SparkFun Forums

[abb51]

OK, so I find myself needing to look at some analogue stuff to debug something I'm building. Hence, an oscilloscope.

I'm not hugely familiar with what specs are important or not, so I'm after a bit of advice.

My current project uses (digital) signals up to about 6MHz.
I also like to sometimes play around with FPGAs , and my next project is likely to be having a crack at DVI (which has a ~25MHz clock).

So I guess my questions are:
how much bandwidth do I need? I know it's good to have more than the bare minimum, so more than 25MHz seems desirable. But how much more? 50, 60, or 100MHz?
There's a fairly big price step from 50MHz to 100MHz.

"equivalent" sample rate?
this is some marketing thing, only useful for periodic signals, right?
I've been pretty much ignoring it and just looking at the real sample rate, as I'll be looking at digital signals which tend to be not very periodic.

storage?
I've been looking at the Atten ADS1102 and Rigol DS1102E, which are both 100MHz units. The $550 Atten has 4k of memory, the $720 Rigol has 1M (or is it 16k, the ebay listing is a bit vague).

Are scope prices tending to decrease as tech improves? Should I get a just-barely-good-enough one today and upgrade down the track if I need to go faster?

I'm leaning towards a stand-alone unit rather than PC-attached, although I remain open to a USB unit if it's cheap.

Any advice would be appreciated, especially if you've used any of the cheap eBay scopes (Atten, Rigol, or Owon... have I missed any?)

Oh, and before anyone says "get a 2nd hand tektronix off craigslist", well, I'm in Australia. There doesn't seem to be much of a market for 2nd-hand scopes on craigslist or ebay, and shipping from USA is huge for those big hunks of glass and metal.

And finally, I already have a logic analyser so don't need that in the scope. LAs are really great for protocol and timing debugging, but much less so for checking signal integrity or power supplies.

[mac]

As promise, a quick test of DSO5200USB from Hantek.

Test setup:
Generator: Agilent 33250A Note that Agilent generator is not very good for noise and jitter issues.
Reference scope: Agilent MSO6054A (50ohm connection).

Hantek is using its standard 1/1:10 probe. Hooked on a dell D630 in usb2.0 connection (Hantek DO NOT WORK on 1.1 ports), winXP SP3.
Dual usb connection used (power and data). Software is latest (7.0.0.0).
All sampling is realtime no average, sin(x)/x interpolation, unless stated. No calibration has been done (I will do one later).


Probe in 1:1 setting:

1khz 1vpp sine: clean but seems there is some jitter for trigger (note: low at low freq, begining to be anoying at high freq (>1MHz). Measures are ok,

1MHz 1vpp: same as above, no attenuation with 1x probe

10MHz 1vpp: probe show some attenuation, as espected as 1:1 is only 6MHz bandwith.
There seems to be more distortion that are not happening on the Agilent.
With equivalent sampling, distortion is gone (but with some artefacts). We are begining to see sampling limits here!
It is better with CH2 turned off.
With 1:10 probe, no attenuation is visible.

80Mhz 1vpp 1:10 probe: this is the limit of the generator.
Voltage measure show quite the same voltage as Agilent (+/- 10%), this is quite ok as a BNC T is used and there is some impedance mismatch.
You should turn off CH2 to have enought sampling, which is quite normal for a 200MSa/s.

So I least you can espect 80MHz of bandwith, I cannot go further, and quite frankly not very useful since range is max 100Mhz at 200MSa/s.

Overall the Hantek suffers from quite a lot of bugs:

- ATS (equivalent sampling) suffers from a lot of glitches and is too slow.
- There is some trigger jitter visible
- There is a lot of bug when switching inputs or sampling mode: sometimes re-switching input is ok, sometimes not. Could be either S/W or H/W. It will aquire something like two sinus one high freq and a slow (50hz?) one. Since I have a lot of HF here, it can be an open input.
- [V6 had true graphical FFT, which was dropped to a table on v7, this is quite a regression!] oops, in fact graphical is displayed over waveform. Less intuitive in think.
- refresh is not high speed. Is it USB limitation?
- Sometimes measurements (esp frequency) are wrong on some time scale, but not on other.
- Only 10ksamples are usable in realtime mode, so stating 14k sample memory is misleading.
- No sampling frequency is shown on UI, this is important to know if distortions comes from sampling or from waveform. Switch to step interpolation show the sampling frequency.

So at the end I will say that this scope is quite decent for the price, but it is clearly not mature.
It is not a replacement for a full featured table scope (both ergonomy and precision suffers), but can be handy for hobby use.

EDIT: the error in measurment is in fact the 400ns/div having an issue (wrong sampling?). 200ns and 1us are ok.

[mac]

Second round:

I switch back to latest v6 software (6.0.0.3).

The ETS glitches are gone, so it's quite usable. I measure a 5ns rise time of a 1MHz square wave correctly. No more issue also when switching on/off channels and ETS.

The FFT is much better, and less buggy (no 2nd hamonic reported as the first).

External trigger (from Agilent MSO) is ok, same jitter as internal.

To be honest on the jitter, Agilent MSO has a way to display it quite misleading: instead of having a single trace that jitters, it thicken the trace, which is equivalent of showing noise, more like an analog scope. But you can see the quick measure cursors flickering more like hantek trace.

So I found the v6 software a little bit better, even if it's less ergonomics (use more menu). At least less buggy.

[abb51]

Thanks for the info, that's some good testing you've done there.

It's very strange that the software has gotten worse rather than better, not a good sign...

[winston]

Get a second hand Tektronix digital storage scope. I recently got one for the equivalent of a couple of hundred bucks - 100MHz bandwidth, max 500Msamples/sec. The only real issue is I have to use a floppy disc to get data off it (or a very slow serial link). However, anything with Tek on it seems to be good, and the scope I got (TDS340A) is easy to use, didn't have to even open the manual until I wanted to find out about the fast fourier math functions.

[stevech]

cheap I am. Satisfied with light-use DS-2090 purchased direct from China for about $140.

[abb51]

Get a second hand Tektronix digital storage scope. I recently got one for the equivalent of a couple of hundred bucks - 100MHz bandwidth, max 500Msamples/sec. The only real issue is I have to use a floppy disc to get data off it (or a very slow serial link). However, anything with Tek on it seems to be good, and the scope I got (TDS340A) is easy to use, didn't have to even open the manual until I wanted to find out about the fast fourier math functions.

I look and look, but buggered if I can find any of these in Aus.
There's an old 25MHz analog (no storage) one for $500 on ebay at the moment - for $100 less I can get a brand new Atten 60MHz with storage, colour, USB.

Sadly the recent drop in AUD value seems to have scared the eBay sellers, can't the Rigol 100MHz is now $850 instead of $720 :'(

[ttabbal]

cheap I am. Satisfied with light-use DS-2090 purchased direct from China for about $140.

I've been considering buying one of those. Where did you get that price? Most sellers I've found with google are about 2x what you are quoting.

[planius]

I bought a Rigol DS1052E from China on ebay and I am very happy with it. I needed a pulse width trigger for PWM measurements, that was my main reason for buying the Rigol. Both the scope and the PC software work fine, precision is not so much an issue for me, I mainly need to capture wave forms (I2C, SPI, PWM etc.).

[ganzziani]

I have put together a table to compare all the different digital oscilloscope for hobbyists: http://www.gabotronics.com/resources/ho ... scopes.htm