Recommendations for an Oscilloscope

I have bought an Analog Arts model SA925 a USB type oscilloscope (www.analogarts.com). I bought the device mostly for its spectrum analyzer capability. For about $300 it was the price that initially attracted my attention. Although it has about 200 MHz bandwidth, I hardly need that level of performance. Overall, I am happy with its performance. Nevertheless, I strongly recommend to you to meticulously read the specification of any device you planning to buy. So you do not fall victim to marketing gimmicks.

Hi Bernard,

About the specifications: this scope has a max sampling rate (single shot) of 125MHz, it advertises as 200MHz "real" sample rate, an "effective sample rate" of 100GHz (which means you're taking more samples of a signal, with a variable trigger delay. The signal should be repetitive) and an analog bandwith of 200MHz.

To me those numbers do not compute. For a single shot trigger the analog bandwith can be no higher than ~50Mhz, taking Nyquist into account without safety margins.

I think that these guys also let the marketing misuse the specs....

BTW: someone mentioned the Rigol DS1054Z, I bought it and received it two weeks ago. Haven't used it much yet, but it's certainly worth its money!

Hi Vsluiter,

            Well, If you are talking about the criteria imposed by the Nyquist theorem, who could argue with that. However, for a repetitive signal that criteria can be manipulated. Analog Arts oscilloscope correctly displays 150 MHz sine-wave in my set up. If you wish I can email you a plot of the signal. Now, that you mention it though, I do find it quite impressive. I am not quite sure how they do it. I am familiar with different methods of signal reconstructions, but none of them could work for really high frequencies since you need to generate extremely small delays, which are impossible. The application engineer, whom I talked to, was not open about how they do it, stating they have a patent pending on their technique. He did mention they somehow calculate the frequency of the signal (to .1ppm) before it is reconstructed. Anyway, for the money I paid, i am quite happy with the unit. For a lower price, I am getting higher bandwidth and spectrum analyzer feature.

The way it works is well documented (known as equivalent time sampling). It is useful in the circumstance that the signal is repetitive as you say, which is a limitation.

The spectrum analyzer feature is offered on many oscilloscopes nowadays.

The way it works is well documented (known as equivalent time sampling). It is useful in the circumstance that the signal is repetitive as you say, which is a limitation.

The spectrum analyzer feature is offered on many oscilloscopes nowadays.

Thanks for that link! There's also an explanation here: http://www.signalrecovery.com/Support-Center/FAQ/FAQ-Hardware/Effective-Sampling-Rate.aspx

Still, the numbers do not compute well to me; a single shot update with a different sample rate than the "real sample rate"? And what about undersampling / aliasing? The analog bandwiith will allow signals to pass that cannot be correctly visualized on screen, and you won't be able to see whether you're undersampling or not.

To be honest: I haven't checked my own scope's data so thoroughly, but I found these numbers to be raising some questions.

Hi Victor!

I read an article a while back (sadly I can't remember where now) but basically although equivalent time sampling is highly useful, doing it well is a hard thing (requires very low jitter from the delayed periodic signal (generated from the input signal) which will cause the ADC to sample. And, as you suggest, there is the risk of misinterpreting the output when the signal actually isn't periodic, or only has some periodic elements - the resultant waveform display quality could look wrong for any of these cases (jitter, non-periodic or periodic with non-periodic components) and probably others too if I could remember it in detail.

Fpga4fun (terrible site, I can't recommend their products because they have expensive "development" boards that they do not supply the entire schematics nor entire source code for) have a design using a 100MSPS ADC which uses equivalent time sampling to achieve higher than 50MHz waveform capture. They have a block diagram showing the periodic output from their ADC boards (no schematic of course..).

I think this is the "secret sauce" in many scope designs and I'm sure that most signals tend to repeat at some point (exactly where seems to the the art).

How exactly do you tell when the scope is making a good job of this and when is it just making funny pictures for your entertainment  seems to be even harder to decipher.

OT: I was looking at the DVI output generator stuff sometime ago on FPGAs4fun and they do have some VHDL, have you tried any of their stuff?? I'm shure there is something missing on many of them?!

After I discovered source was incomplete for some of their projects, and they stated that some parts were closed source (and their refusal to supply schematics) I didn't use their boards any more to be honest. I'm glad there are plenty of higher quality alternatives.

Indeed, I agree about the quality of trace thing. Nowadays it is well established that some particular math manipulations need to be done to convert the raw samples into a waveform and that you can't just dump to a screen and hope for the best, but probably early products, and possibly nowadays for some of the very small outfits, may not do this.

At the end of the day there is no substitute for a stupidly fast ADC

Thanks Shabaz. Well, as I mentioned I am familiar with the technique you mention. But achieving  consistent 10 picosecond time , 100 GHz, resolution (what Analog Arts oscilloscope does) is very challenging. I would appreciate any explanation.

Hi Bernard,

I'm afraid I can't justify or explain it - perhaps Analog Arts may have a paper published where they explain their design?

As you may know, a sample rate of 2.5 times the analog bandwidth is all that is needed to reproduce (with accuracy) the input waveform.

As a result, the advertisement of 100GHz effective sample rate on a 200MHz analog bandwidth and 125MHz real sample rate scope is

"marketing bluster" perhaps.

Furthermore bear in mind, if there is jitter on their 10ps time between samples (which is conveniently ignored in their specs), that

results in a bad (i.e. misleading) shaped waveform.

I recently bought from Analog Arts. Their devices perform as advertised.

Hi! Welcome to the forum.

Personally I wouldn't buy it. Just a personal, subjective opinion of course.

This is because they are quite silent on waveform quality because it is well known that if there is jitter inside their product (conveniently absent from their specification) then

equivalent time sampling will produce a poor result. Furthermore, equivalent time sampling only works well when the input signal has no  glitches or jitter too, let alone needing to be periodic. That's a fact, well established.

The 125Msps of the Analog Arts oscilloscope only provides 50MHz of real-time bandwidth (i.e. usable for non-periodic signals). That's a fact too and not a subjective opinion of course.

For near-similar amounts of money to the Analog Arts 125Msps 2-channel USB scope, it is possible to get a full-blown 4-channel 1Gsps oscilloscope with the same real-time bandwidth (DS1054Z) and built-in display, an order of magnitude more memory, and 30k waveforms updates per second, or a 2-channel 250Msps USB scope with twice the real-time bandwidth(Tenma  72-1016572-10165 of the Analog Arts product(i.e for non-periodic signals

I'm not saying the Analog Arts product is bad, but what I am saying is that it is seems to be clearly easy to recommend alternatives which appear better, without looking too hard. This is just an opinion.