c2n turbo loader experiment: 4kbit/s data on tape

Basic and Machine Language

Moderator: Moderators

eslapion
ultimate expander
Posts: 4162
Joined: Fri Jun 23, 2006 7:50 pm
Location: Canada
Occupation: 8bit addict

Re: c2n turbo loader experiment: 4kbit/s data on tape

Postby eslapion » Thu Apr 27, 2017 5:12 am

pixel wrote:What's the effect of the filter anyway? To me on the software side it makes no difference.

I'll go with this loader at 3.6kbit/s for the syncing part alone. :)

The effect of the filter (bandpass) is to considerably reduce the amplitude of signals that are out of a very specific range of frequencies and to perform an analog integration (which boosts amplitude) of the signal that's in the sweet spot (of 2.5kHz).

Signal higher or lower in frequency are considerably attenuated and so datasettes can no longer read them if they get just a little weaker over time or for whatever reason.

Since the normal Commodore datasette protocol operates exclusively within the peak frequency, tapes with data saved with this protocol remain readable even if the signal has considerably degraded over time or for whatever reason.

https://en.wikipedia.org/wiki/Band-pass_filter

Since the filter only applies to playback, a tape carrying data that has become unreadable may seem fine when inspected with an oscilloscope and audio tape deck.
No one can do everything. Everyone can do something.

groepaz
Vic 20 Nerd
Posts: 548
Joined: Wed Aug 25, 2010 5:30 pm

Re: c2n turbo loader experiment: 4kbit/s data on tape

Postby groepaz » Thu Apr 27, 2017 6:19 am

if the signal has considerably degraded over time

however, when the signal degrades, amplitude doesn't change considerably - its the relative flux _timing_ that changes (which is why pre-compensation exists on high capacity floppy drives). and typically homecomputer tapes are written without pre-comp, because the frequency is so low that it doesn't matter much (like C64 floppy disks).

eslapion
ultimate expander
Posts: 4162
Joined: Fri Jun 23, 2006 7:50 pm
Location: Canada
Occupation: 8bit addict

Re: c2n turbo loader experiment: 4kbit/s data on tape

Postby eslapion » Thu Apr 27, 2017 8:26 am

groepaz wrote:
if the signal has considerably degraded over time

however, when the signal degrades, amplitude doesn't change considerably ...

You are correct. However, this claim doesn't account for the attenuation that may have been added by the filter because the protocol is using a frequency that's out of the peak range.

It may take very little reduction in amplitude for the tape to become unreadable.

I posted this in may 2015...
eslapion wrote:However, I found another very interesting use for that test point. I stored a 2.5 kHz square wave signal onto a tape using a datasette that is in extremely good shape. This little guy was able to read a few tapes from the Commodore Educator system which other datasettes couldn't read properly.

With the best aligment, the analog signal had an amplitude of 1.2v pk-pk.

Then I recorded a 5 kHz square wave, something close to the frequency used with TurboTape. On read back, altough I got a good clean 5 kHz square wave on the datasette normal port, the amplitude on the analog test point fell to 400mV pk-pk. This means the filtering built into the datasette makes these tape accelerator very "borderline" for reading back.

The way the datasette works "hides" from the user that the signal acquired prior to turning into a square wave might be very weak. This can be caused by head misalignment, degrading tape signal or too high a frequency.


Mind you, I did these investigations because I expected I could store data faster on a datasette by adjusting the motor to spin faster. It turns out the real bottleneck was the bandpass filter.

When doubling the frequency, the signal was suddenly 1/3rd the amplitude. That's a drop of 4.77dB per octave or 16dB per decade.

If a 4kbit/s protocol encodes data at 4kHz then that's close to 0.7 octave higher than 2.5kHz, the signal will have a 3.33dB attenuation compared to the reference of 1.2Vp-p so it will now be 560mV p-p. Your signal is now 2.15 times weaker.
No one can do everything. Everyone can do something.

User avatar
R'zo
Vic 20 Afficionado
Posts: 350
Joined: Fri Jan 16, 2015 11:48 pm

Re: c2n turbo loader experiment: 4kbit/s data on tape

Postby R'zo » Thu Apr 27, 2017 1:32 pm

Sometimes it doesn'the hurt just to give something a try. :)
R'zo
I do not believe in obsolete...

groepaz
Vic 20 Nerd
Posts: 548
Joined: Wed Aug 25, 2010 5:30 pm

Re: c2n turbo loader experiment: 4kbit/s data on tape

Postby groepaz » Thu Apr 27, 2017 2:15 pm

However, this claim doesn't account for the attenuation that may have been added by the filter because the protocol is using a frequency that's out of the peak range.
It may take very little reduction in amplitude for the tape to become unreadable.

thing is, it doesnt just magically change. magnetism is very very stable once something has been magnetized.

User avatar
Mike
Herr VC
Posts: 2925
Joined: Wed Dec 01, 2004 1:57 pm
Location: Munich, Germany
Occupation: electrical engineer

Re: c2n turbo loader experiment: 4kbit/s data on tape

Postby Mike » Thu Apr 27, 2017 4:44 pm

groepaz wrote:thing is, it doesnt just magically change. magnetism is very very stable once something has been magnetized.

At a temperature near 0 K this is very much true, but we're at 300 K room temperature, and this disturbs the elementary magnets over time. The effect may be small, but accumulates over time and becomes faster with an exponential law near the Curie-Point.

...

I'd like to bring another point into the discussion here regarding the datasette schematic: what is recorded by the head is *not* what is being read back by the head! It is the reason for the analog amplifier of the read electronic to *reconstruct* the signal that was used to write the tape:

I'm referring to schematic CAS-12R on zimmers.net.

First, the write circuit:

There we have one gate of IC3 as buffered inverter, and then the signal is split and one branch is inverted once, the other one inverted twice. Means these two signals are inverses of each other. They're biased against two pull-ups. In write mode, the head coil is connected in series with a 10K resistor to this bridge: when a 0 is on the write signal from the VIC-20, the coil is polarised in one direction; with a 1, the coil is polarised in the other direction. These polarisations are reflected in two possible magnetic flux orientations recorded on tape.

For ease of the following explanation, let's say a 1 writes a upward flux (U), and 0 writes a downward flux (D).

However, when read back, a constant flux doesn't do anything on the head. Regardless whether it is up or down. The only thing that effects is a magnetic flux change. In case of down to up, the head produces a small positive voltage spike; in the other case (up to down) it produces a small negative voltage spike. Now what's the function of the 4 op-amps?

IC1-B simply pre-amplifies the spike. The next stages are coupled to the pre-amplifier with C1 to remove any DC bias.

IC1-A and IC2-A act as 'leaky' integrators. They transform the spikes first into a step, and then into a ramp. C5 once again removes the DC bias going into the last stage.

There, IC2-B is driven as comparator. A small positive voltage on pin 6 will drive the output of the op-amp into negative output (which is returned to pin 5 with R17) and 'clamps' the op-amp there. With a small negative voltage, a clamped positive output results. The result is an analog flip-flop!

Two TTL inverters of IC3 complete the pulse-forming circuit. The op-amps are also all inverting, i.e. the positive spike is formed into a high level TTL signal, the negative spike is formed into a low level TTL signal.

...

The whole reason behind the read circuit, I repeat, is to reconstruct the original two-level write signal with as sharp edges as possible, without it being sensitive to much noise. That requires the circuit to be tuned to the speed of the actual flux change on write. As long as the flux changes on tape don't overlap, it is always possible to reconstruct the original write signal.

That also explains why the azimuth of the head is critical: when the azimuth is wrong (or different from the recording azimuth), the spikes are broadened when read back and won't register any more good to the pulse-forming circuit. But the record amplitude itself is much less important - the pre-amplifier and the comparator-flipflop in the last stage take care of that.

Any questions?

eslapion
ultimate expander
Posts: 4162
Joined: Fri Jun 23, 2006 7:50 pm
Location: Canada
Occupation: 8bit addict

Re: c2n turbo loader experiment: 4kbit/s data on tape

Postby eslapion » Thu Apr 27, 2017 7:19 pm

Mike wrote:IC1-A and IC2-A act as 'leaky' integrators. They transform the spikes first into a step, and then into a ramp.

Every op-amp based integrators aka low pass filter has a cut-off frequency. Did you check it for both of these ? Their effects add up.

Also, C5 and R15 together act as a high pass filter or differentiator. The same with C1 and R6, I believe.

The whole reason behind the read circuit, I repeat, is to reconstruct the original two-level write signal with as sharp edges as possible, without it being sensitive to much noise. That requires the circuit to be tuned to the speed of the actual flux change on write. As long as the flux changes on tape don't overlap, it is always possible to reconstruct the original write signal.
...

Any questions?

Systems like that have an optimum frequency at which they were designed to operate and do what you just said, otherwise they considerably attenuate the amplitude of what comes out.

I didn't run a SPICE simulation on this but empirically, I found it to be around 2.5kHz.

Do you come to the same conclusion ?
No one can do everything. Everyone can do something.

groepaz
Vic 20 Nerd
Posts: 548
Joined: Wed Aug 25, 2010 5:30 pm

Re: c2n turbo loader experiment: 4kbit/s data on tape

Postby groepaz » Thu Apr 27, 2017 9:13 pm

At a temperature near 0 K this is very much true, but we're at 300 K room temperature, and this disturbs the elementary magnets over time. The effect may be small, but accumulates over time and becomes faster with an exponential law near the Curie-Point.

compared to other factors, eg mechanical stress on the tape, its probably still pretty irrelevant. seeing how modern hard disks use (almost literally) single elementary magnets to store a bit, and they run at typically twice that temperature.... oh well :)

eslapion
ultimate expander
Posts: 4162
Joined: Fri Jun 23, 2006 7:50 pm
Location: Canada
Occupation: 8bit addict

Re: c2n turbo loader experiment: 4kbit/s data on tape

Postby eslapion » Fri Apr 28, 2017 7:28 am

groepaz wrote:... and they run at typically twice that temperature.... oh well :)

Twice 300K would be 600K, that's 327 Celcius...

I don't know of any hard drive that runs at the temperature of a soldering iron...
No one can do everything. Everyone can do something.

groepaz
Vic 20 Nerd
Posts: 548
Joined: Wed Aug 25, 2010 5:30 pm

Re: c2n turbo loader experiment: 4kbit/s data on tape

Postby groepaz » Fri Apr 28, 2017 8:46 am

i was looking at "room temperature" and °C, obviously.

Vic20-Ian
Vic 20 Nerd
Posts: 882
Joined: Sun Aug 24, 2008 1:58 pm

Re: c2n turbo loader experiment: 4kbit/s data on tape

Postby Vic20-Ian » Fri Apr 28, 2017 12:06 pm

eslapion wrote:
groepaz wrote:... and they run at typically twice that temperature.... oh well :)

Twice 300K would be 600K, that's 327 Celcius...

I don't know of any hard drive that runs at the temperature of a soldering iron...



2 x 21C = 42C
Vic20-Ian

The best things in life are Vic-20

Upgrade all new gadgets and mobiles to 3583 Bytes Free today! Ready

groepaz
Vic 20 Nerd
Posts: 548
Joined: Wed Aug 25, 2010 5:30 pm

Re: c2n turbo loader experiment: 4kbit/s data on tape

Postby groepaz » Fri Apr 28, 2017 12:14 pm

that'd be pretty nice and cool for a HD actually :)

eslapion
ultimate expander
Posts: 4162
Joined: Fri Jun 23, 2006 7:50 pm
Location: Canada
Occupation: 8bit addict

Re: c2n turbo loader experiment: 4kbit/s data on tape

Postby eslapion » Fri Apr 28, 2017 1:48 pm

Vic20-Ian wrote:
eslapion wrote:
groepaz wrote:... and they run at typically twice that temperature.... oh well :)

Twice 300K would be 600K, that's 327 Celcius...

I don't know of any hard drive that runs at the temperature of a soldering iron...


2 x 21C = 42C

Pfew... then I guess all of this is completely false... https://en.wikipedia.org/wiki/Thermodynamic_temperature

That being said, we still don't know the optimum encoding frequency for the datasette filter integrator.
No one can do everything. Everyone can do something.

groepaz
Vic 20 Nerd
Posts: 548
Joined: Wed Aug 25, 2010 5:30 pm

Re: c2n turbo loader experiment: 4kbit/s data on tape

Postby groepaz » Fri Apr 28, 2017 2:26 pm

lol

Vic20-Ian
Vic 20 Nerd
Posts: 882
Joined: Sun Aug 24, 2008 1:58 pm

Re: c2n turbo loader experiment: 4kbit/s data on tape

Postby Vic20-Ian » Fri Apr 28, 2017 3:30 pm

eslapion wrote:Pfew... then I guess all of this is completely false... https://en.wikipedia.org/wiki/Thermodynamic_temperature

That being said, we still don't know the optimum encoding frequency for the datasette filter integrator.


315K if you prefer. (21C room temp x 2 + 273C for room temperature in C offset from absolute zero)

I have not seen a laptop hard drive run at around 600k or 300C yet, so I don't understand why you made that particular post, you normally make more sensible arguments.
Vic20-Ian



The best things in life are Vic-20



Upgrade all new gadgets and mobiles to 3583 Bytes Free today! Ready


Return to “Programming”

Who is online

Users browsing this forum: No registered users and 1 guest