What is it about?
I have read all too often about things that are supposed to be just as good but ten times cheaper. I’m not a fool. Even though cheap tool can often do the job, I can see the difference all the same. However, often a compromise has to be made.
While at my day job, at my workbench, I can have a nice LeCroy Waverunner, it’s hardly something I enjoy lugging around in field. For field use I want a handheld scope. It has to run on battery, has to be reasonably rugged, and so on. However, at the same time, with portability comes stealability, droppability, wettability and all world-happened-abilities that seem to be present outside of the nice, dry, warm lab. Field equipment has to be a compromise between usability and price.
Since I have bought for my own hobby workshop a chinese-made handheld scope, I thought I’d compare it to a bench LeCroy anyway. I want to focus on what signal faults I’m blind to when using the handheld at home. The difference between a Mercedes SLK and a Fiat Panda is obvious, both in price and in comfort, but the both get you there, don’t they? But if the road is bad, how much worse are we off in the Panda?
Looking for faults
The thing about scopes is, they are not for looking at signals that are right. They are for finding signals that are wrong. In the digital age more and more often problems are one-off pulses, that throw communication off-sync, gate voltages that rise too slow or too fast etc. LeCroy has made a very nice demo tool to help selling their oscilloscopes. It can generate some of the typical ‘nasty’ signals that the current LeCroy oscilloscopes can oh-so-easily detect. The tool is, of course, tailored for LeCroy, but it still delivers a few of the common suspects in ‘digital crime’:
- a ‘runt’ – a single pulse of reduced amplitude in a crowd of normal pulses. To make it harder, the base signal has high jitter (35%). Think bus communication, where one transmitter speaks out of turn every now and then.
- a ‘glitch’ – single, very narrow pulse close to an edge of a square-wave signal.
- a one-off slow-rising edge in an otherwise clean square wave.
- a non-monotonic rising edge (think mosfet/igbt turn-on with Miller capacity turning it back off again).
Now, my bench LeCroy will find most of these automatically. Automatically like in saying ‘find what’s wrong with this signal’ and getting a response ‘there’s a one-off pulse that looks like this’. Sure, this level of comfort is not available in a cheap handheld. But can these anomalies be found at all?
Test S1-S3: Digital transmission
Well, you can watch it, but the handheld won’t do the decoding for you.
Test S4: A – Runt
Although the scope does not have an automated solution for this sort of anomalies, it has a very useful trigger mode that lets you find most of them fast. It’s… slope trigger(!). Meant to trigger on a slope faster/slower than a given steepness, it is very useful in pure digital domain too: it observes two levels and triggers if the signal crosses them within / outside of a specified time. Just set Level#1 at 10%, Level%2 at 90%, trigger when delay greater than your normal rising edge time (say, 1µs, as in the picture) and it will catch all pulses that start but don’t quite make it all the way up. Select negative slope for pulses that start towards zero, but pause on the way.
Test S4: B – Slow rising slope
After previous test: peanuts. Just lower the trigger limit to see what shows up.
Test S4: C – Glitch
Pulse Trigger works, triggering (in this case) on pulses shorter than 1µs.
Test S5: Persistence and long memory
This is not really a trigger test, but a demo of LeCroy dynamic persistence shading:
On a modern scope the trailing edges should be shown in varying intensity, as they occur at different rates. This small scope has infinite persistence, but does not shade the signal. You will see all the variations of the signal after some time watching, but you won’t get a hint about which are more frequent than the others.
By the way: there’s a bug in the scope here: turning menu on and off resets the persistence buffer. Annoying if you have menus set to auto-off after some time…
This is about the limit of what the small handheld can handle: bursts of 500kHz and 750kHz signals in a time frame of 60ms. You can zoom in enough to see there are pulses and roughly determine the frequency, but to see more you have to set up a trigger and capture at higher sample rates. Sure, a LeCroy with 10M+ of memory will shine, but you won’t be quite that blind with this handheld either.
Test S6: A – Non Monotonic Edge
This is a hard one and, honestly, unless you suspect this (turn-on problems on MOS-FET/IGBT) you can catch it only by accident, when checking for rise time issues/runts. If you suspect a miller turn-off, it’s easy to trigger on it: find a trigger level below the Miller plateau and trigger on negative pulses shorter than, say, 1µs. Or set up persistence and trigger on the beginning of the rising edge:
Once you see it, it’s easy to set up trigger to catch it.
Test S6: B – Runt (negative)
You can catch it using the slope trigger method (for negative-going edges that can’t quite make it) or, as shown here, by using a pulse trigger set to catch low-pulses narrower than expected.
So: how to figure out what to look for?
Method 1: Persistence
Set infinite persistence and watch for funny things. If your signal has jitter, check for both – rising and falling edge trigger.
Method 2: Pass/Fail test
Set up a pass/fail test and set it to beep on fail. If it starts beeping for no obvious reason, there’s something there…
OK. So I can catch what’s there. But is what I see really what is there?
About any scope can show a 1kHz sine wave. But how does this scope perform for higher bandwidth signal?
To check it I have connected a 33MHz clock generator (encapsulated crystal oscillator from Kyocera) as a signal source:
The importance of good probes shows immediately. The original 60MHz probes won’t cut it, obviously. The white trace is a nice, japanese-made Iwatsu 100MHz 1:10 probe, the yellow line is LeCroy PP008 500MHz 1:10 passive probe.
For comparison: The same wave captured on 5GS/s, 600MHz B/W LeCroy, sampling in equivalent time (100GHz equivalent sample rate):
speaking of which, the Hantek offers sampling in equivalent time, but it doesn’t work. I’m not sure what does it do exactly, but it sure isn’t doing it right…
So, how does the Handheld scope compare to LeCroy?
The blue line is Hantek, the pink: LeCroy, both equipped with LeCroy PP008 probes. Limitations in bandwidth on the handheld Hantek (Gibbs’ effects) are obvious, but the signal is usable.
What I want to point out is, that a pair of PP008 500MHz probes is worth more than the Hantek scope itself. Since in normal use cheaper and lower-bw probes are likely to be used on it, they are going to be the limiting factor in scope’s HF imaging performance and not the scope itself.
nice compare. Not sure if you know that, but there is a way to increase bw (actually improve RF response)
of these DSOs by a small backdoor.
If you create a file named
with this content
and reboot your DSO the bandwidth filter will be set to 500MHz. Unfortunately the
total capacity of all these 8 ADCs connected together is not creating a low pass, so you will get
anyway only 250MHz -3db bw, but then the filter is not that sharp so a compare with LeCroy
should looks better.
I know this hack, I have used it, however, as you said, the improvement of BW is not that significant, as the input stage is anyway limited to some 250MHz…
Thanks for input anyway!
Which hantek model did you actually do the testing with?
DSO1062B patched up to DSO1202B
I want buy the DSO1062B and would like to hack it till 200MHz.
Can you please share the hack method you used (maybe the reference link).
here’s a lot info.
I won’t write exactly how I did it, I have access to equipment that would be a) overkill for the task if you had to get it for this purpose only and b) anyone having it already won’t need my description on how to use it. Look in that forum, there’s plenty of how-to laid out.
Thanks for the review. I was on the fence about the Hantek product and you’ve really helped me out.