Commodore PET 2001

[ajgriff]

Some very useful thoughts and ideas. Further investigation is ongoing but I can confirm that Vcc is spot on at pin 8 of both op-amps.

Having had a closer look at the down at heel C64 type unit it's PCB layout doesn't correspond exactly to any of the versions in Commodore's 1984 service manual (see post#120) but the circuit does match the schematic for the #NP-090 board aside from the use of a different manufacturer's quad op-amp. This in turn is virtually identical to the PET datasette circuit except that the two dual op-amps have been replaced by the single quad op-amp. Images of the two schematics are attached for the purposes of comparison. Helpfully the first schematic shows expected waveforms at different points in the circuit.

All this does open up some possibilities for comparative testing including the option of board swapping. Unfortunately they are not directly compatible largely because of the different positioning of the read/write switch. The battle between me and the PET continues. I apologise on behalf of us both that we are such slow workers!

Alan

[Mark1960]

The data format is designed to remove the dependence on frequency response required to reproduce a square wave output.

The write circuit uses a push/pull driver to drive the head with plus or minus approx 0.5ma, recording two opposite flux polarities to the tape.

In read mode the transition between flux polarities will generate a positive or negative pulse from the head. If the transitions are close enough together it will look like a sine wave. The first op amp in the read circuit is a differential amplifier. This is ac coupled to the next stage to remove any dc bias due to input offset voltages in the first op amp and the second op amp includes a low pass filter in the negative feedback. The third op amp is additional filtering. The final op amp is configured as a comparator with positive feedback to generate hysteresis, so a positive pulse will switch the output high and a negative pulse will switch the output low. This recovers data similar to the data recorded though though it might have some slight lengthening or shortening of the mark/space ratio.

The data is coded in the delay between high to low transitions. The longer delay is logic 1 and a short delay is logic 0. Low to high transitions are ignored but obviously required to set up for the next high to low transition. This method of coding is tolerant of tape speed variation between record and playback as it only measures the delay since the last transition, unlike the more traditional cassette interface using a simple serial protocol which is dependent on timing since the last start bit.

One disadvantage is the length of a block of data on the tape will change depending on the ratio of 1s to 0s in the data, so its not a good scheme for recording blocks of data, and then going back to rewrite only selected blocks. Though this isn’t a problem for the intended purpose of recording programs.

I’m not sure why they didn’t include low to high transitions in the coding to double the data rate. Maybe this would have made the software more complicated.

[ajgriff]

Interesting background information and will help me to check the waveform shapes. That's if I ever manage to find any sensible looking signals of course.

Although I think the PET datasette's tape head is ok I thought I'd compare the out of circuit head resistances of the heads in the two machines. The PET's head reads more or less exactly 310Ω whereas the C64's is 195Ω. I appreciate that the operational impedance of the head coils would be a more relevant measurement but it does seem odd that the resistances are so far apart bearing in mind that they both feed identical input circuitry. Is this another red herring?

Alan

[julie_m]

The difference in DC resistance most probably is just due to different manufacturers using different gauges of wire. Under AC conditions (are they generating the bias frequency in software?), the inductive reactance is going to dominate the impedance, so another hundred ohms won't make much difference (especially as it's the vector sum of two orthogonal components).

You would need a universal component tester, a multimeter with inductance ranges or an old-fashioned bridge to measure the inductance.

[SiriusHardware]

I don't think it's going to have bias at all, it looks like the record process just lays a series of forward / reverse 'magnets' on the tape.

DC or AC Bias is used in audio recorders to overcome the reluctance of the tape to be magnetised by a small amount of signal, which is a problem when you are trying to record audio with its large variations in amplitude, but here it looks as though the unit just flat-out magnetises the tape with one polarity or the other.

[Timbucus]

Originally Posted by SiriusHardware

I don't think it's going to have bias at all, it looks like the record process just lays a series of forward / reverse 'magnets' on the tape.

DC or AC Bias is used in audio recorders to overcome the reluctance of the tape to be magnetised by a small amount of signal, which is a problem when you are trying to record audio with its large variations in amplitude, but here it looks as though the unit just flat-out magnetises the tape with one polarity or the other.

Sort of like they were trying to create core memory on a piece of mylar tape...

[ajgriff]

The PET seems determined to keep on teaching me new things and now we're on to techniques for digitising magnetic tape. By the way Julie's quite right about the head resistances being irrelevant as I've managed to prove today. Also there's no head bias used in these machines as outlined by Sirius.

Having reconnected the heads in the two datasettes and recorded a new test tape with much longer recordings of sine (instead of square) waves, I set about comparing the signals at various points in each circuit. To my surprise the PET datasette started yielding sensible signals at the outputs of the first two op-amps. It would seem that there was a poor connection around the wiring to the head. Either that or the read/write switch is still operating intermittently despite the squirts of contact cleaner. On the downside the signal dies at the third op-amp. Progress of sorts and I'll be checking for dry joints and faulty components next. At least I now know that the heads are indeed ok in both machines.

Alan

[Slothie]

Originally Posted by SiriusHardware

I don't think it's going to have bias at all, it looks like the record process just lays a series of forward / reverse 'magnets' on the tape.

DC or AC Bias is used in audio recorders to overcome the reluctance of the tape to be magnetised by a small amount of signal, which is a problem when you are trying to record audio with its large variations in amplitude, but here it looks as though the unit just flat-out magnetises the tape with one polarity or the other.

The tape in my PET which was the older design with the white keys definitely used a DC bias because you had to demagnetise the head regularly. Later ones didnt have this problem to the same degree, but they did revise the electronics several times.

[ajgriff]

I gather that the early PET datasettes were rebadged versions of ordinary Sanyo audio cassette recorders. Interesting (to me anyway!) video about Commodore datasettes:

https://www.youtube.com/watch?v=5kRmlG9cCMI

Might try and get hold of the booklet mentioned at the beginning.

Alan

[Slothie]

Originally Posted by ajgriff

I gather that the early PET datasettes were rebadged versions of ordinary Sanyo audio cassette recorders. Interesting (to me anyway!) video about Commodore datasettes:

https://www.youtube.com/watch?v=5kRmlG9cCMI

Might try and get hold of the booklet mentioned at the beginning.

Alan

The booklet is a free pdf download. You do need to put in your email address to get it...

[ajgriff]

Yes, found it ....

Alan

[ajgriff]

This is a link to the 'Commodore Tape Recorders' booklet:

https://www.oldcomputr.com/product/c...ape-recorders/

The free download offer is a tempter to buy one or more of the author's Commodore related books. Nothing harmful about the site as far as I can tell.

Alan

[ajgriff]

Quote lifted from the PROM Adapter thread:

Originally Posted by Slothie

Its looking like if I come up with a semi-useful diagnostic tool then I will stick it on a ROM somehow and send it to AJ to test (presuming he hasn't passed on his PET to some other lucky soul by then!).

Not much chance of the PET going anywhere soon although it could go wrong again at any moment, especially if I sneeze near it. It's more than two months since I started this merry little game and the joust between us isn't over yet.

Alan

[ajgriff]

I’m not thinking in terms of starting a new thread about Commodore C2Ns as the work I’m doing on the PET’s datasette is just part of the overall refurbishment project. However, if there seems to be enough interest I’ll perhaps do a Success Stories write up once the datasette’s working properly. I could include some background information about the various versions plus some useful reference material. This might then provide a more unified resource for the benefit of future restorers.

Getting back to the datasette itself, I’ve checked the components around the third op-amp where the signal on read (play) currently vanishes and have found no evidence of a fault. As a result a pack of replacement LM358s is on order plus some 8 pin DIP sockets as I’ve none left in stock.

In the meantime I’ve been experimenting with head alignment as the C2N’s azimuth adjustment screw has clearly been twiddled in the past. There are commercial alignment tapes with associated software for the C64 but there doesn’t seem to be anything similar available for use with a PET. However, by chance I found a document (see attached) via Google which gives guidance on how to adjust C2N azimuth using an oscilloscope and a test tape. Unfortunately there’s no information provided about what was actually recorded on the test tape but I’ve assumed that it would have been a sine wave at some frequency or other.

Of course I could only contemplate trying this alignment method as the machine is now working as far as the first dual op-amp is concerned. For the purposes of testing functionality I’ve so far been using a pre-recorded test tone of 500Hz as most heads will pick up a lowish frequency signal even when quite badly misaligned. For accurate azimuth adjustment I needed to record a higher frequency tone. I know that 6kHz is often suggested as being appropriate for inexpensive portable cassette recorders but the datasette is a fairly basic device so I opted for 4kHz. To adjust the azimuth screw the chassis of the datasette needs to be key side up which makes it difficult (not enough hands) to probe the first op-amp’s output at pin 7 so I soldered a flying lead to a convenient PCB pad and attached the scope probe’s gripper to it. With this set up I was able to get a nice strong 4kHz signal that could readily be peaked which was not the case when I was using the 500Hz tone. Clearly I won’t know whether or not this has done the trick until the rest of the read circuit’s working correctly.

All fairly basic stuff but I thought an update might be of interest.

Alan

[ortek_service]

If Commodore C2N etc. Datasettes always record full-amplitude squarewaves, rather than using usual sinewaves, then a standard audio alignment sinewave recording may not be the best one to use.

And there may be a computer test program, that you can play the signal into
- I seem to recall seeing an on-screen display with others, although they were being fed with an analogue signal rather than Commodore's already digitised one.

With stereo cassette players, apparently you adjust the head alignment for the correct phase between the channels on the 'scope using X-Y Lissajous figures.

But for mono tape recorders, I always found just adjusting the head for best highest frequency response (max. treble), by ear generally worked OK, on either the continuous-tone at the start of most computer save recordings or the alternating tones data part.

So may have to just scope a suitable point in the C2N circuitry (whilst still analogue, and not clipping?) and adjust for max. level using a few kHz tone / maybe and actual Commodore computer recording.
It may be best to use a pre-recorded Commodore cassette (even a C64 one, if format is similar), otherwise the head alignment of what you record the test tape on will then be a factor if it isn't quite right.
If you do have access to a C64, then maybe able to connect the PET's Datasette to that, and use some available software / test tape.

[ajgriff]

Originally Posted by ortek_service

If Commodore C2N etc. Datasettes always record full-amplitude squarewaves, rather than using usual sinewaves, then a standard audio alignment sinewave recording may not be the best one to use.

Compact cassettes are problematic when it comes to square waves because of the limited frequency range inherent in the media and the recording/playback method. That’s why Commodore designed the datasette’s circuitry in the way that it did. The waveforms illustrated in the first image of post #201 are instructive. Note that no attempt is made to record actual square waves.

Originally Posted by ortek_service

And there may be a computer test program, that you can play the signal into - I seem to recall seeing an on-screen display with others, although they were being fed with an analogue signal rather than Commodore's already digitised one.

The PET 2001-8 doesn’t give any visual or audio indication of what’s happening when writing or reading programmes and I haven’t yet found any reference to a test programme for datasette head alignment.

Originally Posted by ortek_service

With stereo cassette players, apparently you adjust the head alignment for the correct phase between the channels on the 'scope using X-Y Lissajous figures.

Yes but this a mono cassette recorder and phasing between channels is not an issue.

Originally Posted by ortek_service

But for mono tape recorders, I always found just adjusting the head for best highest frequency response (max. treble), by ear generally worked OK, on either the continuous-tone at the start of most computer save recordings or the alternating tones data part.

Agreed that the best high frequency response is the aim of the game. However, I didn’t see the point of rigging up some kind of audio output particularly when an oscilloscope is more accurate than my ears.

Originally Posted by ortek_service

So may have to just scope a suitable point in the C2N circuitry (whilst still analogue, and not clipping?) and adjust for max. level using a few kHz tone / maybe and actual Commodore computer recording.

This is the method I used as prescribed by Commodore in the image attached to post #214. I'll test the datasette's actual performance once it's fully functional.

Originally Posted by ortek_service

It may be best to use a pre-recorded Commodore cassette (even a C64 one, if format is similar), otherwise the head alignment of what you record the test tape on will then be a factor if it isn't quite right. If you do have access to a C64, then maybe able to connect the PET's Datasette to that, and use some available software / test tape.

I don’t have ready access to a C64 or C64 software tapes and was happy to use Commodore’s recommended method of alignment. The test tone was recorded using a good quality Yamaha deck which plays commercially recorded music tapes exceptionally well, for a cassette player, so I’m not worried about possible head misalignment at source.

Alan

[ScottishColin]

Just to let you know I'm watching this with interest. As you know I'm still battling to get my PET working, but I do have 3 C2N tape drives here to work on onoce the PET is working. I've opened them all and they all have 320275 boards in them.

Two are black, and one white (well, it was white) with a tape counter.

I've ordered some new drive belts as they are all unsurprisingly perished. One of them still has a tape left in by a previous owner....

Colin.

Originally Posted by ajgriff

I’m not thinking in terms of starting a new thread about Commodore C2Ns as the work I’m doing on the PET’s datasette is just part of the overall refurbishment project. However, if there seems to be enough interest I’ll perhaps do a Success Stories write up once the datasette’s working properly. I could include some background information about the various versions plus some useful reference material. This might then provide a more unified resource for the benefit of future restorers.

Getting back to the datasette itself, I’ve checked the components around the third op-amp where the signal on read (play) currently vanishes and have found no evidence of a fault. As a result a pack of replacement LM358s is on order plus some 8 pin DIP sockets as I’ve none left in stock.

In the meantime I’ve been experimenting with head alignment as the C2N’s azimuth adjustment screw has clearly been twiddled in the past. There are commercial alignment tapes with associated software for the C64 but there doesn’t seem to be anything similar available for use with a PET. However, by chance I found a document (see attached) via Google which gives guidance on how to adjust C2N azimuth using an oscilloscope and a test tape. Unfortunately there’s no information provided about what was actually recorded on the test tape but I’ve assumed that it would have been a sine wave at some frequency or other.

Of course I could only contemplate trying this alignment method as the machine is now working as far as the first dual op-amp is concerned. For the purposes of testing functionality I’ve so far been using a pre-recorded test tone of 500Hz as most heads will pick up a lowish frequency signal even when quite badly misaligned. For accurate azimuth adjustment I needed to record a higher frequency tone. I know that 6kHz is often suggested as being appropriate for inexpensive portable cassette recorders but the datasette is a fairly basic device so I opted for 4kHz. To adjust the azimuth screw the chassis of the datasette needs to be key side up which makes it difficult (not enough hands) to probe the first op-amp’s output at pin 7 so I soldered a flying lead to a convenient PCB pad and attached the scope probe’s gripper to it. With this set up I was able to get a nice strong 4kHz signal that could readily be peaked which was not the case when I was using the 500Hz tone. Clearly I won’t know whether or not this has done the trick until the rest of the read circuit’s working correctly.

All fairly basic stuff but I thought an update might be of interest.

Alan

[ajgriff]

Originally Posted by ScottishColin

Just to let you know I'm watching this with interest. As you know I'm still battling to get my PET working, but I do have 3 C2N tape drives here to work on once the PET is working. I've opened them all and they all have 320275 boards in them.

Two are black, and one white (well, it was white) with a tape counter.

I've ordered some new drive belts as they are all unsurprisingly perished. One of them still has a tape left in by a previous owner....

Colin.

After the work you've been doing on the PET you'll have no trouble sorting out the datasettes. You might even want to have a go at returning the off-white one to its original colour using the Retrobright process which Tim has been experimenting with recently:

https://www.vintage-radio.net/forum/...d.php?t=176883

I certainly made life difficult for myself by 'losing' the original datasette and then buying a wreck as a replacement.

Alan

[ScottishColin]

If one of the three I have will help you, you're more than welcome to it.

Colin.

Originally Posted by ajgriff

After the work you've been doing on the PET you'll have no trouble sorting out the datasettes. You might even want to have a go at returning the off-white one to its original colour using the Retrobright process which Tim has been experimenting with recently:

https://www.vintage-radio.net/forum/...d.php?t=176883

I certainly made life difficult for myself by 'losing' the original datasette and then buying a wreck as a replacement.

Alan

[ajgriff]

Originally Posted by ScottishColin

If one of the three I have will help you, you're more than welcome to it.

Colin.

That's very kind Colin, thank you. At the moment I'm hoping that the parts currently on order will complete the restoration of the datasette and indeed the PET itself. However, I'll keep your offer in mind just in case.

Alan