Rigol HDO1000 and HDO4000 12bit oscilloscopes launched in China

[mawyatt]

There is an adjacent volume (profitable) market in CMOS image sensors, which china now has fully domestic capability end to end. It wouldn't be difficult to imagine the industrial/military businesses working with those same design houses to come up with GS/s ADC. A few iterations and they'll catch up, and those iterations are likely to come much quicker than the ASIC lifecycle of say HP/Agilent/Keysight.

CMOS image sensors are made by just about every CMOS fab house worldwide, these are routine today. However, the higher performance CMOS image sensors come from a few sources like Sony and Samsung but not any Chinese sources we are aware of. Sure some of the leading performance cameras from Sony, Canon, Nikon are assembled in China but the sensors are from the outside. Same goes for CMOS ADCs, various flavors of ADCs are available from many sources worldwide, however the highest performance high speed high resolution ADCs come from AD and TI, and captivity within Keysight and LeCroy. We can't find any Chinese sourced high performance HS HR ADCs that are competitive with those companies mentioned, and they've had ample time to study, reverse engineer and copy such designs which hints at the extreme difficulty of creating such!!

Recall some time ago when Fujitsu Microelectronics shocked the IC design world with their amazing 56 GSPS 8 bit ADC back in ~2009, that was the last we've seen of any significant ADC surprises, maybe there are others lurking in China that we aren't aware of!!

Anyway, think it'll take China more than a few iterations to catch up, and meanwhile the ASIC lifecycle at KS may slow down but don't count on AD or TI to become stagnate

Best,

[EEVblog]

Anyway, think it'll take China more than a few iterations to catch up, and meanwhile the ASIC lifecycle at KS may slow down but don't count on AD or TI to become stagnate

The Keysight Megazoom IV ASIC is now 11.5 years old! 
It's still the standard ASIC in the 1000, 2000, 3000, and 4000 seriees scopes.

[EEVblog]

It´s no secret/surprise, that siglent would benefit from the "partnership" with lecroy

Keysight hugely regretted their technical partnership Rigol 

But is Lecroy happy with partnering with Siglent? It would not surprise me if Siglent buys Lecroy just to get their high-end market share. Siglent likely dumps all Lecroy designed hardware when that happens though.

Not sure that would be allowd to happen, a Chinese company buying a high end (maybe even strategic?) US tech company. AFAIK Lecroy do military and other exotic stuff.

[Martin72]

Yep,
Try to buy a scope from them directly..
You get a form to fill out back where you have to declare who you are, what you do and for what you will use the product.
And then it must be signed from the one who is responsible for the security of the company.
No joke...

[Someone]

Recall some time ago when Fujitsu Microelectronics shocked the IC design world with their amazing 56 GSPS 8 bit ADC back in ~2009, that was the last we've seen of any significant ADC surprises, maybe there are others lurking in China that we aren't aware of!!

Lurking is probably the right word, much of the Chinese talent/engineering is busy/fully utilised by domestic demand so they have no reason to reach outside. When you have Chinese nationals go looking, all sorts of interesting businesses appear that have zero english/foreign exposure. The Rigol ASICs are likely fully domestic as these world "super powers" know the value of maintaining access to dual-use technologies.

But is Lecroy happy with partnering with Siglent? It would not surprise me if Siglent buys Lecroy just to get their high-end market share. Siglent likely dumps all Lecroy designed hardware when that happens though.

Not sure that would be allowd to happen, a Chinese company buying a high end (maybe even strategic?) US tech company. AFAIK Lecroy do military and other exotic stuff.

This is it, the US holds much of the worlds semiconductor "pixie dust" on the side of design and partnered with Taiwan to scale up production. Japan has standalone domestic capability, aggregate EU has design and technology but lacks scale to meet consumer demand. China is continuing building up to be less dependent on US whims/effects of sanctions.

[Someone]

Yep,
Try to buy a scope from them directly..
You get a form to fill out back where you have to declare who you are, what you do and for what you will use the product.
And then it must be signed from the one who is responsible for the security of the company.
No joke...

That used to be the case with thermal cameras. Once there were suppliers popping up around the world the difficulty in buying them evaporated (even at the US based companies)....  after the horse has bolted etc.

[tonyget]

CMOS image sensors are made by just about every CMOS fab house worldwide, these are routine today. However, the higher performance CMOS image sensors come from a few sources like Sony and Samsung but not any Chinese sources we are aware of.

OmniVision，GalaxyCore，SmartSens are all Chinese CIS manufactures. OmniVision and SmartSens are fabless，most of their chips are fabricated in SMIC.

GalaxyCore used to be fabless as well，and outsourced their CMOS chips to Samsung. Later they break up with Samsung，because Samsung deliberately delay the chip delivery to GalaxyCore，as they felt that GalaxyCore was taking market share from Samsung. So now GalaxyCore build their own 12-inch fab in Shanghai, thus become an IDM.

[EEVblog]

I've been informed that the HDO4000 is on the DHL plane, and the HDO1000 with come two weeks later.

[hhappy1]

I'm looking forward to Rigol's UltraVision iii.
The advantage of 12-bit is the higher dynamic range, the smaller noise floor.

Users who need 12 bits will carefully observe the fft function.
The difference between 8-bit and 12-bit is because the dynamic range of fft can be easily distinguished.

I want max-hold and average in the fft view.

[Fungus]

https://ecmobile.rigol.com/goods-154.html
Price of 70MHz HDO1072 : 3999 CNY = 593 USD
They are going for the throat. Building your own ADC chipset has its perks.

Under US$600 for a 12bit 1.5Mwfm/s 100Mpt 10" 1280x800 screen scope that almost certainly will be hackable to 200MHz 
They'll have the market to themselves.

I wonder if there'll be any grey imports? Maybe the Chinese version can be had on Aliexpress and changed to English.

[2N3055]

I'm looking forward to Rigol's UltraVision iii.
The advantage of 12-bit is the higher dynamic range, the smaller noise floor.

Users who need 12 bits will carefully observe the fft function.
The difference between 8-bit and 12-bit is because the dynamic range of fft can be easily distinguished.

I want max-hold and average in the fft view.

As it was said before, 12 Bit doesn't mean less noise implicitly.. First, 12 bit is only ADC specification, and you can pair it with front end that is noisy and you still get suboptimal results. Second, you can have 12 Bit converter that has 5 bits of noise.

Of course I'm not saying Rigol did that. I'm just saying that just looking at 12Bit specification is misleading. Scope is a system, and should be evaluated as such.

But you made interesting point, that adds to my previous posts on this topic. Just having good front end means nothing if you can just look at nice waveforms. Like one poster said, there is Owon hires scope on a market for few years now. Apart from hires ADC it has no capabilities to speak of. And it seems not really a bestseller.

So even if Rigol did it right this time, hardware wise, they have to fix many things in their analysis package and measurements for these scopes to play in market segment. You can't make DS1000Z a 12 Bit with big touchscreen and say it is now same as RTB2000 or SDS2000X HD. It's not, it's missing lots of other stuff that make difference to usability and capabilities. So they either have to upgrade software to parity, or they will have to drop prices. They do seem to have made completely new touch GUI/OS/scope application platform, so maybe they are working on software side too..

Interesting times..

[hhappy1]

I'm looking forward to Rigol's UltraVision iii.
The advantage of 12-bit is the higher dynamic range, the smaller noise floor.

Users who need 12 bits will carefully observe the fft function.
The difference between 8-bit and 12-bit is because the dynamic range of fft can be easily distinguished.

I want max-hold and average in the fft view.

As it was said before, 12 Bit doesn't mean less noise implicitly.. First, 12 bit is only ADC specification, and you can pair it with front end that is noisy and you still get suboptimal results. Second, you can have 12 Bit converter that has 5 bits of noise.

Of course I'm not saying Rigol did that. I'm just saying that just looking at 12Bit specification is misleading. Scope is a system, and should be evaluated as such.

But you made interesting point, that adds to my previous posts on this topic. Just having good front end means nothing if you can just look at nice waveforms. Like one poster said, there is Owon hires scope on a market for few years now. Apart from hires ADC it has no capabilities to speak of. And it seems not really a bestseller.

So even if Rigol did it right this time, hardware wise, they have to fix many things in their analysis package and measurements for these scopes to play in market segment. You can't make DS1000Z a 12 Bit with big touchscreen and say it is now same as RTB2000 or SDS2000X HD. It's not, it's missing lots of other stuff that make difference to usability and capabilities. So they either have to upgrade software to parity, or they will have to drop prices. They do seem to have made completely new touch GUI/OS/scope application platform, so maybe they are working on software side too..

Interesting times..

You're right.

It is also true that the Rigol mso5000 and 7000 have more noise than the competing models.
We'll see what the 12-bit performance is like in two weeks.

[Fungus]

As it was said before, 12 Bit doesn't mean less noise implicitly.. First, 12 bit is only ADC specification, and you can pair it with front end that is noisy and you still get suboptimal results. Second, you can have 12 Bit converter that has 5 bits of noise.

Of course I'm not saying Rigol did that. I'm just saying that just looking at 12Bit specification is misleading. Scope is a system, and should be evaluated as such.

"Low noise" is a bullet point for this one:

[mawyatt]

As it was said before, 12 Bit doesn't mean less noise implicitly.. First, 12 bit is only ADC specification, and you can pair it with front end that is noisy and you still get suboptimal results. Second, you can have 12 Bit converter that has 5 bits of noise.

The core ADC is as good as it gets and everything in front will degrade the performance as you said it's a System and must be designed and considered as a System.

Best,

Edit: Remember a Keysight Engineering Fellow telling us 10 years ago that the Measuring Instrument needs to be more than 10dB better than the measurement, and the Instrument Components need to be more than 10dB better than the Instrument!! This is why they were designing all the critical chips in-house and many on there own fabs in various technologies, because nothing commercially available was even close to what they needed!! Good example is the Griffin 14 bit DAC from ~2009. Unbelievable NPR back then, over 74dB and we witnessed this performance first hand back then!! BTW the core design team had been together working toward these goals well over a decade!!

[nctnico]

5 times better... that is little over 2 bits. IIRC the ENOB of the Rigol MSO5000 is around 5 or so. So this new '12 bit' scope would have an ENOB around 7 to 8 bit.

[Sighound36]

I would suggest the Rigol should able to match the Siglent 12 bit noise performnace, although this may well be with the 20Mhz BW limiter engaged.

A 4000 rigol will be heading this way as well in the not distant furture.

[mawyatt]

5 times better... that is little over 2 bits. IIRC the ENOB of the Rigol MSO5000 is around 5 or so. So this new '12 bit' scope would have an ENOB around 7 to 8 bit.

Let's hope they do much better than just a 7~8 bit ENOB, they should perform on par with ADCs from LeCroy, KS, AD & TI. If they don't then this may lead some credence that this new chip set design was done in-house as some believe.

Best,

[Fungus]

5 times better... that is little over 2 bits.

Says who? How do you know what they mean by "5 times"?

IIRC the ENOB of the Rigol MSO5000 is around 5 or so.

Maybe it's 25 bits...   

[nctnico]

5 times better... that is little over 2 bits.

Says who? How do you know what they mean by "5 times"?

That is very obvious ofcourse! 5 times better means they are seeing a noise level that is 5 times lower. As usual with datasheets this will be a very optimistic number so very likely it is measured with 20MHz bandwidth limit enabled using RMS calculation over a sweet spot where it comes to memory depth / sampling rate.

[coppice]

Recall some time ago when Fujitsu Microelectronics shocked the IC design world with their amazing 56 GSPS 8 bit ADC back in ~2009, that was the last we've seen of any significant ADC surprises, maybe there are others lurking in China that we aren't aware of!!

If you look at the presentation material for that device you'll see the actual ADC was not much of a technical challenge. All the complexity in the device is handling the huge flood of data.

[mawyatt]

Know little something about that chip, and the 64GSPS version. They employed ~400 SAR ADCs of higher resolution that ran at ~200MSPS, each individual ADC had quite a few error correction DACs (recall 17 per ADC), so something like 400 ADCs with 13,000 error correction DACs. These chips were quite good and recall had a very respectable ENOB at Nyquist. The chips were specifically developed for the high end Fiber Optic Communication market and Fujitsu imposed a "Not for Military Use" requirement on any potential customer, and this really ticked some folks off!!

So don't think you can discount the chips complexity as "just handing the data flow", there was quite a bit of clever engineering within these chips not to mention getting all these ADCs and the input charge redistribution network to work together. Also this early chips were implement in-house with 65nm CMOS, they really did a masterful job of squeezing the performance from that in-house process!!

Best,

[Martin72]

5 times better... that is little over 2 bits. IIRC the ENOB of the Rigol MSO5000 is around 5 or so. So this new '12 bit' scope would have an ENOB around 7 to 8 bit.

Specsheet of the 4000 claims greater 8 bits, without any details, not very professionel.

[Fungus]

Specsheet of the 4000 claims greater 8 bits

It also mentions a 16-bit "high resolution" mode. 

[switchabl]

Recall some time ago when Fujitsu Microelectronics shocked the IC design world with their amazing 56 GSPS 8 bit ADC back in ~2009, that was the last we've seen of any significant ADC surprises, maybe there are others lurking in China that we aren't aware of!!

If you look at the presentation material for that device you'll see the actual ADC was not much of a technical challenge. All the complexity in the device is handling the huge flood of data.

I guess it's no big secret that if you want to build a very high-speed ADC, you probably need an RF sampler up front, a massive demux stage and then you just interleave a couple hundred slower ADCs. You can draw the basic architecture on a napkin. Well, and then you just need to solve the endless practical problems that come with it...  Like mismatch (gain, offset, timing) calibration, clock distribution/alignment, the concrete demux/subsampling/buffering topology. I think one of the current Keysight ADCs has a secondary SiGe die that is mostly buffers for signal distribution. Power/thermal budget will be also a big concern all the time. A lot of small things have to go right in order to have a low-noise, low-distortion output instead of a complete mess. Sadly, marketing material tends to focus on impressing potential customers, not on providing interesting technical details to would-be-competitors. 

[tautech]

Recall some time ago when Fujitsu Microelectronics shocked the IC design world with their amazing 56 GSPS 8 bit ADC back in ~2009, that was the last we've seen of any significant ADC surprises, maybe there are others lurking in China that we aren't aware of!!

If you look at the presentation material for that device you'll see the actual ADC was not much of a technical challenge. All the complexity in the device is handling the huge flood of data.

I guess it's no big secret that if you want to build a very high-speed ADC, you probably need an RF sampler up front, a massive demux stage and then you just interleave a couple hundred slower ADCs. You can draw the basic architecture on a napkin. Well, and then you just need to solve the endless practical problems that come with it...  Like mismatch (gain, offset, timing) calibration, clock distribution/alignment, the concrete demux/subsampling/buffering topology. I think one of the current Keysight ADCs has a secondary SiGe die that is mostly buffers for signal distribution. Power/thermal budget will be also a big concern all the time. A lot of small things have to go right in order to have a low-noise, low-distortion output instead of a complete mess. Sadly, marketing material tends to focus on impressing potential customers, not on providing interesting technical details to would-be-competitors. 

Although retired from IC design/layout mawyatt is still bound by NDA's. 
Sad it is too as there is so much to learn of his experiences in crafting the silicon we all use.