ChanServ changed the topic of ##yamahasynths to: Channel dedicated to questions and discussion of Yamaha FM Synthesizer internals and corresponding REing. Discussion of synthesis methods similar to the Yamaha line of chips, Sound Blasters + clones, PCM chips like RF5C68, and CD theory of operation are also on-topic. Channel logs: https://freenode.irclog.whitequark.org/~h~yamahasynths
futarisIRCcloud has quit [Quit: Connection closed for inactivity]
racoon has left ##yamahasynths [##yamahasynths]
ugla has quit [Ping timeout: 265 seconds]
ugla has joined ##yamahasynths
andlabs has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<Foone>
great. now we just need to figure out what title generator they used
<cr1901_modern>
Maybe a Video Toaster?
<Foone>
supposedly the ad is from 1990, which predates the video toaster
<cr1901_modern>
Oh, I... didn't know that. lol
emily has joined ##yamahasynths
andlabs has quit [Ping timeout: 264 seconds]
andlabs has joined ##yamahasynths
<whitequark>
cr1901_modern: hey i just realized something
<whitequark>
remember when we talked about higher data densities on unmodified floppy drive media?
<whitequark>
and you (maybe someone else) mentioned that e.g. SMR would be infeasible because the erase head is wider than the read/write head
<whitequark>
i just realized that this entire line of argument is dumb.
<whitequark>
the *reason* the erase head is wider than the read/write head is that the floppy doesn't have a servo! so you use it to prevent previously written data from becoming a coherent noise source, which would tank your BER
<whitequark>
basically it adds guard bands
<whitequark>
but. if we're doing anything clever with a floppy, involving an increase of tpi, we're generally OK with an increase in crosstalk... because we're handling it in some other way, or ways. maybe we're using a servo system, or data whitening, or both
<whitequark>
this means that we can, simply, just not use the erase head at all
<cr1901_modern>
I'm a bit out of it tonight, but sounds reasonable. I think maybe implicit in me saying "SMR can't be done" was "the erase head is on while a write is taking place (to add the guard bands)".
<cr1901_modern>
But if just don't use the erase head (maybe mod the drive?), then more opportunities for fun open up.
<cr1901_modern>
Also, yes you'd need something better than a stepper motor to position the head finely, unless you come up w/ a mechanism to knock the drive out of alignment each time you move the head :).
<whitequark>
i mean you could do heavy microstepping
<cr1901_modern>
... oh... TIL (I haven't played much w/ steppers over the years :P)
<whitequark>
you could go even further
<whitequark>
you could treat the stepper as a two phase BLDC motor and have effectively infinitely fine positioning in the ideal case
<Sarayan>
Is the write head powerful enough to overwrite if the erase head didn't clean things up beforehand?
<whitequark>
loaded question!
<whitequark>
obviously the write head can coerce the medium to flip to either orientation, since the erase head just has dc current applied to it
<whitequark>
past that we need to get into the question of what "erase" even means
<Sarayan>
true
<whitequark>
for example, it turns out that unless you specifically avoid it, your write head will have side fringing effects that do a small amount of trim erasing on its won
<whitequark>
*own
<whitequark>
which is sad, because it means no SMR for me :(
<Sarayan>
I was just thinking that going neutral -> direction probably required less power than one direction -> the opposite one
<whitequark>
well... probably no SMR
<whitequark>
there's no "neutral" is it?
<Sarayan>
there is. degaussed/unmagnetized is neutral
<whitequark>
yeah but (looking at shugart 460) that's not what the erase heads do
<whitequark>
actually hold on
<whitequark>
on shugart 860 and shugart 460, the erase heads don't even erase the data about to be written
<whitequark>
they're trim erase heads
<Sarayan>
Perhaps it's not, I'm reaching the limits of what I know and jumping into what I guess, which is not good :-)
<whitequark>
those are 8" and 5 1/4" drives though
<whitequark>
i'm not sure if 3 1/2" drives still use trim erasing
<whitequark>
who did the earliest 3 1/2" work? shugart?
<Sarayan>
Wasn't it sony?
<cr1901_modern>
I don't know, I'm afraid
<Sarayan>
somebody from our world will need at some point to do a systematic study of flux migration though if we ever want reliable floppy write capability
<cr1901_modern>
Idk what they mean by a "yoke" (analogous to CRTs?)
<whitequark>
i also had another crazy idea
<whitequark>
so the trim erase head in 460 is constructed somewhat similarly to the read/write head
<whitequark>
what if i used the trim erase head for servo positioning?
<whitequark>
this is where things *really* start to depend on construction of specific heads, and so far i don't know how 3.5" ones are made
<ej5>
theoretically you could just use the analog output from the read head for that
<ej5>
it's the old "balance a ball on top of a hill" controls problem
<ej5>
then to figure out what track you're on, you use the data encoded in the sector header
<whitequark>
ej5: my thinking is that the read head gives me error, but i want error and direction
<ej5>
the tunnel erase heads won't give you direction, they're just wired in series internally iirc
<whitequark>
yeah i've just discovered that
<ej5>
btw there's a trick for tracking with just a single head
<Sarayan>
oh, in perpenticular recording the erase head is in front of the write head
<Sarayan>
(the 2.88M floppies)
<ej5>
you purposefully introduce a little bit of wobble into the head positioning signal
<ej5>
tracking the amplitudes at each endpoint so you can tell how well centered you are
<ej5>
so you go a tiny bit inward, watch the amplitude drop, then go a tiny bit outward, watch the amplitude go up and then drop again
<whitequark>
hmm
<ej5>
and then if the amplitude is higher at one side, then you're too far off in that direction, so you correct
<whitequark>
wait is this effectively the same as the compact disk servo tracking system, just with a... time multiplexed detecting element?
<ej5>
basically yes. you have to turn the system off during a write
<whitequark>
that is damned clever
<ej5>
but you might be able to interpolate from an eccentricity previously measured
<ej5>
this trick is often used for laser tracking systems, but they do it in 2 dimensions
<whitequark>
yeah, or introduce my own guard bands between multi track zones and rewrite zone at a time
<whitequark>
... so
<whitequark>
ej5: do you know the detailed construction of the HD head?
<whitequark>
is it still just tunnel erase?
<ej5>
high density or hard disk?
<whitequark>
former
<ej5>
not really, you might be able to dig up some patents though
<whitequark>
so i'm not sure which keywords or even companies to look for
<cr1901_modern>
I could personally try harder to find 3.5" literature. Probably some stuff on bitsavers I haven't found yet. I found the 5.25" stuff b/c vcfed member Chuck(G) pointed me to it.
<ej5>
sony or teac probably have patents
<cr1901_modern>
whitequark: 1.2MB 5.25" counts as high density for you?
<whitequark>
ej5: ohh fascinating. so still tunnel erase
<ej5>
yeah their innovation is to add a third head in the mix for doing low density vs high density recording
<ej5>
dunno if that was ever used
<ej5>
but it does give you an idea of the geometry of the erase head
<ej5>
US4644435A is a classic, wow
<cr1901_modern>
The IBM PC/AT floppy drive- high density- was "YE-DATA YD-380"
<cr1901_modern>
Never heard of this company
<whitequark>
ej5: ... according to US4803571A, it might be actually possible to do something vaguely SMR-like by writing with the erase head
<whitequark>
since, if you stagger the tracks written by that head so as to avoid side fringe crosstalk, you could write much narrower tracks than with the normal r/w head
<whitequark>
that assumes that the erase head is otherwise similar enough to the read/write head, of course
<ej5>
but it would write two tracks at once
<whitequark>
yeah, that's the SMR-like part
<whitequark>
you have to rewrite an entire zone at once
<ej5>
going afk for a few hours. lmk if you come up with other interesting floppy drive hacks
<andlabs>
seems like the answer is no, looking at the nibtools source
<andlabs>
but I'm not sure
<andlabs>
ZrX-NoMs: what is this
<ZrX-NoMs>
something I put together around christmas knowing there were few hundred amiga disks coming that need to be read without too much fuzz.
<andlabs>
cool
<ZrX-NoMs>
ended up working such that insert disk, gets read with retries to sd card, then insert next and it repeats without any other user intervention.
<andlabs>
ooh
<andlabs>
that would be convenient though I hope you don't get confused labellinga ll those files =P
<ZrX-NoMs>
00001_nd.adf, 00002_d0.adf, 00002ad0.adf, 00003_d0.adf was very descriptive. need a better fat library next time. :)
<Sarayan>
andlabs: write probably needs a better model of flux migration than what we have
<andlabs>
oh not writing flux
<andlabs>
writing images
<andlabs>
specifically I need to write my own amiga 3 disks because the one I got with my a1000 are broken
<cr1901_modern>
whitequark: Cool patent. It's older stuff, but the author of that patent has a bunch of cool stuff that may be relevant: https://patents.google.com/?inventor=Frank+J.+Sordello (Magnetic track following servo system, Sectorized data path following servo system)