23 replies to this topic

[Mclaneinc]

Avery has kindly released a new Altirra beta with drive sound support, I'll copy from the blog the release info that contains a request for a little help.

After reading the info if you think you can provide the needed help then please do, I've bolded the help issue.

http://www.virtualdu...2.10-test10.zip

---------------------------------------------------------------------------------------------

Sorry, guys, I'm a bit busy in real life so updates are going to be slow... but I might as well share some goodies I already had cooking:

Adds drive sound emulation. The sound samples suck since I don't have a real drive and had to synthesize up some samples in Audacity, but that can be improved later. It includes the drive rotation sound, and if you have accurate sector timing on and SIO patch off, seek sounds too. The step rate depends on the drive emulation profile used. Drive idling is also now emulated so you will see the drive indicator dim before the drive shuts off.

Currently, the drive emulation profiles mainly control which method and rate of high-speed SIO is used and whether certain commands are supported (high speed commands, format skewed, read/write PERCOM block, etc). The plan is to improve this over time. XF551 mode already does a couple of other things like adjust the PERCOM handling, the timeout reported by the status command, and the rotation speed.

What the drive emulation modes don't do is actually run code on the drive. This is the main reason that I don't want to do full Happy, Speedy, or Indus emulation. (Synchromesh uploads code, I believe, because the version that's on the drive is fatally broken.) Besides the extra cost, the hardware is also quite varied, since the various drives use 6507, 65C02, 8040, or Z80 CPUs. I haven't looked yet at what the Happy software needs though since it's possible it mainly just wants a few more extended commands, and some of them like read/write track wouldn't be too bad to add.

If anyone can get clean sound recordings of a floppy drive spinning, stepping a single track at a time, and stepping over multiple tracks, I'd really appreciate it. Also, I could use verification on whether the step rates for the various drives are correct. The 810 mode uses 5.3ms/track, the 1050 modes use 10ms per half track, the XF551 uses 6ms per half track and Speedy is ~4ms/half track.

[Mclaneinc]

bump.....

I'd have thought people would jump in to help Avery?

[Rybags]

My 1050s are packed away - probably the case with many others too.

Plus one makes way too much vibration and racket, the other is uncharacteristically quiet.

[Mclaneinc]

Okey dokey....

I hope someone out there can step in at some point, my 1050 died so I'm out.

The drive sound emulation is a nice 'in the zone' addition to the emulator...

All we need now is the sound of the different drive openings because closed / locked for the full effect

Now that would be amazing

And a little OTT

[Rybags]

I might see if I can get it out at some point.

I've only got the one microphone and it's a cheapy never used so no idea if it'd be any good.

[jacobus]

My mic is rather crappy so the sounds are very quiet - are these usable or should I source a new microphone?

 810 Disk Noise.wma.txt   145.17K   74 downloads
 810 Seeking and Stepping.wma.txt   706.42K   88 downloads
 810 stepping.wma.txt   434.57K   77 downloads
 1050 Disk Noise.wma.txt   167.1K   79 downloads
 1050 Seeking and stepping.wma.txt   395.11K   81 downloads
 1050 Seeking.wma.txt   1.1MB   82 downloads
 1050 stepping.wma.txt   395.11K   69 downloads
 XF551 disk noise.wma.txt   259.18K   76 downloads
 XF551 seeking and stepping.wma.txt   377.57K   72 downloads
 XF551 stepping.wma.txt   364.41K   62 downloads

(renamed to bypass the upload filter)

[bbking67]

These don't plat for me (and yes I removed the .TXT).

[jacobus]

I created them using Windows 7 sound recorder - the first few seconds of most of them are blank, and the rest of the sounds are very quiet.  Try the "810 seeking and stepping" from about 12 seconds onward.

[bbking67]

Not audible on my Windows or Linux box at any volume...

[Mclaneinc]

I can't hear them either, don't tell me windows media player encodes them to stop them being played on other machines?

[andym00]

No idea what you 2 are doing but the WMA samples from Jacobus play just fine..
I think you're just expecting something more raucous than it actually is..

[Mclaneinc]

Wow, they are really faint, I turned my speakers almost fully up and then heard it..Lol

At normal volume they just sound blank, mind you my daughter is playing on her Wii and its quite close to me.

[ijor]

Mclaneinc, on Fri Jan 27, 2012 4:21 AM, said:

What the drive emulation modes don't do is actually run code on the drive. This is the main reason that I don't want to do full Happy...I haven't looked yet at what the Happy software needs though since it's possible it mainly just wants a few more extended commands, and some of them like read/write track wouldn't be too bad to add.

The Happy doesn't have any built-in read or write track commands. All the "Backuper" functionality of the Happy is uploaded at run time from the computer to the drive. Conceivable, you could emulate the commands that are added after the upload. But there are variations across the different versions. And the software doesn't expects (obviously) a mismatch because the software uploads its own version of the firmware. So probably it would need to be version specific.

Quote

step rates for the various drives are correct. The 810 mode uses 5.3ms/track, the 1050 modes use 10ms per half track, the XF551 uses 6ms per half track and Speedy is ~4ms/half track.

1050 (standard firmware) is correct. Regarding 810, I understand it is correct from reading the ROM, I don't have any real 810 to do any kind of verification. XF551 is 6ms per track (not per half track), those drives can't do half stepping. Speedy don't know, I seem to recall it is configurable, but not sure about the default.

[phaeron]

jacobus, thank you, those samples are enormously helpful! There's a bit of WMA artifacting in them after normalization, but I can get a lot of info about stepping noises out of these. The 810 sound files confirm that it does park the heads when the drive goes idle, which is something I'd found in the ROM dump and remembered from long ago but needed verification on. They also confirm that the 810 actually does step at 5.25ms/track... still unbelievable that the oldest drive is also one of the fastest seeking. However, the sound is much lower because it appears to be either dominated by resonance effects or by some phases being much louder than other and thus sounds two octaves lower (and another octave below Altirra's current sound, due to a bug). The track step sound I'm using is also far too percussive for the 1050 and XF551, which have fairly bassy sounds.

What confuses me is the 1050 drive seek. From the sample, the drive appears to be seeking at 26ms/track, which is considerably lower than the 20ms/track that it should be doing according to the firmware. The only thing I found that could account for this is two calls to reset the FDC per step, but I'm not sure that would account for 3ms of delay for each half step.

I also can't get over how loud the 1050 disk rotation sound is... it sounds more like sanding than spinning.

[flashjazzcat]

phaeron, on Tue Jan 31, 2012 1:15 AM, said:

I also can't get over how loud the 1050 disk rotation sound is... it sounds more like sanding than spinning.

This struck me when I first got a 1050. It sounds like an old washing machine.

[Fox-1 / mnx]

flashjazzcat, on Tue Jan 31, 2012 5:39 AM, said:

This struck me when I first got a 1050. It sounds like an old washing machine.

It varies.  I have some that are as silent as an XF551 but most are noisy, like the drive wants to cut up the disk to 3,5" form factor.

[morelenmir]

Its good to see so many people helping out with Altirra!  After the 'desert of the real' around 'Atari800win plus!' it is nice to see so much community surrounding an atari 8bit project.

I am sad to hear about the complexities of Happy emulation.  I really had my fingers crossed for that one and I doubt now that Avery is going to be able to add it in any time soon.  Ah well.  Maybe in the future.

[Mclaneinc]

Ha ha ...The old 1050 grind, the grrrrrrr grr was deafening on my original machine, I always expected it to spit the disc out in little bits when I closed the lever on the front. And when the 810 hit a bad sector, wow..

Happy days, no pun intended...

[Mclaneinc]

Fox, I know I'm old and senile these days but I never spotted what your avatar was til now............LOL.....

[Fox-1 / mnx]

Mclaneinc, on Tue Jan 31, 2012 11:40 AM, said:

Fox, I know I'm old and senile these days but I never spotted what your avatar was til now............LOL.....

:-)

I used these ones      for many years and just a few months ago I decided it was time to draw a new one :-)

[flashjazzcat]

Mclaneinc, on Tue Jan 31, 2012 11:38 AM, said:

Ha ha ...The old 1050 grind, the grrrrrrr grr was deafening on my original machine, I always expected it to spit the disc out in little bits when I closed the lever on the front. And when the 810 hit a bad sector, wow..

I only experienced an 810 once in my life - there's something wrong with the track 0 sensor, but it does spin and sounds like a high-pressure hose.

[ijor]

phaeron, on Tue Jan 31, 2012 1:15 AM, said:

What confuses me is the 1050 drive seek. From the sample, the drive appears to be seeking at 26ms/track, which is considerably lower than the 20ms/track that it should be doing according to the firmware.

Where on the samples do you see the 26ms? Or there are some more samples other from the ones posted here?

[phaeron]

I got the 26ms/track from measuring the waveform period in Audacity for the 1050 seek samples, and by comparing the pitches by ear between the sample and Altirra. I had to substantially slow down the seek rate from 20ms/track to get them to match.

[ijor]

phaeron, on Mon Mar 19, 2012 9:18 PM, said:

I got the 26ms/track from measuring the waveform period in Audacity for the 1050 seek samples, and by comparing the pitches by ear between the sample and Altirra. I had to substantially slow down the seek rate from 20ms/track to get them to match.

Well, I obviously can't argue with you about waveform periods, pitches, etc. Certainly not my stuff. But I can tell you that I actually measured the 1050 step rate sometime ago, and it is just slightly over 20ms, not 26 ms.

The half step rate is about 10.16ms, or 20.32ms per track. To be precise, when seeking, there are approx. 10.16 ms between each physical movement of the stepper.

Note that the total seek time is not the same as the step rate multiplied by the number of (half) tracks. There is a significant one time overhead per each seek. And some firmware versions sometimes perform an extra back and forth half step.