Re: TG501 20ns 10ns 5ns
|
Hello Kerry,
toggle quoted message
Show quoted text
I don't own a TG501 so I don't really qualify as an expert, but I downloaded the service manual from the Tekwiki (here: ) and, by just looking at the block diagram, it doesn't seem a simple tweak (e.g., the turn of a trimpot) to change the time marks of 20ns to 25, the 10ns to 12.5 and 5n to 6.25. Still, it's technically possible that only the 20n and faster ranges are off, since all the slower ranges are derived by direct division of the 1MHz reference clock, while the faster ranges, rely on a 100MHz clock internally generated by PLLing the 1MHz reference clock, which is divided by 100 (in successive divisions, ¡Â5, ¡Â2, ¡Â5 and ¡Â2) and phase compared to the reference 1MHz clock. If the error is exact (i.e. if the 20ns is turned into exact 25ns) a good theory could be that something wrong with the 1st stage of dividing by 5 (U310A, U310B and U315A) is making it divide by 4 instead. This would make the PLL to generate an 80MHz frequency, instead of 100MHz. First hypothesis that came to my mind was that a decade divider IC could have been replaced by a binary divider... But not applicable in this case, as the ¡Â5 is obtained by a discrete combination of 3 ECL flip-flops. My second hypothesis is that, maybe a false contact, (a bad IC socket comes to mind) is interrupting some signal in this ¡Â5 logic, causing it to misbehave and divide by 4. In face of the known issues that some Texas sockets used by Tektronix on the equipment of this vintage, it's worth to give it a check. See this link () for more information on the dreadful Texas IC socket. If, however, the error is not exact... than it may just be that the PLL is failing to generate the 100MHz clock. In this case, it can be easily diagnosed (even with an uncalibrated scope), by just displaying simultaneously (on 2 channels), the 1MHz reference clock, at the same time with any of the clocks generated from the - supposed - 100MHz clock. If you trigger in one and the other roll on the screen, or vice-versa, than it's the PLL that it's not locking. Hope this helps, Rgrds, Fabio On Thu, Aug 16, 2018 at 01:14 PM, <kburns@...> wrote:
|
Fabio Trevisan