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
andlabs has joined ##yamahasynths
ej5 has joined ##yamahasynths
<cr1901_modern>
ej5: If you are taking requests, play this theme on your new Amiga :P
<TD-Linux>
fortunately the most common psu failure won't damage anything. the primary side transistor fails short, which cooks itself, the primary side transformer winding, and eventually blows the mains fuse
<fseidel>
guess I'll get a PCB fabbed, this seems like a no-brainer
<fseidel>
for safety/longevity
andlabs has quit [Ping timeout: 258 seconds]
andlabs has joined ##yamahasynths
andlabs has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<ej5>
(it has a synth track so technically it's relevant)
<TD-Linux>
ej5, do you think dremel or hacksaw is more appropriate for a nicd battery
<ej5>
none of the above
l_oliveira has quit [Quit: ChatZilla 0.9.92-rdmsoft [XULRunner 35.0.1/20150122214805]]
nukeykt has joined ##yamahasynths
cr1901_modern has quit [Ping timeout: 250 seconds]
cr1901_modern has joined ##yamahasynths
<whitequark>
lol
glowcoil has quit [Read error: Connection reset by peer]
glowcoil has joined ##yamahasynths
l_oliveira has joined ##yamahasynths
<cr1901_modern>
ej5: More sampled music that's good (this is without the SNES compression- someone located the actual hardware and rerouted the music driver's commands and rerouted it to the hardware): https://www.youtube.com/watch?v=-5rAjOjTGtc
<cr1901_modern>
ej5: Also, I looked into (a long time ago- 2013-ish) how one would convert FM synth audio capture from the Genesis to samples one could use for the SNES. I was told the SNES compression ruined any attempts to do that. But the Amiga seems fine generating FM synth samples that sound good (no compression on Paula/whatever the sound chip is called?)
futarisIRCcloud has quit [Quit: Connection closed for inactivity]
<cr1901_modern>
nukeykt: I don't have much time to elaborate today, but I think I just realized something important re: envelope generator attenuation
<cr1901_modern>
The YM2608 manual lists the maximum attenuation as 96 dB, but experiments show it's actually 48dB
<cr1901_modern>
is this solely because when operator output is fed into another input, the LSB is discarded?
<cr1901_modern>
And if so, wouldn't this mean that the maximum feedback possible for operator 1 is 2*pi and not 4*pi (because again, LSB is discarded)
<nukeykt>
More like typo in manuals i think(it's stated to be 96db in OPL manual as well), but in reality it's 48db yeah. No LSB is dropped, sine LUT has higher precision than EG output
<nukeykt>
Maximum FB is 4*pi, this is correct in manuals
<cr1901_modern>
I meant the LSB is dropped when the operator output is fed into the input of a new operator
<cr1901_modern>
that would therefore halve the effect of everything
<nukeykt>
Isn't halving by 2 is equal adding 3db?
<cr1901_modern>
wait... no, that's now how decibels work ._.
<cr1901_modern>
6db
<nukeykt>
ah, ok
<cr1901_modern>
or 3db
<cr1901_modern>
nukeykt: It's 6dB... the 3dB comes from bode plots if you've ever done one of those
<cr1901_modern>
3dB is cutoff frequency b/c it's half the power
<cr1901_modern>
Anyways I'm not feeling well today, so I'll elaborate next week. But I've been trying to duplicate the effects of operator feedback in my own experiements and I can't get the waveforms to look right
<cr1901_modern>
This was just me minddumping before I forget :P
<nukeykt>
Hm, if we assume it as 6db then yamaha manual is correct
<cr1901_modern>
The convention for dB is a log10 ratio... multiplied by either 10 or 20. I almost _never_ see 10*log10.
<cr1901_modern>
20*log10 is appropriate for comparing two "voltages", or magnitudes
<cr1901_modern>
nukeykt: Can you give me the magnitude of the smallest value that the envelope generator generates and the largest value?
andlabs has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<cr1901_modern>
My personal opinion right now is that "halving a signal is in fact 6dB of attenuation, and the maximum attenuation is 48dB" :)
<nukeykt>
You mean in operator output magnitude? like if sine lut attenuation level is 0?
<cr1901_modern>
yes
<cr1901_modern>
I want the largest and smallest values
<nukeykt>
For OPN it ranges from 8168 to 0
<cr1901_modern>
smallest positive* value
<nukeykt>
It would be 1 then, no?
<l_oliveira>
nukeykt, you free right now?
<cr1901_modern>
oh probably... I was just curious
<nukeykt>
l_oliveira, yeah
<cr1901_modern>
nukeykt: Anyways let me think about this... decibels are _by convention_ typically ratios of _power_
<l_oliveira>
it's a question regarding the VRC7 testmode, it is drawing a lot more current than the usual
<cr1901_modern>
if they are ratios of power, the 96dB in the manual makes sense. But Yamaha might've represented decibels as a ratio of magnitude
andlabs has joined ##yamahasynths
<l_oliveira>
which pin enables it's output?
<l_oliveira>
I think it might be overloading my bus here
<l_oliveira>
I need it to be locked into write only mode
<nukeykt>
Pin 2 and pin 47 IIRC
<cr1901_modern>
P = V^2/R log10(P) = 2*log10(V/R)
<cr1901_modern>
(in bels, multiply by 10 to get decibels)
<l_oliveira>
which way I need to set them to make it write only?
<nukeykt>
Let me check
<l_oliveira>
the computer I am using is not crashing only because the CPU is buffered
<l_oliveira>
if I bypass the buffer (for testing) the computer malfunctions
<l_oliveira>
that does not happen when the VRC7 is outside of testmode
<nukeykt>
Hm, hard to tell. What's is current state of these pins?
andlabs has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<l_oliveira>
lemme look
<nukeykt>
Probably you can check this experimentally, only 4 combinations
<l_oliveira>
2 =GND 47 = +5v
<nukeykt>
ok, thanks
<cr1901_modern>
Wohali: Paid for my Snoot Boppers :D
andlabs has joined ##yamahasynths
<nukeykt>
l_oliveira: I dunno, try to set pin 2 to +5v
<ej5>
i was aware of open compute but didn't realize they have the designs up for anyone to download.
<whitequark>
that video series rocks
<whitequark>
i don't like the cargo cult parts of his approach but i acknowledge that the way he applies those are eminently practical
<whitequark>
and he explicitly calls it out too, cf "you see people do this for high-speed signals on a server motherboard, so if you want something similar to work, you can either just copy it or spend weeks simulating it to understand it better, your choice"
<ej5>
oh haven't gotten to that part, but typically those are encoded as design rules, like diff pair spacing to avoid crosstalk, etc
<whitequark>
no, it's not just that
<whitequark>
it's things like stitching vias, or zigzagging to account for inhomogenity of pcb substrate
<whitequark>
also a lot of things that are NOT done go into that category
<ej5>
oh i see, basically doing things without understanding why
<whitequark>
yes
<whitequark>
or rather, incomplete understanding
<whitequark>
i know what zigzagging is in this context, but do i have any tools to evaluate if my board needs it? not really
<ej5>
one of the things that always bugged me about schematics of this complexity level is that a lot of the pages (like DDR4 channel wiring) is just a big table of pins to nets which really ought to be done using a spreadsheet, or even just a script
<Wohali>
i can't handle watching youtuve videos for that stuff. gimme a blogpost or a paper or slides or something
<Wohali>
too goddamn slow
<ej5>
yeah this guy talks slow too, time for 1.5x or 2x XD
nukeykt has quit [Quit: Page closed]
<Wohali>
yeah, my only designs with really complex stuff like that were all done using expensive, professional pcb layout tools
<Wohali>
and they give you the tools to evaluate if your board needs those things, mostly.