Very strange problem with a crystal oscillator

[patroclus]

Hello

I've been hours trying to solve a problem, with no luck.
I've got a 6502 CPU system feed by a 27MHz clock.
I bought a 32MHz crystal oscillator to raise the speed (something I already did some time ago with another 32MHz oscillator sucessfully) , the one with 4 pins; power, GND and output, with the orher one not conected.

The strange thing is it does not work now with this new oscillator.
Well, it is strange because it DOES work when I hook a logic probe of my logic analyzer to its output. System runs fine. 100%
As soon as I disconect the probe, it crashes...
When the logic probe is conected, it measures 32MHz exactly.

I'm really stuck.. please, any help!?

 

[rigdoctor99]

Did you check the soldering? Or was it a plug in device? Sounds like a dry joint but if you did not have to solder it, make sure all the pins are contacting correctly........ Thats all I can think off for now.............. Maybe you need a decupling cap on it, and the prob makes it work because it sees a capacitance

 

[tulit]

Sounds like it doesn't have sufficient load capacitance. Check the datasheet for the oscillator. Or simply try swapping out whats there already for something slightly more.

 

[William At MyBlueRoom]

Umm isn't the 6502 a 2MHz cpu (using a 4MHz crystal)? The 8088 only ran at 4.58 (using a 14MHz crystal) and the 8088 was a much newer CPU at the time.

Heck a 80486 maxed out at 100MHz and that was clock doubled.

27MHz sound way too fast for a stock 6502

 

[JimB]

The oscillator does seem a bit fast for a 6502!!!

But regarding the stopping and starting, some of those 4pin oscillators use the 4th pin as an enable.
Could it be (dont ask me how) that when the logic probe is connected to the output, the 4th pin drifts to the enabled state and the oscillator gives an output?

JimB

 

[patroclus]

This 6502 feed from an external clock and then divides internally by 12.
So, stock 27MHz is 2,25MHz
I try to pull it up to 2,66MHz. It's OK, as I've done it previosly with great results.

But this oscillator does not seem to be working properly.
I tried to put a 10pF capacitor between clk output and ground.
I tried to put a 100kohm resistor between clk output and ground.
Nothing changed.
The output is enabled.

Why the hell should it work 100% fine when a logic probe is conected to ANY of its pins (power, ground or output??)


I don't know what else to try, and I have no replacement here now..

 

[Nigel Goodwin]

This 6502 feed from an external clock and then divides internally by 12.

As I recall the 6502 divided by a MUCH lower value, four seems to sound vaguely familiar? - although perhaps I'm thinking PIC?.

The divide by 12 sounds more like a Z80, which was why a 6502 was as fast as a Z80 while using a much lower clock crystal.

I'll check when I get home, I still have all my old Mictotan 65 gear!.

 

[patroclus]

this 6502 is a slight modification used in Nintendo NES videogame console.
The stock NES clock is 21MHz, and 6502 runs at 1.75MHz according to specs docs.

 

[Nigel Goodwin]

Right, so it's NOT actually a 6502! - unless the clock is divided down before it's fed to the processor?.

 

[ZIGGY_DAN]

It sounds like your logic probe is changing the functioning of the board, probably changing the potential of the pin, i don't really know much about crystal oscillators, you could try to measure the potential of your logic probe with a digital volt meter, and then apply the same potential to the crystal (through a large resistor)

hope this helps.

 

[patroclus]

It is a 6502, but this version internally divides clock by 12.
I know it is 6502 as I made some programs for it.

 

[Nigel Goodwin]

It is a 6502, but this version internally divides clock by 12.
I know it is 6502 as I made some programs for it.

You mean it runs 6502 code! - this doesn't mean it's a 6502, does it say 6502 on the chip?.

 

[patroclus]

It isn't a standar 6502. It is a slightly modified version.
But the problem is that last time I did this it worked just fine. same circuit, same board, just different oscillator, but both of them 32MHz...
Why does it work well only when probe is conected???!!

 

[Nigel Goodwin]

As suggested, perhaps the extra capacitance?, perhaps injecting noise into the oscillator?, perhaps it's loading the circuit?.

 

[patroclus]

I'm not sure if I understand. should I try to add capacitance or something?? I already tried to add 10pF parallel capacitor between output and GND...

 

[Nigel Goodwin]

Try posting the circuit of the oscillator section, so we can see what it consists off.

 

[patroclus]

So far, it's as simple as
pin 1 - NC
pin 7 - GND
pin 8 - CLK Output, to CPU clk in pin
pin 17- +5V

Not using any other element, and so, it is isolated from the board circuit (as I previosly cut the track that delivered original clk signal to CPU).

 

[Nigel Goodwin]

Does the clock ouput perhaps require a pull-up resistor?.

Do you have a scope to check what's going on?.

 

[patroclus]

I've got no oscilloscope.
I have a logic analyzer. It has an internal frecuencimeter, and it detects perfect 32MHz clock (32,000,600 Hz to be exact). but as soon as I take the probe out, it hangs. also, sometime it runs at a erratic speed when no probe is conected but I hold the oscillator in my hands. As soon as I release it, it hangs!!

I tried pull down resistor, but not pull up... maybe I could try.

 

[William At MyBlueRoom]

I've got no oscilloscope.
I have a logic analyzer. It has an internal frecuencimeter, and it detects perfect 32MHz clock (32,000,600 Hz to be exact). but as soon as I take the probe out, it hangs. also, sometime it runs at a erratic speed when no probe is conected but I hold the oscillator in my hands. As soon as I release it, it hangs!!

I tried pull down resistor, but not pull up... maybe I could try.

**broken link removed**

The HuC6280 CPU in the NEC Turbographics is only a 7.16MHz device. The oscillator will probably oscillate at 32MHz it doesn't mean it can internally clock at that speed.

Most 6502s use /2 over crystal not a /12 (thats the internal instruction clock) so your maximum crystal is 14.32 MHz.

What's the part # on the CPU?

PS how do you know it's a perfect clock without a scope?