Rigol's New DHO800 Oscilloscope unbox & teardown

[2N3055]

not a gamechanger

The DHO800 is a massive step. It defines what every manufacturer needs to do if they want to sell oscilloscopes in the future.
How is that not a game changer?

Now in the sub $400 price bracket that other manufacturers will be forced to follow:
12bit
Touch screen
HDMI
VESA mount
USB-C (arguable downside for some, big feature for others)

I'd call that game changer. As always YMMV.

Yes if you look at it like that it makes sense..
Except...
Did Keysight (or R&S or Tektronix) made any reply in that segment lately?
They didn't. They don't care for sub 500USD scopes market.

I do know Siglent has response to DHO1000 series by Rigol.
As for 800/900 it remains to be seen.
Their initial response to DS1000Z was SDS1000X-E that was better and more expensive and still is... And sells very well still..

Race to the bottom is not always the best answer.

But I'm not saying it would not be nice that it does start a fierce competition, with awesome hardware coming from all sides for more affordable prices...
It's just I won't hold my breath.

Also remains to be seen how will Rigol handle marketing for their now very eclectic portfolio. Maybe some model streamlining?
Stuff that doesn't sell makes no sense to make.

[Fungus]

Maybe cheapest "proper" scope DS1000Z we all keep recommending in that price range will be even more affordable, good news.

Don't get your hopes up: The DS1052E isn't an awful lot cheaper than a DS1054Z.

(and it's still selling!)

On the bright side: We finally got you to admit the DS054Z is a proper oscilloscope... 

[Fungus]

Did Keysight (or R&S or Tektronix) made any reply in that segment lately?

That's like asking if Fluke has responded to Brymen yet.

[Fungus]

Their initial response to DS1000Z was SDS1000X-E that was better and more expensive and still is... And sells very well still..

It sells less than the Rigol DS1052E on Batronix.

Race to the bottom is not always the best answer.

The DHO800 is the polar opposite of a "race to the bottom".

[Nikki Smith]

For those who weren't around earlier:
User Azusa already installed DHO900 firmware in a DHO800 and got 50mpts of memory. It's not a hardware thing.

Was this just an SD card copy?
Is there a 900 SD card dump available?

I think it is just a very small vendor.bin file that was needed? and @hubertyoung shared in earlier in this thread:
https://www.eevblog.com/forum/testgear/rigols-new-dho800-oscilloscope-unbox-teardown/msg4976533/#msg4976533

[Fungus]

Was this just an SD card copy?
Is there a 900 SD card dump available?

Look a couple of posts up from that one...

They changed "vendor.bin" using ADB. You can probably do it on the SD card if you know what you're doing.

nb. The logic analyzer UI also appeared and took up some screen space even though there was no extra hardware installed.

[2N3055]

I'm still having a hard time grokking that the sample data is going into main RAM and the FPGA is processing/decimating it there. There's no obvious bus connection between the FPGA and the RAM chips. It could be on inner layers of the PCB i suppose:

I guess the clue is in the fact that memory depth doesn't quite degrade linearly as you turn on more channels. Maybe there's more bus transactions for the interleaving...

Those RAMs are connected to application processor. You can't have it both way. They probably do data packets to CPU in bursts, with dead time inbetween.

[2N3055]

The DHO800 is the polar opposite of a "race to the bottom".

It is cheapest 12 bit scope available with all the compromises made to make is so cheap.
That is literally the definition...

[Fungus]

The DHO800 is the polar opposite of a "race to the bottom".

It is cheapest 12 bit scope available with all the compromises made to make is so cheap.

What "compromises" have been made, exactly? All I see another level of integration via some very-expensive-to-develop ASICs.

Race to the bottom would have been "a 12-bit DS1054Z", not this.

[2N3055]

The DHO800 is the polar opposite of a "race to the bottom".

It is cheapest 12 bit scope available with all the compromises made to make is so cheap.

What "compromises" have been made, exactly? All I see another level of integration via some very-expensive-to-develop ASICs.

Race to the bottom would have been "a 12-bit DS1054Z", not this.

Only integration is (very good) front end amplifier chip. Otherwise it is ADC-FPGA-Aplication processor system as any other.....

[Fungus]

Their initial response to DS1000Z was SDS1000X-E that was better and more expensive and still is... And sells very well still..

It sells less than the Rigol DS1052E on Batronix.

Looking at today's Batronix sales rankings:

a) The DHO800 and DHO900 are above the Rigol MSO5000 even though they haven't shipped any units yet. 
b) The most popular models are the 2-channel ones (DHO802 and DHO812)
c) The Rigol MSO5000 is still above Siglent SDS2000
d) Even the old Rigol DS2000 is still selling more than Siglent SDS2000
e) The R&S 1k series sells more than the Rigol MSO5000, go figure. 

[Fungus]

Only integration is (very good) front end amplifier chip. Otherwise it is ADC-FPGA-Aplication processor system as any other.....

Front end and ADC are both Rigol ASICs now.

[2N3055]

Only integration is (very good) front end amplifier chip. Otherwise it is ADC-FPGA-Aplication processor system as any other.....

Front end and ADC are both Rigol ASICs now.

They are NOT integrated together.  You still have preamp (rigol chip) -> ADC (rigol chip but not important -> FPGA > application processor. We are talking scope architecture. It is same as before and other scopes.

If ADC and FPGA were combined together then it would be something different.

[gjvdheiden]

Only integration is (very good) front end amplifier chip. Otherwise it is ADC-FPGA-Aplication processor system as any other.....

Front end and ADC are both Rigol ASICs now.

They are NOT integrated together.  You still have preamp (rigol chip) -> ADC (rigol chip but not important -> FPGA > application processor. We are talking scope architecture. It is same as before and other scopes.

If ADC and FPGA were combined together then it would be something different.

Highly unlikely combi (ADC and FPGA) wouldn't it? No flexility in the range of products in PCB / circuit design. And you have to roll you own FPGA. For what gain? It's just digital between ADC and FPGA.

I'm sorry, I don't get your point. I do think it is impressive, investment wise mainly.

[nctnico]

Their initial response to DS1000Z was SDS1000X-E that was better and more expensive and still is... And sells very well still..

It sells less than the Rigol DS1052E on Batronix.

Looking at today's Batronix sales rankings:

Again: these aren't sales rankings. Batronix would be utterly stupid to reveal data like that. More likely they make most margin on the top listed models. Every shop works that way. Put the items with highest margin in front.

[Fungus]

Batronix would be utterly stupid to reveal data like that.

Many shops have lists of best sellers, top ten, most watched, etc.

More likely they make most margin on the top listed models. Every shop works that way.

You're arguing that Batronix makes more from selling the Rigol DS1052E than from selling Siglents?

That works for me.   

[nctnico]

Batronix would be utterly stupid to reveal data like that.

Many shops have lists of best sellers, top ten, most watched, etc.

And you believe that is based on actual purchase data? Ask yourself: how does a shop benefit from such lists?
Quite often the manufacturers pay shops to put products in particular places in a shop. It is perfectly possible Rigol is paying Batronix to list products in a certain order.

[2N3055]

Only integration is (very good) front end amplifier chip. Otherwise it is ADC-FPGA-Aplication processor system as any other.....

Front end and ADC are both Rigol ASICs now.

They are NOT integrated together.  You still have preamp (rigol chip) -> ADC (rigol chip but not important -> FPGA > application processor. We are talking scope architecture. It is same as before and other scopes.

If ADC and FPGA were combined together then it would be something different.

Highly unlikely combi (ADC and FPGA) wouldn't it? No flexility in the range of products in PCB / circuit design. And you have to roll you own FPGA. For what gain? It's just digital between ADC and FPGA.

I'm sorry, I don't get your point. I do think it is impressive, investment wise mainly.

Obviously you don't..

I wasn't sayin it should be done but that was example of integration. If you make your own ADC it is not new architecture, but just another ADC. Which is nice acomplishment fo rthem because they don't depend on anybody. But it is just an ADC chip..

[gjvdheiden]

Even integrating ADC within a FPGA is just moving the same main blocks in a single IC in the end. But yeah, the blocks are the same.

[nctnico]

Even integrating ADC within a FPGA is just moving the same main blocks in a single IC in the end. But yeah, the blocks are the same.

I don't think that is even possible. It is more likely to have two dies in one package. AFAIK analog processes are not compatible with high speed digital processes. Ofcourse you can shoehorn one into the other but I doubt the result is worthwhile. I could be wrong ofcourse. Xilinx has some really expensive FPGAs (as in tens of thousand dollars each) which combine high speed ADCs and FPGAs but I don't know whether these are single or multiple die devices.

[Mortymore]

Looks to me that the DHO800 heatsink is made of casted aluminum and not machined aluminum that I believe has higher thermal conductivity.
If made from machined aluminum, the fan could run quieter then.

https://www.cnclathing.com/guide/machined-aluminum-vs-cast-aluminum-whats-the-difference-between-them

Just a thought. I'm not an expert in the field.

[souldevelop]

Looks to me that the DHO800 heatsink is made of casted aluminum and not machined aluminum that I believe has higher thermal conductivity.
If made from machined aluminum, the fan could run quieter then.

https://www.cnclathing.com/guide/machined-aluminum-vs-cast-aluminum-whats-the-difference-between-them

Just a thought. I'm not an expert in the field.

but Looks to me that the  whether it is cast aluminum or other high thermal conductivity aluminum, the bottleneck of heat dissipation in the entire system is not heat conduction, but more air heat exchange from the outside world.

[souldevelop]

I  teardown DHO914S The point is on the AFG module, if it is possible to copy it and install it to DHO900 or DHO800?
AFG core chip is AD9744.

[mawyatt]

Even integrating ADC within a FPGA is just moving the same main blocks in a single IC in the end. But yeah, the blocks are the same.

I don't think that is even possible. It is more likely to have two dies in one package. AFAIK analog processes are not compatible with high speed digital processes. Ofcourse you can shoehorn one into the other but I doubt the result is worthwhile. I could be wrong ofcourse. Xilinx has some really expensive FPGAs (as in tens of thousand dollars each) which combine high speed ADCs and FPGAs but I don't know whether these are single or multiple die devices.

Believe those Xilinx FPGAs with high speed ADCs are all on a single die. This was extremely difficult to do but before we retired some folks were learning how to implement high performance analog on the same leading edge digital CMOS processes. This became necessary as the silicon foundries were not investing in pushing analog friendly processes, but putting all their investment into advancing digital CMOS. This forced the rethinking of analog chip design and folks began to employ some techniques they learned from the digital folks, basically use only the smallest geometry devices available and throw lots of them at the problem. The ADC architectures have changed also, leaning more towards the digital side to take advantage of such, and utilize charge redistribution techniques (a version of Discrete Time Continuous Amplitude) rather than propagate continous voltages or currents within the chip.

An interesting ADC chip design that emerged from Fujitsu research labs over 2 decades ago started this trend. Remember when we found out about their developmental 56GSPS 8 Bit ADC single chip in digital 65nm CMOS, with dual ADCs. This was based upon DTCA charge redistribution and recall it employed ~250 SAR (charge redistribution) ADCs running at ~250MHz for each final ADC rendering (two per chip!!), and each SAR had multiple digital trim DACs (also charge redistribution). Recall each ADC had over 5000 total trim DACs per final ADC rendering!!

This was all done in a 65nm digital CMOS process, nothing conventional analog about it, with large numbers of the smallest geometry transistors involved. It got a lot of attention from the folks at AD, TI, Keysight, LeCroy, Tek, USG and so on when they found out about it!!


Best,

[Someone]

I  teardown DHO914S The point is on the AFG module, if it is possible to copy it and install it to DHO900 or DHO800?
AFG core chip is AD9744.

For the price delta between those models, it's barely worth trying to source and install (let alone clone or build) the board. No money to be saved.