instead the amplifier were added to these simple AttoPilot boards this would also necessitate adding a voltage regulator and additional passives. A voltage regulator that accepts up to 50 volts and supplies an IC at say 3.3V is a large Vdrop for a compact cheap linear regulator and would waste approx (50-3.3)/50 of power drawn through the Vreg circuit powering the ADC amplifier… switching regulator would be more expensive and most likely require inductor and complication of more expensive high speed switching diode to avoid excessive leakage at the 120+ deg F likely to be experienced by this PCB… versus a cheaper Schottky diode in the switch Vreg circuit. Basically, the goal has been to keep this sensor small and simple as possible. I’m am, however, very open to new components in the future that can improve this PCB.
It’s recommended to flood the pads with solder between shunts and your leads. These boards receive manual solder flooding as final assembly step, however variations in amount of flooding do occur.
Could be difficult: INA169 is powered by in+ to GND, and 70V is above spec’d limit of INA169. A crude approach that I wouldn’t recommend is cut the trace between +in and INA supply input, then use resistor divider between +in and supply to INA169… would have to stay within power dissipation limits of added divider while supplying enough current to INA169. Cans of worms. Better approach would be voltage regulator, however at 70 V would probably need to keep Vdrop across regulator small. Another option is between +in and supply to INA169 insert several silicon diodes in series to drop the voltage about 1.5V/diode, using the # of diodes to reduce voltage to under INA169 upper limit. However that still requires cutting a tiny trace but doable with a razor blade tip. Please take my comments with a grain of salt, depending on your personal comfort level doing ad-hoc modification of circuits. I cannot accept liability for undesirable results to performance or safety.
Thank you, that was helpful. The added OpAmp serves as amplifier into external circuits, overcoming impedance unknowns of external circuits and delivering the full output analog signal into user’s ADC. The AttoPilot sensor is designed per spec sheet of the components with focus on small size for use in small UAVs. Additional circuitry that’s not strictly required, per se, has been omitted. However - given the wide use in some popular UAVs, it’s smart to ensure ease of use w/o need to consider unknowns. I will take this into serious consideration for a design update post-depletion of my present stock of PCB panels. Thank you very much for the feedback!
about 5 months ago
No, though I’m unclear on the reason behind the question. In effect, there is already a resistor chain between the high side of shunt (+ in) and ground, which forms the divider circuit that outputs analog sensed voltage to your ADC. Are you referring to a power resistor of some sort? I’m not very clear in what you are asking. I may not see possible reply for a few days, so feel free to email me at email@example.com. I’m Dean, creator of this product.
about 2 years ago
As-sold, the bandwidth for current sense is approximately 16 kHz. This means it is plausible to catch ripple of the motor IF this is your desire. You might even be able to monitor the rapidly changing current across one pair of wires to a brushless motor.
As-supplied, this PCB also has the 0.1uF capacitor on output of the ADC, so it has a BW of about 16 kHz, which is reduced from about 60 kHz if it weren’t added and the INA169 chip’s internal capacitance were the limiter along with the external chosen 73.2 kOhm resistor.
But you used the word “cope” - so I assume you DON’T want ripple to be measured but instead want a much lower bandwidth such as 10 Hz. If this is the case, you essentially want to measure the DC component of the current without seeing any AC ripple. The way to do this is if using an ADC and micro, simply average successive readings. In my work I sample the ADC channels at a high rate such as 10 kHz but then average 16, 128, 256, or more samples. By doing this, all of these measurements happen to catch the ripple at a nice sampling of all points across it’s periodicity and therefore they cancel each other.
It is rated for 180 A continuous… but that depends on how it is integrated into your project. Also, I don’t recommend any “burst” over 180A no matter how good the cooling is… the fact is that this board has LOW thermal capacity due to very small size and if the waste heat generation were to jump from a max of 8 Watts to 10 or 12 for even a fraction of a second, the temperature could exceed the limits of the shunts or INA169. The shunts have masses of mere milligrams - so even a very brief excursion of over-power (2 Watts each shunt) could conceivably raise their temperature by several 10’s of Celsius. The shunt manufacturer is wise in stating 2 Watts as limit… they know the minutia of detail regarding how quickly heat is conducted out and away from the metal film resistive element. Also, think in terms of an exploding bridgewire detonator…. a little wire heats from ambient to > boiling point of the metal and into a plasma on the order of a millisecond or less; it emulates the shockwave of a chemical explosive well enough to initiate a chemical explosive. Thin metal films of 100 micrograms can heat and melt VERY rapidly if heat generation outstrips heat loss. In other words, I know you were asking about “burst” in terms of using the full stated 180 amp range, but here I want to 1) assure you it’s OK to use this sensor at 180 amps if component temperatures are kept in spec, but 2) never try a burst > 180 amps.
If you wish to use this unit up to 180 Amps, you certainly can, however this requires proper engineering work in the integrated design of your project + this PCB. There some guiding principles and helpful equations detailed below, and employing good engineering methods will guide integration of this PCB. The process might require extra cooling be added above and beyond your initial guess of what an adequate cooling amount is.
Based on spec sheets of sub-components, this PCB is rated to a maximum 180 Amps provided that heat is removed so as to not exceed upper temperature limits of the components. The shunts are rated to 2 Watts each and there are 4 of them. It doesn’t matter that they are stacked in this PCB, the fact is their electrical conductivity is high and it follows that the thermal conductivity within the shunt itself will be high also (heat conduction in conductor is dominated by the “gas” of conduction electrons). I realize “high conductivity” is a subjective term here. The shunt metal film resistance element will generate heat across the voltage drop area and those are on an insulating substrate, so in fact the hottest part of the shunts will be the metal film mid-point between the leads, and for the PCB will be the area between the stack of shunts. However this area is only about 2mm wide and spread out over a nice longer distance of about 6mm (I’m glad the chip was designed this way). As a general principle of PCBs, the unit will heat above surrounding ambient until heat loss = heat generation. The only requirement is to have this temperature not exceed upper temperature limit of the components. Spec sheets of the parts show: INA169 +85C, LVK25 0.001 Ohm shunt +125C. The 0603 passives (0.1uF, 10k, 1k, 4.7k, 73.2k) are rated to +85C. Thus, the limitations are +125 for shunts and +85 for everything else. Because the shunts are the heating source, you might find in your board they equilibrate hotter than the other parts, and most probably will be.
Waste heat power of a resistor scales as the 2nd power of current (power at 180 amps = 4x that at 90 Amps, or in this case 2 Watts at 90 amps, and 8 watts at 180 amps). In practice, if your installation has this sensor confined in shrink wrap and/or the heavy gauge power leads are short, then you might not be able to use it at 180 amps. Loss of heat to the environment is only by 1st power of temperature difference between hot object and the environment. This applies to heat conduction out through the heavy leads, and from board to the air. The air and heavy leads might settle at different temperatures, but the loss equation is still by 1st order temperature delta through each route respectively. So the fact that heat loss is 1st order, and heat generation is 2nd order by current level, I can deriving a simple back-of-napkin equation to estimate the temperature of the board at various current loads… you could use this to take a measured temperature at a known current load in order to estimate the temperature at some other current… a guesstimate is possible to determine max operating current load without exceeding upper temperature limits of the components based on measured current and temp at some lower value. I stress “guesstimate” because complicating factors arise and in reality systems like this are not completely linear…. greater heat delta = greater convection of air and this affects heat loss to environment… other affects change etc ad inifinum.
Heat Loss (Watts) = k1(Delta Temp) where k1 is a constant that defines relationship between the units.
Heat Creation (Watts) = k2(I2) k2 is a constant defining relationship between units… in this case would be resistance in Ohms.
The temperature equilibrates to stable value when the loss = creation, thus at that equil temp:
k1(Delta Temp) = k2(I2)
Rearrange to: DeltaTemp = (k2I2)/k1, and we now see the temperature will vary as the square of current to a decent approximate (excluding 2nd order affects).
Going a little farther, and because k1 and k2 are constants (more or less), rearrange to (I2)/(DeltaTemp) = k1/k2, thus plugging in DeltaT1 at Current1 and wanting estimate for DeltaT2 at hypothetical Current2, we’d us this relationship: DeltaT2 = ((DeltaT1)(Current22))/(Current12). If instead you want to estimate current required to cause another DeltaT, then use: Current2 = ((DeltaT2*Current12)/DeltaT1)0.5
If ambient is +25C and the PCB stabilizes at +35 under a load of 100 Amperes, we have a DeltaT of +10. Considering an upper limit of +85C (DeltaT of +60), then the estimated current flow to reach this DeltaT would be around: ((65C*(100Amps2)/10C)0.5 = 254 amps… which being above 180 amps means that we can safely pass 180 Amps because 180 amps would result in an estimated DeltaT < +60 and thus Temp2 less than 85C. In fact, the estimate of Temp2 at 180 amps in this case is 57C if ambient is 25 C. Yes 57C would cause a little pain to the touch (again DON’T touch a live circuit!), BUT is almost 30 deg C below upper limit of the INA169.
If you desire to push towards upper limit of 180 Amps, then I suggest a thorough evaluation of the integrated design under max load. For this you’d need a temperature sensor, such as one of those IR guns, or you could adhere a thermistor to both the INA169 and the shunt stack. Measure the temperature under full load and typical cooling airflow conditions, and simply ensure the INA169 doesn’t exceed +85 C and the shunt stack +125C.
Note: +50C is roughly the threshold of heat pain to your flesh… so testing if the part is too hot by merely touching it (I don’t recommend touching any powered circuit) is not a quantitative analysis to be relied on if you need to use the upper current range. Consider that +125C is well above boiling point of water and +85 would rapidly cause a nasty 1st/2nd degree burn… it would be very easy to wrongly judge the part is ‘burning’ up or too hot but it might still be 20 C under max limit. Of course magic smoke would be a clear sign of overheating, but that’s more of a concern for semiconductor materials because of a semiconductor’s inverse temperature coefficient of electrical conductivity and their ability for thermal runaway. But in this circuit the shunts are the heat source, and because they’re normal conductors, they have positive resistance temperature coefficient. Rather than thermal runaway, they would tend to pinch off current as they heat up… they are in this sense self-limiting to some degree versus how semiconductors work. I’m NOT saying they do in fact self limit too much current, I’m simply stating that I’d be surprised if they didn’t have lower conductivity at higher temperature like normal conductors. On the other hand, by virtue of their function as a tight-spec shunt and wide operating temperature range, the manufacturer is forced to use an alloy for the metal film that is relatively flat for resistivity as function of temperature. At high enough current the metal film would eventually generate heat faster than it is cooled out, and the film would melt. I don’t know if that would be gentle or an explosive failure… wait a sec I’m going to short one of these across a large LiPo 3s pack and let you know….
about 2 years ago
I have not looked at the spec sheet to see what material is used for the window. It looks identical to the silicon used in the Melexis thermopiles I used to use for an attitude sensor of my first autopilot sold back in 2008. If it is in fact silicon, then it acts as a band pass filter with SHARP cutoff for all wavelengths below about 1050 nm. Without being rigorous here, we know the sun looks orange/yellow so I’d assume the dominant photon wavelength is around 600nm… with plenty in the red, and less in the blue end. So again if window is silicon, this means NONE of the visible nor half of the NIR even make it into the sensor. All that can get through is NIR longer than 1050 nm as well as all the FIR. Check out http://www.flickr.com/photos/imager/3380554807/
Also, then sun is very tiny in the sky relative to the total field of view of these sensors: sun = 0.5 degrees, sensor field of view is > 90 degrees. By ratio of area of those disk sizes, this means the sun is basically NOTHING. So, sun = tiny, and sensor can’t even see the dominant wavelengths of the sun.
My autopilot product used an array of 6 of these sensors pointing 1 each up/down left/right and forward/backward to “see” the thermal difference between sky and earth. Blue sky looks very cold as in what it is in the stratosphere up where passenger jets fly (let' say -60 centrigrade). Even if it’s overcast with low clouds, those clouds are most certainly going to be colder than the ground. So these sensors with their large FOVs can be used to calculate tilt angles of the UAV relative to the horizon… and it is surprisingly accurate on the order of 1 to 2 degrees of angle. Where I’m going with this is that even here in Phoenix Arizona during a heat wave in 2008 with stark blue sky and full bright sun, the UAV autopilot was still rock solid (as always) measuring tilt of the UAV and flights were perfect.
I wouldn’t at all fear trusting these sensors outside during a sunny day.
about 3 years ago
I’m Dean Goedde, creator of this product. The power source would have its “+” output soldered to the board’s “+ In” pads, the “+ Out” connects to the motor’s + input. The return ground of the motor solders to the large GND pad of this sensor as well as the battery’s GND is soldered to the same large GND pad of this sensor. Simply put, this sensor is placed in the + supply line between power source and load. The GND pad of the sensor is simply to get a GND reference for the voltage divider, and as Vss on the supply of the INA169 IC that determines current flow.
about 3 years ago
Hi, sorry no this is only for measuring DC current up to 90 Amps. However, the spec sheet for the INA169 IC by Texas Instruments shows a circuit layout using two INA169 for measuring AC current. You can find that spec sheet by going to www.digikey.com and searching for “INA169”.
Disclaimer: I have not made an AC current sensing prototype using the INA169.
No public wish lists :(
Forgot your password?
No account? Register one!