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 joined ##yamahasynths
<mofh>
Lord_Nightmare: even monophase KBD-windowed sinc() is pretty decent
<cr1901_modern>
Wonder where nukeykt has been
andlabs has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
andlabs has joined ##yamahasynths
andlabs has quit [Ping timeout: 250 seconds]
andlabs has joined ##yamahasynths
ahihi has quit [Read error: Connection reset by peer]
ahihi has joined ##yamahasynths
andlabs has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
andlabs has joined ##yamahasynths
ej5 has quit [Read error: Connection reset by peer]
<whitequark>
is it me or does opl2/opl3 output need an equalizer to sound well through a modern system?
<whitequark>
i feel like what i get isn't quite what the composer intended, but i may be wrong
<cr1901_modern>
different LPF cutoff?
<cr1901_modern>
compared to sound cards of the era
<superctr>
the OPL3-SA2 in my old laptop has a very bright sound, i don't mind it
<ValleyBell>
The OPL3 on my YMF744 sounds a lot like the (unfiltered) emulation. The SB16 in my old PC is pretty bass-heavy though and cuts off a fair amount of heights.
<whitequark>
second, of the non-percussion tones, mid-frequency often (but not always) tends to drown out higher frequencies
<whitequark>
and that feels unintende
<Sarayan>
oh, whitequark
<Sarayan>
if I send you a votrax sc01a, would you put it on the internet?
<whitequark>
how is it driven, and what is on the output?
<Sarayan>
caveats: it's an hybrid digital/analog chip, with analog output, so you'd have toplonk an adc in there
<whitequark>
>Votrax International, Inc. (originally the Vocal division of Federal Screw Works)
<whitequark>
i'm sorry, federal WHAT works?
<Sarayan>
the voltages may be annoying, but it may only be 0/=(V, lemme check
<Sarayan>
+9V
<whitequark>
uhh, that's mildly irritating
<Sarayan>
og.kervella.org/VotraxSC01.pdf
<Sarayan>
also, it's a chip that's not made anymore, and rather static-sensitive
<cr1901_modern>
what a strange voltage
<Sarayan>
metal-gate cmos
<Sarayan>
so it's not rare, but it's not common either
<cr1901_modern>
>SSI told us then that due to the types of processes used in this analog chip, it was unlikely to be made again--too complicated to transfer the technology
<Sarayan>
we already have die shots, vectorization and schematics though
<cr1901_modern>
>(However, we never had really good communication with SSI, so take this with a grain of salt.)
<Sarayan>
need to switch it to leaflet or groupxiv
<whitequark>
that's tiny
<cr1901_modern>
>All transitions between phonemes are handled automatically.
<cr1901_modern>
The hell does that mean?
<whitequark>
Sarayan: so ideally, could you put it on a PCB with all the support circuitry and whatever SPI ADC you have in the junk pile?
<Sarayan>
wq: that would be the idea
<whitequark>
it always takes me a long time to obtain components because of the whole "too much anxiety to go outside" thing
<whitequark>
if i got something that i just need to write a glasgow applet for, it's trivial
<whitequark>
ideally make 9V on board from 5V
<Sarayan>
cr1901: all the characteristics of the analog part are controlled by a dozen 3-5bits values which activate or not a number of capacitors. The digital circuitry does interpolation of these values when transiting from one phone to another
<Sarayan>
wq: it uses some current even when idle because it makes the internal 5V from a resistor ladder on the 9V
<whitequark>
why does it even need 9V
<Sarayan>
for the analog
<whitequark>
ah
<whitequark>
somehow i see a christmas tree of glasgows connected to a huge usb hub in my life
<whitequark>
but that's okay, it was designed to do that
<Sarayan>
wait, I'm not sure it's 9V
<cr1901_modern>
12V would be more common I'd think
<cr1901_modern>
based on a sample size of 1: the SID
<whitequark>
the SID is 12V?!
<cr1901_modern>
it's dual 12V and 5V
<Sarayan>
ok, it can be anything you want from 9V to 14V
<Sarayan>
see page 8
<whitequark>
Sarayan: so you know the Glasgow IDC connector?
<cr1901_modern>
I'm guessing that b/c SID is NMOS, the "resistor ladder trick" doesn't give enough power, so 12V is used for the analog frontend
<whitequark>
Sarayan: so I'm thinking you'd bring out MCX, STB, AR from this device, then DSCK, DSI, DL for the shift register, and ASCK, ASO for the ADC
<whitequark>
the idea being that the chip inputs are connected to a 74xx shift register that latches on DL
<whitequark>
chip parallel inputs*
<whitequark>
and chip audio output is connected to a serial ADC.
<whitequark>
or wait
<Sarayan>
beware that the I inputs are realtime
<whitequark>
aha
<whitequark>
yep, I was about to say that DL is not needed
<whitequark>
but it is
<whitequark>
because having explicit input latch avoids glitches on I input.
<Sarayan>
the 6 phone numbers input are taken on stb, but I has an immediate effect
<whitequark>
yeah, but it's easier to just wire them all together to a 8-bit shift register
<Sarayan>
agreed
<Sarayan>
you need to 720KHz clock somewhere
<whitequark>
because then we fit into 8 pins and it becomes a nice one-port daughterboard for glasgow
<whitequark>
sure, glasgow generates all kinds of clocks
<Sarayan>
oh, that's MCX
<whitequark>
yep
<Sarayan>
you've taken it into account, cool
<Sarayan>
so, the amusing question, is it possible to send something to you form france so that it doesn't get lost on the way?
<whitequark>
yes, use either registered mail or EMS
<whitequark>
EMS is faster but a bit more expensive
<whitequark>
anything with a tracking number is not generally going to get lost
<whitequark>
EMS is going to get extra not lost
<whitequark>
so for a rare chip EMS is worth it
<Sarayan>
ok
<whitequark>
they also tend to not smash EMS packages with something heavy on top
<whitequark>
or at least do it less
<Sarayan>
oh, it's chronopost in france, so it's going to be easy
<whitequark>
yep
<whitequark>
the cool thing about having an IDC socket there is it prevents me accidentally miswiring it
<whitequark>
if this chip had any outputs I'd also ask you to add 200 ohm or whatever series resistors
<whitequark>
I do this if I'm worried the FPGA might misbehave and drive an output
<whitequark>
(they've actually saved my GameBoy Color once when I loaded a bitstream for a completely wrong Glasgow revision manually. doh...)
<Sarayan>
it has one output
<whitequark>
oh yeah
<whitequark>
AR
<Sarayan>
yup
<whitequark>
put a series resistor there :)
<whitequark>
oh, it has a 0.5 mA driver
<whitequark>
so a 10k resistor then? to make it always within a safe margin.
<Sarayan>
I have no idea, I suck at electronics :-)
<whitequark>
10k works
<whitequark>
not really a question
<whitequark>
... how does one reverse engineer multiple ICs while "sucking at electronics" I would like to know.
<whitequark>
a level of suck i can only aspire to
<Sarayan>
because analyzing and designing is different
<Sarayan>
for instance I know decoupling caps exist, in the theory I kinda know what they're useful for, but I wouldn't have a clue where to put them of with which value
<Sarayan>
I also have no clue about EMI, I just know it exists
<whitequark>
usually, i read the datasheet, which is quite explicit on where to put with what value ;D
<whitequark>
or some kind of application note
<whitequark>
semiconductor vendors (other than yamaha) tend to write profusely on the topic of running their circuits just right
<Sarayan>
IOW, I can understand what something works, but I couldn't create it
<whitequark>
yeah, i get it
<Sarayan>
s/what/how/
<Sarayan>
while in software I can create anything, but that's another story :-)
<whitequark>
i'm guessing you will want suggestions on what ADC to use then
<whitequark>
i assume you'll figure out how to wire a shift register
<Sarayan>
actually I'm thinking sending you the chip and letting you do your magic :-)
<whitequark>
eh, i guess that works
<whitequark>
i guess i'll order parts online for this particular project to make it faster
<Sarayan>
oh, there is no hurry
<whitequark>
well for one I want to hear it
<Sarayan>
:-)
<Sarayan>
that I get
<whitequark>
mmyeah, should be easy enough
<whitequark>
not entirely sure how to couple it to the ADC, but it'd doable
<whitequark>
i think i need to bias the ADC input so the voltage always stays nonnegative
<Sarayan>
the votrax output is always nonnegative
<Sarayan>
it is dc-biased already
<whitequark>
why is there an LPF on the output?
<whitequark>
figure 8
<whitequark>
there's a dc blocking cap and an LPF
<whitequark>
to be clear, i could just connect the ADC to Ao
<Sarayan>
figure 8 is not a lpf, it's a rc osicllator for the clock?
<whitequark>
4.7k/0.05uf
<whitequark>
to the right of rc oscillator
<Sarayan>
figure 8 single message system?
<whitequark>
er
<whitequark>
there are two figure 8 s in this datasheet
<whitequark>
figure 8 on page 10.
<whitequark>
who the fuck writes these things.
<Sarayan>
the second figure 8 seems to be a note added by someone
<whitequark>
there are also two figure 6 s, both printed...
<Sarayan>
huhuhu
<Sarayan>
american quality
<Sarayan>
hey, both figure 6 are identical
<whitequark>
oh I see
<Sarayan>
same for 7
<whitequark>
anyway there is an LPF
<whitequark>
i'm not sure if it's necessary for correct operation or not
<Sarayan>
no, it's not
<whitequark>
so, i would just sample the output?
<Sarayan>
and I'd rather not have it if possible
<Sarayan>
yes
<whitequark>
oh
<whitequark>
how fast?
<whitequark>
sample rate
<Sarayan>
boh, the usual 44100?
<Sarayan>
it's not actually producing anything over 4K or so deliberately
<Sarayan>
but it's analog, so it can
<whitequark>
16 bit 48000?
<whitequark>
yikes, my local electronics retailer is fucking greedy
<Sarayan>
ouch, I'd have thought a normal audio range adc wouldn't be expensive
<whitequark>
they are just greedy period
<Sarayan>
heh
<whitequark>
a few years ago an acquaintance caught them scraping the entire farnell catalog (this is very prohibited) and reselling it all
<whitequark>
with precisely a 3x markup
<Sarayan>
ouch
<whitequark>
unfortunately they are also the least stressful option in many ways
<whitequark>
and the only ones who maintain any sort of physical presence anymore that isn't a booth at a flea market
<Sarayan>
ah yeah
<whitequark>
Sarayan: so there are cheap ADCs
<whitequark>
but they are all 8 or 10 bit
<Sarayan>
otoh, finding a physical electronics shop in *Paris* is almost impossible. I know of *one*
<whitequark>
there are cheap 16 bit ADCs... but they have a 16 Hz sample
<whitequark>
rate (not a typo)
<Sarayan>
that's ridiculous
<whitequark>
it's a low noise instrumentation ADC
<whitequark>
no idea why is it so cheap
<whitequark>
500 Hz...
<whitequark>
this is silly
<whitequark>
aha
<whitequark>
ADS8320EB
<Sarayan>
TI has some nice ADCs iirc
<whitequark>
yes but I went to the retailer and sorted by price
nukeykt has joined ##yamahasynths
<cr1901_modern>
Well sleeping was a complete disaster right now. Guess I'll... get ready for the day only to fall asleep in 2 hours
<cr1901_modern>
true, and I think I have breadboard adapters for those too.
<cr1901_modern>
Lord_Nightmare: Is the "B" marking in whitequark's link for SSOP or "Revision B" or something else? It's not on the wiki
<cr1901_modern>
I anticipate a client will want me to make a MIDI application. Dunno if I can shoehorn a Yamaha chip in, but if I can, I figure I'd try out a chip that makes it easier to create instruments :)
<cr1901_modern>
(and by that I mean, "I don't have to create instruments at all" :D)
<nukeykt>
I think YM2413B is a CMOS variant of OPLL
<cr1901_modern>
oh right, I think that brought up
<whitequark>
hm
<whitequark>
still a builtin DA converter :S
<cr1901_modern>
time-division multiplexed at that
<whitequark>
wait, does it need some filter on the output?
<cr1901_modern>
IIRC yes, because time-division multiplexing has lovely high frequency content that needs to be suppressed
<whitequark>
ugh
<whitequark>
wait
<whitequark>
why does it have TWO outputs?
<whitequark>
is this so you can easily bass boost the rhytm instruments?
<cr1901_modern>
can you link me the datasheet?
<whitequark>
it's in my library
<cr1901_modern>
ahh, right I have it then
<whitequark>
wait, it uses an XTAL?
<whitequark>
i guess it's just an inverter between those pins, like usual
<whitequark>
and you can feed something to XI
<cr1901_modern>
opl3 (sic) has 2 outputs as well, for up to 4 channels of music
<whitequark>
opl3 has digital output
<whitequark>
and you can assign operators to channels somewhat arbitrarily
<whitequark>
opll just has rhytm and melody.
<cr1901_modern>
what do you mean by "inverter between those two pins"?
<whitequark>
the usual way an XTAL interface is implemented is as an inverter from XI to XO, with clock taken from XI
<whitequark>
it works like an inverter connected with feedback to itself, but crystal-stabilized
<whitequark>
so you can just ignore XO if you don't need it
<whitequark>
and feed a clock directly into XI
<whitequark>
like if you look up "xtal driver" or something on mouser it's literally just an inverter that has been specced for driving an xtal
<cr1901_modern>
cool
<cr1901_modern>
I was never good w/ crystals/oscillators
<cr1901_modern>
I love figure IV-2 in the application manual. It seems that Mouser doesn't carry passive components corresponding "plus sign".
<cr1901_modern>
I think it's supposed to represent a direct connection, but IME "summation" means "stick a summing op amp there".
<whitequark>
i'll just do it in digital domain.
<whitequark>
absolutely can't bother to figure out what they meant
<cr1901_modern>
the passive components attached directly to Ro and Mo are also filters; I don't feel like calculating the cutoff. It's probably meant to deal with TDM artifacts.
<cr1901_modern>
ValleyBell: Do you know anyone who has done pictures of the Master System FM addon and/or posted a schematic?
<cr1901_modern>
(Also, that schematic is hard to read)
<whitequark>
cr1901_modern: lol what
<cr1901_modern>
whitequark: It took me a good two minutes to realize that network attached to Ro/Mo in the schematic is the same as presented in the application manual (except w/ the capacitor shared between both Ro/Mo)
<cr1901_modern>
I like my circuits running from left to right, thanks
<whitequark>
ETIMEDOUT for me
<whitequark>
seems like they blocked ru via ip range
<ValleyBell>
cr1901_modern: For anything related to SMS stuff, looking on SMSPower should be a good idea: http://www.smspower.org/forums/
<ValleyBell>
especially the "Development" forum and wiki sections
futarisIRCcloud has quit [Quit: Connection closed for inactivity]
andlabs has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<Sarayan>
yay, finished the pads
<Sarayan>
I need to do something about the die image scaling though
andlabs has joined ##yamahasynths
nukeykt has quit [Quit: Page closed]
mz` has quit [Ping timeout: 245 seconds]
mz` has joined ##yamahasynths
<Lord_Nightmare>
cr1901_modern: no
<Lord_Nightmare>
there are two different products: the ym2413 and ym2413B
<Lord_Nightmare>
afaik both have the same patch set
<Lord_Nightmare>
but ym2413B is CMOS, not nmos, and may be the 2-metal layer version we know as FHB013
<Lord_Nightmare>
the output impedance is different because it is cmos instead of nmos, i believe
<Lord_Nightmare>
yamaha calls ym2413B OPLL, they should have called it OPLLC
<Lord_Nightmare>
and should have given it a ym3xxx number
<Lord_Nightmare>
but someone didn't get the memo about consistency, so...
<Lord_Nightmare>
re: above
<Lord_Nightmare>
I'm NOT SURE i'm correct that ym2413B is cmos
<Lord_Nightmare>
but given the fact that it draws so much less current than ym2413, i'm inclined to say it is
<whitequark>
OPLLC or OPLL-C
<Lord_Nightmare>
so whitequark if you get a ym2413B off ebay or alibaba etc, it probably has the same die as FHB013 on siliconpr0n but may have a different instrument set masked
<Lord_Nightmare>
iirc the FHB013 that digshadow decapped like 6 years ago was dead, so we don't know which inst set that one has
<Lord_Nightmare>
it was also re-marked with a bogus label
<whitequark>
Lord_Nightmare: we'll see soon enough
<cr1901_modern>
https://segaretro.org/Copera Interesting, a mystery console that's not emulated. Is the FM chip also a mystery?
<Nerionaya>
huh, I'd never heard of that before; interesting