Should I buy Siglent SDS1202X or Rigol 1054z

[miguelvp]

For someone's first experience with a scope, 7 days might not be enough anyways.

[coppice]

Chinese IC fabs will be happy to MPW 90nm chips for you at as low as $10k, and there are low cost EDA tools rental services, or even free service if your company resides in an IC industry park.

An MPW only gets you development samples. Its $1M for a 90nm mask set to get into production.

[Wuerstchenhund]

There's little point doing this for entry level scopes for which lots of generic parts are available at low cost.

Low end DPO2k and MSOX2k both use ASICs, but they are selling well. For the same range, MSO/DS4k are not sold that much, so making ASICs for them, and taking risk of (highly possible) bugs is not making economic sense.

I'm not sure that the old Tek MSO/DPO2k scopes really sell that well (most of them go to the edu market which pays so little that there isn't much margin left, out there in the commercial market they actually don't sell very well at all) but Tek's ASICs for their low end scopes are pretty much based on designs from 15 years ago and the investment costs have been paid off a long time ago already, so the current costs are pretty low.

But I really think Rigol should ASIC their low end DS1kz/DS2k to save cost. After all these are super popular models and the cheapest FPGAs will be more expensive than ASICs, providing the NRE can be ignored at that shipping volume. In addition, hacking will be harder as a side benefit unless someone is really having too much money to spend on FIB/SEM the chips.

There's no way that would be cheaper, in fact it would cost Rigol a lot more to roll their own ASICs. Commodity parts are cheap through huge volume (magnitudes more than Rigol sells through their products, which are tiny in comparison), they are well understood and documented, and they come as ready-to-use package that has been specified, qualified and is guaranteed by its manufacturer.

Rolling your own ASIC means you have to come up with a design (which means spending $$$), you have to test/verify it (which means spending even more $$$), you might have to adapt your software development toolset (even more $$$), and if everything works you have to fab it out to a supplier who will simply be unable to give you a really low price because the volume is too low.

There's a reason why for entry level scopes even the big brands (which do have higher margins than Rigol) either use old and long paid of ASIC designs (i.e. Tek, Keysight) or just go with commodity parts (i.e. LeCroy, R&S), all companies which have tons of experience in designing their own ASICs. There's little point in re-inventing the wheel for the bottom end of the market.

Chinese IC fabs will be happy to MPW 90nm chips for you at as low as $10k, and there are low cost EDA tools rental services, or even free service if your company resides in an IC industry park.

There's a good article about why custom ASICs aren't really cheap:
http://www.planetanalog.com/author.asp?section_id=526&doc_id=559840

Of course that assumes you do it properly. You could probably also get away with some cheap EE students using some free EDA tools that come up with something, and then have it produced by one of these cheap IC fabs in an ancient 90nm process, but it still won't be anywhere near as cheap as the commodity part.

[coppice]

I meant that Rigol is unlikely to have the capability to design something like a complex high speed ADC hybrid or signal processing ASIC like the ones you find in modern mid-range/high-end scopes (there's little point doing this for entry level scopes for which lots of generic parts are available at low cost).

I doubt their expertise would be the limiting factor.

It certainly will be. There's a lot of proprietary IP in the ASICs from Keysight, LeCroy, R&S & Co, coming as the result of pretty large investments in R&D, and this IP is generally kept under lock and protection. Especially the high speed ADC hybrids (i.e. 20GSa/s and up) that you can find in a modern high end scope are highly complex, and there are only just a few companies that can actually make them.

I think you seriously underestimate the technical capabilities that Rigol could call upon. Converters running at 20Gsps or more are a serious systems design exercise. They end up being mostly digital parts, trying to handle the storm of digital data in real time. Some real ingenuity is needed to come up with architectures that stop the complexity getting out of control. Once people are funded to sit down and work through things like that, you might be surprised at the progress which can be made. It all comes down to finance. Keysight have fast converters not because there is something special about being Keysight, but because they had a sufficiently high value niche to justify the investment.

Rigol would need to do their own research (which they don't do and are unlikely to be able to afford at a similar scale anyways). And even if they did they'd a very hard time to turn it into profits at least in the T&M market (they might do in other markets like defense in China, though, but the question is if this is large enough to justify the costs).

And there you have the issue. Is there enough market to generate enough cash to justify enough investment to have people solve the problems? The small number of players in high end instruments has a lot more to do with the size of the market than technical capabilities.

[coppice]

Of course that assumes you do it properly.

This is the main problem. No one wants to put a buggy design in silicon.
But I highly doubt if it will be more expensive to make your own chips.
There is a company called eASIC, they offer semi customizeable silicon as well as low cost EDA tool (I mean $10k tape out and $10k/year tool, OEM by Magma), and you do not have to care about low level things, you start from RTL and end at P&R, just like FPGAs. The mass production cost is cheap enough that Esagate can afford to put them into hybrid HDDs.

Isn't eASIC just a sea of gates company, where your only NRE with them is the metal masks? That's a lot like mask programming an FPGA, and the FPGA suppliers can offer you some of the functionality, and cost benefit, today.

[nctnico]

I think you seriously underestimate the technical capabilities that Rigol could call upon. Converters running at 20Gsps or more are a serious systems design exercise. They end up being mostly digital parts, trying to handle the storm of digital data in real time. Some real ingenuity is needed to come up with architectures that stop the complexity getting out of control. Once people are funded to sit down and work through things like that, you might be surprised at the progress which can be made.

We never said Chinese are bad at making hardware. It is the software they can't wrap their head around!

[miguelvp]

An ASIC in high volumes will be cheaper in cost.

They also will have a lower power consumptions and better performance.

On the other hand the FPGA can be changed so you are not committed to a particular design.

At least that is my understanding.

[Wuerstchenhund]

I think you seriously underestimate the technical capabilities that Rigol could call upon.

Might well be, but I doubt that (but I'd be glad to be proven wrong). Just look at their products, which pretty much only excel through their low price. As far as technology goes, none of Rigol's products are anything to write home about, even their top end products are pretty basic. Also, their products are cheap because they use commodity parts in their designs, which keeps costs low.

This raises a few questions, like where all the supposedly advanced technical capabilities within Rigol are showing. Or why a manufacturer that has built a very successful business model on building cheap bottom-of-the-barrel and entry-level T&M and lab gear using commodity parts would invest money in acquiring the technical expertise required to roll their own complex ASIC designs, especially when it's not beneficial to their market segment?

It might well be that Rigol wants to get away from being a producer of cheap kit to a reputable brand that is considered not just by hobbyists, but then the first steps certainly would be improving their current offerings, especially in terms of firmware maturity and support.

So no, I don't believe I underestimate Rigol's capabilities. But as I said I'm open to be proven wrong.

Converters running at 20Gsps or more are a serious systems design exercise. They end up being mostly digital parts, trying to handle the storm of digital data in real time. Some real ingenuity is needed to come up with architectures that stop the complexity getting out of control. Once people are funded to sit down and work through things like that, you might be surprised at the progress which can be made. It all comes down to finance. Keysight have fast converters not because there is something special about being Keysight, but because they had a sufficiently high value niche to justify the investment.

While of course I agree with what you say about the complexity of 20+GSa/s converters, I don't with your last statement about the niche. In fact, one of the (not T&M related) markets I work in is pretty much screaming for high speed ADC hybrids, the faster the better. The margins in this market are a lot bigger than say for high end scopes. Still, there are very few vendors who can offer fast ADCs, because as you stated correctly the complexity is pretty high which limits the number of manufacturers that can deliver them to a handful of high tech companies (plus 20Gsa/s is the lower end of the spectrum, most of the time we're talking in excess of 80GSa/s). Keysight has a ton of IP and patents on their designs, as has Tek, R&S and LeCroy, which is because all invest a lot of money into R&D.

Keysight has over 9k employees and over $2bn revenue, Tek has still some 4k+ employees and over $1.5bn revenue, R&S has almost 10k employees and some $2bn revenue, and LeCroy has approx 450 employees and some $200M revenue. I find it pretty hard to believe that Rigol (for which I couldn't find any revenue numbers, but I'm sure they are well below $200M) which has some 400 employees for their whole product portfolio (which includes scopes, spectrum analyzers, PSUs, DVMs, RF and frequency generators plus their lab gear) will have the same technical ressources as say LeCroy who's 500+ employees are pretty much dedicated to upper midrange/high-end scopes only. The number and type of patents Rigol is holding seem to confirm that.

But as I said I'd be glad to be proven wrong.

[Wuerstchenhund]

An ASIC in high volumes will be cheaper in cost.

Yes, but "high volumes" usually means volumes in the Millions, and even then it's not always assured.

They also will have a lower power consumptions and better performance.

Not necessarily, especially when produced in older processes like 90nm. Like many things, it depends, and commodity parts are pretty good these days.

On the other hand the FPGA can be changed so you are not committed to a particular design.

Yes, that's one thing. The other is that the part you get has been specified, qualified and tested, so you know what you're getting. With your own ASIC that is up to you, which can make for a substantial parts of the cost.

[diyaudio]

I meant that Rigol is unlikely to have the capability to design something like a complex high speed ADC hybrid or signal processing ASIC like the ones you find in modern mid-range/high-end scopes (there's little point doing this for entry level scopes for which lots of generic parts are available at low cost).

I doubt their expertise would be the limiting factor.

It certainly will be. There's a lot of proprietary IP in the ASICs from Keysight, LeCroy, R&S & Co, coming as the result of pretty large investments in R&D, and this IP is generally kept under lock and protection. Especially the high speed ADC hybrids (i.e. 20GSa/s and up) that you can find in a modern high end scope are highly complex, and there are only just a few companies that can actually make them.

I think you seriously underestimate the technical capabilities that Rigol could call upon. Converters running at 20Gsps or more are a serious systems design exercise. They end up being mostly digital parts, trying to handle the storm of digital data in real time. Some real ingenuity is needed to come up with architectures that stop the complexity getting out of control. Once people are funded to sit down and work through things like that, you might be surprised at the progress which can be made. It all comes down to finance. Keysight have fast converters not because there is something special about being Keysight, but because they had a sufficiently high value niche to justify the investment.

Rigol would need to do their own research (which they don't do and are unlikely to be able to afford at a similar scale anyways). And even if they did they'd a very hard time to turn it into profits at least in the T&M market (they might do in other markets like defense in China, though, but the question is if this is large enough to justify the costs).

And there you have the issue. Is there enough market to generate enough cash to justify enough investment to have people solve the problems? The small number of players in high end instruments has a lot more to do with the size of the market than technical capabilities.

+1 Could not say that better myself.

[coppice]

While of course I agree with what you say about the complexity of 20+GSa/s converters, I don't with your last statement about the niche. In fact, one of the (not T&M related) markets I work in is pretty much screaming for high speed ADC hybrids, the faster the better. The margins in this market are a lot bigger than say for high end scopes. Still, there are very few vendors who can offer fast ADCs, because as you stated correctly the complexity is pretty high which limits the number of manufacturers that can deliver them to a handful of high tech companies (plus 20Gsa/s is the lower end of the spectrum, most of the time we're talking in excess of 80GSa/s). Keysight has a ton of IP and patents on their designs, as has Tek, R&S and LeCroy, which is because all invest a lot of money into R&D.

Do you really think that if reasonable volume applications existed for a well defined device the likes of TI, ADI, Linear and Maxim couldn't turn out these products? There are 80Gsps converters in the merchant market right now, but they are embedded in comms devices, where the storm of sample data doesn't have to escape the die. Fujitsu has no history as a state of the art ADC maker, but they have such a device. I note that one, because you can find an extensive slide set about it on line. Its an interesting read, especially when you think beyond what it is saying about the problems caused by that storm of sample data, and consider its implications for trying to make a general purpose device. I don't know the details of what Keysight and LeCroy's fast ADCs look like, but I remember someone from Keysight saying they are anything but general purpose.

[tggzzz]

I tinker more with microcontrollers during my spare time and I do have a logic analyzer. However I'd like to get back to analog and spend some time understanding op-amps and mosfets and RC circuits. Hence the oscilloscope.
I definitely dont think i'll be working with anything that might need a 100+mhz scope.  I am also trying to avoid my bad habit of buying the "shiny new toy".

You might be surprised.

Even 74LVC logic has >500MHz components in its outputs, and can be problematic if you don't have appropriate signal lines and terminations. If, as I hear rumoured, things like the RPi have 2ns transitions times, they will have 200MHz components. IMNSHO, a good used for a scope is to ensure digital signal integrity, an once that's good you should debug in the digital domain with logic analysers and printf statements

Opamps and mosfets can have high frequency tranisents and/or oscillations ("amplifiers oscillators, oscillators won't").

If by RC you mean radio control, then RF is best examined using a spectrum analyser; scopes are insufficient. Consider RTL-SDR dongles as a first step for both wideband power measurements and narrowband modulation measurements.

[ziplock9000]

Is the Rigol DS1054Z still the sub £500 "go to" scope for almost 2017

[ProBang2]

As an versatile stand-alone oscilloscope for beginners or for the hobby: Yes.
(If you consider the possible and easy "hack": It can´t be beaten.)

[tautech]

Hell, we're not even there yet and there's a whole 12 months for competitors to to release a model that surpasses it.

[ProBang2]

With todays knowledge I can´t give a different opinion.

Problem: Any competitor is aware of the "Osborne effect".

E.g. a SDS1104X, including decoding and roughly at the same pricepoint as a DS1054Z would be for sure a bestseller.

[Blisk]

Whole page of what one company should do and another can't.
And what they mabe will do or not.
But much less info which scope is better.