05-08-2012, 10:45 AM
It's a curious circuit. Although it might not look complex at a glance, I'd suggest that the operation is rather more subtle than first appears. As is often the case with circuits from this era...
I've no-doubt that most people have worked this out already; perhaps the technique is already familiar to my, ahem, more experienced friends (how's that for diplomacy?). However, I thought I'd have have a go:
In a 'scope that uses a Schmitt trigger to derive the trigger pulse, the Schmitt trigger doesn't need to work at the bandwidth of the 'scope. Rather, it just needs to respond at the right time to start a sweep - remembering that there might well be many cycles during the sweep. And if it's a bit slow, the 'scope might get a bit dimmer at higher sweep speeds as the duty cycle is reduced - probably not a big problem...
In the manual for my trusty Telequipment D83, the timebase circuit explanation uses an expression like "normal dividing action". In other words, the simple 2-transistor Schmitt circuit doesn't have to ackle along at 50MHz - and because of the storage time of the transistors, it certainly won't. But it doesn't matter if it takes a while to reset; sure a few clock pulses might be missed, but providing the Schmitt triggers at the right DC threshold, all is well on the 'scope display.
The dividers appear to rely on similar principles to achieve the division, by arranging the time constants to intentionally miss 9 pulses, responding on the 10th...
The transistors within (TR4, 5 and TR7, 8) are connected as SCRs. Looking at the first one, I'd suggest that when the emitter of TR4 reaches -3.6V, it turns on and drives TR5 on, which locks on the pair, of course, via R9. I'm less sure about how the pair are driven off, but wonder if it's from a pulse fed via C17? The arrows show that signal flows from the dividers to the pulse shaper, but I wonder if that's wrong, given that presumably quite large signals appear at that inductor at the pulse shaper?
So, I suspect that the critical factor is the time taken to charge the capacitors (C16, C22) to the SCR threshold voltage. And I note that the series capacitors appear to have parallel SOT (select on test) elements, so perhaps these can be fine-tuned to ensure it takes the right number of pulses to sufficiently charge the capacitors and trigger the transistors...
All this is conjuncture - but my starting point in any fault-finding is always to establish how it might be expected to work first of all. It might help if I had copies of the waveforms.
Anyway, given that you've changed all the transistors, I'd be suspicious of the capacitors. And perhaps the diodes. The resistors are probably OK, given that they are no high voltages and high values around, and I assume you would have spotted and checked any that run hot. It would be lovely to see immediately where it's all going wrong on the 'scope screen, rather than "blanket" repairing it, but that could be easier said than done in this case...
The potential (bad pun!) trouble with changing this to TTL is the supply voltages (OK, a 4017 decade counter would be happy with 15V), but also the shape and amplitude of the waveforms. It would require some work - potentially more than just fixing the original fault by changing every component on the board
I'm on leave from the middle of next week, so if it's still not going by then, I wouldn't mind escaping the kids for an hour or so. Being in front of the thing with a 'scope is much easier...
Cheers,
Mark
I've no-doubt that most people have worked this out already; perhaps the technique is already familiar to my, ahem, more experienced friends (how's that for diplomacy?). However, I thought I'd have have a go:
In a 'scope that uses a Schmitt trigger to derive the trigger pulse, the Schmitt trigger doesn't need to work at the bandwidth of the 'scope. Rather, it just needs to respond at the right time to start a sweep - remembering that there might well be many cycles during the sweep. And if it's a bit slow, the 'scope might get a bit dimmer at higher sweep speeds as the duty cycle is reduced - probably not a big problem...
In the manual for my trusty Telequipment D83, the timebase circuit explanation uses an expression like "normal dividing action". In other words, the simple 2-transistor Schmitt circuit doesn't have to ackle along at 50MHz - and because of the storage time of the transistors, it certainly won't. But it doesn't matter if it takes a while to reset; sure a few clock pulses might be missed, but providing the Schmitt triggers at the right DC threshold, all is well on the 'scope display.
The dividers appear to rely on similar principles to achieve the division, by arranging the time constants to intentionally miss 9 pulses, responding on the 10th...
The transistors within (TR4, 5 and TR7, 8) are connected as SCRs. Looking at the first one, I'd suggest that when the emitter of TR4 reaches -3.6V, it turns on and drives TR5 on, which locks on the pair, of course, via R9. I'm less sure about how the pair are driven off, but wonder if it's from a pulse fed via C17? The arrows show that signal flows from the dividers to the pulse shaper, but I wonder if that's wrong, given that presumably quite large signals appear at that inductor at the pulse shaper?
So, I suspect that the critical factor is the time taken to charge the capacitors (C16, C22) to the SCR threshold voltage. And I note that the series capacitors appear to have parallel SOT (select on test) elements, so perhaps these can be fine-tuned to ensure it takes the right number of pulses to sufficiently charge the capacitors and trigger the transistors...
All this is conjuncture - but my starting point in any fault-finding is always to establish how it might be expected to work first of all. It might help if I had copies of the waveforms.
Anyway, given that you've changed all the transistors, I'd be suspicious of the capacitors. And perhaps the diodes. The resistors are probably OK, given that they are no high voltages and high values around, and I assume you would have spotted and checked any that run hot. It would be lovely to see immediately where it's all going wrong on the 'scope screen, rather than "blanket" repairing it, but that could be easier said than done in this case...
The potential (bad pun!) trouble with changing this to TTL is the supply voltages (OK, a 4017 decade counter would be happy with 15V), but also the shape and amplitude of the waveforms. It would require some work - potentially more than just fixing the original fault by changing every component on the board
I'm on leave from the middle of next week, so if it's still not going by then, I wouldn't mind escaping the kids for an hour or so. Being in front of the thing with a 'scope is much easier...
Cheers,
Mark
