HNHacker News
TopNewBestAskShowJobs

Ephil012

381 karma · joined January 30, 2017

etp256 at nyu dot edu
submissionscomments
Ephil012··on Crazy Thin ‘Deep Insert’ ATM Skimmers
Carrying cash is more common in NYC due to many small businesses there being cash-only. For instance, halal food carts are cash-only. As a result, there is higher ATM usage in NYC where these ATM skimmers were found.
Ephil012··on How to Make a CubeSat
Yeah that is true as well. I talked about this a bit in another comment I made about how much easier aluminum is to work with. The thing is even for the raw sheets it’s still very pricey compared to aluminum which is pretty good at illustrating the price difference. Once you look at molding them into the shape you want the price only goes up.
Ephil012··on How to Make a CubeSat
Oh yeah, forgot but to clarify you might be able to get something better than a smartphone camera with the right sensors and clever space saving in a 1U cubesat. It's just most of the 1Us I have seen don't use anything beyond smartphone level cameras due to their internals (e.g. reaction wheels, computer system, etc) taking up too much space and the fact they don't need something more advanced. Though if you were very clever about laying out the space and removing as many components as possible you could fit a small zoom lens in there and get something better quality. I suspect though there are people who have done this. I just don't think it would be anything super different from existing images taken from the ISS or other things in orbit. If you went for something bigger then maybe you could get something more substantial and get cool images of planets in our solar system close up. A 24 U might be able to pack quite a camera system, but as far as I know the general max size is 12U and anything past that is rare/theoretical. I know though at even a 2U you can get some pretty high quality images of Earth, but I don't know if that would really translate to anything super amazing of space given the distance most planets/stars are. I think those images would still look like they were taken with a high quality smartphone or maybe a dslr camera just due to the fact they are far out (even though technically 2Us pack way better camera systems than most smart phones).

As a whole though these camera system are mostly used for Earth as it's an easier target. Space images are more like astrophotography grade and not Hubble grade.

If you want an example of a good cubesat camera here are some. https://dragonflyaerospace.com/products/ I should note though that these would not generally fit inside a 1U given their smallest one takes up 1U worth of space.

Ephil012··on How to Make a CubeSat
Depends on the size. If you’re doing a small one (1U) you are probably just getting smartphone quality pictures. The reason why is because in my experience a lot of the electronics and stabilization hardware takes up enough space inside you can’t fit a super large camera system there. Although, I do think there have been upgrades to camera tech recently so maybe you could actually fit something with a decent zoom on earth if you’re careful with how you design it. Although, I doubt something similar in quality to a DSLR camera would be able to fit in a 1U. If you upgrade to bigger sizes you can get satellites that are high enough quality that they could see decently close up to earth. For instance, a bunch of CubeSats are able to get good enough images to monitor deforestation. Another commenter asked a variation of this deep space planets. That’s harder because it’s a much smaller target to aim for in the sky and would probably require a much better camera system.

https://dragonflyaerospace.com/products/ Here is an example of some CubeSat cameras to give you an idea of resolution.

Ephil012··on How to Make a CubeSat
Pretty much what the other commenters said. Nobody cares whether your cubesat lives or dies in orbit. However, the radiation tests done on these open source boards have been quite extensive and there are a bunch of cubesats in orbit with these boards right now. They haven’t died from radiation so it’s generally assumed it’s fine. Even NASA has a project using consumer hardware in orbit (KickSat-2). Anyhow, generally you only need to pass the “mechanical” tests like vibration, vacuum, outgassing, etc. A lot of people do radiation testing to ensure that their satellite won’t die in orbit, but as a whole these open source boards like PyCubed pass them enough where you can be sure you’ll get at least a decent amount of time out of it.

If you’re curious about the testing procedure here is a paper one open source project published about it. It’s by no means a comprehensive paper on testing but it gives a general idea. https://roboticexplorationlab.org/papers/PyCubed-SmallSat.pd...

Ephil012··on How to Make a CubeSat
There are CubeSats with various propulsion systems like ion thrusters and pulsed plasma thrusters in development. It's relatively new though iirc. These thrusters can last for an insanely long time. So maybe in theory it could reach a planet one day, but it would take ages. So in practice probably not practical and it just makes more sense to launch one from a rocket that is going to that planet.
Ephil012··on How to Make a CubeSat
Yeah that is also true. The CubeSats don't spend too much time over each ground station. You can technically get around this by swapping between ground stations and enable semi continuous communication, but it would require a ton of them to work well. Also, it wouldn't work over the ocean. There isn't really a way to partner with enough ground stations to get the needed coverage while over land. I don't think there are any ground station networks out there with enough coverage for this type of thing to be constantly swapping for near constant connectivity over land. We didn't actually need amazing connectivity. We were fine with just livestreaming occasionally, but the other issues I mentioned around data speed prevented it. I think if data speeds were better you could get live video temporarily (like 15 mins).

Also, for the satellite transmission via Starlink, I don't think Starlink / OneWeb offers that yet for CubeSats (might be wrong though). I do know Iridium allows for it though. However, as I mentioned earlier in the prev comment, the data speeds might suck still (could be wrong here too) and it costs a lot of money. I do think some ground stations would have wanted money too. We didn't actually need to livestream as it was just some fun PR trick. We ended up deciding it wasn't worth it in terms of cash / effort.

As far as over subscription, it depends. If you are using a public ground station network then its an issue. If you are partnered with no public ones then it's not as big of an issue. We were thinking of private ones only because public ones would not allow this kind of thing. So over subscription just didn't even matter for us really.

Ephil012··on How to Make a CubeSat
I don't know if I am technically allowed to say since it was given to us on a contract. I don't think I can name them specifically until we do a joint press release announcing the mission. However, I will say that it was a startup that basically had a lot of VC money to subsidize. It was to sun sync orbit though. Also, they have increased prices for future launches to like 25k or 30k or something like that. Mainly because of they are trying to wean off of subsidizing they said.
Ephil012··on How to Make a CubeSat
I forgot to note that the Carbon fiber comparison isn't entirely comprehensive. Carbon fiber is stronger so you could get away with a thinner sheet which saves money. However, I will say it still somewhat illustrates the significant cost difference there. If you do have a big enough cubesat then I do think it could make a difference to launch costs. However, the advantages of aluminum and significant drawbacks to carbon fiber mean that using it isn't practical regardless.
Ephil012··on How to Make a CubeSat
I should note though that due to the strength differences this analysis may not be complete. Technically carbon fiber is stronger so you can get away with a thinner sheet while having similar strength. Using a thinner sheet might save on carbon fiber costs. However, I still believe this analysis illustrates the significant cost difference.
Ephil012··on How to Make a CubeSat
So we are actually adding imagery to our satellite so I think I can answer these pretty well.

For question 1, I do think you could use it for astrophotography. You could probably capture pictures of the stars if you have stabilization on the CubeSat. However, the photos would probably be similar to the photos taken of space from the ISS. What I mean by this is you might be able to take photos of the stars, but it wouldn't be able to see anywhere as far out as Hubble. I know a bunch of CubeSats use smartphone quality cameras so you would probably get a decent photo that might make a nice desktop wallpaper, but it wouldn't really compare to a bigger satellite. It would show an overview of the stars, but it wouldn't be significantly different from existing images imo except maybe it being from a different perspective. As far as zooming in, that would actually be kind of hard due to space constraints. CubeSats are small, like really small (10 cm x 10 cm x 10 cm in some cases). So it would be hard to fit any substantial zoom lens system in there. That might make getting ultra close ups hard. Maybe as camera tech improves in the future what you are suggesting would be possible though. Aiming it at a specific planet also might be a bit of a challenge. CubeSats mostly don't have propulsion (although we are adding an ion thruster in ours). They can use reaction wheels to rotate though. However, accurately rotating it towards a tiny planet in the sky could be hard. However, I will say you could go for a really big CubeSat design and fit a good camera system in that. Then if you solve the aiming issue you could get some really cool images that are zoomed in on planets / stars.

For question 2, we actually considered trying this. I had the idea to live stream the satellite on Twitch. We discovered that livestreaming it was hard. If you are sending data from the CubeSat to earth, the data transmission is very slow. That basically rules out livestreaming. Maybe you could transmit it to a big communications sat in orbit which then sends it to Earth. However, I suspect the data transmission speed might still be less than ideal. Plus, using a communications sat network (like Iridium) could be expensive. Let's say though communication tech gets better and you can transmit directly to Earth at a fast speed. There is still are two major issues. Regulations and getting ground stations. Every single ground station (basically an antenna on Earth that gets data from space) requires you list it on your FCC application. If it involves ground stations in other countries (especially countries the USA isn't friendly with) it can get very messy from a regulations perspective. Also, getting those partnerships to begin with can be messy as well. There are some networks like SatNogs that do allow for receiving data from volunteer ground stations in other countries. Although, I don't know if they would really be practical for livestreaming due to coverage. I know Amazon has some offerings where they let you access their ground station network that might allow for livestreaming one day. However, I haven't really explored too much into it. I also haven't heard of any teams using AWS Ground Station personally though. Honestly, you'd be forced into using Iridium regardless due to coverage being shoddy for ground stations. Overall though, I think the data speeds just aren't there yet to livestream, at least not with LoRa (which is the protocol a lot of teams use for transmission).

Ephil012··on How to Make a CubeSat
I work on a CubeSat team and there are a few reasons why CubeSat teams choose aluminum.

1) CubeSat spec rules. The rules guiding CubeSats technically only allow aluminum. "3.2.15 Aluminum 7075, 6061, 5005, and/or 5052 will be used for both the main CubeSat structure and the rails." You can get a waiver, but it makes more sense to just keep to the spec for practical reasons. The CubeSat spec doesn't matter as much to private companies I think, but it does matter if you are a non profit or student team trying to get a launch with NASA.

2) Ease of use. This is the biggest reason (besides the CubeSat spec) most teams use aluminum. Aluminum is just so easy to build with compared to say carbon fiber. You can easily bend and drill aluminum to make the shapes you want for cheap. Carbon fiber is just difficult to work with.

3) Temperature resistance. Carbon fiber doesn't handle extreme temps as well. Outer space is extremely cold and in some cases it can cause carbon fiber to weaken. Meanwhile, titanium is a horrible choice because of its high temperature resistance. NASA explicitly tells people to not use it because Titanium has such a high melting point it can survive re-entry to Earth and accidentally hurt someone or destroy property.

4) Air trapping. Sometimes air can get trapped in composites like Carbon Fiber. This can result in it trying to escape into space and damaging the CubeSat. It's a major risk factor and one of the big reasons to avoid carbon fiber unless it's made in a specific way to ensure that doesn't happen. Unfortunately, it is hard to find carbon fiber that meets those standards.

5) Cost. For a 1/32 thickness 12" x 12" sheet of carbon fiber, it is 35 dollars. Same sheet size / thickness for aluminum is 16 dollars. That is over half the price. However, CubeSats are small (10 cm x 10 cm x 10cm in some cases) so material cost isn't as big of a deal as you might think. The bigger concern is the added weight might increase your launch cost. However, in practice the weight savings are negligible at the smaller CubeSat sizes. I should also note that the aluminum is mostly for the frame / chassis and that doesn't consume that much material. NASA charges nothing for their student / non profit launches so weight doesn't matter there as long are you aren't over the limit. For private launches, they usually round off the weight anyhow. Sometimes to the nearest whole KG, other times to the first decimal place. Either way, you probably aren't too concerned about weight due to the limited material used. I know for ours using Carbon Fiber wouldn't actually save on launch costs due to the rounding. If we were near the border or were doing a bigger sized CubeSat, then sure it might save us, but there are other things we can do first to save weight before switching materials. I think if we were building a very big CubeSat then yeah we might become concerned over the weight costing us extra. However, weight isn't the biggest factor for us simply because of the other reasons. Even if we wanted to, we couldn't use Carbon Fiber due to the air trapping and temperature issues.

Overall, the benefits of a lighter weight material like carbon fiber just isn't worth the difficulties that come with working with it. As a result, the standard is to use aluminum because it's cheap, works well, and is easy to use.

Edit: I said a 1U sat was around the size of a Rubik's cube before. However, I underestimated how small Rubik's cubes are. A Rubik's cube is 5.6cm on each size vs 10cm for the 1U CubeSat.

Ephil012··on How to Make a CubeSat
I haven't really seen this to be the case personally.

For a 1/32 thickness 12" x 12" sheet of carbon fiber, it is 35 dollars. This is actually this company's cheapest carbon fiber sheet too at that size. I have seen other companies charge more iirc. https://dragonplate.com/EconomyPlate-Solid-Carbon-Fiber-Shee...

Same sheet size / thickness for aluminum is 16 dollars (although I saw a shadier website selling for 12 dollars). https://www.mcmaster.com/aluminum-sheets/aluminum/multipurpo...

Now for cubesat grade stuff you might want something better than what I linked, but still its a good illustration of the price difference. That is over half the price. I think the prices are maybe closer than in the past, but there is still a significant difference.

Ephil012··on How to Make a CubeSat
Yeah, it would be useful for that. In general, anything involving tracking changes on the Earth is fair game for CubeSats. I haven't heard of any organizations / companies using CubeSats for El Niño, but I think that could be a use case in the future. I think there is also a pretty good use case for gov agencies tracking hurricanes that might form due to the influence of El Niño.
Ephil012··on How to Make a CubeSat
If anyone is curious, I believe the big provider for this is Nanoracks. Never used them ourselves for our CubeSat due to our specific goals. However, they are extremely useful for stuff like sensor data, tech demos, research, and testing.
Ephil012··on How to Make a CubeSat
For commercial, the big use case is satellite imagery. Satellite imagery is pretty useful for mapping, agriculture, and even planning out large construction projects. CubeSats are especially useful if you need a cluster of satellites for cheap to blanket the Earth for imagery or sensor data. I think that there are also a bunch of defense companies using CubeSats for sensor data. I believe in general the US gov is investing heavily in defense companies using CubeSats. There are also other use cases involving communications. I have heard of companies looking into using them for IoT and phone communication. That might not be fully developed because right now there are some limits on data transmission speed from CubeSats. Another big use case is asset tracking. You can keep track of exactly where cargo ships are at sea. You can also use them to track planes in the air. As a whole, though a lot of these use cases I mentioned are very early stage / theoretical since CubeSats are a relatively new tool for companies to use. In terms of practical use right now, I think it's mostly satellite imagery / sensor data that it's being used for. However, I expect it to expand in the future to the other use cases I mentioned and many more.

In terms of scientific, I think it's best to think of CubeSats as a floating lab. You can basically stick whatever you want on there (size permitting). One thing you can do is stick cameras on there to track climate change through a cluster of tiny cheap CubeSats. Another is to put sensors to measure different particles in space. CubeSats are also useful for testing out new forms of propulsion like ion thrusters. You can also use them to test new materials in space. I have even heard putting bacteria inside CubeSats to see how they react in space. In general, CubeSats are just insanely useful for collecting sensor data for research and putting items in space to see how they react.

The big appeal though is you can do all of this stuff I mentioned for the fraction of the cost of a normal satellite. As a result, CubeSats are great for smaller companies or researchers who don't have unlimited cash to spend on a big satellite.

Hope all this info helps!

Ephil012··on How to Make a CubeSat
One interesting development in the area of cubesats is that most teams are shifting towards open source designs. In the past, cubesats were mostly using old and expensive hardware (like the pumpkin board). Now a lot of teams are using open source designs like PyCubed and building the electronics themselves. As a result, the price to enter the cubesat space has dropped dramatically. This has enabled a bunch of university teams to start building their own cubesats.

Many of these open source projects actually use consumer / hobbyist grade components instead of using commercial aerospace-grade components. The commercial hardware for cubesats is made with a specific manufacturing process that makes them resistant to radiation in space. However, this increases manufacturing costs. Also, most of the radiation resistant hardware being sold is sort of slow / outdated. The solution some open source projects came up with was to just take components from Mouser and run their own radiation tests on it then choose the ones that fared good enough to be used in space. That way they can make satellite boards for way cheaper.

Another interesting development is there are some startups now offering sub 50k launches for cubesats, such as launcher space. They essentially buy launches from SpaceX then outfit it with hardware to launch a massive amount of cubesats at once. As a result of the number of cubesats they’re able to fit on one launch, the prices are now relatively cheap compared to the hundreds of thousands if not millions required in the past. However, most of these startups are heavily subsidizing the cost using VC money so these low prices won’t stay forever.

Also, if anyone has any questions about cubesats feel free to AMA. I work at Isomer space which is a not for profit building cubesats. I would love to answer any questions about cubesats or our work.

Ephil012··on How to Make a CubeSat
The other figures are correct. However, the cost to build cubesats are coming down rapidly. I work at a not for profit called isomer space which builds cubesats. We originally were looking at 50k-100k for ours, but there are a few open source projects now that provide design files to manufacture your own for way cheaper. For us, we are able to get the costs down to under 10k by doing it all in house. This is including multiple prototypes. A single satellite itself is probably 3-4k though for us. If you’re curious, the big open source project in the space is called PyCubed. A bunch of university teams use their designs now because the hardware is relatively affordable.

As far as launch prices, I think the 50k figure given is accurate, but only through companies that resell / broker space launches. Getting your own launch outside of a reseller / broker would cost more. For instance, if you buy from SpaceX's rideshare program directly it can cost 1.2 million. This is because SpaceX's base pricing starts at a 200kg payload. Anything that weighs less is still charged at 1.2 million. However, if you go through a third party that resells the launches then it can be 50k or lower. I know I have seen one as low as 20k for a 1U launch. The main reason why its cheaper is a combination of economy of scale (launching lots of cubesats at once) and/or VC funding. The best launch prices in general tend to come from startups who are heavily subsidizing the launches with VC money.

Also, if anyone is curious about cubesats or our work in general feel free to AMA.

Ephil012··on Ask HN: Who is hiring? (August 2022)
I didn't see any contact info in your profile, but I am definitely interested. Feel free to email me Etp256@nyu.edu
Ephil012··on Yesterday the FBI signed its first public contract with Clearview AI
That’s what’s interesting. My state has a law going into effect soon that covers requesting the removal of facial data, but at the moment it hasn’t started yet. Presumably once it goes into effect, Clearview would be legally obligated to remove my data.

Also, forgive me for asking this dumb question, but what do you mean by send a letter with a CD? What does CD stand for? Is it some kind of certificate of receipt? I tried looking it up, but didn’t find much about it.

Also, in terms of asking what they have on me, I have basically been unable to have any communication with them. Others seem to have the same issue. In general, Clearview takes months to respond to any sort of email (at least for me). I sent one semi-recently from a different email address. Hopefully they respond to that.

Ephil012··on Yesterday the FBI signed its first public contract with Clearview AI
Yes, Clearview requires it. Technically, the laws can only cover residents of those areas so that is how Clearview is able to require residency.
Ephil012··on Yesterday the FBI signed its first public contract with Clearview AI
Yeah same, it feels very backward to me. I mean I can get that they want an ID to make sure you aren’t filing a request on someone else’s behalf without them knowing. Still, it is a bit disturbing they need an ID. Supposedly they don’t do anything with the ID info, but you never know I guess. I sent the ID picture in hopes they’d remove the facial data, because they did for others. However, when they denied removing the data I instantly regretted sending the picture. A bit of a stupid move on my part to be honest.
Ephil012··on Yesterday the FBI signed its first public contract with Clearview AI
Typically with these types of requests it isn’t abnormal to have to give over an ID. I don’t like the idea of doing it, but usually it’s the only way to do it. Even if you’re a California resident, you’re forced to upload your ID to remove data. It is kind of a no-win scenario. Either they keep your facial data or you give over highly sensitive data to get the facial data removed. I only did it because it seemed initially they were willing to remove the facial data (despite their very delayed response).
Ephil012··on Yesterday the FBI signed its first public contract with Clearview AI
The worst part about Clearview is there is no way to easily opt-out if you do not live in California or Illinois.

I wrote Clearview an email a long time ago saying I would like my data removed. They asked for my ID to verify who I was. Clearview responds saying thanks for sending your ID, but you aren’t a California resident so we don’t have to remove your data. I wrote follow up emails still asking if they could consider removing it. After my follow up email, they just stopped responding. It’s been months now and no word back.

I get that they have no legal obligation to, but they should provide people a way to opt-out in other states. It felt like a bait and switch to ask for my personal information then just ghost me. However, I guess nobody can expect Clearview to be a moral company.

← PreviousPage 3 of 3