Atari Heavy Sixer - Crystal Issues

[+DrVenkman]

I started a thread in the 2600 forum but I'm still struggling with this issue so thought I'd post here in case anyone in the wider retro/hardware community has run into an issue with a crystal oscillator circuit running below-spec. 

 

The TL;DR here is I bought an interesting Taiwan-made Atari Heavy Sixer. Internals appeared to have been original and untouched. Powered on to a black screen. I've been diagnosing it for a couple weeks without complete success.

 

The longer version:

• Good power and ground at all IC's, no obvious shorts or damaged components.
• Socketed IC's work in another system, and swapping replacements into this one made no difference.
• Oscilloscope showed that I had no clock input on OSC, pin 11 of the TIA. Instead I had a solid 1.14V with no oscillation at all. 

 

Here's the clock circuit schematic from the Atari VCS Field Service Manual:

 

 

• I removed and replaced both 3906 transistors in the clock circuit but no change. Both old ones tested good anyway once removed.
• I removed the two 1N914 diodes and replaced with modern 4148's (like-for-like modern equivalents), but no change.
• I ordered replacement crystals and installed one. Now the system works but ... color spazzes in and out and won't lock-in. 
• I scoped the OSC pin with the new crystal and found I was about 860 Hz slow, way out of spec for the NTSC colorburst signal
• Thinking it was an off-spec new crystal I replaced it with another new one and no change.
• So then I pulled a vintage crystal from another working system and that one still runs slow in the Heavy.
• I then replaced C202, C203, C208 and C209 ceramic caps with new ones but again no change.

 

I am about reduced to pulling all the resistors from the clock circuit off the board to measure them out of circuit even though they rarely fail, but there's frankly nothing else I can imagine which would drag down a crystal oscillator circuit unless one of them is vastly out of spec somewhere. I have modern replacements for all of them except R211, which is the 500K color pot. But that one, at least, is not failed short and when the system is powered, results in approximately the correct range of values for Vdel measured at Pin 10 of TIA, so I really doubt that's the cause of the entire clock running slow like this. 

 

Here's a link to my post in the 2600 forum that has some scope pics and such. 

 

 

Anyway, if anyone has any thoughts, I'm all ears

 

 

 

[+Stephen Moss]

Just speculating here but...

Have you checked that R200 and R210 are getting the coreect 5V supply at the resistors and not just that is it correct out of the regulator, it is possible that if the voltage is incorrect may throw of the biasing of the transistors therby throwing off the frequency a little.

 

Also, I know that with regard to PIC Mocrosontrollers there is usually a part in the ocsillator section that states.... 

The oscillator design requires the use of a parallel cut crystal.

Note: Use of a series cut crystal may give a frequency out of the crystal manufacturer’s specifications.

 

So maybe the new crystal is of the wrong cut and that is throwing the frequency off, although if that were the case I do not know if the resulting frequency would be off by more of less than the 0.2% you are reporting.

Additionally, I am not sure which type you need but as crystals used with a PIC are typically connected across two Clock pins (unless an oscillator module is used) there would be some logic in thinking that a serial cut crystal would therefore need to have a lead to GND and so given the circuit design in the schematic it would seem likely a parallel cut crystal is required.

Unfortuately, in my experiance the majority of crystals do not appear to specify which cut they are, which is not very helpful and so why I use an oscillator module instead when driving a PIC with an external clock source.

[+DrVenkman]

3 hours ago, Stephen Moss said:

Have you checked that R200 and R210 are getting the coreect 5V supply at the resistors and not just that is it correct out of the regulator, it is possible that if the voltage is incorrect may throw of the biasing of the transistors therby throwing off the frequency a little

I’m not sure if I have made those specific measurements but I will definitely do so later this afternoon.

 

3 hours ago, Stephen Moss said:

So maybe the new crystal is of the wrong cut and that is throwing the frequency off

I’m certain that’s not an issue because I’ve actually gone so far as to transplant a working vintage crystal from another 2600 into this board and I’m experiencing the same slow-down. 

[+DrVenkman]

I've now verified good +5V at R200, R201 and CR200. Well, each one is right at 4.94V, but pin 3 of the VR is 4.97 so it's what I would expect, and nothing that should affect the clock.

 

I've also lifted and tested R00, R204 and R205. All measure within a couple ohms of nominal value. I've previously removed and tested R203 and it's good, too. Basically the only ones I have not lifted or removed to test are R202 and R201. I'm pretty sure R201 is good because it forms part of a voltage divider which should give just a little under 4V at transistor Q200 and at the positive side of C202, both of which track with my tests on the board with power applied. I guess for the sake of completeness I should lift it and check it for sure though, along with R202.

 

But so far, nothing else I've checked has actually been wrong with this board - it's incredibly frustrating.