Hi Torch
Do you think that it might be related to the fact that my Ultrascope software was fairly recentversion 2.08 I believe and that by downgrading my version fron 02.05 SP2 to 02.02.SP2 the Ultrascope software doesn't work with this version?
I didn't realize this was asked in two different threads, but Chet T16's post looks like a very good answer.
Anybody found a simpler way, or know where all the darned files are?
There couldn't possibly be a simpler way!
Direct link to attachment
Thanks Chet for the link, guess I'll have to read all 49 prev pages first, Ultrascope does not run either with missing dll error, another google required.
Thanks Chet for the link, guess I'll have to read all 49 prev pages first, Ultrascope does not run either with missing dll error, another google required.
You don't need to read the whole story. You need to read page 1, post 1. All the critical information has been continuously updated in that post through the dedicated efforts of Polossatik.
To make things easier, I put all the files together in one package that you can download from
DS1052 Upgrade tools zip file
EXCEPT Ultrascope (which comes from Rigol)
Great stuff torch, have got everything, even Ultrascope now functions ok, cheers mate!
Great stuff torch, have got everything, even Ultrascope now functions ok, cheers mate!
Solution: increase the pulse duration so it's much longer than the rise time. This ensures that the rising edge is over before the falling edge appears at the input. A charge line can be used to increase this pulse duration
Unfortunately, he doesn't actually give his measurements in that post that I can see, which may be important since I'm not sure he's using the same stuff.
The app note specifies 40" of 50 ohm "hard line or at least teflon-based & rated for HF transmission", but he says he used RG-174/u, which is polyethylene based. (I think "hard line" refers to the thick direct-bury co-ax with the solid shield?) RG-178/u is teflon based though -- maybe it was a typo? Or maybe it's not actually that important?
What is actually more important: will this technique work with the AN47 circuit or does it require the AN94 version? There are some differences I assume the most important is the change in the cap value to 10pf and the addition of the damping resistor? The array of 200 ohm resistors in parallel add up to equal the 50 ohm termination of the AN47, why would he splay them out like that? Spreading the current distribution to minimize RF or something?
First, I switched out the probes. Instead of the supplied Rigol 150MHz probe, I used a 250MHz probe (Coline M12W). I carefully adjusted both matching pots according to the directions. I realize that this is not fair, given that the purpose of this was to test the system supplied by Rigol as a whole, but thought it might make a difference.
I also selected "Dots" and turned on persistence. Here is the result:
You can see the calculated rise time is just about the same as before. Notice that the line is a touch thick in places -- my circuit is still on the breadboard which I *think* is the cause of the jitter.
While in both cases, the leading edge of the pulse is clearly shown, in the 5ns shot, the *trailing* edge has dropped further back to 0. Using the manual cursors, It appears that the Rigol is using the displayed trailing edge when determining the minimum voltage for the rise-time calculation, NOT the displayed leading edge. I don't think this is the right way to do it. Correct me if I'm wrong, but rise time should be looking at the pre-trigger minimum, not the post-trigger minimum, should it not?
So, allowing for manual error and the fudge factor caused by the jitter in my circuit, I think it's fair to say the rise time is 2.5ns +/- .06ns? Which suggests a Gaussian bandwidth of 3.4\2.5*1000 = 136MHz?
EDIT: Forgot two things: first, thanks to tnt and alm for teaching me not to trust the automatic measurements. And second, this seems to confirm that telling the scope it's now a 150MHz DS1152E-EDU model means bumpkiss. Once the input filter is defeated with the 100MHz mod, that's all you get. Anything more requires a hardware mod, if it's possible at all.
but for the purpose of evaluating the bandwidth of the scope, I would ignore the probe. Ideally I would directly connect the pulser to the input with a feed-through terminator in between, I think this is how Jim Williams did it. Testing probes is also useful, but basically a separate issue.
So no issue with interpolation or sampling. My initial thought was that the automated measurement maybe used the uninterpolated signal, and someone has shown that the sin x/x interpolation is not perfect.
I hope that's not a solderless breadboard, I would build a circuit like this on a piece of copperclad if I didn't have a proper PCB. Any parasitics may severely impact the performance. Something like a solderless breadboard would have parasitics one or two orders of magnitude worse than the parallel termination resistors Jim Williams is worrying about.
There's definitely something strange about the measurement. Does moving the trigger point to the right at 2ns/div so the low signal is longer help at all?
The usual conversion factor is actually 0.35, not 0.34,, but I don't think it makes much sense to calculate this with many digits of precision, since the scope's response is unlikely to be perfectly Gaussian anyway. I would say that according to these results, the bandwidth is probably around 140MHz at these settings.
Any idea how this compares to your earlier measurements with the semi-Dirac pulse? Do you have a saved screenshot that you can use to estimate the rise time from? I'm wondering how much difference the very short pulse makes in this system.
Unfortunately, he doesn't actually give his measurements in that post that I can see, which may be important since I'm not sure he's using the same stuff.
I have some ideas how to improve this. BNC connection seems to be difficult thing to do optimally (at least for my taste), as most PCB connectors seem to have a quite long area of uncontrolled impedance. Soldering components directly to the BNC jack leaves relatively large loops 3D-wise (maybe I'm getting little paranoid here). I think it would be best to begin the 50 ohm transmission line right from the transistor emitter, where the termination resistors are located. Maybe an edge mount SMA connector is the solution:
Regards,
Janne
Jahonen, I have a question for you. The picture of the pulse-generator, seems to me like a factory-made device. Can you tell me where you bought it?