FrontierSat's 3U frame was manufacturing using CNC & Laser-Cut Sheet Metal. This had a few advantages & disadvantages: Pros:

• 1/4 the cost of a fully CNC'd frame
• Much shorter lead & iteration times
• It was easier to accommodate mounting cut-outs

Cons:

• Less rigid than fully CNC'd frame
• Can be harder to hit your tolerances
• There is a risk of deformation if using bending processes.

Overall, this hybrid design is worth doing again - its ease of manufacturability & rapid iteration cycles being essential for a student team that saw its design requirements constantly change. A caveat to this is that it's not worth doing commercial bent sheet metal rails; it's incredibly difficult to hit the +/-.1mm specification. Even being a degree off-parallel makes for a bad day.

SendCutSend

The surface roughness of the sheet metal delivered by SendCutSend was 1.5-1.6um, meaning no additional surface prep was required. This led to it being a straight line of being received -> A286 stainless steel PEM insertion -> Epoxying the PEMs in place -> Using clear chromate (instead of anodizing, with that being cheaper than expected) -> Ultrasonic cleaning -> Being installed on the satellite.

4 frames were ordered from SendCutSend during development. 2/4 were usable, and if we knew we didn't need to be as careful as we thought for removing rail material, it might've been 3/4. These 4 frames cost less than CA$1000, with its longest lead time of 7-10 business days & shortest of 2 days. They're not the cheapest per part, but they're incredibly quick in comparison to our local suppliers. Our CNC parts were done in-house.

Issues with the CubeSat Design Specification

Between your launch provider & the CubeSat Design Specification (CDS), go with your launch provider's specifications. Our focus on the CDS led to problems with integration. For example, sheet metal bending was chosen to meet the fillet requirement - this is hard to do on a CNC. Our launch provider, Exolaunch, didn't have that requirement, meaning chamfers were fine.

This unawareness of our launch provider's specifications also led to us losing out on 33% more volume & mass than stated in the CDS at no extra charge. For SAT-2, we've told the team to try to use this extra volume, preferably filling it with extra deployable solar panels. The worst case for SAT-2 then becomes an extra U of solar panels that they can do without if Exolaunch winds up not being its launch provider. Exolaunch has some key, non-negotiable requirements:

• Must be 100x100mm rail width +/-.1mm
• Must exit the deployer during a ground deployment test using spring force alone
• Must have no visible damage to the rails from being inserted/removed for the deployer

If a requirement doesn't make sense to you, keep asking why it exists until it does or it gets removed.

For your own satellite design - look at your 3 most likely launch providers & design to their specs at your level of risk tolerance. They all defer to the rocket provider, which in our case was SpaceX & their Rideshare Payload User's Guide (RPUG). The CDS is a comparatively static guideline compared to the launch provider industry's volatility - 2/3 of our providers fell through, & Exolaunch didn't go without delay!

Fasteners

Due to our extremely sensitive sensors & expensive ADCS, we needed to use a combination of high-quality fasteners that satiated a strict no-magnetic-materials requirement. These wound up being:

• Nitronic 60 Helicoil Threads
• A286 age-hardened stainless steel PEM pressfit insert nuts/studs
• 316 stainless socket heads/nuts
• A few aluminum standoffs/spacers & M2 countersinks Altogether, we spent ~CA$200 on fasteners. Make sure to be careful with ones from China, as you get what you pay for.

Epoxies & Adhesives

We had to use several epoxies & adhesives on our frame. These included:

• Blue Loctite 242
• Red Loctite 272
• 3M Scotch-Weld 2216 B/A Grey Epoxy. Be careful mixing this, as the lid colours are very misleading. The Blue Loctite & 3M Scotch-Weld satisfied our requirement of a maximum total mass loss of 1%.