Measurements with a repetitive time sampling scoop like the HP54503A

[Arjan Emm]

I have a HP54503A 500MHz scoop, it only does 20Ms/s. So, to get say a 10Mhz or higher repetitive signal to display properly, it has to do several acquisitions to build up the signal and display it. You can actually see trace being build up since it only does a few acquisitions a second. No questions so far.

When you get to long time bases or low frequencies I expect it to become a realtime sampling scoop but that doesn't seem to be the case.

Even at 1kHz it seems to keep adding data from previous acquisitions to the screen. I added two images to show this.
Channel 1 and 3 are two free running function generators within a Hz or so. triggered on 1, persistence minimum and connect the dots on, to exaggerate the issue. you can see at least 3 acquisitions showing up. 20Ms/s is plenty for the maximum 512 point of data on the screen in one go. Is this normal for a scoop like this?
The second picture is a single shot after clearing the screen. And looks perfect and will always do after clearing the screen. If i don't clear in between the single shots it will become a mess like picture 1 after 2 or 3 times.
Is this normal or is there something wrong with this scoop?

[David Hess]

I think this may be just a case of the HP54503A being weird.  As you point out, at such slow time/div settings it should be able to fill the entire 500 point record in one pass.  There are some odd things in the specifications though:

When doing a simultaneous acquisition on two channels, maybe it can only fill half of the points on each channel on each acquisition.  If you trigger on a third source like the line so both channels are asynchronous to the trigger, then this should show up on both channels.  I think this is the most likely explanation. (1)

Alternatively the HPIB record length it 500 points in single shot mode and 1000 points in repetitive mode at slow time/div settings indicating that it always take two acquisitions to fill.  If the display record is derived from the HPIB record, then what you observed could make sense.

The display record is 500 points but the "data display resolution" is only 451 horizontal x 256 vertical.  How does that work?  Does the record extend beyond the bounds of the display?  How many horizontal points are displayed per division?

As another test, you might try using line triggering so the acquisitions are triggered asynchronously to all input signals.

Apparently it can only simultaneously acquire channels 1 and 4.  What happens with other channel combinations?  Does it also simultaneously acquire on channels 2 and 3?

(1) The early Tektronix DSOs like the 2230 and 2232 work this way however they also halve the record length for each channel when 2 channels are used so the half length records still get filled in one acquisition.  This is not as bad as it sounds because their display with one channel is 100 points/div and even at 50 points/div, they still have higher horizontal resolution than many DSOs now.

[TAMHAN]

Phase shift between the two waveforms maybe? Pls do this: get a DPO, and trigger it to Chan1. Then plot both chans.

[David Hess]

Phase shift between the two waveforms maybe? Pls do this: get a DPO, and trigger it to Chan1. Then plot both chans.

The signals are asynchronous and have no constant phase relationship.

Channel 1 and 3 are two free running function generators within a Hz or so.

And they *must* be asynchronous to show the problem.  What Arjan Emm is saying is that during an acquisition where real time sampling should be possible, the DSO is apparently using ETS or only filling part of the record on each acquisition.  This behavior would not show up if the signals were synchronous.  That is occurs at all is weird.

[Arjan Emm]

Hi david, you totally get my point.
If the hp545 would clear it's memory on low timebases after each acquisition,  it would be much more usefull as a realtime 2MHz scope. But it doesn't. I just tried between chan1 and 2,3 or 4 the problem remains

[Arjan Emm]

Phase shift between the two waveforms maybe? Pls do this: get a DPO, and trigger it to Chan1. Then plot both chans.

Yes, certainly out of phase, I try to throw something hard at it to see how it performs. The only way to know if I can use it for fault finding later.