We decided to build a rover! The Caltech/Jet Propulsion Laboratory has put out plans for the Open Source Rover, which is a mini version of the Mars Curiosity rover. Check them out! JPL is a research lab funded through NASA, but managed through a contract by Caltech. It’s the place where many of NASA’s robotic spacecrafts, such as rovers and orbiters, are built and monitored. I work physically from JPL as a contractor, though I am not a staff employee (I am employed the Planetary Science Institute.)

For obvious reasons, I’m a huge nerd for space robots and we thought it would be really fun not only to build the rover, but to bring it with us camping, to maker faires, and other places. I end up doing a lot of science outreach activities, and who wouldn’t love getting to drive a ‘lil rover! Maybe I’ll even do some field research with it eventually. I guess building a rover also seemed like a very timely project. Last year Mars had a giant dust storm that forced all of our active surface rovers to hunker down. Unfortunately, one of the rovers did not wake back up. NASA officially declared the mission of Opportunity, one of the Mars Exploration Rovers, to be complete. Opportunity was a fantastic little rover – its nominal mission duration was only 90 days, but it lasted for ~15 years. Talk about a well-built machine, and a rover that stole our very hearts.

While I have a lot of craft robots, I’ve never actually built a robot before. The build involves all elements of STEM including fabrication/machining, electronics work, and some basic software programming, so it has definitely been a good project for learning new skills. Ian and I are both creative in different ways, and our overlapping but separate skill sets have made for a very complementary partnership for this kind of project. The OSR is well designed, and the instructions are very detailed and thorough. The developers tried to keep the costs down to enable school groups, maker spaces, and other interested folks to build their own. There is also an active forum where builders can discuss problems, modifications, and other rover topics. The OSR developers are also very engaged on the forum, and working with the community have made upgrades and improvements to the plan since we’ve started building. These get incorporated into the Git repository and the build instructions for new builders. Overall, it’s a very fun, welcoming, and active community!

Planetary scientists (and NASA in general) use acronyms for everything, so we decided right away the rover needed a good name. We have named it PARSLEE, for Planetary Analog Remote Sensor and ‘Lil Electronic Explorer. (In planetary science, we sometimes do research in certain Earth environments that can be thought of as analogous to other planets, which we call planetary analogs or analog environments.)

When PARSLEE is complete we’re looking forward to taking it on a few missions:

Outreach Programs

Projects like this are really great for science outreach. Bring a mini rover to an event and kids (and adults!) are guaranteed to get excited! I also focus on doing outreach in other under-targeted communities, mostly through art. I both make and organize exhibitions for planetary science/astronomy related and data-driven artwork and we would like to use PARSLEE as a kind of interactive art experience. We want to use it to enable people to experience how it is that we can explore other worlds through remote sensing. When most people think of space exploration they think of sending humans places or taking images with a Mars rover. But you don’t have to see something to explore it, and an instrument like a seismometer helps us understand our physical environment in a whole different way. By having multiple instruments onboard for people to use and explore their own environment, we hope it will give a more personal and real understanding of space and scientific exploration.

(Skip to the bottom to hear about PARSLEE’s trip to the Burning Man arts festival!)

Engaging The Maker Community

On a more practical side, when we send rovers to other worlds the camera is often a really important tool. On Mars, the rover operations team sends instructions telling the rover where to drive. When it arrives, they look at the images it sends back, the science team decides what kinds of measurements or data they want to take, and then it is sent off to drive to a new location. This type of stop-and-go exploration works fine in some places, but there are some space missions in development that are planning to explore really challenging terrains or in which the mission lifetime is extremely short. One of my favorites is Moondiver, which plans to explore lava tube caves on the Moon. I’m not involved in this mission, but it got me thinking about how we could use PARSLEE to help us develop new ways of exploring. For these kinds of missions, navigation can be really hard due to the time delay in sending/receiving data to/from a rover. We’d like to engage the maker community in developing some technologies that facilitate more flexible and rapid navigation in pseudo-real-time, and plan to use the rover as a practical tool for testing them. We also have some other projects brewing around accessibility issues, and using rovers as a way for scientists who can’t physically get out into the field to participate in field research.


Finally, on the research side, I do plan to use the Raspberry Shake for some experiments. One of my main areas of research is on mechanical weathering processes on other worlds. Stresses induced in rocks from heating and cooling can cause fracturing and breakdown of rock into dust. We think this is really important on asteroids, but also on the Moon and Mars. In some places, the signature of this on the surface is really clear, but on the Moon it’s hard to see because there are other weathering processes happening at the same time.One of my collaborators, Dr. Renee Weber from Marshall Spaceflight Center, has analyzed the geophone data from the Apollo 17 landing site and discovered that there are thermally driven shallow moonquakes that happen every (lunar) day. We think these quakes may reflect the signals either from a thermal fracturing process or from soil/boulder shifting and settling due to thermal creep. Dr. Weber is on the Mars InSight team, so they’ll be looking for similar quakes on Mars. With this in mind we’d like to do some experiments heating and cooling boulders and soils in the lab and in the field, and analyzing their seismic signature. We don’t have funding for these experiments yet, but we’ll hopefully be able to do some preliminary analysis in order to write a grant proposal for it. For this work we’ll likely use the Shake both on and separate from the rover.

UPDATE: The rover is now finished! Peruse the whole process below, or skip to the end to hear about its inaugural outing the Burning Man arts festival.

Also, check out the articles from the MagPi / Raspberry Pi blog and the Raspberry Shake blog about our build, and find us on twitter for any recent PARSLEE news!


Feb 23:

First, we ordered all the things on the parts list which came in over a couple of weeks, then spent an evening sorting and labeling them. The instructions refer to things by numbers (S22, B7, etc) and there are a lot of pieces, so it really pays to be organized!

Feb 24-25:

We started with the rover wheel assembly, which came together pretty easily! I was pleased with myself for using a real compass (the drawing kind) to make a jig for drilling the holes in the rims. This really isn’t a huge accomplishment, I just enjoy being efficient and precise. The wheels are pretty sizeable, should be good for climbing things.

March 15:

We built the head and did some cutting and other fabricating work for upcoming components. We haven’t done the electronics yet, but the head has an LED array in it that should be fun to play with.


March 29-30:

We built part of the main body today, but realized we couldn’t finish with starting the electronics. We’ve been holding out because the rover developers are working on a new PCB board that will simplify a lot of the wiring. It should make the build go faster and be more robust. Then we hopped over to the differential pivot. This involved drilling holes in an aluminum pipe, which was tricky to do. They didn’t come out perfectly straight, and we may have to redo one of the parts. The differential pivot has to do with the steering, but I can’t tell if the crooked holes are going to affect its performance until we integrate all the parts together. So for now we’re going to move along.

We spent most of our build time this weekend on the rocker bogie joints. These proved the trickiest part to put together so far. There are some minor errors in the instructions that made it a little difficult, but mostly it just involved some trial and error to get through. There are a lot of small parts you have to hold together at the same time while you screw them in. We found in some cases it was best to do this alone, in others we worked together on the same part. The pattern of holes on the aluminum pieces is cleverly done so that all the parts can be screwed together at 45 or 90 deg angles, but more than once we put something on at the wrong 45 deg angle and had to redo it. This was partly because some of the images in the instructions are mirror images of what they should be, and partly just because making mistakes is easy. We eventually decided to build both rocker bogie joints in parallel so that we could double check each other and not have to figure out the instructions more than one time.

April 1:

We bought instruments! I know it’s a little early to be thinking about a payload for a rover that hasn’t been finished yet, but I just can’t help myself. We know that we want weather instruments, a camera, and a seismometer.

For the weather data, we didn’t want to have to build a whole custom, heavy data logger system so we bought a portable Kestrel 5000. I have a 3500 model I carry with me all over the place which I absolutely love. My friend dubbed it the Nerdomometer. It measures temperature, pressure, elevation, wind speed, and humidity. They are convenient because they’re extremely well built, rugged for outdoor environments, and really lightweight. It has a screen, but this new nerdometer can also communicate to a remote device via bluetooth or wifi. I’m going to want to write the data to a file, so we’ll probably route it directly to a tablet rather than having it go through the rover’s Raspberry Pi just to save on processing power. For the camera we went with a GoPro. There are cheaper solutions out there, but one of the places we plan to bring PARSLEE is an ancient, dry lake bed. Playas often have extremely fine dust that can get into and destroy anything, and we just don’t expect a webcam to last out there. We are already talking about how to proect the rover’s electronics from the dust, more on that later.

IMG_1002 2 copy

Finally for the seismometer we got the Raspberry Shake. There are a few versions you can choose from. They also have an infrasound monitor, but we decided the current paylod is enough for now. The one we got is built with a geophone and 3 accelerometers, and runs on its own Raspberry Pi. Like the other two instruments, we’ll power it directly from the battery and send the data to a tablet where we can view it live. It wasn’t too hard to set it up by itself, though it took some doing to be able to see the data from a tablet, as it was designed to just connect to a regular wifi network. Eventually, we plan to build a more integrated software system to control all the instruments, so I won’t go into that now. But here you can see some readings from the Shake during testing.

April 20:

Building the corner steering assemblies tonight. Update: The actual design of these turned out to cause us problems later. The assembly is put together such that the shaft of the motor is attached to the leg assembly via a clamp, however the vibrations from the ground we were driving on caused the leg and shaft to fall apart. This hadn’t seemed to be a problem on flatter ground, but I guess the roughness was too much. We will have to redesign this part of the rover before our next big trip.

May 4-11:

Time to do the electronics! Look at this adorable mini rover on the rover PCB ❤


So I have not soldered in a very long time. I broke out the PCBs and my old soldering iron from college and got to work, and it was *really* hard!! I had a minor existential crisis thinking I was just terrible at doing it, but eventually learned from doing some research that it was probably a problem with the soldering iron tip since it was ~12 years old. I did not know that they oxidized, plus the whole thing had only cost $20 at Radioshack. I bought a new soldering iron that was much nicer, had smaller tips, and weighed a LOT less and it was MUCH easier to use. It turns out, in fact, I am perfectly fine at soldering haha. There were a lot of components to solder on, but it was just a matter of plugging away and following the build instructions. After the soldering, we went through a series of tests to make sure all the components worked.

May 18-25:

Now that the smaller components are soldered on, it’s time to attach the ROBOCLAWs. I’ve never used these before, but (pro tip) it is incredibly fun to shout ROBOCLAW while you are working. These are motor controllers that control each of the drive and corner steering motors. They are kind of black boxes to me, but they do the job so that is great. We first attached them to the main PCB, fiddled with the settings, and then tested plugging the motors into each one. Each controller connects to two motors, so there are five ROBOCLAWs total (6 drive motors, 4 corner steering motors).

This video doesn’t exist

June 1:

We put the wheels on and did some final* testing of the motors today! We also mounted the PCB inside the body to prepare for full assembly. [*Narrator’s Voice: It was NOT their final testing of the motors, dear readers, but they were happy just in that moment.]

June 8:

We put the body together! We spent hella time filing the metal rod that goes through the center because the tolerance on the part was too high and it didn’t fit properly. But, such is the way of things- now, PARSLEE is finally starting to look like an actual rover. You can see our dog in the background, who is thoroughly annoyed at how much time we’ve been spending in the garage. I have to say, I really did not fully understand how the rocker bogie and differential pivot assembly worked until we put the whole thing together. The engineering design is really impressive, and the way that the pivot works to keep the rover’s motion smooth and balanced while the different wheels are being used is really elegant. [Update: I really meant to take a video of this, but forgot and now one of the legs is broken. I will get to this once we fix it! In the meantime, you can get a sense of how the components work together from this blog post and the animations at the bottom.]

June 15-25:

It turns out we had neglected to include time for wiring in our build schedule, which was a gross miscalculation on our part. It is making us nervous about finishing in time, but we can only try. I did not anticipate how loooooong it would take to splice all the wires together that run from the motors to the PCB and to the Arduino in the head. First, we soldered connecting wires for the head to the pins on two D-sub connectors. I’ve never made cables before! It was fun to learn how to do this (thanks to this website for instructions). There were a few components to solder onto the Arduino PCB and then we could assemble the head.

Here is a nice shot of the entire body put together. The angle makes PARSLEE look really big, but you can tell from the other photos it’s only got a ~2×3′ footprint. The 3D printed parts we ordered ended up a lot more pink than purple as we intended, but I don’t mind the bright color.


June 28:

Thanks so much to Raspberry Shake and the MagPi Magazine for featuring our rover build!! It was fun to talk with them even though PARSLEE is not complete. We have plans the folks who make the Shake to do a follow up post down the road.


June 29-July 7:

As previously mentioned, splicing all the wires to the motors took a very loooooong time to do. There 5 to 6 wires that run from each of the six motors to the PCB. It took us several evenings to finish all the wire bundles. Then we ran them from the motors, inside the legs, and in to the main body via the pre-drilled holes. [Update: It turned out later that we had to connect and disconnect the wires from the PCB several times which was annoying, and that due to its size the rover is hard to transport. In the future, we plan to run the wires to D-sub connectors so that we can just plug and unplug each entire leg whenever we need.]

July 8:

FINALLY, FIRST POWER TO THE ROVER. WILL IT DRIVE??? [Narrator’s Voice: It would not, dear readers. Our heroes still have a long road yet ahead of them….]

July 9 – August 8:

It has been at this point in the build that we ran into our biggest roadblocks of the project. Like any real engineering project, PARSLEE did not work right away. We first had problems connecting the XBox controller to the Pi OS. Once that was sorted, the  rover still did not drive. It was at this point we realized there was a whole procedure in calibrating and setting up the motors and the ROBOCLAWs that we had not done. This was partly a result of the fact that we’ve never done anything like this before, so we didn’t realize we were missing steps.

It was also a result of the Open Source Rover project being in an active state of evolution, and we had outdated documents. The advantage to working on an open source project is that it can be improved by community input, and so far the rover community has been very active and very collaborative! The downside is that the folks running thee project have limited time to actually work on it, and as a result the documents have been in an active state of evolution all summer as we’ve been working. This has led generally to numerous occasions where we were following partly new and partly old instructions, and so we did something incorrectly and it took us a long time to figure out what the issue was. There is no great way around this issue, so it was just a fact of the build. I won’t pretend it wasn’t frustrating, but on the upside the project developers and the other builders have all been there to help each other out whenever anyone gets stuck. In that sense, I’m appreciative of all the support we’ve gotten from a great community.

Once we finally got the missing instructions, we were able to perform the calibration! Our problems didn’t stop there, but it was really satisfying to FINALLY get the wheels to turn for the first time.

At this point, we started having issues with the motors breaking inexplicably. We were not the only builders to have this issue, and we all spent many many hours discussing the issues, trying to connect common parameters and behaviors and narrow down what the possible issue could be. We spent a long time talking with the support folks at Pulolu who were *great* in helping us troubleshoot. In our case especially, we were confused because we had never even driven the rover carrying its own weight before (only turning the wheels while upside-down), so it couldn’t possibly be related to bearing capacity or some of the other issues we considered. Then we thought it could be an electrical issue or a problem with the ROBOCLAWs, but that didn’t yeild anything either. I think the rover developers were also getting frustrated with not finding a solution, since they had not had any issues at all. This fact right here pointed exactly to what the issue was, which was lack of experience, my friends.

There turned out to be two issues. First, after we sent some of our failed motors to Pololu, they pointed out in one of them that there were gouge marks inside the gearbox on the top of the motor. These must have been from the small screws that attach the wheel to the top of the motor casing. Since we had tightened the screws too much, it would have only allowed the motors to turn slowly or prevented them from turning at all. We were warned that the motors can be sensitive to hand-turning them and that doing so could damage them, thus we had no way to (or didn’t think to) check if the screws were too tight. Then when we sent a signal for the motor to turn while it physically could not, it put the motor in a stalled condition, leading to spike in the current and the gears busting. This is a mistake that could have happened to anyone, but especially to folks without much experience with motors. The second issue was directly related- it turns out that calibrating the corner steering motors was pretty hard. We had used the auto-calibration option in the ROBOCLAW software, but it had not done a good job. As a result, we believe that some of the motor failures resulted from bad settings causing the motors to turn too slowly or to change directions too fast, which would have caused a similar current spike and stalled condition and, again,  busted the gears. Both of these issues were occurring at the same time, which is part of what made it so difficult to untangle. It is unclear to what extent the tight screws and bad calibration worked together to cause failures, but they easily could have. From that point on, we always calibrated motors using a manual method, and without the wheels attached and we stopped having failures.

All in all, this was a good learning experience but it was INCREDIBLY frustrating. We were working on a timeline to finish PARSLEE in time to bring it out to Burning Man with us. Combined with our lack of experience and needing help troubleshooting, we spent the last month of the build pretty stressed and unhappy. But, we eventually got there. Here’s a video of driving it for the first time. You can see the wheels aren’t turning quite properly yet- we had to update some of the actual code that tells the wheels how to move to reverse some RIGHTs and LEFTs, which fixed it pretty easily.


August 11:

I made PARSLEE a clean suit! This is to protect the electronics from the fine alkaline dust of the Black Rock Desert while we’re at Burning Man. There’s a separate post here.

August 23:

Here’s a picture and video of PARSLEE driving around our yard before we head out to the desert. It’s a FINE looking rover, if I may say so myself. Very proud to have done this- it was, by far, a much more challenging project than I expected and was worth every moment.


And because I’m me, I obviously could not resist making some PARSLEE stickers, patches, flags, and some buttons to hand out to interested bystanders. (FYI the Fandango flag is referring to the name of our camp at Burning Man, which actually predates the company.)


Burning Man Post-Mortem

Well folks, we had a fantastic time at Burning Man with PARSLEE. Everyone loved it!! We took it to all kinds of different camps and met lots of really interesting people who were excited to learn about the rover, and about space exploration generally. We’ve been going to Burning Man for a lot of years, and one of the things that really makes the event what it is is the harsh desert environment. It is actually a lot like Mars! It’s warmer, and certainly easier for us to breathe, but the landscape is a similar mix of dry playas and dusty terrains, interspersed with rocks and mountains. There are also dust storms and dust devils all the time, which add an element of surprise and discomfort sometimes when you least expect it. We call these kinds of environments planetary analogs, because they are similar enough to other bodies to enable us to do research and test technology but obviously much easier to reach and work in than actual extraterrestrial surfaces. Plus, it’s really beautiful out there and everyone we met commented on how well suited the desert was as a backdrop.

Everywhere we went, PARSLEE piqued people’s curiosity. They came to talk with us about the rover, asked questions about planetary explorations and current NASA missions, wanted to learn about the engineering of the rover, and asked to test drive it. As an outreach effort, it was wildly successful! The thing about Burning Man is there is art, events, people, and generally stuff everywhere, so you just have to go towards what you’re interested in seeing and what catches your eye. So it drew people in who were really interested in learning about it, but at the same time people who may not otherwise have ever been exposed to something like it. It isn’t like going to a sci-fi focused or ComicCon type event where the people attending already tend to be interested in nerdy things. At Burning Man, there are ALL KINDS of people, and it gave them an opportunity to see the rover and express interest without any expectations or assumptions about their prior knowledge or interests. Some people stopped for half a sentence, and others for a half an hour. People go to the event from all over the world, so we also met a lot of international folks, many of whom were interested in space exploration but whose countries of origin did not have space programs.

First we took it to Sk8 Camp to drive on the skate ramps and met a guy who’s last name is actually Parslee (he was stoked). It did a great job at climbing things, and the rocker bogie system made it a breeze to drive over small obstacles, so long as they were not bigger than ~2/3 the wheel height.

It drives pretty slow, so we had a little bike cart we used to transport it around the city.

Then we brought it to Black Rock Roller Disco (or as we’re now calling it, Black Rock Rover Disco) and asked it we could drive it around the roller skating rink. We stopped at Arctica, which is where you buy* ice, so that we could see that PARSLEE discovered water on Mars. This is kind of a joke for planetary folks, since it seems like there’s a new headline every few months that we’ve discovered water on Mars…again. (*Burning Man has a gift-based economy and so there is nothing for sale by anyone at the event, with the exception of ice for food storage and coffee from the Centercamp Cafe.) We did had some issues with PARSLEE, but nothing Playa Bike Repair could help us with.

In fact, we only broke it three times! Hahaha. Really though, if you’ve ever been to the event you know that the Black Rock Desert is a physically harsh place to survive. We expected to run into problems while we were out there, and there is nothing like like soldering a new motor into a robot behind a dive bar in the middle of the desert to make you feel really capable and proud of everything you’ve learned. We had to replace one of the motors after going over a giant pothole with the rover in the cart. We think the turning of the wheel against the canvas side of the cart just busted the gears.

Later in the week, we had problems with one of the legs falling off. Since the motor shaft is only couple to the leg via a clamp, we think the vibrations from driving on the bumpy, gravelly terrain just caused it to keep coming loose. We’re going to look into how we can redesign this part of the rover for future trips. This is a natural part of engineering and technology development! I do want to use PARSLEE for some fieldwork eventually, but I’m going to have to address this issue first to make it more robust. We have plans to return to the builder forum to look for ideas, but just haven’t gotten to it yet. On the upside, the Clean Suit worked like a charm! The electronics stayed clean as a whistle.


We also brought PARSLEE to all the space- and robot-themed camps that we could find. Our favorite was meeting this giant robot (it was actually an art car, not a robot) that lived down the block.

We had a tablet with us that displayed the output data from all of PARSLEE instruments. People are generally less interested in data than they are in robots, but I had wanted to use this a way to introduce people to how we use data to explore. The camera and the weather meter worked just fine. The Raspberry Shake also worked, but unfortunately did not give us the result we had hoped for. There are a lot of camps with loud sound systems out there, and we thought it we were close enough we could pick up the signal on the Shake. What we hadn’t considered was that the rubber wheels of the rover would be really isolating. We wouldn’t actually pick up any signal on the Shake except for from the rover moving itself. All the other signals coming through the ground were dissipated in the wheels. This is why we test technology before sending it to another planet! This is, again, one of those things that is just a natural outcome of any engineering project. Sometimes you just don’t know what is and isn’t going to work until you try it out. For this year, we’ll have to build it a little deployable arm to lower the Shake to the ground.

On that note, I’d encourage everyone to check out the latest news from NASA’s InSight and OSIRIS-REx missions. InSight is a Mars lander spacecraft with a seismometer for detecting Marsquakes. The data it is collecting not only show the first quakes ever detected beyond Earth and the Moon, they also provide invaluable information about the planet’s interior and active fault systems. OSIRIS-REx is an asteroid sample return mission, meaning it’s been tasked with collecting rocks from the surface of the asteroid Bennu and returning them to Earth. I am a part of the OSIRIS-REx science team, and we’ve been working hard all year analyzing images of Bennu’s rocky surface. Below is a picture of the site selected to collect our sample, named Nightingale [Credit: NASA/Goddard/University of Arizona]. (All the asteroids features are named after birds, check out my embroidery of all four candidate sample sites.)


One of the most interesting discoveries we’ve made at Bennu is that small pebbles are being ejected from the surface! The asteroid is what we call a microgravity environment, that is it’s so small (only about a half mile wide) that it’s gravitational force is low enough to make ejecting stuff from the surface pretty easy. So most of these particles completely leave the asteroid’s system, though others fall into sub-orbital trajectories and eventually fall back down onto the surface. But what is causing this phenomena? Well, the process I study is called thermal stress weathering, the fracturing and mechanical breakdown of rocks into dust from stresses caused by heating and cooling. It’s the same idea as when you leave something out in the sun for a long time and it breaks down, only in space the heating and cooling it experiences is much more intense. This causes boulders on the asteroid surface to shed material through a process called exfoliation, and we think that particles get flung off the surface as this occurs. I have a scientific publication coming out on this soon, so keep your eyes peeled for news.

Some colleagues and I have been discussing how much we could learn about the properties of the asteroid’s boulders and its interior if we had a seismometer on the surface. I have some funds to do laboratory testing of heating and cooling rocks in a vacuum chamber to study how quickly fractures develop in planetary environments. I’m going to try using the Shake inside the vacuum chamber so that I can detect emissions from the fracturing events at the same time. We’ve also made some plans for field experiments trying to measure seismic and acoustic emission events from rocks fracturing here in the mountains of CA and eventually down in Antarctica. Once we have a better understanding of how to interpret the signals and relate them to rock breakdown we can propose to send a seismic network to an asteroid surface. That will be down the road, but it’ll happen someday.

Anyways, I digress. In the end, building PARSLEE was an incredibly challenging experience. Even two smart people with mostly complete instructions had to learn a lot to accomplish it. We wasted a lot of time doing things wrong, learning how to fix things, learning how to communicate better, arguing with each other, and generally being confused about robots. It was a really fun and a great learning experience, though at times a very stressful and frustrating project. But we did it! And you can too, if you want! It just takes some tools, elbow grease, and patience. What would you do with your rover? Where would you take it? How would you modify it? Rove onward, friends- I hope you explore our world and others in whatever you do.