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
Xyz_39808 has quit [Ping timeout: 276 seconds]
<cr1901_modern> I need the 8051 module for the PB-10
Xyz_39808 has joined ##yamahasynths
andlabs has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
andlabs has joined ##yamahasynths
andlabs has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
SceneCAT has quit [Ping timeout: 272 seconds]
andlabs has joined ##yamahasynths
<Lord_Nightmare> Foone_: that sound card with the 2716 chips on it: make sure you dump them as intel 2716; tms2716 are 3-voltage parts and will blow up a jedec (intel) 2716
<Lord_Nightmare> if you dump as tms2716
<Lord_Nightmare> ti did a stupid back in the day and we're still paying for it today in blown up eproms
<Lord_Nightmare> this is assuming the soviet chips are clones of the jedec 2716
<Lord_Nightmare> also you may have luck dumping the flaky eproms with the voltage slightly too low
<Lord_Nightmare> like 4.95v instead of 5v
<Lord_Nightmare> talk to kevtris about it
<ej5> i figured that might work.
<Lord_Nightmare> iirc the voltage lowering trick works with bit-rotten eproms, but theres a point of diminishing returns where the eprom starts to do weird stuff from too low a voltage
<Lord_Nightmare> it won't erase it
<Lord_Nightmare> but i think it will stop working or start spitting out garbage/wrong address data or something
<Lord_Nightmare> at some point
<Lord_Nightmare> so you can try 4.99 4.98 down to 4.85 or so and do 10x dumps at each voltage
<Lord_Nightmare> and find the electrical 'sweet spot'
<Lord_Nightmare> then do 100x dumps at that voltage and binary AND them together
<Lord_Nightmare> since a jedec eprom reads as 0xFF if it is erased
<Lord_Nightmare> though this being a soviet clone, they may have not followed that part of the spec...
cr1901_modern1 has joined ##yamahasynths
cr1901_modern has quit [Ping timeout: 248 seconds]
ej5 has quit [Quit: Leaving]
_whitelogger has joined ##yamahasynths
kode542 has joined ##yamahasynths
kode54 has quit [Ping timeout: 245 seconds]
Xyz_39808 has quit [Ping timeout: 276 seconds]
SceneCAT has joined ##yamahasynths
cr1901_modern1 has quit [Quit: Leaving.]
cr1901_modern has joined ##yamahasynths
Stilett0 has quit [Quit: leaving]
Stilett0 has joined ##yamahasynths
<andlabs> [22:02:23] <Lord_Nightmare>ti did a stupid back in the day and we're still paying for it today in blown up eproms
<andlabs> tempted to say "ti did a stupid back in the day and we're still paying for it today (end of statmenet)"
andlabs has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
andlabs has joined ##yamahasynths
andlabs has quit [Client Quit]
andlabs has joined ##yamahasynths
andlabs has quit [Ping timeout: 258 seconds]
andlabs has joined ##yamahasynths
<cr1901_modern> Foone_: Feel free to publicly smack me :P
Xyz_39808 has joined ##yamahasynths
<cr1901_modern> I usually like my ISA cards with the thick gold fingers, but this is a good idea too: https://twitter.com/Foone/status/1153718213449904128
<TD-Linux> Foone_, so I also thought of doing this but last time I mentioned it someone said that the io ranges exposed via lpc on that header are restricted usually
<TD-Linux> aha one hour ahead of me
<TD-Linux> for maximum shitposting, power9 also has a lpc bus
<cr1901_modern> Still a fun way to get clones of old hardware (or an adapter) working on new
<TD-Linux> now that I know it's programmable I'm more inclined to make an adapter
<TD-Linux> I could make a c-bus adapter instead
<TD-Linux> then you could plug in an OPNA :)
<cr1901_modern> I like the idea of a "Boneless LPC" (discussed somewhere in the chat logs) too. Actually, I understand "Boneless LPC" more than I do actual LPC :P
<whitequark> why didn't we use actual LPC anyway?
<whitequark> >01:45 <whitequark> it's like LPC but without all the stupid ass parts
<whitequark> oh. right.
<cr1901_modern> There was some extra functionality you didn't like about LPC AIUI.
<whitequark> because LPC has a lot of stupid ass parts
<cr1901_modern> Ahhh :)
<whitequark> but honestly we could make it a firmware change
<cr1901_modern> Probably a cheap bridge hiding in there too
<whitequark> hey cr1901_modern TD-Linux Foone_ i'd be pretty interested in the LPC<>ISA bridge for glasgow purposes
* cr1901_modern will try to convince Foone_ to fix his ssh issues
<whitequark> ... actually
<TD-Linux> are the stupid parts for legacy PC reasons or just general stupidity
<whitequark> aren't those actual chips people use
<whitequark> TD-Linux: like "firmware hub"
<whitequark> oh
<whitequark> ohhhhhh
<whitequark> cr1901_modern: i'm a dumbass
<cr1901_modern> ?
<whitequark> you can just get an LPC<>ISA bridge chip at arrow, for 7 euro https://www.arrow.com/en/products/w83627uhg/nuvoton-technology
<TD-Linux> whitequark, aha that's probably the chip on those boards
<whitequark> wait
<whitequark> oh wait that's a multi io controller
<whitequark> i mean it would work but is an overkill
<TD-Linux> yeah most nuvoton stuff is superio
<cr1901_modern> I have seen this chip used tho
<cr1901_modern> on a homebrew 65xx design
<cr1901_modern> to get "easy" floppy and parallell I/O
<whitequark> lol
<whitequark> ok so i actually want F85226 but they don't exist anymore
<TD-Linux> I would be willing to make a pcb for this
<TD-Linux> the superio garbage is probably not so bad as it's in address ranges that peripheral cards wouldn't use anyway
<whitequark> TD-Linux: what about putting a small FPGA on it instead?
<whitequark> seems way more versatile & i'd totally write gateware for it
<TD-Linux> whitequark, I would also prefer that, I just am already backlogged on gateware
<TD-Linux> if you'd write the gateware than that is preferred :)
<TD-Linux> would an ice40 be fast enough
<cr1901_modern> My hobby ISA project would be to buy an FOSS ISA backplane, a prototyping ISA board, stick a glasgow and some SRAM on it, and picorv32 (or boneless!) and boneless LPC.
<whitequark> excellent, looking forward to the board
<whitequark> TD-Linux: yes absolutely, even up5k probably will do
<whitequark> I figure you could put HX8K on it and then it could have an entire SoC
<whitequark> and a Glasgow-compatible header to debug/drive/etc it
<TD-Linux> the hx1k is my normal goto for this sort of thing but maybe a 5k/8k would be better?
<whitequark> up5k won't work as it has too few pins i think
<whitequark> sg48 has ~36 pins so that's 20 address + 16 data and oops
<whitequark> so we're stuck with a BGA or one of those cursed QFPs and I obviously prefer a BGA
<cr1901_modern> Even 8 data is pushing it (I only am getting an 8-bit backplane for now)
<whitequark> yeah sticking to 8 bit ISA doesn't help
<TD-Linux> the tqfp100 package of hx1k is pretty reliable for me
<whitequark> sure, your board your rules
<TD-Linux> the bigger ones get a bit grosser. I can do bga
<whitequark> CB256 is easiest to buy, CB121 is a bit of a pain to get since it's new
<TD-Linux> do you want 8 or 16 bit isa
<TD-Linux> also 5v or 3.3v isa
<whitequark> it'd be silly to not route all 16 bits
<whitequark> um
<cr1901_modern> I've never heard of 3.3v isa
<whitequark> there is 3.3v isa?
<cr1901_modern> it might exist, but I've never heard of it
<whitequark> i'm pretty sure it doesn't exist
<TD-Linux> might be mixing it up with something else
<cr1901_modern> ISA cards will almost certainly accept 3.3v logic without complaint, but they need 5v power
<TD-Linux> oh right legacy pci is 5/3.3
<whitequark> anyway yes that's why a custom board is needed: it needs a lot of level shifters with direction input
<TD-Linux> yeah it's going to be 90% level shifters
<TD-Linux> but that's the norm for my projects
<whitequark> fortunately wide shifters are cheap
<TD-Linux> some day I will design a board that isn't just an ice40 plus 10 level shifters
<whitequark> oh btw this board has an interesting side effect
<cr1901_modern> and unlike Glasgow, this board does NOT need individual dir on each line
<whitequark> it would enable glasgows to read parallel ROM *very* easily
<cr1901_modern> (I think)
<whitequark> cr1901_modern: correct. only 1 direction pin really
<TD-Linux> whitequark, oh yeah interesting. is there something I can add to make it easier
<whitequark> TD-Linux: what form factor do you think it should be in? backplane itself ($$$) or a card ($$ and you need a backplane) ?
<whitequark> TD-Linux: also dumb idea
<whitequark> what if we put PCI *and* ISA slots there
<TD-Linux> can you map PCI to LPC bus?
<whitequark> more or less
<cr1901_modern> whitequark: Ngl, this morning I thought about "what would happen if you took the FOSS ISA backplane, and added an ECP just for the PCIe PHY"? https://github.com/skiselev/isa8_backplane
<TD-Linux> it would be interesting, but if there are no useful PCI cards to actually use it with I would skip
<cr1901_modern> (besides being a horrible cursed idea)
<TD-Linux> I could put a ZIF footprint on it for ROM reading, unfortunately every ROM has a different pinout
<cr1901_modern> There's two standard ROM pinouts AIUI
<cr1901_modern> pre-JEDEC 23xx series and the JEDEC 27xx series
<cr1901_modern> passive adapters between them are very doable
<whitequark> TD-Linux: so like, we're going for CB121 or CB256 with tons of pins
<whitequark> and i'm thinking a card + backplane makes more sense
<whitequark> so
<whitequark> what about making it a dual interface card? PCI on one side, ISA on the other
<TD-Linux> PCB area isn't too expensive if I use pcbway
<whitequark> like, you're paying extra for gold fingers anyway, might as well put 200% of card per card there
<TD-Linux> whitequark, how does this end up physically? the card sits on the lpc connector, and then a ribbon cable to the backplane?
<TD-Linux> or do you mean jsut an ISA backplane and an ISA card drives it
<whitequark> TD-Linux: I'm thinking the card would be a standard IBM half sized card that has an FPGA, ISA on one side, PCI on the other, and a ribbon connector at the side that has the metal bracket thing
<whitequark> it could either be an ISA device or a PCI device or ISA host or PCI host
* cr1901_modern would like to subscribe to this newsletter
<whitequark> now, what would be on the ribbon connector...
ej5 has joined ##yamahasynths
<cr1901_modern> Well, if the 5150 expander card is any indication, you can put any ISA signal on the ribbon connector and be safe ::P
<cr1901_modern> also, hi ej5!
<TD-Linux> just the LPC bus ribbon?
<TD-Linux> if the backplane can be only 2 layers that drops the cost a lot
<TD-Linux> (which is ez for an isa backplane)
<cr1901_modern> TD-Linux: FWIW, even the IBM 5150 was 4 layer
<cr1901_modern> (though I agree)
<TD-Linux> also looking at ebay there are lots of backplanes you can get. so maybe we just don't even make that
<whitequark> TD-Linux: the TPM bus ribbon?
<TD-Linux> whitequark, yea
<whitequark> it's kind of obnoxious because it's a 20-pin connector like the one on glasgow but it has tons of power and such
<TD-Linux> it would be neat if this could mount inside a normal computer case *with* the main motherboard but I don't know how that would fit :(
<whitequark> I think that's definitely doable
<whitequark> (for the lpc/glasgow adapter card that is)
<TD-Linux> one thing we could do is make it (ATX - microATX) sized
<TD-Linux> e.g. if you put a microatx in a normal atx case, our backplane mounts below the microatx board
<whitequark> TD-Linux: i think it makes sense to do this adapter card first and then look at improved backplanes
<whitequark> there's tons of backplanes and also they're easier to make anyway
<whitequark> hm
<cr1901_modern> https://github.com/skiselev/isa8_backplane And you can remix this one
<whitequark> yes
<ej5> whatcha trying to mount on an ISA bus
<whitequark> ej5: glasgow, of course.
<cr1901_modern> it's not 16-bit but
<cr1901_modern> 1. You can still get those connectors
<cr1901_modern> 2. It's not hard to extend
<ej5> ahh that's your FPGA + level shifter thing right?
<whitequark> yeah! though it's more than just that
<whitequark> TD-Linux: i'm thinking maybe two connectors on the adapter card make sense? a tpm bus connector and a glasgow connector
<whitequark> the tpm bus connector would be on the *inside* of the case
<TD-Linux> whitequark, I was just typing that :)
<whitequark> and the glasgow connector would be on the *bracket* side
<whitequark> cuz that's how you want to use them
<ej5> so glasgow would drive the bus or monitor the bus?
<whitequark> ej5: either. the adapter card will have an FPGA on it.
<TD-Linux> fpga is a lpc <-> isa bridge
<whitequark> the adapter card would be able to convert LPC<>ISA, and the default gateware would plug into the TPM header so you can drive floppies or whatever from modern chipses
<whitequark> i also want to put PCI on the adapter card, so it'd be a cursed dual-side one
<TD-Linux> pretty sure we can find a source of 16 bit isa connectors
<TD-Linux> now I want floppy emulator gateware for it too
<whitequark> TD-Linux: so for the glasgow connector, here's my suggested pinout
<whitequark> LRESET# LCLK# LCLK2 LFRAME# LAD3:0
<whitequark> why two clocks? simple. for PCI, we're going to need to Go Fast.
<whitequark> this means source synchronous transfers.
<ej5> oh i see, so you're converting LPC back to ISA
<TD-Linux> do we need a local clock for the fpga? or can it just be clocked directly by lpc clock
<whitequark> mmm, good question
<cr1901_modern> LFRAME# is the out-of-band signal for "LAD3:0 describes the data xfer type", right?
<cr1901_modern> I.e. same role as previously described?
<whitequark> I would put a footprint at least there, in case cr1901_modern wants to put an entire SoC on the card
<whitequark> cr1901_modern: yeah =
<whitequark> s/=//
<whitequark> ej5: sort of! that's the original idea
<whitequark> but I want to make Glasgow also a PCI/ISA multitool.
<whitequark> because clearly supporting just *dozens* of protocols is not enough
<whitequark> specifically, we need to go wider.
<TD-Linux> does isa or pci support any sort of multi mastering?
<whitequark> this card would *also* allow people to super easily dump parallel ROMs, and we're thinking about putting a JEDEC ZIF socket on it
<ej5> pci does, isa you could kinda of hack it i think
<cr1901_modern> This^
<ej5> whitequark, tough thing is suppling power to arbitrary JEDEC roms
<whitequark> ah hm
<ej5> they can use hundreds of ma for the older types
<whitequark> for readout even?
<ej5> it's why switching circuitry on device programmers tends to be complex
<ej5> yeah
<whitequark> hm ok
<TD-Linux> not really an issue if we only support one pinout
<TD-Linux> but then the utility is rather limited
<whitequark> yeah i'm not sure if it makes sense
<whitequark> maybe not for revA?
<cr1901_modern> the original IBM 5150 power supply was 63W. Easily powers at least 5 ROMs
<cr1901_modern> fwiw
<cr1901_modern> and I don't think that comes close to the limit
<cr1901_modern> I thought the switching circuitry was mainly to generate the voltages required to reprogram
<cr1901_modern> not to read ROMs
<ej5> that's part of it too
<cr1901_modern> (5 ROMs- 4* BASIC plus BIOS)
<ej5> some ROMs use like 200mA depending on the part
<cr1901_modern> there's an empty spot for a sixth ROM
<cr1901_modern> >in case cr1901_modern wants to put an entire SoC on the card
<cr1901_modern> So, the ISA bus technically has two clock outputs- 14MHz and system clock. I don't know many cards that use the 14MHz one, but AFAICT, it's the mainboard's responsibility to provide it.
<cr1901_modern> In the case of an ISA-SoC, I think the SoC needs to provide it
<ej5> 14.318MHz ;)
<cr1901_modern> hehe, indeed
<ej5> the CGA uses it, original SB uses it, adlib uses it...
<cr1901_modern> ahhh, good to know
<cr1901_modern> (I didn't know :P)
<cr1901_modern> the 14.318MHz clk it's not in phase w/ the system clk AIUI; on the original IBM PC, the 4.77MHz clock _is_ derived from the 14.318 clk. Idk how it works for turbo clones or the 286.
<whitequark> right, so we need to put two crystals on it, right
<whitequark> one to generate 14.318 and one to generate 33
<whitequark> (at least the footprints)
<ej5> iirc 286 has a separate clock for the cpu
* cr1901_modern nods I'll look later... ISA is weird in the 286 era because it was before they agreed to stop increasing the bus speed after 8.33MHz
<cr1901_modern> they==computer vendors
<ej5> yeah it's separate.
Xyz_39808 has quit [Ping timeout: 276 seconds]
<cr1901_modern> So one xtal for 14.3MHz, and another for the CPU/rest of the system?
<ej5> yep
<ej5> 12MHz for the AT
<cr1901_modern> Ahhh, divide by 2 for 6
<ej5> there's an intel clock chip that manages it
<ej5> the 14.318mhz is totally separate and only goes to the ISA bus pins
<ej5> they are not guaranteed to be in sync with the sys clock
<cr1901_modern> Ohhh right, I remember something about that. I think that's one of the larger LSI chips IIRC
<TD-Linux> that's now 3 xtals if we are also counting 33mhz
<cr1901_modern> why?
<ej5> for example, the 5161 has a separate free running 14.318mhz oscillator inside
<TD-Linux> oh if you derive system clock from 14.318 that's fine
<cr1901_modern> TD-Linux: I think the intent is to derive from the 33MHz clk
<cr1901_modern> ej5: What does Snort Bladder use the 14MHz clock for?
<whitequark> cr1901_modern: TD-Linux: no the idea is to use 33 for PCI and 14.318 for ISA
<whitequark> since you can divide by 3 and get your ISA clock
<cr1901_modern> oh right
<TD-Linux> we're still below 8 gbufs so we're ok :)
<whitequark> yep
<whitequark> ahhh shit, i just realized something
<whitequark> PCI uses reflected wave signaling and the electrical part of the spec is REALLY weird
<ej5> cr1901_modern, it runs the adlib and cms chipset clocks
<whitequark> uh
<cr1901_modern> whitequark: I'd still put PCI on and improvise later :P
<whitequark> PCI should be able to drive HOW MUCH current
<whitequark> EACH driver should be able to drive a HUNDRED MILLIAMPS?
<whitequark> that's fucking nuts
<whitequark> "the open circuit voltage range at the pins of PCI devices will exceed the ground-to-Vcc voltage by a considerable amount"
<whitequark> how the fuck did any of this crap ever work
<cr1901_modern> In practice, ISA cards will have their own buffers to refresh the signal. Do PCI peripheral controllers tend to have the same (I don't know)?
<cr1901_modern> Meaning I wouldn't worry about drive strength problems w/ _existing_ ISA cards
<whitequark> the input buffers of a PCI device should tolerate transients up to ELEVEN VOLTS?
<cr1901_modern> ej5: Is the system clock used anywhere?
<TD-Linux> whitequark, how else do you implement Power over PCI
<cr1901_modern> ... that's not a joke, is it?
<ej5> whitequark, that's why they didn't stay with it
<whitequark> ej5: I haven't realized it was *this* insane
<ej5> i have a book on it somewhere. *digs*
<TD-Linux> what are the receiver high/low levels
<whitequark> TD-Linux: normal cmos levels
<whitequark> Vil -0.5 to 0.8, Vih 2.0 to Vcc+0.5 for 5V
<whitequark> Vil -0.5 to 0.3Vcc, Vih 0.5Vcc to Vcc+0.5 for 3V
<ej5> ahh here we go: PCI system architecture, third edition
<TD-Linux> if we are very careful with level translators we could probably share them for isa and pci
<ej5> chapter 4: intro to reflected-wave switching
<whitequark> TD-Linux: i feel like it'd be safer to use separate
<whitequark> this isn't a high volume board anyway and level translators are pretty cheap
<whitequark> as in, we'd lose out more on a respin
<TD-Linux> I guess they can also be no-populate if you don't care about it
<whitequark> exactly
<ej5> haha there's some good stuff in this book "the usage of incident-wave switching consumes a significant amount of power and violates the green nature of the PCI bus"
<whitequark> violates the green nature of the PCI bus
<ej5> you must manufacture the PCI bus on boards with a green soldermask.
<whitequark> TD-Linux: did you hear that
* TD-Linux sets the soldermask to black, the color of coal
<whitequark> pfhahaha
<whitequark> TD-Linux: yeah no let's not bother with PCI
<whitequark> I'm reading the spec and like, all electrical section is batshit insane
<cr1901_modern> ISA directly to PCIe
<whitequark> you could make a career doing nothing but meeting that spec most certainly
<ej5> people did.
<ej5> there are/were FPGAs available with built-in PCI pin drivers.
<whitequark> oh yes, I remember that
<cr1901_modern> I would not recommend this, but I know someone (elm-chan?) made a working PCIe device using a GAL
<whitequark> a what
<cr1901_modern> PAL/GAL
<whitequark> that seems unlikely
<cr1901_modern> oh fuck
<cr1901_modern> PCI*
<whitequark> oh.
<cr1901_modern> not PCIe
<whitequark> oh yeah tons of these
<cr1901_modern> Freudian slip
<ej5> wow.
<whitequark> xc9572 post cards
<whitequark> hm
<whitequark> actually
<whitequark> does that have PCI pin drivers?
<cr1901_modern> Not sure. I didn't know xc9572 was a popular post card
<whitequark> it's not a post card
<whitequark> it's a CPLD
<cr1901_modern> err was popular for post cards*
<whitequark> that i found on a few post cards i tore labels off
<whitequark> that's why glasgow has a programming module for it
<ej5> hmm
<whitequark> like, they also might just be doing something noncompliant
<cr1901_modern> elm-chan's GAL PCI card was certainly noncompliant
<cr1901_modern> he even said so in his writeup :P
<whitequark> yes, xc9572 IS in fact PCI-compliant
<cr1901_modern> (but it worked)
<cr1901_modern> wow
<ej5> interesting.
<cr1901_modern> They still make xc9x, right?
<whitequark> yes
<whitequark> i just really don't want to think about PCI after reading this spec
<whitequark> fuck that, i'll write ISA gateware for TD-Linux and consider whether i hate myself enough for PCI
<cr1901_modern> Right, I knew about the reflection stuff, I didn't know about the other stuff. And I've long since forgotten how to calculate how long it takes for a wave to come back and what the peak voltage will be.
<cr1901_modern> (Used to be a fun party trick... I could tell you where a fault in a cable was from looking at how long the reverse wave took to come back.)
<ej5> 8 inches per nanosecond
<ej5> in a typical PCB
<ej5> sometimes as low as 6
<cr1901_modern> (it was coax :P)
<ej5> then you divide by 2 because it has to go out and then back again.
<ej5> coax is closer to 10 inches per nanosecond.
<cr1901_modern> Do you remember how to derive that from the "real" characteristic impedance, incl the imaginary component?
<cr1901_modern> (it's like 50 + j0.1)
<cr1901_modern> huh, I actually can't find a theoretical imaginary component for rg58. I sincerely thought it was spec'd
<ej5> nope, it's been almost 20 years
<cr1901_modern> So a Glasgow ISA system has gotta have a: Glasgow, a Snark Barker, maybe XTIDE, an Ethernet card (custom?), the TPM ribbon cable, and... anything else? No I'm not projecting my own plans, why do you ask?
<whitequark> cr1901_modern: that'd be just a Glasgow ribbon cable
<whitequark> which is pretty much any IDC ribbon cable with 20 pins
<ej5> npt thread adapter so you can hook it up to the kitchen sink
<whitequark> lol
<whitequark> ej5: when are you getting a glasgow? :p
<ej5> hehe dunno
<cr1901_modern> I'll get one eventually
<cr1901_modern> whitequark: Oh no kidding, TPM _does_ fit on a ribbon cable (that'll show me for not blowing up the photo: https://twitter.com/Foone/status/1153718213449904128/photo/1)
<TD-Linux> personally I want it for the PC-FXGA DOS/V
<cr1901_modern> yea you could do some potentially cool/cursed things with the LPC connection
<cr1901_modern> DOS/V is the Japanese name for DOS, correct?
<whitequark> cr1901_modern: yes, both glasgow and tpm work with the same kind of ribbon cable
<TD-Linux> cr1901_modern, specifically for ibm pc compatibles
<TD-Linux> vs pc-98 and fm towns
<cr1901_modern> Ahhh yes :3, 11+ inches of ISA goodness
<whitequark> cursed phras
<cr1901_modern> :)
<ej5> for cursed ISA cards, i ought to dig out my hard card.
<ej5> 20mb hard disk bolted at the end of an ISA card.
<cr1901_modern> There was a good vid released recently on how to fix those
<cr1901_modern> apparently the bumper that stops the r/w head on a decent number of those hard drives wears out
<cr1901_modern> so it confuses the drive into thinking the r/w head is in the wrong position
<cr1901_modern> https://www.youtube.com/watch?v=kXMFbHx8s0Q Yes I watched the whole thing. I regret nothing.
<ej5> hmm, mine works, but good to know.
<ej5> keep thinking it might be fun to take an old drive like that and put a clear window on the lid.
<TD-Linux> I want to do that but with a microdrive
<cr1901_modern> 3d printing prob makes it very easy to do now
<TD-Linux> laser cut acrylic is what you want for that
<cr1901_modern> I don't have a good track record w/ opening drives tho, so I wouldn't do it
<ej5> i've got a prototype connor hard drive that's partially clear plastic
<ej5> i should tweet some pics.
<cr1901_modern> oooh, plz 2 post pics?
<cr1901_modern> TD-Linux: Oh right, I have a full length ISA card that's used for wire wrap or point-to-point prototyping. I'm waiting for a good project to use it on (something mostly thru hole).
<cr1901_modern> It was a gift from a professor ("free to a good home" sort of deal). I want to genuienly make the most of it.
<cr1901_modern> A custom multifunction card (serial, RAM, floppy, parallel, game, RTC) would be a good candidate.
<cr1901_modern> ej5: So could I use this to create an 8051 programmer for my PB-10? https://twitter.com/TubeTimeUS/status/1153802848733257730
<cr1901_modern> s/programmer/module/
<cr1901_modern> or is it EMP-2x only?
<ej5> EMP-20 only
<cr1901_modern> damn
<cr1901_modern> Guess I could RE the pinout at some point
<cr1901_modern> I'm very unlikely to find the expansion modules in 2019
<cr1901_modern> I've been looking since 2011
<ej5> i don't see any expansion slots on the pb-10
<cr1901_modern> They were adapters that plug into the 40 pin ZIF socket
<cr1901_modern> for whatever reason, 805x and other families need an adapter
<cr1901_modern> I call them "expansion modules"
<ej5> wow, this is really cheap. $40 for 20 boards, 24 hour turn!
<cr1901_modern> I just realized, after all this time: the PB-10 really is a glorified "two I/O expanders ICs plus programmable voltage rails"
<cr1901_modern> curiously it's using Motorola bus I/O expanders instead of Intel chips (which ISA bus would understand more readily)
andlabs has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<TD-Linux> ej5, welcome to 2 layer boards :D
<ej5> indeed.
<ej5> first 2 layer board i ordered was nearly exactly 20 years ago.
SceneCAT has quit [Ping timeout: 268 seconds]
<cr1901_modern> mine was about 6 years ago (Early revision of: https://github.com/cr1901/AT2XT)
<TD-Linux> mine was in 2010-2011 ish
<TD-Linux> my other plan for the pc-fxga dos/v was to write a modern linux driver and run it in an old 486
<TD-Linux> ideally in time for vcfwest. but I have to miss vcfwest this year :(
<cr1901_modern> oh that's a cool idea
<ej5> awww
<TD-Linux> that said it would be hard to get it done in time. same for midiori cards. I'll have to show them next year
<whitequark> TD-Linux: anyway what's the plan for the ISA cad
<whitequark> card*
* whitequark is too excited
<cr1901_modern> Same, tbh :D. I have... plans.
<whitequark> i just want to add one more pipe to the kitchen sink
<cr1901_modern> I should say "GlasgISA" accelerates my plans. The most immediate one being talking to a Sound Blaster from a custom non-x86 CPU (riscv, Boneless, doesn't matter to me- I'm sure it was done in the 80s/90s already).