Closed LumaTOLB - LumaFix with TOLB combo

Status
Not open for further replies.
I assume you've removed the clock board of your SX-64 because you wanted to convert it from PAL to NTSC or because there was a problem with the board present.

However, please note I do not recommend using LumaTOLB on the SX-64 because of the way the VIC-II is cooled. The heat sink is rather large and that's perfectly normal and LumaTOLB slides under this IC and elevates it even further possibly causing physical conflicts or short circuits. I don't recommend using a smaller heat sink.

The reason for wanting to use LumaTOLB is because the 8.182MHz crystal is faulty and I've found them to be very much unobtainium.

Very good point regarding the heatsink. Yes, that's likely to cause problems I hadn't considered. Will have to look at alternatives.

Thanks!
 
Last edited:
The reason for wanting to use LumaTOLB is because the 8.182MHz crystal is faulty and I've found them to be very much unobtainium.

Very good point regarding the heatsink. Yes, that's likely to cause problems I hadn't considered. Will have to look at alternatives.

Thanks!
There is no such thing as a 8.182MHz crystal. In the SX-64 as well as older C64 breadbins 326298 and 250407, a crystal of 14.31818MHz is used and then a group of discreet components (a PLL and various logic ICs) are used to make a 4/7 syncromesh conversion. Of course, we are talking about NTSC machines here.

In later C64, the C64c and the C128, the 8701 performs this conversion. It is very important the 2 frequencies be syncromesh or there will be a video 'swimming' effect on screen as the color encoding becomes async with the pixel clock..

On PAL machines, the crystal is 17.7344MHz and the conversion ratio is 4/9 but it's generally the same principle.

LumaTOLB is, of course, a possible solution to a failing clock architecture on the SX-64 but there are other possible alternatives.

Would you have a photo of the board which generates the clock signals on the SX-64 ? I could make a custom version of TOLB specifically for that machine.
 
Last edited:
The reason for wanting to use LumaTOLB is because the 8.182MHz crystal is faulty and I've found them to be very much unobtainium.

Very good point regarding the heatsink. Yes, that's likely to cause problems I hadn't considered. Will have to look at alternatives.

Thanks!
There is no such thing as a 8.182MHz crystal. In the SX-64 as well as older C64 breadbins 326298 and 250407, a crystal of 14.31818MHz is used and then a group of discreet components (a PLL and various logic ICs) are used to make a 4/7 syncromesh conversion. Of course, we are talking about NTSC machines here.

In later C64, the C64c and the C128, the 8701 performs this conversion. It is very important the 2 frequencies be syncromesh or there will be a video 'swimming' effect on screen as the color encoding becomes async with the pixel clock..

On PAL machines, the crystal is 17.7344MHz and the conversion ratio is 4/9 but it's generally the same principle.

LumaTOLB is, of course, a possible solution to a failing clock architecture on the SX-64 but there are other possible alternatives.

Would you have a photo of the board which generates the clock signals on the SX-64 ? I could make a custom version of TOLB specifically for that machine.
Interesting. There's a 14.31818 crystal, then there's another crystal immediately below that (see photos). I had replaced the 14.31818 and now at least get some coloured raster bars on screen, but not much else (previously just white screen). On the scope, I now get a clean 14.318... On the output that's supposed to provide 8.182MHz, I get a signal that's fluctuating all over the place. I'd assumed the crystal beneath the 14.31818 at top was 8.182 and was also faulty. I haven't been able to find a schematic for the clock module itself, but I guess someone that's familiar with these circuits would find it simple enough. I've been trying to wrap my head around oscillator circuits but sadly, I have much to learn..

I'll work on this board tomorrow, but it sounds like a good start would be to remove that lower crystal to see what it's meant to be. The parts list shows a 16MHz crystal (Y1) that I haven't located on the CPU board. Perhaps that's it? I'd certainly be very relieved if that were the case. The logic ICs are all commonly available (74LS74AN, 74LS161A and 74LS04P).

Thanks!

b527de747fa524b7aa5cd218eea2ca53.jpg
72a0ddded66c043863738dd0ceec015d.jpg
706b31008fa2a77e10d00ffe8b3e1b0c.jpg
 
Interesting. There's a 14.31818 crystal, then there's another crystal immediately below that (see photos). I had replaced the 14.31818 and now at least get some coloured raster bars on screen, but not much else (previously just white screen). On the scope, I now get a clean 14.318... On the output that's supposed to provide 8.182MHz, I get a signal that's fluctuating all over the place. I'd assumed the crystal beneath the 14.31818 at top was 8.182 and was also faulty. I haven't been able to find a schematic for the clock module itself, but I guess someone that's familiar with these circuits would find it simple enough. I've been trying to wrap my head around oscillator circuits but sadly, I have much to learn..

I'll work on this board tomorrow, but it sounds like a good start would be to remove that lower crystal to see what it's meant to be. The parts list shows a 16MHz crystal (Y1) that I haven't located on the CPU board. Perhaps that's it? I'd certainly be very relieved if that were the case. The logic ICs are all commonly available (74LS74AN, 74LS161A and 74LS04P).

Thanks!
I am familiar with these circuits and I don't find making clock conversion, especially those with odd factors simple at all. They can be pretty unpleasant to deal with because they almost always involve doing analog effects with digital circuits. Proper operation and values of the resistors and capacitors involved is cryptic. Sometimes you even have small inductors in there and when they are broken, you have to have exactly the same part number from exactly the same manufacturer.

A 16MHz clock is necessary for the 1541 drive portion to operate properly so I would assume that's why you get a 16MHz crystal in there. I also assume the crystal that's under the one marked as 14.318 is in fact the 16MHz one you can't find. The logic circuits and surrounding discreet analog parts must be responsible for making the conversion to 8.182MHz and finding what's wrong in there usually takes the same amount of time as replacing all parts with new ones... (does anybody really want to do that??)

Thank you for the detailed photo. I don't know what connection does what exactly in there. I assume you know exactly which connection does what so I suggest I make you a TOLB NTSC with a description of the pin functions and then you can simply take this board off completely and replace it with this customized TOLB.

For the 16MHz clock, well, I do have a spare 16 MHz standard format 5V oscillator that's been busy collecting dust for the last 20 years.
 
Last edited:
I am familiar with these circuits and I don't find making clock conversion, especially those with odd factors simple at all. They can be pretty unpleasant to deal with because they almost always involve doing analog effects with digital circuits. Proper operation and values of the resistors and capacitors involved is cryptic. Sometimes you even have small inductors in there and when they are broken, you have to have exactly the same part number from exactly the same manufacturer.

A 16MHz clock is necessary for the 1541 drive portion to operate properly so I would assume that's why you get a 16MHz crystal in there. I also assume the crystal that's under the one marked as 14.318 is in fact the 16MHz one you can't find. The logic circuits and surrounding discreet analog parts must be responsible for making the conversion to 8.182MHz and finding what's wrong in there usually takes the same amount of time as replacing all parts with new ones... (does anybody really want to do that??)

Thank you for the detailed photo. I don't know what connection does what exactly in there. I assume you know exactly which connection does what so I suggest I make you a TOLB NTSC with a description of the pin functions and then you can simply take this board off completely and replace it with this customized TOLB.

For the 16MHz clock, well, I do have a spare 16 MHz standard format 5V oscillator that's been busy collecting dust for the last 20 years.


OK, the plot thickens. I've just lifted the top crystal and the one beneath is indeed 8.1818MHz (see photo)!

clock.jpg

For now, I have managed to source another complete CPU board complete with clock module, so i'll give that a go (I'm hoping the crystals on that one are both fine).

I do currently have an NTSC TOLB here, so i can hook it up with a breadboard to the required pins.

Thanks!
 
Last edited:
OK, the plot thickens. I've just lifted the top crystal and the one beneath is indeed 8.1818MHz
Well, well... what do you know! Instead of having a conversion circuit such as the one on the 250407, they used 2 different crystals.

I understand what the 74LS04 would be used for then but what about the 74xx74 and the 74xx161, a flip-flop and a counter. What about the cluster of resistors and capacitors around these ?

For now, I have managed to source another complete CPU board complete with clock module, so i'll give that a go (I'm hoping the crystals on that one are both fine).
This must have been quite expensive to get. SX-64 and related components are getting pretty tough to get.

I do currently have an NTSC TOLB here, so i can hook it up with a breadboard to the required pins.
Pins 2 and 9 are ground. Pin 15 is 5Vdc IN, pin 6 is the dot clock (8.181 MHz) output, pin 8 is the color clock (14.31818MHz) output. All other pins are NC.

'-)
 
I understand what the 74LS04 would be used for then but what about the 74xx74 and the 74xx161, a flip-flop and a counter. What about the cluster of resistors and capacitors around these ?

The ~1MHz signal out of the VIC is fed back into that clock module, so perhaps that's all there to clean up the noise?

Pins 2 and 9 are ground. Pin 15 is 5Vdc IN, pin 6 is the dot clock (8.181 MHz) output, pin 8 is the color clock (14.31818MHz) output. All other pins are NC.

'-)

OK. So, it's a pretty ghetto hack (for now), but I have a working SX-64. Yay!

Now I can continue work on refurbishing this beauty. Many things to do (including trying.my hand at a simple PCB adapter).

Thanks so much!
8178db4c2be19e9f3881f458132c0241.jpg
6c26885ed57dd13409a4cbc2fb3deef6.jpg
 
I was jumping up and down when I saw your photos!

Yes! A TOLB was used to fix an SX-64 !! Am I correct in understanding you connected both grounds ?

I see +5Vdc connected to the lower left solder point, the dot clock to the top center (under the TOLB module), the color clock to the lower right, one ground to the top right and another to the bottom center. Correct ?

I would suggest you make your ground wires as short as possible, preferably less than 1cm. You should be able to move the TOLB 1cm to the right to make them much shorter. This would also make the wire carrying the 14.31818MHz signal shorter which would make it better, less ringing.
 
I was jumping up and down when I saw your photos!

Yes! A TOLB was used to fix an SX-64 !! Am I correct in understanding you connected both grounds ?

Yep, both connected.

I see +5Vdc connected to the lower left solder point, the dot clock to the top center (under the TOLB module), the color clock to the lower right, one ground to the top right and another to the bottom center. Correct ?

Probably best if I post a picture;

clock.jpg


I would suggest you make your ground wires as short as possible, preferably less than 1cm. You should be able to move the TOLB 1cm to the right to make them much shorter. This would also make the wire carrying the 14.31818MHz signal shorter which would make it better, less ringing.

Thanks. I'll shorten everything up tonight and see how that goes. There is some noise (ringing?), but it's nothing unusual. I'm going to try my hand at KiCAD later and maa little adapter board for it.

Cheers!
 
I am proud to announce there will be a new batch of LumaTOLB for the C64.

New boards should become available within the next 2 months.

Thank you all!
 
It took quite a long time but I finally got the missing parts. The PCB has been updated too.

I expect a new batch to be available early May.
 
declaring interest for 1. Located in Toronto Canada. Have original unit in NTSC. Is it fair to assume this will replace the discrete logic under the metal plate (with the 2 pots?) and should rid of most of the jailbars I currently see with the original ceramic VIC?
 
declaring interest for 1. Located in Toronto Canada. Have original unit in NTSC. Is it fair to assume this will replace the discrete logic under the metal plate (with the 2 pots?) and should rid of most of the jailbars I currently see with the original ceramic VIC?
Yes, it replaces the discreet logic generating the dot clock and color clock present on early C64 boards.

LumaTOLB gives slightly better results than LumaFix as it uses the faster 74AC14 logic IC but the difference is not very noticeable. AFAIK the ceramic VIC-II generates vertical striping not present on the plastic version which LumaFix may not be able to solve.
 
Status
Not open for further replies.
Back
Top Bottom