• Please review our updated Terms and Rules here

Repairing a CT-1020 datasette

geekdot

New Member
Joined
Feb 15, 2022
Messages
6
Location
Hamburg, Germany
Hi lovely folks of the vcfed forum,

I'm about to finish my PET 2001-8C restoration. The last non-working bit is the "CT-1020" or "white keys 2001 datasette"
This is the model I mean, just in case...
...

It simply refrains from saving data - sometimes when reading back, it gets close to the given filename, e.g. "TEST-123" will be read as "Found T!STZTTs#$3" but most of the times it's just nothing. That said, reading data written with an C2N works fine.

What I've done so far:
  • Clean the whole thing. Inside out. Heads, gears etc.
  • Checked the PCB and all cables for discontinuity.
  • Adjusted azimuth
  • Then swapped the read/write and erase-head from a working C2N into it
Nothing of that helped. So besides scratching my head, you guys are my last resort 😔

Thanks a ton in advance for any tip, hint, idea I can get this last piece of the puzzle solved!

Cheers, Axel
 
If you could post a link to the schematic for this one. Many forum members could explain where to start fault finding the record circuitry.

Most of the schematics I can find for other datasette models show the arrangement for record is incredibly simple. They use a DC on the erase head and the record amplifier that drives the record/playback head, is either a simple transistor amplifier/buffer or some units, logic gates, without the AC bias typically seen in audio grade recorders.

The most likely problem is always electro-mechanical first, in other words the playback-record changeover switch is the first thing to check, that the play/record head is being correctly switched onto the output of the record amplifier's output in record mode and that the DC is being switched onto the the erase head, for that you could put the DVM across the resistor in series with the erase head and measure & calculate if the expected current is there.

If the switch was ok (and you know your heads are ok) then it would be a matter of looking at the incoming record data on the scope and following that through to the point that it arrives at the record head.There could be a fault in the record amplifier.

I would guess that you would have confirmed on the scope that the record data coming from the computer is normal, or tried another recorder that makes a successful record/playback of that.
And I would guess that you have tried to play the defective recording back in another player and that it remains defective on playback in the other player, meaning the fault is located in the record amplifier/switching area in your CT-1020. And you know that reading data written in the C2N is ok, so the playback amplifier is fine, as is the record/playback head.

(There were some recommendations about de-magnetizing the head (I seldom have to do this), but you already have excluded the record playback head and the erase head by substitution. Also even if that were the case, it would unlikely affect record, as the energy pulses to the head are relatively high, unlike playback).
 
Last edited:
Hmm. That video looks familiar.
Let me see if I understand ....
This white keys tape drive CAN read tapes saved on another datasette, but can't read tapes written by itself?
Do other datasettes save OK when connected to this same PET?

The first thing I would look at is the record/play switch on the PCB. The switch has many different contacts and some can oxidize making a bad connection on the write side.
A bad op amp in the write circuit is also a possibility but the switch is more likely. I spray the switch on both sides with deoxit and exercise it back and forth over and over.

Other thoughts ...
There are two datasette ports and two data read inputs but only one data write output, shared by both ports.
Maybe the 6522 that controls the write signal is flaky? Pin 13 on the VIA.

If you feel like getting out the scope, you could check the read signal strength.
I don't think it has anything to do with alignment since the drive should always be aligned with any tape written by itself,
but, on the first test point, you should be getting ~50mv p-p IIRC. If that signal is low, it seems likely the signal on the tape was recorded at a lower level than it should be.
Link
Or just look at the signal with an audio player to see if it's lower than a working tape.

The schematic(s) are here ... http://www.zimmers.net/anonftp/pub/cbm/schematics/datassette/index.html
Note the switch S1 which has a number of different connections all over the schematic, but they are all the same physical switch with multiple contacts.
If one of those contacts is dirty or oxidized and not making a good connection, you can run into problems.
I've even had one with the erase head stuck on so it would erase tapes when you were trying to play them.

Maybe S1-2 is making contact but S1-1 isn't, resulting in a poor record signal and distortion ... ??

1648178036216.png
 
Last edited:
A bad op amp in the write circuit is also a possibility but the switch is more likely. I spray the switch on both sides with deoxit and exercise it back and forth over and over.



View attachment 1239607


There are no OP amps in the write circuit, only in the read circuit.

And the diagram posted has errors.

IC3 is a 74LS14, which is an inverting Schmitt trigger. So it should have circles drawn on the outputs and the Schmitt symbol inside the triangle. Corrections attached.

The idea in this case, is that they provide an anti-phase (push pull ) signal to the record head. Rather than the other versions where they simply have a buffer stage capacitively coupled to the record head.
 

Attachments

  • record.jpg
    record.jpg
    127 KB · Views: 3
Last edited:
If you could post a link to the schematic for this one. Many forum members could explain where to start fault finding the record circuitry.
Attached below. It's from http://www.zimmers.net/anonftp/pub/..._Manual_Preliminary_314002-002_(1984_Oct).pdf (Page 10).

Screenshot 2022-03-25 081321.jpg
Most of the schematics I can find for other datasette models show the arrangement for record is incredibly simple.
Absolutely... especially the recording part.
The most likely problem is always electro-mechanical first, in other words the playback-record changeover switch is the first thing to check, that the play/record head is being correctly switched
Checked ✔ All three "positions" of the switch do work as expected.
, for that you could put the DVM across the resistor in series with the erase head and measure & calculate if the expected current is there.
Will do so! Thanks for the reminder!
If the switch was ok (and you know your heads are ok) then it would be a matter of looking at the incoming record data on the scope and following that through to the point that it arrives at the record head.There could be a fault in the record amplifier.
According the to schematic, the OpAmp (LM358P) is just used in the read circuit. Saving just uses two transistors (Q3 2N4401, Q5 2N3904)
I would guess that you would have confirmed on the scope that the record data coming from the computer is normal, or tried another recorder that makes a successful record/playback of that.

Just to clarify my test-setup (sorry for being confusing before):
2 machines: PET 2001-8C (the target) and a C64 to counter-test
2 datasettes: CT-1020 (the target) and a C2N to counter-test

Both datasettes behave the same on each computer. That is:
CT-1020 does read fine on the PET and the C64 but saves garbage on both. This gets the PET out of the equasion.
C2N works fine on both machines. This proved C2N is fine.

As the C2N heads put into the CT-1020 did not change the drives failure to save data the remaining issue is the CT-1020 PCB.

Hmm. That video looks familiar.

It was just meant as 'visual description' of that rather rare model and shows more of the internals than a simple piccy.
The video is not mine as is the shown drive.
Let me see if I understand ....
This white keys tape drive CAN read tapes saved on another datasette, but can't read tapes written by itself?
Do other datasettes save OK when connected to this same PET?
Both "yes"
The first thing I would look at [...]
All very valid points but as written above, checked and fine.
If you feel like getting out the scope, you could check the read signal strength.
Took that on the to-list (y)
Or just look at the signal with an audio player to see if it's lower than a working tape.
All gone :cry:
Waaaaaayyy to modern ;-)
There are no OP amps in the write circuit, only in the read circuit.
IC3 is a 74LS14, which is an inverting Schmitt trigger. So it should have circles drawn on the outputs and the Schmitt symbol inside the triangle. Corrections attached.
The TC-1020 has no LS14... it's simpler than simple :sneaky:
Sadly I don't have fitting transistors in my "stuff drawer"... so just for the fun of it, I'll order some and replace them.
Besides that dreaded switch and some resistors/caps, there's not much more we can blame.... well, you could blame me for being a lazy dude wasting your time.

Anyhoo, thanks a metric ton Hugo & Hutch - you already gave me valuable to-do's for my list.
I'll come back as soon I checked all the points you mentioned.

Cheers, Axel
 
"Checked ✔ All three "positions" of the switch do work as expected."
The record switch only has two positions. Playback and Record

"It was just meant as 'visual description'"
My comment was because that's my video

"There are no OP amps in the write circuit"
My bad. But there are active devices that could be generating a lower than normal signal.
I figure it has to be a bad connection somewhere or a bad component in the record path.
How's the edge connector on the PET and the plug? Is it making good contact with the PCB?

"Waaaaaayyy to modern ;-)"
That's all of the schematics for all the different datasets types. Not sure which one is for which datasette.
But of course this model is an older version.

Since you swapped the heads, I just thought I'd mention that the signal phase is critical.
Since you are reading data from the other datasette, you must have got it right. If the connection to the read head were backwards you wouldn't read anything.
 
Since you swapped the heads, I just thought I'd mention that the signal phase is critical.
This is very interesting.

Looking at the schematic versions in here:


The two schematics, the one with the transistor record amplifier overall, vs the one with the 74L14 IC record amp;

In the version without the 7414, the playback amplifier overall, is non-inverting (as it has an even number of signal inversions, 4, in the playback amp chain), in terms of the signal amplified from the head's white wire.

In the other version with the 7414, the input is to a differential amplifier and the polarity, considering the amplifier overall is inverting for the white wire & non-inverting for the red (even though the white feeds the + input of the OP amp and the red the -, because there are an odd number of inverting stages 5, after that before the signal gets to the output .

The record amplifiers are inverting, in terms of the signal passed to the white wire leading to the head, in both cases.

This might indicate that the polarity of the recovered (playback) signal at least would be the opposite in the two versions of the machine, but they would both record the same polarity on the actual tape.

Probably because the playback signal is AC coupled in the playback amp, the signal polarity might not matter too much.

I have not looked into the tape system in the PET yet. In the SOL for example, playback signal inversion has no effect on the recovered data from tape because the system works on a zero crossing principle, but its likely different in the PET as the data has already been digitized(clipped to a rectangular wave) out the recorder output.
 
Last edited:
I have not looked into the tape system in the PET yet. In the SOL for example, playback signal inversion has no effect on the recovered data from tape because the system works on a zero crossing principle, but its likely different in the PET as the data has already been digitized(clipped to a rectangular wave) out the recorder output.
Here's a link. The data is based on timing the waveform crossing positive to negative. That triggers an interrupt.
Negative to positive transitions are ignored.
Short duration between two triggers is zero, long duration is 1.
Link: https://wav-prg.sourceforge.io/tape.html
If the signal is inverted, the bit timing will be incorrect.
1648246303082.png
 
Interesting though, how it appears, at least looking at the schematics, that the two different circuit versions would recover opposite polarity pulses, making it look as if the polarity of the pulses, on playback, doesn't matter.

But I guess I could have messed up what "looks like" the total number of signal inversions from the head on playback on the two schematics. Perhaps somebody could chime in to check my assessment of it.
 
Regardless of the number of cascaded inverting stages I guess the polarity would ultimately depend upon which way around the head coil was wired. The encoding scheme as used in the PET, VIC20 & C64 is certainly polarity sensitive. I have a scan of an official Commodore technical bulletin as the authoritative reference, which I can't seem to find at the moment. This bulletin showed how to check the polarity of the serial data stream on a 'scope.

Some time ago I made a "deluxe" cassette port interface to essentially allow me to use any audio recording device for program storage, that will clean up and resolve even the crappiest of signals. A number of of designs were published back in the 80s, in various electronics magazines, for interfacing generic cassette recorders to Commodore computers , but every one of these designs that I've seen lacked provision to invert the signals if required. That limitation is a potential hindrance as the designer of your generic audio cassette recorder might not have payed particular attention to absolute phase!

I equipped my interface with the provision to independently invert the phase of both the read and write paths. Yes, this has proved necessary with some audio recording devices and no, the computer will not copy if data is returned to it in the wrong polarity.
 

Attachments

  • cas3.jpg
    cas3.jpg
    191.9 KB · Views: 6
  • cas4.jpg
    cas4.jpg
    211.9 KB · Views: 7
  • cas2.jpg
    cas2.jpg
    212.7 KB · Views: 6
  • cas5.jpg
    cas5.jpg
    139.9 KB · Views: 6
  • cas1.png
    cas1.png
    96.4 KB · Views: 6
Last edited:
Regardless of the number of cascaded inverting stages I guess the polarity would ultimately depend upon which way around the head coil was wired.

Yes I thought about that too.

But since the polarity of the drive to the head coil, is the same in both cases of the two different schematics, it would not matter about the head coil polarity, in terms of the fact that, even if that were reversed on the two versions, the playback data would still have been reversed polarity on one version of the schematic vs the other. It would just mean that the two recorded versions on tape had reversed polarity.

If you look at the recorded data, it would not appear to matter in terms of data recovery, because with an inversion it only phase shifts the data a half cycle in time, it does not alter the overall nature of the serial data.

So, I am suspicious, that whoever concluded that the recovered data polarity was important , might have got that wrong, because, obviously, both versions of the circuit work in the actual players.

To confirm this what I would do , as an experiment, with my own pet and datasette , is to invert the data and test, if it is still recovered normally by the computer, or not.
 
Last edited:
If you look at the recorded data, it would not appear to matter in terms of data recovery, because with an inversion it only phase shifts the data a half cycle in time, it does not alter the overall nature of the serial data.


Although I know from building my own computer that the data cassette won't work if the signal polarities are wrong, it has been a while since I built my PET 2001 clone and the cassette interface and I still can't find that technical bulletin I mentioned, but just to reassure myself that I am not going mad I quickly put pen to paper.

The Commodore encoding scheme measures the time interval between high-to-low edges (or it could be the other way around as far as I can recall - doesn't necessarily matter). This timing information does not necessarily remain the same but simply shifted in time if the signal is inverted. I just drew waveform A and its inverse B. At and Bt are the timing intervals between high-to-low edges of waveforms A and B respectively. I'm not saying that this is an accurate representation of the PET data stream (I'll have to find that bulletin) but it's enough to demonstrate the principle.

timing.jpg
 
Well, my guess on graph paper wasn't too far off the mark!

I've just come back in from the shack, having fired up the PET 2001 to capture some data cassette serial data. You mentioned the SOL. If its data cassette is not polarity sensitive and it is not return to zero (like, say the TRS-80), then my guess is that is uses frequency shift keying modulation.

This is not what Commodore does. The PET uses fixed frequency duty cycle modulation - See the timing diagram attached, which I produced from the captured data on my DSO. The PET uses a fixed frequency of approx 2400 Hz and it is the duty cycle of each pulse, not the period as per FSK, which determines if 0 or a 1 is being represented. The PET doesn't do the 0 Vs 1 cycle timing perfectly - it alternates between approx 400uS and 420uS per cycle, depending on if a 0 or a 1 is being represented, but this is not how the information is conveyed.

It is the time interval between high-to-low (negative-going) edges, which selects either a 0 or a 1. I don't know which way around it is, but one is represented by an interval of 340uS and the other by 480uS. If you invert the signal and trigger off the positive-going edges instead, then the information content is lost!


pettiming1.jpg
 

Attachments

  • pettiming2.jpg
    pettiming2.jpg
    301.9 KB · Views: 5
Last edited:
It is interesting.

Before I fixed up my datasette, I played back the one and only PET commodore tape I had in another player and examined the output on a scope, I wanted to see if there were any dropouts. It was interesting to look at the "data". It was, just as the image in the post indicated with the red dots, basically sinusoidal, though I only saw it over a hand full of cycles.

By the look of the circuitry in the datasette, though I have not gone through with the scope yet, while playing a tape, the signal is amplified up and amplitude limited. The implication of inversions would be a phase advance of the data by half a cycle time, and that might matter.

It still looks to me, studying the two versions of the schematic, that, for the same recorded tape, swapped between the two different machines, the polarity of the recorded signal would be played back 180 degrees out of phase (inverted) for one machine and not the other, yet we know the machines are compatible and both work.

But, obviously to be 100% sure, especially if there was some sort of schematic error, not impossible, I would have to have both the machines in front of me and check it out.

When I can I will study the circuit like I did for the SOL-20:


Of course in this system, the tape players were pot luck in terms of the signal polarity they produced and the SOL's circuitry used a zero crossing method and and makes no difference at all if the signal is phase reversed on different machines.In fact as shown on page 7, depending on how the SOL boots up, the baud rate generator, which had no initial power on resets can come up in two states resulting in the phase of the record signal getting flipped, or sometimes not as pot luck. But it has no effect in this system.

Just saw your post above, it must mean that the schematics between the two datasettes, one cannot be accurate, because from what you say, the waveform polarity is critical.
 
Looks like we cross-posted - I think my previous post is fairly definitive now. I would suggest that if there is a schematic of a datasette indicating a reversal of polarity from other datasettes (there were several circuit revisions over the years) then that schematic probably has a drafting error on the play/record change-over switch wiring.
 
Looks like we cross-posted - I think my previous post is fairly definitive now. I would suggest that if there is a schematic of a datasette indicating a reversal of polarity from other datasettes (there were several circuit revisions over the years) then that schematic probably has a drafting error on the play/record change-over switch wiring.
I have attached the diagrams, the most likely error would be that the two 10k resistors that feed the head signal into the version with the differential amplifier's plus & minus inputs, have their inputs switched.
It would be fun to actually look in that recorder and check it.

On the other hand the plot thickens to the extent that another diagram in that manual shows, again inverted drive to the heads white wire, but the white wire again (or what should be it for the same record polarity) feeding the first op amps plus input and again an odd number of inversions after that.

So all the decks drive the head with the same polarity, but in the case with the more simple circuit, there is no overall signal inversion, but in the other two circuits there are. And its not obvious how the more simple circuit could have a drawing error.
 

Attachments

  • petcass.jpg
    petcass.jpg
    209.1 KB · Views: 5
  • petcass2.jpg
    petcass2.jpg
    226.2 KB · Views: 5
Last edited:
Perhaps the simple circuit is the correct one and both the others have the DPDT switch drawn wired wrong (the draftsman probably just copied the previous design).
 
Perhaps the simple circuit is the correct one and both the others have the DPDT switch drawn wired wrong (the draftsman probably just copied the previous design).
I don't see that the switch could be incorrect, or it would reverse the record data polarity too, and all the circuits appear to have that in the same uniform manner.

I was wondering, although I have been able to find all the schematics for the PET on Zimmers, there seems to be a paucity of technical descriptions of the commodore circuitry in general.

In the SOL for example there was moderately elaborate circuitry to get around tape speed errors with a PLL and a good description of it in the SOL manual.

In the PET, the amplitude limited data from the cassette merely goes directly into the PIA, so all the tricks are done in the software, with a lot less in the hardware it would seem.

Is there a link anywhere to a technical manual for the PET with detailed circuits & descriptions of them, including the tape data recovery methodology, or were commodore not big on making technical manuals like Processor Technology ?
 
I don't see that the switch could be incorrect, or it would reverse the record data polarity too, and all the circuits appear to have that in the same uniform manner.

Wired wrong in that the two wires from the DPDT switch to the inputs of the differential pre-amplifer are likely swapped. That would only reverse the playback polarity. You suggested the same thing already.

Hardware documentation for the PET overall is poor. I don't think that what you are looking for exists - well not authored by Commodore anyway. There were some good technical books written independently, such as PET and the IEEE488 Bus (GPIB) (see archive.org) but, that's only for a specific aspect of the machine.
 
Hardware documentation for the PET overall is poor. I don't think that what you are looking for exists - well not authored by Commodore anyway. There were some good technical books written independently, such as PET and the IEEE488 Bus (GPIB) (see archive.org) but, that's only for a specific aspect of the machine.


It would be good with a collaborative effort to create what would have been the Commodore manual for the PET mobo variants and the technical description of the logic circuitry, say to create what Commodore could have done, if they had bothered to do it.

So far I have just about finished one for the PET 9" VDU's, it is a very large document.

The only part of the PET logic circuitry I have examined in any detail, so far, is the interesting 4 state machine (image attached) which uses narrow (100uS wide) drive pulses that are difficult to see with the scope time-base set up to see the overall waveform. But also there is other very interesting logic circuitry in the PET and it would be good to create a detailed description of it and some timing diagrams. These sorts of things would really help a lot for people doing repairs.

( I wrote up a detailed manual for the IBM-5155 power supply as that did not exist either, it is on minuszerodegrees)

If find it interesting how things go missing over time too, even from the original factory and documentation on circuitry goes in a dumpster, by accident or not. One classic example (not in computing) the Gretsch guitar company threw away their records of the early transistor treble boost amplifier called a "Sonic Boom" used in some of their early electric guitars from the '60's era.So I had to reverse engineer it:

 

Attachments

  • CBM4STATE.jpg
    CBM4STATE.jpg
    191 KB · Views: 5
Back
Top