Microdrive Unit Design

Nagging hardware related question? Post here!
RWAP
RWAP Master
Posts: 2834
Joined: Sun Nov 28, 2010 4:51 pm
Location: Stone, United Kingdom
Contact:

Re: Microdrive Unit Design

Post by RWAP »

I have 3 microdrive units with just a single key (instead of the normal 2) - does anyone want one of these units to experiment with and compare?


User avatar
Cristian
Aurora
Posts: 960
Joined: Mon Feb 16, 2015 1:40 pm
Location: Veneto

Re: Microdrive Unit Design

Post by Cristian »

RWAP wrote:I have 3 microdrive units with just a single key (instead of the normal 2) - does anyone want one of these units to experiment with and compare?
It could be interesting. Can you tell something about their condition and price please?
thanks


RWAP
RWAP Master
Posts: 2834
Joined: Sun Nov 28, 2010 4:51 pm
Location: Stone, United Kingdom
Contact:

Re: Microdrive Unit Design

Post by RWAP »

Cristian wrote:
RWAP wrote:I have 3 microdrive units with just a single key (instead of the normal 2) - does anyone want one of these units to experiment with and compare?
It could be interesting. Can you tell something about their condition and price please?
thanks
Condition - they have been stored in a box with other spare microdrive units for a few years - they are pulled from systems and untested.

I was asking £8 each plus postage but for the sake of expanding the knowledge about microdrive units, you could have one for the cost of postage.


martyn_hill
Aurora
Posts: 909
Joined: Sat Oct 25, 2014 9:53 am

Re: Microdrive Unit Design

Post by martyn_hill »

Hi Tomas and crew!

So, prompted by your thorough investigation in to these great wee devices (contentious, I know, but I love the MDVs!), I have been carrying out some testing myself. More to follow, but I thought that you might appreciate some oscilloscope images to replace those poor-quality ones from the original Service Manual.

Please see attached for an updated Appendix D5 page, produced with help from my cheapo-but-capable DDS-120 digital USB 'scope, plugged in to an MDV test-rig I knocked-up on breadboard and hanging-off the end of my QL.
QL Service Manual - Appendix D5.pdf
Updated Appendix 5 from the QL Service Manual
(517.53 KiB) Downloaded 147 times
Some specs for reference (though most of these shouldn't have much impact on the results shown in the attached PDF):

Issue 6 QL mainboard with custom external 512KB SRAM memory upgrade and Min v1.98 + TK2 v2.23
Running the small SB routines from the Service Manual (with some typos corrected)
Charles' vMAP adapter replacing the (removed) internal MDV units - so as not to foul the external test-rig, configured as MDV1_
Issue 7 MDV unit, (variant 2 moulded base) that was originally MDV2_ - known good unit, that typically formats to 225/230+ good/found sectors on even the most troublesome cartridges

Regards,
M.


tcat
Super Gold Card
Posts: 633
Joined: Fri Jan 18, 2013 5:27 pm
Location: Prague, Czech Republic

Re: Microdrive Unit Design

Post by tcat »

Hi Martin,

This is excellent work, just studying the write waves for DATA1-2. I see some asymmetry here in shapes, I wish to ask about first, as my next post will be about asymmetric bias for FM recording.

Y-axis is in Volts, but there are two rasters for Channel 1 (0,419 .. 5,18V) and 2 (0,968 .. 4,968V), and the wave of Channel 1 seems taller?

Tomas


martyn_hill
Aurora
Posts: 909
Joined: Sat Oct 25, 2014 9:53 am

Re: Microdrive Unit Design

Post by martyn_hill »

Hi Tomas

I too noted the slightly different (absolute) voltage levels between DATA 1 and 2 during writing. Couldn't explain it myself, but it is quite possible that the offset at least is due to the 'scope itself needing zero-level calibration (I didn't find that UI option until later in the testing).

The 2V/DIV is identical for both - with the following ranges, as you noted already:

DATA 1: 0.6V to 5.2V
DATA 2: 0.97V to 4.97V

So, the differential voltage-swing was 4.6V for DATA 1, vs 4.0V for DATA 2.

As some of the reference labeling produced in the UI of this software (SainSmart DDS-120 v1.3) gets confusing and is sometimes misleading (for example, if you choose to change the scale to view AFTER sampling), I wouldn't place too much credence in the absolute value labels. Fortunately, its the general shape of the waveforms that gives the most information from a troubleshooting perspective...

Furthermore, the DATA 1/2 pins are supposed to be TTL levels, so, ignoring the voltage offset (due to need for calibration), these particular voltage-swing differences won't impact the logical data-stream (standard TTL levels: 0 - 0.8V == Logic 0 and 2 - 5V == Logic 1).

If you look at the waveforms rendered during Reading, you'll note precisely the same voltage swings, but offsets back where they should be (near zero.) By that point, I had found the Zero callibration feature in the UI :-).


tcat
Super Gold Card
Posts: 633
Joined: Fri Jan 18, 2013 5:27 pm
Location: Prague, Czech Republic

Re: Microdrive Unit Design

Post by tcat »

Hi Martin, All,

Thank you for clarifying. I need to get myself a scope, I wish it be plugable into my RPI, I already saw some DIY tips.

Now I also went through `ZX Interface-I & Microdrive' service manual. This caught my attention.
DATA 1 and DATA 2 are used to drive current sources for track 1 and track
2. When DATA 1 is h i g h , for example, current is pul led in one
di rect ion through the head, when it is low current is p u l l e d in
the other direct ion. Wr i t i n g is done wi th an asymmetric write
current. Because the erase function is carried out wi th a d.c. f ield
it is necessary to write 'ha rde r ' in one direction than in the other
to overcome residual magnet ism. Resistors R4 and R5 on pins 11 and 7
of IC1 are chosen to al low this action. Should a di f ferent tape be
introduced or a di f ferent head type be fitted these resistors would
also need to be changed.
This asymmetry is caused by the fact, that in one direction a higher coercive force is needed that in the other, to overcome erased record which is a singular in the top left quadrant. `Bre' - remanent magnetism of erased record, `Hce' - its coercive force.
Coercive Force
Coercive Force
Knowing this, I can now correct the earlier diagram, now showing two currents with different bias. The current switching is done by MD ULA amplifiers? As the erase current is some 30mA, I just estimate that in one direction twice as high current is needed in writing through R/W coils.
asymmetric recording
asymmetric recording
Tomas


martyn_hill
Aurora
Posts: 909
Joined: Sat Oct 25, 2014 9:53 am

Re: Microdrive Unit Design

Post by martyn_hill »

Hi again Tomas!

One small correction to the frequency/cycles in your 2nd diagram (doesn't change your analysis of the asymmetric currents needed):

When we read the Dicken's QL Advanced User Guide, we see that the nominal time per byte is 40uS. As this is across 2 tracks, that equates to 80 uS per byte/track and thus the actual 'bit-frame' - which is constant - is 10us, regardless of 1 or 0 levels. This allows for 1/2 cycle at 50 KHz (==0) and 1 full cycles at 100 KHz (==1). Sort of, but not quite FM encoding.

Instead, the detection of 1 or 0 is effectively based on the number of transitions that occur in any given 10us bit-frame (1 or 2 transitions), or more precisely, whether or not a transition occurs both at the start of the frame as well as in the middle. Its more helpful, then, to discuss in terms of Manchester Encoding (ME), rather than FM.

The Wiki page on ME shows how easy it is to decode/encode the signal to or from tape (XOR data with CLK): https://en.wikipedia.org/wiki/Manchester_code

-----

On a different but related note... The Service Manual lists a simple SB program to write/read 100 KHz across the surface of the tape. This is what I used to generate the signal-traces to replace Appendix D5.

Curiously, the program simply writes just one byte of 0x01 to the transmit register (pc_tdata), after enabling the MDV unit/motor, then setting the Erase & Write bits of the pc_tctrl register, before then switching back to Read mode after 8 secs (ensuring the full tape surface is covered.)

I would have expected, even if the ULA actually writes the same byte continuously until pc_tdata is re-written, and to both tracks, that the procedure would have generated the encoded version of %00000001 - a signal at 50 KHz for 7 bit-times, then finally 100 KHz - repeatedly. Yet instead it produces a continuous 100 KHz.

EDIT/revelation:
If we assume that the final '1' bit of the single byte written to the register sets to 100 KHz, in the absence of any further data, then we might assume that the ULA simply leaves the last 'bit' signal free-running until it is later disabled (by resetting the Erase/Write bits in pc_tctrl.)

Anyone else here know or ever worked-out the 'truth'' about this behaviour? :-)


tcat
Super Gold Card
Posts: 633
Joined: Fri Jan 18, 2013 5:27 pm
Location: Prague, Czech Republic

Re: Microdrive Unit Design

Post by tcat »

Hi Martyn,

Thank you for reading my post. You are possibly right, in my 2nd diagram I just wanted to show how current is biased in one or the other direction through R/W coils. Frequencies just show that series of "1" ones need twice as many transitions to series of "0" zeroes.

The actual encoding I tried to show earlier with this picture, in naming it, I was lead by ZX and QL service manuals referring to it as FM.
FM/ME encoding
FM/ME encoding
fm-modulation.png (4.56 KiB) Viewed 3720 times
In the record mode the modulator in IC23 converts the ones and zeros in the data into FM (frequency modulation) where there is always a transition at the beginning of the bit cell. If the data is a one there is a transition at the beginning and in the middle of the bit cell, which means essentially that the frequency doubles if the data contains ones.
The decoding/encoding, albeit FM or ME, seems easy as you explained using XOR function. I like the picture you shown at wiki, copying it here for clarity.
ME encoding
ME encoding
ME decoding
ME decoding
FM-ME_decoding.png (5.11 KiB) Viewed 3720 times
Tomas


tcat
Super Gold Card
Posts: 633
Joined: Fri Jan 18, 2013 5:27 pm
Location: Prague, Czech Republic

Re: Microdrive Unit Design

Post by tcat »

Hi,

Now giving it a second thought. There is also `Differential Manchester Encoding', also known as, Biphase Mark Code (BMC), Frequency Modulation (FM), or F2F (frequency/double frequency).
Unlike with Manchester encoding, only the presence of a transition is important, not the polarity.
Which helps to eliminate noise on tape.
BMC / DME / FM - encoding
BMC / DME / FM - encoding
But how to decode this stream?

Tomas


Post Reply