HP 1720A Oscilloscope, loss of bandwidth

[dzseki]

I've faced now with a strange problem. It seems my old HP 1720A oscilloscope lost some bandwidth... A few months ago I was able to measure rise times around 1.5ns (driven by my pulse generator which has 750ps self rise time). But since the past weeks I can't measure risetime below 2ns with the same setup. Apart from this the scope behaves similarly as before and the signal shape is also not worse than before, only the rise time slowed somewhat -strangely.
Of course my pulse generator would have been suspected as the source of the problem, but to rule that out I quickly built Jim Williams' pulse generator of with I also got the "now-unbreakable" 2ns rise time.
There aren't much trimmers in the scope to adjust pulse response and while I played with them found that they can't make the things any better.
Any ideas what could have happened?

[grumpydoc]

It could be anywhere in the vertical signal path

Is it the same for both channels?

[tggzzz]

My HP1740A had a common fault which sounds similar to yours: its risetime had risen from 3.5ns to 7ns.

The cure was as easy as it was surprising... without removing it from the scope, gently fondle, caress and squeeze the delay line. But do be careful of high voltages etc.

Apparently the cause is that chemical decomposition inside the delay line causes the wires in the braid to become insulated from each other. Why that should cause the effect, I don't know. But I did cure it

Let us know if it works for you.

[dzseki]

@grumpydoc: Yes, it's the same.

@tggzzz:  thanks will give a chance to the delay line, by simply twisting it gives a permament cure?

[tggzzz]

@grumpydoc: Yes, it's the same.

@tggzzz:  thanks will give a chance to the delay line, by simply twisting it gives a permament cure?

Get the individual wires in the braid to move against each other, by squeezing the coil of cable while it is still inside the scope.

[grumpydoc]

@grumpydoc: Yes, it's the same.

In that case the problem is probably the delay line driver, the delay line itself (as has been suggested) or the final Y amplifier.

[dzseki]

Well, after molesting a few things around the preamp area the rise time is back to normal again, but I'm not happy since I did not find any exact cause, but I don't think the delay line is the culprit. In fact I resoldered some spring fingers of the Channel A attenuator's where it connect to the main board, but that would not affect the other channel's response.

[dzseki]

So in the past years I went through several other digital oscilloscopes with higher bandwidth. Scopes come scopes go but the 1720A remains the same, I guess we grown together...
That said, this problem is still not solved ultimately.

A few addition, it turned out that the self rise time of my pulse generator is ~400ps and considering the scope should have 1.3ns rise time with this setup I should be able to see 1.4 ns rise time on the screen, so this is the goal.

In 2016 I bought a complete vertical amplifier board for this scope, and verified that isn't the problem.

Tggzzz's suggestion looks to be true indeed as poking around the delay line seems to solve the problem, but only temporarily...
But then look into this problem deeper, how could the delay line affect the bandwidth? When I got the replacement vertical amp board it came with a piece of the delay line cut, so I could examine it a little.
In this scope the delay line is constucted to have 165 Ohm impedance according to the manual. it is a differential coax (twinax?) cable. The dielectric material is rather  hard plastic with opaque color, the inner inductors are solid, by the look of it silver plated copper. The shield is a generic tin plated copper braid.

Now what is exactly the role of the shielding in terms of bandwidth?
- Since the drive is differential it should not carry useful signal.
- As I see its purpose mainly to define the impedance of the cable, but then if the impedance is not matched the rise time is not affected (as I experienced so far). Anyway from this point of view it makes no difference whether the braid inductors are isolated from eachother or not, so long they are connected to ground at either end, right?

[dzseki]

Old thread, but it doesn't mean it is forgotten

So I still have this scope, and while its bandwidth performance is somewhat stabilized but it is still not where it should be...
With my 400ps rise time pulse generator I can make only 1.6ns rise time on the scope's screen, which is more than the theoretically expected 1.4ns, so far nothing new...

Lately I have acquired a levelled signal generator, and attempted to record the bandwidth of the scope. As it looks like the vertical amplifiers rather need a complete compensation readjustment. Unfortunately the service manual is not going into much details regarding the HF adjustments, but from the recorded curve some breakpoints appear clearly. It does not look like a hopeless quest.
BTW the scope's trigger works up to 450Mhz.