I'm interested that they went for an FPGA for the display processor. It's an obvious choice of course for 50k waveforms per second. Except, does the LCD actually do that many (best case LCD refresh rate is 3-4ms right now), or is it a DPO effect? Also, the FPGA costs $35 in 100 units, I'll bet much less in 1k to 10k - I wonder if Agilent considered it but found ASIC design to be cheaper (thinking they'd get the chips for $5-10, but invest several million in developing it.)
My guess would be it does the DPO type stuff too, hence the extra big memory around it. You don't need a memory that big for just a frame buffer et.al.
If they hired former Tek engineers, won't they be expecting a similar salary? I was under the impression that salaries would be lower in China. After all, everything else is.
They wouldn't go if it wasn't the same or better.
Good technical talent (including programming) is not cheap or easy to get in China.
Ask Altium!
Dave.
[quote ]
Except, does the LCD actually do that many (best case LCD refresh rate is 3-4ms right now), or is it a DPO effect?
[/quote]
It will be using a standard framerate LCD, and combining the results of multiple acquisitions over the LCD frame time to make an intensity map - essentially a per-pixel histogram of the waveform position in each acquisition.
A simple way to think of it is where a traditional scope just sets a particular pixel in the display memory, a DPO type scope increments it, and does so for every acquisition which can be many times per displayed frame.
Another aspect is that at slower sweep speeds, where there is more than one sample per pixel, it can potentially combine multiple samples into the intensity of each pixel.
The intensity control then controls the scaling of the accumulated pixel values to the displayed pixel intensity.
I'm wondering if the 'tape' you saw on the clock chip might be a piece of ferrite for noise reduction? Makes sense on a clock chip.
As I already wrote in my response to the “EEVblog #360 – Rigol DS2000 Oscilloscope Teardown” video, the input stage can easily be demystified and is very common.
The programmable gain amplifier seems to be a re-labeled LMH6518 with a MMBFJ309 (JFET) followed by a MMBTH10L (VHF amp 650MHz).
The circuit is close to what is given in the datasheet of the PGA/VGA and is often used in low cost and midrange DSOs. I expect that none of the Rigol labeled ICs is something special.
Bandwidth limitation seems to be definied only by the programmed value inside the LMH6518 (20/100/200/350/650/750MHz or full bandwidth), no varactor diode or anything else.
I wonder why they don't use something like the NE3508M04 with much lower noise figure. They could have benefit from the developments on the Welec DSO project.
The programmable gain amplifier seems to be a re-labeled LMH6518 with a MMBFJ309 (JFET) followed by a MMBTH10L (VHF amp 650MHz).
The circuit is close to what is given in the datasheet of the PGA/VGA and is often used in low cost and midrange DSOs. I expect that none of the Rigol labeled ICs is something special.
Bandwidth limitation seems to be definied only by the programmed value inside the LMH6518 (20/100/200/350/650/750MHz or full bandwidth), no varactor diode or anything else.
That will explain the selectable 100MHz (and 20MHz) bandwidth limits in the software, but does not explain the 70MHz bandwidth model?
Dave.
Is it possible Rigol have learned their lesson with the DS1052/1102E hacking and there are actual physical differences between the models?
Is it possible Rigol have learned their lesson with the DS1052/1102E hacking and there are actual physical differences between the models?
The evidence seems to be starting to point in that direction.
i.e.
a) they do not offer it as an upgrade option, even though they have gone that router for all other features.
b) If the chip is as branadic claims, then they must be doing something else physical for the 70MHz limit. I can't believe they do it as a software implemented limit.
But different hardware doesn't make sense from a construction point of view, it adds to the cost. That is why most companies have gone the identical hardware route.
It's all rather confusing.
Dave.
Is it possible Rigol have learned their lesson with the DS1052/1102E hacking and there are actual physical differences between the models?
I would assume they've gone to some length to prevent upgrade hacks. The DS2000 costs enough that the loss of the price difference will be felt, but at the same time there won't be enough sales to make up for it.
Is it possible Rigol have learned their lesson with the DS1052/1102E hacking and there are actual physical differences between the models?
No there aren't , doesn't worth, trust me
a) they do not offer it as an upgrade option, even though they have gone that router for all other features.
Not yet, but there will be "100M BandWidth" and "200M BandWidth" upgrade options available for DS2
(..and "Bandwidth 200M", "Bandwidth 350M" and "Bandwidth 500M" for DS4).
I would assume they've gone to some length to prevent upgrade hacks. The DS2000 costs enough that the loss of the price difference will be felt, but at the same time there won't be enough sales to make up for it.
At the beginning of the product lifecycle (after all the research and development), it likely would slow down their recouping of investment costs if people were buying lower BW models and hacking them into higher bandwidth ones. But once they've made their initial money back, I'm not sure that a hack doesn't help them more than hurt them. I can't imagine how many extra DS1052Es Rigol sold (taking prospective customers away from other brands) with the revelation of the hack - and the possibility of 'free' BW.
I can't imagine how many extra DS1052Es Rigol sold (taking prospective customers away from other brands) with the revelation of the hack - and the possibility of 'free' BW.
Probably quite a lot, but even the 70MHz model is too expensive for the more casual hobbyist market that went after the DS1052s.
That will explain the selectable 100MHz (and 20MHz) bandwidth limits in the software, but does not explain the 70MHz bandwidth model?
There are a few unassembled parts between the mystic Rigol labeled LMH6518 and the differential gain amp, the filter section, maybe this section is assembled in the 70MHz Version? As I already wrote, I can't find any varactor.
That will explain the selectable 100MHz (and 20MHz) bandwidth limits in the software, but does not explain the 70MHz bandwidth model?
There are a few unassembled parts between the mystic Rigol labeled LMH6518 and the differential gain amp, the filter section, maybe this section is assembled in the 70MHz Version? As I already wrote, I can't find any varactor.
So, do you believe it is software implemented?
I'm not sure if it is just software defined, I'm still looking for that.
Today a Rigol DS2072 test device arrived. I took a deeper look onto the circuit and found some further interesting news.
The parts between the relabeled LMH6518 and the garbled IC are also not assembled, so this is not the answer for the bandwidth limitation. The garbled IC's output directly inputs the ADC, there seems to be nothing in between.
Interesting is, that by wetting the garbled IC with some Isoporopanol I found the marking "21AB" (some Rigol code?) and "6552". It's obvious that if they use a relabled LMH6518 (today a TI product) the garbled IC also is a TI one. Also the package seems to be some TI special, which they call 8WSON. Bingo, there is a fully differential opamp from TI named LMH6552, that matches the package, the pinning and the code fragment "6552".
Also interesting is, that the ADC08DL502 from TI (Low Power, 8-Bit, Dual 500 MSPS A/D Converter) seems to match the 144LQFP pinning of the Rigol labeled one. Is this again an overclocked device?
branadic
The clock source is identified. Rigol has milled or etched the surface of the LFCSP-packaged IC. The package identifies, that this could be some IC fron Analog Devices. By knowing what you're searching for I found that the pinning matches the ADF4360 and I would expect that it is a ADF4360-7, to provide a 1GHz differential clock source.
By the way, the datasheet of the ADC08DL502 mentions the guaranteed performance with a differential clock of 1GHz. With the fact that the pcb is full of TI labled ICs it's nearly 100% sure that the ADC is also a TI product.
I can't imagine how many extra DS1052Es Rigol sold (taking prospective customers away from other brands) with the revelation of the hack - and the possibility of 'free' BW.
Probably quite a lot, but even the 70MHz model is too expensive for the more casual hobbyist market that went after the DS1052s.
If you just looking for something basic and inexpensive try this.
http://triosmartcal.com.au/1764-uq2062c-digital-oscilloscope-60-mhz-500-msas.htmlWe've got about 5 left. They are nothing special but they work OK.
A 60MHz scope with a locally supported warranty is not bad for what we are selling it for. There a Silicon Chip review on it too.
That doesn't have intensity graduation, high sampling rate or high waveform capture rate... you may as well buy a DS1052E.
You have a point. The price is lowered. Thanks for the suggestion.
Also interesting is, that the ADC08DL502 from TI (Low Power, 8-Bit, Dual 500 MSPS A/D Converter) seems to match the 144LQFP pinning of the Rigol labeled one. Is this again an overclocked device?
I'm not quite sure how they could overclock to double the 500 MS/s rate. It's quite possible that Rigol has taken a page from Owon's success with the SDS series and is rebadging the Ruifeng MXT2001 ADCs (
http://www.fjrfdz.com/e_products3.asp?ProId=99) - which are a 1GS/s pinout-compatible 'clone' to TI's ADC08D500 series, slightly noisier, but much cheaper (
http://www.fjrfdz.com/download/MXT2001.pdf). That would provide the 2x 1GS/s for the interleaved 2GS/s single channel rate. Extensive tests run on the Owon SDS scopes seem to prove they work very well to specifications.
I think it's fairly clear by now that the entire DS2000 series is identical hardware - just firmware-locked to specific bandwidths.
I don't fully agree that, because the DS2000 series comes with "reduced noise" (~500µVpp in the smallest vertical diversion), what is contrary to what you've mentioned about MXT2001 and it's higher noise compared to the ADC08DL502.
By the way, we may can expect that the MXT2001 is not just pin compatible, but exactly the same core out of the same asian factory?
http://www.ti.com/general/docs/lit/getliterature.tsp?genericPartNumber=adc08dl502&fileType=pdfQuote: "...Although the ADC08DL502 is tested and its performance is guaranteed with a differential 1 GHz clock, it typically will function well with input clock frequencies indicated in the Converter Electrical Characteristics.... Operation up to the sample rates indicated in the Converter Electrical Characteristics is typically possible if the maximum ambient temperatures indicated are not exceeded. Operating at higher sample rates than indicated for the given ambient temperature may result in reduced device reliability and product lifetime. This is because of the higher power consumption and die temperatures at high sample rates. Important also for reliability is proper thermal management..."
- just firmware-locked to specific bandwidths
I have to verify the real bandwidth of the device, because in the LMH6518 there is 20 / 100 / 200 / 350 / ... MHz selectable, but not 70MHz and up to now I couldn't find any limiting filter stage or varactor circuit.
I don't fully agree that, because the DS2000 series comes with "reduced noise" (~500µVpp in the smallest vertical diversion), what is contrary to what you've mentioned about MXT2001 and it's higher noise compared to the ADC08DL502. By the way, we may can expect that the MXT2001 is not just pin compatible, but exactly the same core out of the same asian factory?
You could well be right. I hadn't looked through the specs of the MXT2001 myself - only heard about it from others. A quick check of a couple of stats vs. the ADC08DL502:
SINAD Signal-to-Noise Plus Distortion Ratio (Typ):
ADC08DL502: AIN = 248 MHz @ -0.5dBFS 47.6 dB
MXT2001: AIN = 248 MHz @ -0.5dBFS 46 db
SNR Signal-to-Noise Ratio (Typ):
ADC08DL502: AIN = 248 MHz @ -0.5dBFS 48.5 dB
MXT2001: AIN = 248 MHz @ -0.5dBFS 47 dB
I have to verify the real bandwidth of the device, because in the LMH6518 there is 20 / 100 / 200 / 350 / ... MHz selectable, but not 70MHz and up to now I couldn't find any limiting filter stage or varactor circuit.
So perhaps there is actually no discernible difference between the 70MHz and 100MHz models (since the 2ns timebase setting and 100MHz BW filter only appear on the DS2202 model)? The only difference in specs that I see from the manual (aside, of course, from the BW) is the calculated rise time - which would just be calculated from the 'advertised' BW.
Perhaps there -is- no difference between the 70 MHz and 100 MHz models, they're both 100 MHz?? Conspiracy???
Let the professional chip crackers loose on a new rigol and this happens ...
Good job guys
I have to verify the real bandwidth of the device, because in the LMH6518 there is 20 / 100 / 200 / 350 / ... MHz selectable, but not 70MHz and up to now I couldn't find any limiting filter stage or varactor circuit.
I wonder if there a digital low pass filter implemented in the fpga.. seems like that would be an easy thing (edit: easy for Rigol) to turn on/off in the firmware.