After some discussion on CPCRulez and emulation issue with some CDT, I've done a CPC side tool that measure the tape speed after motor is switched on or off.
The test is made of 2 parts :
1 - making a test tape : 10 second of 3kHz tone (exactly 334us period)
2 - using that test tape to measure the start/stop speed
Here are the result on 6 different 464
[attach=2]
From top :
- 2 tape drives with old belt
- 2 tape drives with ok belt
- 2 tape drives with new belt
The blue curve show the tape speed (in mm/s) after the motor has been stopped. After a small delay (20ms), the speed will decrease linearly. The curve will not go to 0 due to the tape drive not detecting pulse any more. So the real full stop is a bit afterward.
The Orange curve show the tape speed after the motor has been started. Here as well, these is a small delay before the effective start, then the speed will increase in a linear fashion, with an exception for the new drive belt. Here the elasticity of the belt and motor regulation seem to fight each other.
Note that the few dots at start of the curve are artefact of the measurement method and filtering. These should not exist.
Ideally, emulators should take this behaviour into account.
Great work!
I know no one usually would like to emulate hardware errors, but having good measures of the behaviour of a real machine is indeed useful information.
Quote from: robcfg on 21:05, 24 February 16
Great work!
I know no one usually would like to emulate hardware errors, but having good measures of the behaviour of a real machine is indeed useful information.
It's not really hardware errors, and some copy protection use this behaviour...
Interesting data, now I just need to work out what else this could be useful for :)
Bryce.
Quote from: remax on 21:49, 24 February 16
It's not really hardware errors, and some copy protection use this behaviour...
You're right there. I was thinking on testing programs that have error detection, as it can be educational and, obviously, if nothing fails, the error detection routines will not fire.
Quote from: Bryce on 09:09, 25 February 16
Interesting data, now I just need to work out what else this could be useful for :)
Bryce.
Main target was to show that the drive does not stop/start immediately and provide data to improve emulation.
The attached CDT (A new dump of Last mission) does not work on most emulator just because of that. The loader take the trailer bit of its own block as part of the data it load. The title screen will be shifted and main program will crash.
A second use could be to get an idea of the belt state.
And with a precisely written test tape, evaluate the real tape speed of the drive. But that one is quite useless as you cannot adjust it on the CPC.
Strange behaviour of the motor speed with new belt ? I assume that is it linked to the fact that it must be "grind" for a certain time, then results should be similar to "OK belt" ?
Quote from: Bryce on 09:09, 25 February 16
Interesting data, now I just need to work out what else this could be useful for :)
Bryce.
We (At Work) used to use 3Khz to test the Wow and Flutter on a range of Tape recorders, Maybe test the wow and Flutter ??
Ray
A nice test program. :)
I have been following the thread on cpcrulez :)
Quote from: gerald on 09:28, 25 February 16
Main target was to show that the drive does not stop/start immediately and provide data to improve emulation.
I think I'll clamp my motor with a mad big diode! That will teach the motor know who's in control! :)
Bryce.
Quote from: Audronic on 09:33, 25 February 16
We (At Work) used to use 3Khz to test the Wow and Flutter on a range of Tape recorders, Maybe test the wow and Flutter ??
I don't think the sampling period is precise enough for this. I am sampling at 62.5KHz (every 16us), which give a really bad precision on the 3KHz frequency (20/21 samples per period).
The plot is the result of a moving average on 8 consecutive transition (4 period of 3hKz).
Quote from: Bryce on 10:35, 25 February 16
I think I'll clamp my motor with a mad big diode! That will teach the motor know who's in control! :)
Useless if your belt is not new enough. Better put brakes on the capstan inertia wheel ;)
So, it's time for the big guns... I have some spare Airbag detonators here, maybe I can use them to stop it? :)
Bryce.