We've made magnetic tape at home, and it (sort of) works. I was hoping to improve its quality, but so far I had no progress in the last month, so might as well share what me and @yottatsa managed to do, plus some thoughts on the topic.
Edit: now there's videos: https://tech.lgbt/@nina_kali_nina/111286617790635607
(next posts will be unlisted)
Computers have used magentic storage almost since their conception, and they still widely use it. The most peculiar example of that is how home computers of the 70s and 80s were using domestic tape players for data storage.
The affordability of the medium and the ease of the implementation are very attractive qualities, as you can see! There's also a bonus: magnetic media can store analog data (including audio).
So, it is very tempting to try and build a magnetic storage system. There are three major components to it: magnetic medium, magnetic head, and "glue" (mechanics, signal amplifiers, etc).
Here comes AEG Magnetophon K1, circa 1935, the first commercial tape reel recorder.
What does it have in common with modern devices like reel-to-reel decks? Turns out, a lot!
It uses a very similar medium - and in fact tapes from Magnetophon should be playable on newer devices, and vice versa (but playback speed needs to be adjusted).
It uses magnetic head based on the same principles as modern heads, and it has very similar schematics - though it turns out cheap Chinese walkman knock-offs are closer to AEG than Elizabethan, because no AC bias :D
So, the step one here is to try and make a working magnetic tape that will work with my tape recorder. If there's a way to make new tape, this very method can be used to make floppies and magnetic drums.
The very first magnetic tapes used iron, which was replaced with iron rust Fe3O4 for Magnetophon, and shortly after replaced with another iron rust, Fe2O3. Let's start simple and see where will it get us?
The tape medium itself, since the 50s or so, was based on 3M Scotch product, but before that it was just paper and glue. I want to experiment with both, but I'll start with the simpler one: Scotch tape, 1/4 inch.
I wanted to make an iron tape first. Where do I get iron? A steel nail and a file!
Prototype 1
The tape is grainy and there's lots of gaps, but no more metal can be added to it.
The tape doesn't work, as expected.
What if I use existing magnetic paper from tickets?
Prototype 2.
Doesn't work either.
No matter what they use for magnetic paper, it's not good.
Prototype 3:
Iron grains were too big. Let's use fine iron powder!
Result: the tape is very heavy and responds to magnets a lot, but it doesn't work with the tape recorder.
Hypothesis: it needs both magnetic and paramagnetic material in the tape. Is it too magnetic?
Prototype 4
What is less magnetic then iron but still have somewhat similar properties? Rust!
Perhaps that was the reason why Magnetophon used it?
Let's get find a rusty pan and scrape some rust off. xD It's going to be unpredictable mess of Fe2O3, Fe3O4 and FeOOH and maybe more.
Does it work? No.
Does it look similar to magentic tape? Hell yes. You still can see the grainy texture and pretty brown colour
We already know from the web that many tapes use Fe2O3 (until it was replaced by CrO2 and more complicated stuff). Let's get some in a science lab shop, and figure out how to synthesize it later.
Prototype 5:
Applying Fe2O3 to scotch and paper tape. It is really fine rust, and the surface looks nice.
It looks JUST like a store bought tape
But it doesn't work?! 
Testing it with a magnet (get a magnet and see if it sticks and how much) shows that it is much less magnetic than regular store bought tape. Huh??! Huh!!!?!
Prototype 6
Let's try to add MORE rust. Glueing an extra layer of rust on top of rust made it slightly more magnetic, but it does not work, still.
A magnet in a glove and Fe2O3 rust: it is barely magnetic.
Do we have other options? Hmm...
Prototype 7:
What if we cut our Fe2O3 with some small amount of Fe? It will make the tape as magnetic as we want it.
I made a tape with increasing concentrations of Fe in the mix.
And, it's a miracle! Well, science.
We recorded some radio on the tape, and the signal/noise level seemed to be quite good. The volume level was quite low though 
The BIG question: 50 years ago, did the companies used some other sort of Fe2O3?
So, did the companies 50 years ago used some other sort of Fe2O3?
YES 
The common type of rust, alpha-Fe2O3, is actually not a great ferromagnetic! The type of rust used commercially for data storage has different crystalline structure and called gamma-fe2o3
It should be possible to convert some of a-Fe2O3 to g-Fe2O3 by heating. We might get some side results, though, like atomic Fe.
Nevertheless, why not try?
Another potential option is boiling Fe2O3 in water at 150C. Maybe we'll get to it later.
We made sure to make our material really hot. Maybe it was a bad idea, but...
Look at those magnetic properties, damn!!!
Prototype 9: the tape sort of works, it's very loud, but it has lots of large grains of the magnetic material that result in lots of added noise.
Looking at the fractions, Atsuko realised that there's lots of iron-looking grains in there. We probably overheated the material. If we can figure out how to make the particles finer, that could help, too.
And so the quest for Gamma-Fe2O3 begins! If it fails, we can always return to figuring out proportions between rust and iron xD Or, perhaps, rust and nickel - such a mix should also be ferromagnetic.
Some resources said that we should be able to get gamma-Fe2O3 after baking it with water at 100-150 °C.
The magnetic properties of the rust didn't seem to change (a tape test needed to verify though).
Next step, incidentally useful for just getting Fe2O3 rust:
Nail is attached to + anode, copper wire is - cathode, we run 1-2 amps of low voltage current through a broth made of NaCOOH•nH2O - which is an electrolyte that shouldn't react with Fe at all. The result should be a solution of yellow rust FeO(OH), known as ochre.
FeO(OH) heated to about 200°C should give us Fe2O3, but it's magnetic properties we won't know until we try.
The amount of FeOOH we got THIS time seemed to be insignificant. More research needed. But at least we got decent rust. xD
Perhaps @yottatsa could share more on this experiment.
@nina_kali_nina
Would love to hear the results, do you think you could post the audio?
@yottatsa
@stib @yottatsa I can make a recording, yes, but I am not sure if it is useful outside of any context.
Perhaps the most sensible way would be to get a test audio file (sine, voice, music), and compare commercial tape with home-made tape, so it'd be possible to analyse how much worse DIY tape is compared to commercial one. But the tape runs at 19 cm per second, and I have maybe 1 (one) m of the tape - enough to confirm that it works/doesn't work, but not enough to do long-ish tests.
So, I could either make a 3s sample, or try to prepare a few meters of the tape as imperfect as it is now, and make a longer-running audio test.
= 
log
) @dougmerritt@mathstodon.xyz@nina_kali_nina @stib @yottatsa Regarding "3s sample" -- people have "always" had some way to turn short tapes into loops (e.g. for special FX), somehow. I don't know how they do it, but it's at least a possibility.
@dougmerritt @stib @yottatsa fair enough~
@dougmerritt @nina_kali_nina @stib @yottatsa Yeah, this is super interesting for building a tape Echo as they were built in the 70ies. There was a tape loop or a magnetic disc to produce a delayed version of the original signal. The Echoplex and the Binson Echorec were examples.