AMSAT, the Radio Amateur Satellite Corporation, joined forces with students from Rochester Institute of Technology to create a MPPT attached to a Fox-1B CubeSat. It successfully launched into orbit on November 18th strapped to the back of a Delta II rocket. This analog MPPT, or Maximum Power Point Tracker, is used for optimizing the draw of a power cell in correspondence to the output of solar panels on the 10cm x 10cm satellite. In a nutshell, this works by matching the voltage of the two together. If you haven’t gotten a chance to play around with one of these first hand, Hackaday’s own [Elliot Williams] wrote up a thorough explanation of the glorious MPPT’s efficiency.
This little guy is currently hurtling along in an orbit every 90 minutes. During each of these elliptical trajectories, the satellite undergoes brutal heating and cooling cycles. The team calculated that this package will undergo a total of 29,200 orbits around Earth during its 5 year mission. This means that there are 29,200 tests for it to crack — quite literally — under pressure. To add another level of difficulty, the undergrad team didn’t have funding for automated board assembly. This meant that they had to hand solder over 400 micro components onto this board, adding additional human error to be accounted for in the likelihood of a failure. But so far, this puppy is going strong. This truly shows the struggles that can be overcome with a little elbow grease, hard work, and plain ‘ole good engineering.
They created some sharp-looking documentation for this project. I would highly suggest taking a couple of minutes to read their Fox-1 MPPT document if you are interested in seeing the carefully thought out design in detail. The collection of schematics shown above was used to predict the maximum power point voltage. The analog computer used Y = mX + B to accurately predict the MPPT based off of solar panel temperature. In doing this, they were able to account for radiation in orbit causing bit flips. Understanding all of these factors was essential for the success of this tiny beast.
We would like to say good luck to the RIT team of Bryce Salmi, Brenton Salmi, Ian MacKenzi, and Daniel Corriero in their future endeavors. Sending anything into space and then keeping it running once it gets there is no easy task, and there is already another MPPT set to launch on a Fox-1E CubeSat aboard a Virgin Galactic Launcher One in December, 2017. Check out some of the necessary testing they did on it below to ensure that this would be a successful mission. Let us know if you think this is as awesome of a collaboration as we do.
For more of an overview of the project, all of RIT’s senior design projects are posted on RIT’s website EDGE; it contains a little bit more information of the project as it’s being developed through different stages. Here’s the link for this project: http://edge.rit.edu/edge/P13271/public/Home
“This little guy is currently hurdling along in an orbit every 90 minutes.” So how many hurdles does Fox-1B (Now designated as AO-91) leap over in 90 minutes? Correct word should be “hurtling.”
Hurdling is to leap over (a hurdle, barrier, fence, etc) as in a race. Hurtling is to rush violently; move with great speed: The car is hurtling down the highway to get away from the police.
Thank you!
Some may say you are picky. I say you are a defender of the English language!
As years of xbox owners can tell you, automated soldering processes aren’t a panacea. Careful inspection after soldering is the key.
It might be hurtling along, but I doubt if it’s jumping over anything… Although, given the amount of junk orbiting the earth, that might not be a bad feature to add :P
As a student project, this is fantastic.
From a system engineering perspective, though, I wonder if it’s even remotely worth the trouble. I burns a goodly fraction of its output just running itself, eating one seventh of the solar cell inventory. It sucks a milliamp just to measure temperature with an RTD!
Between the inefficiency and weight, and also the decreased reliability it will inevitably introduce, I wager it’s much better overall to send up its weight in more solar cells.
Most of that loss is in the DC/DC regulator, which you’re going to have to have for any satellite running off of solar power running electronics that needs a stable voltage.
The DC/DC converter here is between the solar cells and the battery. It’s not between the supply and the spacecraft loads, rightfully so. It is definitely in addition to the the DC/DC converter used to supply the electronics. They even explicitly make the point that this converter doesn’t do charge control, so it doesn’t even replace that component.
Paul, what this allows is the assumption that you are obtaining the maximum power available from the solar panels – DC/DC loss when sunlight is present sufficient to enable the MPPT. AO-85 simply uses a LDO regulator so that the solar panel – LDO dropout voltage is always charging the battery until the regulator kicks in to protect from overcharging. This means that on AO-85 the battery voltage + LDO dropout voltage is the panel voltage. This does not place the panels near Maximum Power Point. In reality the MPP voltage of these panels is about 4.5V or higher. A fully charged battery is 4.2V and on AO-85 this means that the panels are never actually producing the desired power. When we need to maximum power to charge the batteries (when discharged to say 3.8V) we actually produce LESS power from the panels than when we don’t need the power (4.2V fully charged).
The AMSAT MPPT trades DC/DC loss such that regardless of battery voltage you can always be producing near the maximum power from the panels. When a battery is discharged after eclipse and is at 3.8V or so then the panel voltage will be allowed to operate near 4.5V-5V where it is most powerful. Then, as the battery charges and we need less charging current the panel is allowed to go out of MPPT since we don’t actually need the power anyways and it’s voltage rises to open circuit panel voltage.
You can see the stability of battery voltage is clearly present between AO85 and AO-91
AO-85: http://www.amsat.org/tlm/graph.php?sat=1&field=BATT_C_V&raw=conv&reset=0&uptime=0&rows=100&port=8080
AO-90: http://www.amsat.org/tlm/graph.php?sat=2&field=BATT_C_V&raw=conv&reset=0&uptime=0&rows=100&port=8080
Hope this helps!
Bryce, KB1LQC
Bryce,
My contention is that the marginal gains provided by the MPPT aren’t worth the cost in weight and reliability.
Without numbers, it’s conjecture, so:
Spectrolabs datasheet says (for two 27 cm^2 cells in series), their panel looks like a 0.460A constant current source out to 4.62V, and MPP is at 4.70V at 0.440A = 2.07W.
Let’s look at two conditions: battery near full at 4.2V, and near empty at 3.8V.
If using a MPPT, the panel delivers 2.07W times the 87% efficient MPPT = 1.80W, independent of battery voltage.
Without the MPPT, the panel delivers to the battery (when full) 4.2V @ 0.46A = 1.93W. This is MORE than with the MPPT.
When the battery is at 3.8V, the panel delivers to it 3.8V * 0.46A = 1.75W.
So the MPPT provides (at most) only 2.8% more power to the battery when empty, and actually provides LESS power to the battery over most of its charge cycle that with just the naked panels.
My contention is that it would be better to devote the mass spent on the MPPT and use it for more panel area. With the bare solar cell mass density of just 84 mg/cm^2, deleting that huge heavy circuit board will buy a LOT of solar cell area.
Paul,
Thanks for the analysis. If you are interested in providing future support for AMSAT satellites from a systems level I highly suggest contacting them at their volunteer contact form: https://www.amsat.org/volunteer-for-amsat/
I believe you are forgetting about the fact that the MPPT accounts for panel temperature changes which vary widely and are designed to be accommodated over -60C to +60C. Per the Spectrolab UTJ datasheet this can shift the MPPV over a volt. If you jump “over the cliff” so-to-speak during any of this operation then the power drops substantially quicker. It is not linear.
That said the MPPT was originally designed for a 3U cubesat with much more current and solar cells to deal with. As you increase the panels and overall power the need for the MPPT increases. Fox-1 is a 1U cubesat and the following goals were achieved:
* Demonstrate the analog MPPT works in orbit
* Provide AMSAT with a scaleable radiation tolerant design for future satellites
* Provide maximum power from panels irregardless of battery voltage
* Allow operation of satellite when batteries fail (solar cell Voc will damage satellite, MPPT works regardless of batteries being present)
Sure, there are scenarios where it isn’t the most advantageous. However. Fox-1 satellites are 1U, you cannot just add more solar cells to them without moving to 2U or larger. This significantly increases costs. The weight of the MPPT is not an issue as long as the satellite meets 1U cubesat allowances. There is no cost difference to my knowledge between a few grams and you might as well use what you paid for to demonstrate technology that allows your satellites to grow in complexity over time.
Bryce,
Fortunately there’s enough voltage reserve here, even at 60C, that it wouldn’t “fall off the cliff”, because that would then require a boost converter, and this MPPT is only equipped with a buck converter.
All good points you raise, and ample justification for flying this. It’s too bad the original intro post here didn’t highlight those more, because most of those steps are important components in systems engineering.
I (and many others here, I’m sure) would love to volunteer at AMSAT. Sadly, US government policies exclude us from helping, even though I’m just across the lake from RIT (I’m a VE3).
Ahhh bummer on the ITAR policies. Yes they are a pain. I agree that the Hackaday post wording is a bit off. They misunderstood a few comments in my original blog post too but I figured it’s better to let people dig in and figure out the issues from the blog post than to comment on every inconsistency. Just happy people get to see the work of AMSAT and a really cool aspect of ham radio I’m fortunate enough to be involved with.
Are these rad-hardened parts ?
It’s fully analog so bitflips are less of a problem.
You pretty much never see rad-hard parts in cubesats. The typical budget for an entire cubesat would barely cover a rad-hard microcontroller, and radiation levels at the low altitudes used by most cubesats are low enough that commercial parts fare pretty well as long as the system is designed carefully. Analog systems, as mentioned, don’t suffer from single-event upsets the same way digital systems do, and most cubesats deorbit before analog parts suffer significant parameter drift from the radiation exposure.
I do wonder if the analog design itself is a deliberate choice to address the potential upsets. Otherwise, you’d just use the myriad of digital-based MPPT ASIC which almost every power semiconductor company made
Yes it was, I don’t believe the Hackaday article mentioned it but the overview documentation should. It’s not as much about being analog as it is about being stateless to avoid single event upsets causing required resets or permanent damage. If this wasn’t the case we would have done an MCU with a simple algorithm like perturb and observe for MPPT or used an ASIC like you mentioned.
Thanks!
Yes it was a deliberate choice. The blog post should state this and if it doesn’t the technical documentation certainly does ;)