Oscilloscopes - What happened to equivalent-time sampling?

[gcewing]

I'm looking for a low-cost scope with reasonable specs. There are a couple of offerings at the moment in New Zealand that look interesting, a Rigol DS1102E and a Siglent SDS1202X-E, for about the same price.

The Siglent appears to be the better one by far in all ways except one -- it doesn't have an equivalent-time sampling mode like the Rigol. And as far as I can tell, neither do any of the other modern scopes in a similar price range.

Does anyone know why? Is ETS considered obsolete technology that's been superseded by something else? Will I be missing anything important by not having ETS?

[ataradov]

What are you using it for now?

I thought ETS was always a marketing gimmick. Am I missing something?

[Marco]

Equivalent time sampling doesn't really have a point unless the signal path and capture and hold have sufficient bandwidth to make use of it, which for the Rigol isn't the case. It was mostly useless marketing wank.

[tautech]

I'm looking for a low-cost scope with reasonable specs. There are a couple of offerings at the moment in New Zealand that look interesting, a Rigol DS1102E and a Siglent SDS1202X-E, for about the same price.

The Siglent appears to be the better one by far in all ways except one -- it doesn't have an equivalent-time sampling mode like the Rigol. And as far as I can tell, neither do any of the other modern scopes in a similar price range.

Does anyone know why? Is ETS considered obsolete technology that's been superseded by something else? Will I be missing anything important by not having ETS?

Welcome to the forum.

Check my profile for a link to my website and email link.
I can offer better direct pricing than listed on Trademe.

[tautech]

What are you using it for now?

I thought ETS was always a marketing gimmick. Am I missing something?

No you didn't miss anything.

If we look at 2 Siglent models, a recent one and an older series:
SDS1202X-E, absolutely no mention of ETS in the website blurb.
https://www.siglentamerica.com/digital-oscilloscopes/sds1000x-e-series-super-phosphor-oscilloscopes/

SDS1000CML+, straight away trying to make something look better than it is.
https://www.siglentamerica.com/digital-oscilloscopes/sds1000cml-series-digital-storage-oscilloscopes/

I've sold both and I know very well which I'd prefer.

[Bud]

Equivalent time sampling doesn't really have a point unless the signal path and capture and hold have sufficient bandwidth to make use of it, which for the Rigol isn't the case. It was mostly useless marketing wank.

In garbage chinese scopes it was useless. It was useful in Tek scopes.

[tggzzz]

What are you using it for now?

I thought ETS was always a marketing gimmick. Am I missing something?

Yes, you are missing something. It is what allows you to see a 50ps risetime with an oscilloscope that runs at <<100kS/s, optionally on a thermal chart recorder - with neither a computer nor an ADC anywhere in sight.

For a more modern example, the first google hit leads to https://www.picotech.com/library/oscilloscopes/equivalent-time-sampling

[ebastler]

Equivalent time sampling doesn't really have a point unless the signal path and capture and hold have sufficient bandwidth to make use of it, which for the Rigol isn't the case. It was mostly useless marketing wank.

In my understanding, ETS dates back to a time when fast A/D converters were really expensive, and the cost for a fast analog front end was small in comparison. So it made sense to derive some work-around to sample the fast signals your front end could handle.

With the cost of A/D conversion having come down as it has, the cost of the front end is no longer negligible in comparison. Modern low-cost scope designs have to control cost on the front end too, so don't provide the generous headroom there -- and equivalent sampling no longer makes sense.

In garbage chinese scopes it was useless. It was useful in Tek scopes.

That's a less technical way to say it. 

[Mechatrommer]

In my understanding, ETS dates back to a time when fast A/D converters were really expensive, and the cost for a fast analog front end was small in comparison.

it still is, its just the BW is increased. this is relative term...

rigol get by at cheap cost because they interleaved several (2 or 4?) lower BW ADC sampling to reach 1GSps, somewhat similar technique to ETS was. since 1GSps is only worth 100MHz in nyquist term, so they match their front end to that BW only, to lower development cost as well, so we get an affordable 100MHz scope like today. otoh if anyone is working in repetitive GHz signal, this will call for higher BW ADC, say if you want to do with 4 X 1GSps ADC chip, to get effective 4GSps or more rate, let alone trying to do development for the matching AFE which need to be tested with a house and a ferrari. go at digikey and look at how much a 1GSps ADCer worth? a single chip cost a DS1054Z already. and then the question of practicality of it, people are now working on intermittent bluetooth or wifi signal making ETS is a useless feature, in the end people who can afford 4 x 1G sampler might as well just go for a house and ferrari making ETS market shrinked like a peanut.

so... ADCs are still expensive real quick, once you go outside the hobbiest market. my 2cnts.

[chris_leyson]

Tektronix used ETS in the 7854, 10-bit ADC ran at a few 100kHz but the sampler before the ADC was high speed 400MHz bandwidth. HP54610B has 750MHz BW (measured) and 20MHz sample rate. Equivalent time sampling or under sampling is only good for repetative signals so that's why you don't see it in scopes anymore.

[Dubbie]

since 1GSps is only worth 100MHz in nyquist term,

Which textbooks are you reading? You’d have to have a pretty crap input filter to be Nyquist limited to 100MHz with 1GS/s sample rate. At least 200MHz should be possible even with a very poor filter.

[Mechatrommer]

since 1GSps is only worth 100MHz in nyquist term,

Which textbooks are you reading? You’d have to have a pretty crap input filter to be Nyquist limited to 100MHz with 1GS/s sample rate. At least 200MHz should be possible even with a very poor filter.

no textbook but practical application. nyquist theoritical limit is 2X sampling. many high end high BW brand scope sample at 2.5X, 10X is a comfortable margin. with proper AFE, DS1054Z (or any brand 1GSps rate for this matter) can be made a 400MHz scope in practice, but in reality some people made a test and at 200MHz, the DS1054Z already show significant lost, we still can eyeball some 400MHz signal though from it.

[jpb]

I really like the ETS on my LeCroy WaveJet, it is certainly not a gimmick.

It only works on repetitive signals of course, and the scope is still limited to its 350MHz bandwidth but this is equivalent to 1 nanosec rise time and under normal sampling the maximum sample rate is 2 GS/s which gives just a couple of points on a rising edge. ETS is equivalent to 100GS/s which give a nice smooth rising edge.

I think the reason ETS is no longer a feature is not because it is superseded but because modern scopes use digital interpolation even for triggering rather than analogue so there is nothing to be gained (that is they always sample at the same integral time points and then interpolate whilst analogue triggers happen at in-between time points if my understanding is correct).

Here is a photo of the Leo's pulser to show what I mean:

[nctnico]

I'm looking for a low-cost scope with reasonable specs. There are a couple of offerings at the moment in New Zealand that look interesting, a Rigol DS1102E and a Siglent SDS1202X-E, for about the same price.

The Siglent appears to be the better one by far in all ways except one -- it doesn't have an equivalent-time sampling mode like the Rigol. And as far as I can tell, neither do any of the other modern scopes in a similar price range.

Does anyone know why? Is ETS considered obsolete technology that's been superseded by something else? Will I be missing anything important by not having ETS?

ETS is the ability to use a lower samplerate to capture high frequency signals. Nowadays most DSOs have a high enough samplerate for their bandwidth so ETS is no longer needed. After all ETS only works on repetitive signals so it is of limited use anyway.

[jmelson]

What are you using it for now?

I thought ETS was always a marketing gimmick. Am I missing something?

ETS was a way to see fine time details of absolutely repetitive signals.  If the signal had a little bit of time jitter or voltage noise, then the trace turned into fuzz.

But, in the 1960's, sampling scopes were the ONLY way to get high bandwidth.  In the '70s and 80s, high bandwidth real time scopes cost a LOT, but ETS made it possible to get high bandwidth (but with the above caveat) quite a bit cheaper.  There really was a niche where ETS could allow you to see what you needed at an affordable price, but you needed to be AWARE of the limitations and especially aliasing effects.  Some ETS scopes put in a jitter so that aliasing would produce muddy traces rather than make beautifully sharp but totally false pictures.

Jon

[ataradov]

ETS was a way to see fine time details of absolutely repetitive signals.

Yes, I know what ETS is. I've used some highly specialized gear that was basically purpose built for ETS-only use.

But why would anyone want it in a consumer grade DSO? Because that's what OP is asking about.

[tggzzz]

ETS was a way to see fine time details of absolutely repetitive signals.

Yes, I know what ETS is. I've used some highly specialized gear that was basically purpose built for ETS-only use.

But why would anyone want it in a consumer grade DSO? Because that's what OP is asking about.

In the context you shipped, it can be seen that people are answering *your* question / assertion - not the OP's.

[ataradov]

In the context you shipped, it can be seen that people are answering *your* question / assertion - not the OP's.

I guess my original question was not specific enough. It should have read "I thought ETS was always a marketing gimmick in low-end DSO."

[tggzzz]

In the context you shipped, it can be seen that people are answering *your* question / assertion - not the OP's.

I guess my original question was not specific enough. It should have read "I thought ETS was always a marketing gimmick in low-end DSO."

Far less contentious

[Bassman59]

In the context you shipped, it can be seen that people are answering *your* question / assertion - not the OP's.

I guess my original question was not specific enough. It should have read "I thought ETS was always a marketing gimmick in low-end DSO."

I think a better way to phrase it would be to write, "I thought that in a low-end DSO, ETS was always a marketing gimmick."

[gcewing]

But why would anyone want it in a consumer grade DSO? Because that's what OP is asking about.

Actually, I was assuming it had some benefit on the DS1102E, and asking whether the newer scopes have something that provides an equal or better benefit.

It seems some people are claiming that it was never any use on the DS1102E anyway, because it already had enough samples for its bandwidth. Can anyone with experience of using a scope from that era confirm this? Seems to me it can never hurt to have a few more samples, particularly since I gather these scopes don't have a particularly sharp cutoff above their rated bandwidth, leaving room for aliasing. Provided it actually works as advertised.

Another explanation put forward is that the newer scopes can't do it because they do all their triggering in software. In that case, it's a tradeoff to keep cost down, but it still leaves the question of how much functionality is being traded away.

I'm wondering whether the digital-phosphor mode gives a similar kind of effect? Or does the software-triggering thing kill that idea?

[tautech]

Sampling in most entry DSOs is generally just 1Gsa/s but there's quite a variation in memory depth between brands and models with more being much more useful than less. Today you're better to consider it just as important as sampling rate.

From phone

[David Hess]

There are a couple of reasons:

1. The economics of increasing integration eventually produced relatively inexpensive ADCs with a high enough sample rate to support the bandwidth without equivalent time sampling.  This is just Moore's law at work.

2. Support for equivalent time sampling requires dedicated non-integrated hardware to measure the trigger to sample time.  If the ADC sampling rate is high enough, then this hardware is not necessary because triggering can be done digitally and again, this processing eventually became cheap enough due to increasing integration and Moore's law.  At that point, digital triggering becomes less expensive than including the hardware to measure the trigger to sample time.  In other words, it is *cheaper* to not support ETS.

With that said, I still prefer to use ETS where possible because:

1. Low sample rates are sufficient for almost all control loops while the higher samples rates provided by ETS are an advantage for repetitive and non-repetitive signal integrity analysis.
2. The increased sample rate prevents measurement artifacts like Gibbs phenomenon.  Try measuring a fast edge using a Rigol DS1054Z with 4 channels enabled to see an example.  That preshoot does not exist and the same corruption is present immediately after the edge.

Tektronix used ETS in the 7854, 10-bit ADC ran at a few 100kHz but the sampler before the ADC was high speed 400MHz bandwidth.

The Tektronix 7854 probably is not the best example because it does ETS in a way very few other digital storage oscilloscopes do; the only comparable example I know of is the equally odd Tektronix 2252.  These instruments simultaneously sample both a vertical signal and the analog sweep to produce a YT display.  In the case of the 7854, this results in sample rate equivalent to 200GS/s at 500ps/div.

Contemporary to the 7854 (1981) are DSOs like the 468 (1981), 7D20 (1983), and 2230 (1986) which have the hardware although I think only the 2230 takes advantage of it to implement ETS while the others use it for jitter reduction of a single acquisition.  I do not remember when HP started using ETS but maybe one of the HP gurus can say.

[rhb]

Here are screen dumps from a DS1102E of the output from Leo Bodnar's 40 pS rise time pulser with a 50 ohm thru termination.  The first is ETS, the second is RTS.

If you're going to buy a scope because it has ETS, I strongly recommend buying one of Leo's pulsers. You'll be able to tell in seconds whether it's of any use.  It's actually a great tool for a quick eval of the  front end of any scope.

[free_electron]

Equivalent tie sampling is only good to study repetitive signals.  for anything else : useless.
Sample faster,wider deeper is the motto these days. ram is cheap.