DamageX:
Time ago i've considered a disk virtualizer that work connecting pc & msx via EAR & MIC jacks. Basically the idea is to patch the disk rom and when the msx read a sector, with a protocol the sector data are suplied from the PC via the AUDIO output. In the same manner when the msx write a disk sector the reverse apply:

Schematically:

MSX READ SECTOR->ROM PATCH->CONVERTION TO PULSE->MIC INPUT OF THE PC->OUTPUT FROM HEADPHONE FROM PC OF THE PULSE DATA WITH THE SECTOR WANTED.


Maybe with this approach the pc could be a server for the msx pratically at 0€ (Simply connect the msx mic/ear to the pc audio board and all work fine)


Unfortunately i've no time to develop this.

While that's certainly possible, I fear the speeds at which it would work would be, well, awful... I think you'd be looking at something like 1 or 2 KB/s, so even slower than an FDD... I'd much rather see an effort into making more obsoNET carts which you could hook up to your router in order to share data between your PC and MSX...

Quote:

While that's certainly possible, I fear the speeds at which it would work would be, well, awful... I think you'd be looking at something like 1 or 2 KB/s, so even slower than an FDD... I'd much rather see an effort into making more obsoNET carts which you could hook up to your router in order to share data between your PC and MSX...

Yes, the speed is on this order of magnitute, of course, but would be very cool to see a drive emulated by the cassette interface at 0 $ (remember hw to add == 0). when reading from disk a simple compression algo could increase the speed-
when writing maybe compressing a single 512 bytes sector would not be so much z80 intensive.....

I've a lot of sw that i cannot try on my old msx1 because is not on tape, only on disk...

For the pc part there is not more to do... only to analyze the audio input signal and decode the bytes.... when reply, situation is more simple, simply synthetize and output from pc to msx...

PingPong:

the half of work in implementing a zero-cost solution is already done in MicroWAVer - on diskless MSX2 you can load any combination of files together with patched DiskROM that emulates RAM drive over part of RAM and/or VRAM. This can work on MSX1 too, if you have a mapper with enough RAM on it.

E.g. you can load and use Turbo Pascal this way.

The missing part is saving the data from MSX to , e.g., PC or other sound recording device.
At the moment I can kinda solve this by using utility CAS.COM I found somewhere that can dump/load files via cassette port.

But, of course, making a virtual drive should be a much cleaner and convenient solution.

On networked Russian Yamaha MSX2 we had such a utility that used network interface on a diskless student machines to access disk drives of the teacher's MSX, via virtual drives.

cax: i think it's possible to achieve rougly about half the same speed of a real fdc on the average, using some kind of protocol that works on 512 bytes chunks (a sector) compressing the sector contents. About 4Kb/sec.

Should be acceptable.
If a people want more, i think should buy some hw board like obsoware, or SD adapters.

PingPong:

MicroWAVer already uses compression, so you can achieve about 1KB/sec on data that packs badly, 2Kb/sec on textual data, and much more on unpacked graphics.

But (and for many people it's a show-stopper) - don't forget that for high-speed loading via cassette port it's very important to set the right volume, and to have clean sound from your PC's soundcard. It's not a secret that most modern motherboards with built-in sound are very noisy. Not every ear can hear this noise, but for MSX it's a real problem.

That's why I like joystick port solution much more. It requires some soldering though, but you get a reliable solution.
And using a breadboard almost anybody can build such a circuit.
(for example, see my "msx-link DIY guide for dummies" in Russian here)

I guess using the joystick ports for data transfer from the PC would be a problem for MSX software using the joysticks... although it would provide good speed I think, when connected via parallel cable. (I did something similar on an Atari a while back and got 12KB/s)

Maybe by using a combination of data lines on the MSX printer port and cassette port we could get a good speed? I know some of the bits are accessed in the PPI and others in the PSG, I forget exactly how it goes...

Of course, the custom DOS would be the harder part.

I can always use sneakernet with LS-120 floppies to transfer files to my MSX

cax: as said by DamageX using psg joy port have problems with sw using joy, that's because i've not considered to work with joy.
AS for noisy ear, i'm able to load at 3200 b/s from pc. not tested at higher speed, but on internet there on youtube i've seen a manic miner load at 19200 b/s. from a notebook.

PingPong:
I don't believe in clean 19200 bps over tape port - the max speed we've reached with [WYZ] while working on mwaver was 5512 bps, the rest is done by 2 levels of compression - one on the signal layer and the other on the data layer.
And we tried a lot of soundcards, DVD and CD players, palm devices, and various MSX models.

cax: see on youtube it.youtube.com/watch?v=E-PPk3ZCimo
it's a 12000 bps
the loader is customized not the bios std loader...

I' have loaded Manic Miner to a real MSX1 (Phillips vg8020) at 19.600 bps (using a SB sound card )
The load that you can see in YouTube is at 12.600 bps.
Both wavs were made with the OTLA tools.
Proyecto OTLA is an open-source project than you download from here: http://code.google.com/p/otla/

The high speed loadings are very realiable too. It's not only a matter of experimentation or curiosity. Otla is based mainly on the Spectrum tool k7zx. Many Spectrum owners use k7zx for laoding games with their mp3 players.

High speed also works for CPC Amstrad. Even using a magnetic cassette adapter

I hope a new version of OTLA will be released soon. New features for MSX computers will include conversion of .CAS files.

About sending data to PC via MIC i'm working in a rutine for Spectrum. Theorically about 12600 bps write speed.

I will give OTLA a try. Maybe we can combine MicroWAVer's compressing and RAM disk emulation abilities and OTLA's speed.

That OTLA is pretty amazing. I guess it's a matter of making a very tiny read out loop. Using the BIOS you don't get higher than 5520 bps. In that video it looks as if it's loaded in one go, but I guess there's a small preloader.

Quote:

That OTLA is pretty amazing. I guess it's a matter of making a very tiny read out loop. Using the BIOS you don't get higher than 5520 bps. In that video it looks as if it's loaded in one go, but I guess there's a small preloader.

Hi, Manuel! Great openMSX.!!! I've tryed to simply flip the ppi bit at very highspeed while saving to wav the results.
Now, i've displayed the resulting wav. Incredible! the wave output is not a simple square wave as i guessed, but instead you emulate the output like in the real machine! I can see the transient part of the signal while passing from 0->1 and 1-0 and looks like in a exponential form! Amazing!



Really openMsx is one step ahead over msx emus.

Anyone know if different msx machines have different waveform in the signal output from ppi when flipping the ppi cassette bit?

I'm suspecting that because i think may be differences in the analog cirquitry used to interface the cassette recorder in different models.
Different components and electronics could generate very different output, far different from the teorichal square wave output.

Anyone have ever sampled the msx output ? How the output waveform looks?