New Owon XDS Oscilloscopes

[nidlaX]

A very immature user who really needs to be banned.

McBryce.

Especially since the formatting is no good.

Keep your dick in a vise!

[Hydrawerk]

http://www.ittsb.eu/forum/index.php?topic=1065.0
I doubt there is a 12-bit 2GS/s ADC. There are even no MSO capability.

[rvalente]

May be an oversampling Technique?

[rf-loop]

May be an oversampling Technique?

Why? How have you arrived at that idea? Some facts or just blind random for nil.

[Wuerstchenhund]

May be an oversampling Technique?

Why? How have you arrived at that idea? Some facts or just blind random for nil.

Well, it's not a completely unrealistic thought, considering that 12bit ADCs with a high enough sample rate (like for example the AD9625) are still pretty pricey.

Plus to process the larger amount of data the processor back end needs to be faster, although this should still be relative inexpensive.

[rf-loop]

May be an oversampling Technique?

Why? How have you arrived at that idea? Some facts or just blind random for nil.

Well, it's not a completely unrealistic thought, considering that 12bit ADCs with a high enough sample rate (like for example the AD9625) are still pretty pricey.

Plus to process the larger amount of data the processor back end needs to be faster, although this should still be relative inexpensive.

I do not know anything about Owon XDS3202A, this product is not avalable, not even demo.
So talking about this model possible 12bit ADC full 2GSa/s sampling rate is just null.
I do not know how this model 12Bit mode works.

Instead, I know XDS3102A. 
They tell its max samplerate is 1GSa/s, there is 12bit resolution, 40M memory etc.

For XDS3102A  facts are:
There is 1GSa/s for single channel in 8bit mode and 500MSa/s for 2 channel simultaneously.
1Ch alone and max memory is 40M
2Ch simultaneously and there is 20M for each channel.

Maximum samplerate for 12bit mode is 500MSa/s for single channel.
(Later I show a table.)

12Bit and 8bit frequency bandwidth's  are around same.  (I think this is not software "hi-res" or "ERES" implementation, it is hardware, there must be ADC what have 12bit resolution mode. I have strong suspect what ADC it is but I need more direct evidences before I can tell it. The rumors, delusions and facts are useful to keep separated from each others.

[Wuerstchenhund]

For XDS3102A  facts are:
There is 1GSa/s for single channel in 8bit mode and 500MSa/s for 2 channel simultaneously.

[...]

Maximum samplerate for 12bit mode is 500MSa/s for single channel.

Well, the sample rate difference does strongly suggest that the 12bit resolution is actually achieved through oversampling.

The presence of an "8bit mode" also very much suggests that this isn't a true 12bit scope but simply an oversampled 8bit ADC (true 12bit scopes run at 12bit all the time, as there's not much point in reducing the resolution, although you can increase resolution further beyond 12bit through ERES).

12Bit and 8bit frequency bandwidth's  are around same.  (I think this is not software "hi-res" or "ERES" implementation, it is hardware, there must be ADC what have 12bit resolution mode. I have strong suspect what ADC it is but I need more direct evidences before I can tell it. The rumors, delusions and facts are useful to keep separated from each others.

There's nothing wrong with speculating, especially when the manufacturer can't be bothered to provide more detailed information.

The thing is that the target price of the scope doesn't leave much room for true 12bit ADCs, even less so for the faster scopes that claim the same sample rate at 8bit and 12bit.

You have to understand that people are skeptical when OWON claims to offer a true 12bit 2ch 200MHz 2GSa/s scope (XDS3202A) for less than $1500 when the cheapest A-brand true 12bit equivalent with similar specs (2ch 200MHz 2.5GSa/s) starts at roughly $9k.

[rf-loop]

For XDS3102A  facts are:
There is 1GSa/s for single channel in 8bit mode and 500MSa/s for 2 channel simultaneously.

[...]

Maximum samplerate for 12bit mode is 500MSa/s for single channel.

Well, the sample rate difference does strongly suggest that the 12bit resolution is actually achieved through oversampling.

The presence of an "8bit mode" also very much suggests that this isn't a true 12bit scope but simply an oversampled 8bit ADC (true 12bit scopes run at 12bit all the time, as there's not much point in reducing the resolution, although you can increase resolution further beyond 12bit through ERES).

12Bit and 8bit frequency bandwidth's  are around same.  (I think this is not software "hi-res" or "ERES" implementation, it is hardware, there must be ADC what have 12bit resolution mode. I have strong suspect what ADC it is but I need more direct evidences before I can tell it. The rumors, delusions and facts are useful to keep separated from each others.

There's nothing wrong with speculating, especially when the manufacturer can't be bothered to provide more detailed information.

The thing is that the target price of the scope doesn't leave much room for true 12bit ADCs, even less so for the faster scopes that claim the same sample rate at 8bit and 12bit.

You have to understand that people are skeptical when OWON claims to offer a true 12bit 2ch 200MHz 2GSa/s scope (XDS3202A) for less than $1500 when the cheapest A-brand true 12bit equivalent with similar specs (2ch 200MHz 2.5GSa/s) starts at roughly $9k.

I can not talk anything about XDS3202A least because this model do not exist or least I do not know where is.
What I know is XDS3102A.


"Well, the sample rate difference does strongly suggest that the 12bit resolution is actually achieved through oversampling."

Really, can you explain? What is there doing oversampling?

 How to do "12 bit resolution" or even  9 bit...  using 1GSa/s 8bit ADC so that one shot real time samplerate is 500MSa/s and real time frequency response is least over 160MHz. (measured using XDS3102A using its 12bit mode) 

I do not know anything what is inside THIS Owon model but lets take just basic trivial example...
Take one Hittite HMCAD1520 and after then start - if nothing else - but thinking. What can do with it.

[rf-loop]

Owon XDS3103A   waveform update speed. This table is only small fraction of all mesurements about wfm/s. (later more)

It need note that Owon memory selection means always true sample buffer lenght. It is not reduced to display width.
Example if 1k is selected and speed is 10ns/div. If samplerate is 1GSa/s there is 1ns sample period. Acquired lenght By time is 1us (1000ns) but in this case display have only around 150ns part, rest of trace is just outside. If stop scope it can zoom out and/or horizontally shift for look)  Even with 2ns/div setting and with 40M memory... display is around 30ns and whole trace is 40000000ns.

If need fastest possible wfm/s user need select 1k
( there can not select less and there is not auto what reduce it automatically for display width.  It need note that this rise visible  blind time.)

If persistence is selected on, it rise some amount wfm/s speed but also with it, it can not speed up to claimed 75kwfm/s. (if there exist some very short time peak values, this I have not tested.)

Later more.
Add: Here better  full table: https://www.eevblog.com/forum/testgear/new-owon-xds-oscilloscopes/msg769599/#msg769599

[rf-loop]

The presence of an "8bit mode" also very much suggests that this isn't a true 12bit scope but simply an oversampled 8bit ADC (true 12bit scopes run at 12bit all the time, as there's not much point in reducing the resolution, although you can increase resolution further beyond 12bit through ERES).

Can you explain this your comment?

Of course it can be true 12bit ADC.  Why you think there is not point for 8bit resolution. There is very strong point to do this.
If ADC can do max 500MSa/s in 12bit mode and 1GSa/s in 8bit mode there is just well enough reason.

And then about this ERES comment. Did you think anything aboit bandwidth?

Can you even small amount think what is possible result if 1GSa/s 8bit adc data is used for ERES so that result is around 12 bit resolution.  Do it have 2MHz bandwidth in 12bit mode? No, it do not have. as told previously, it have around same BW if 8bit or 12bit mode is selected.  So how relevant is your comment in this case.

[Hydrawerk]

I think that DMM on the XDS series back panel is taken from their handheld series. Please note that the screen image is simulated.

[rf-loop]

Quick look to FFT

Signal 86MHz sine (HP/Agilent 8644B), FFT 0-125MHz  zoomed so that one div is 500kHz.


Ref level adjusted so that no clipping, just below 10div height p-p

Then adjusted signal 70dB down, right inserted image (ref -70dB). After then dropped 10dB more and left inserted image (ref -80dB).

All other settings untouched.   

Images directly with PC using Wi-Fi,  scope as Wi-Fi AP. (upper right corner scope wifi symbol)

[rf-loop]

Here is full table about maximum average waveform update rate for single channel in use.
It need note that if persistence is in use then wfm/s speed may be bit higher with some settings.

In this table is used 1s or 10s average speed. There may be some higher wfm/s speed if look highest short time peak speed. I do not use these values.

With some t/div and memory settings user can select between 8 or 12 bit mode for ADC.  If n/a then scope of course use only option.  (If ADC is working in one channel mode then  12 bit mode  max real samplerate 500MSa/s )


Later same using 2 channel

[pascal_sweden]

Seems they still have VGA output, but on the specs it's listed as an option.

When will they go for an HDMI output? =)

[rf-loop]

Seems they still have VGA output, but on the specs it's listed as an option.

When will they go for an HDMI output? =)

If I remember right some villages and towns schools in Nigeria still may have VGA monitors next 5 years....

And..
Because it is more interesting to talk about what feature somewhere do not exist than what  there is. 

But some thing still exist today. If there is customer who need A and wnat pay for A there is manufacturer / seller who do and sell A.  But if there is manufacturer who make A without anybody want A exept some rare hobbyist or other...  sorry but I do not believe this business - I do not buy this idea. Even if A is nice.

[rf-loop]

Here is full table about maximum average waveform update rate for two channels in use.
It need note that if persistence is in use then wfm/s speed may be bit higher with some settings as also with single channel.

In this table is used 1s or 10s average speed. There may be some higher wfm/s speed if look highest short time peak speed. I do not use these values. I think I have seen some very rare and short time peak values up to over 74kwfm/s but if there exist just some peak value between two individual acquisition it is not very honest to write it on promotion material. I do not know if touch panel eat any power from acquistion speed if not touched. I can not test scope without this option to look if there exist 75kwfm/s  without fisherman lies.

I do not know if there is some advantage possible in later FW versions. I wonder why persistence on give bit better speed. Example two channel 1k mem, 100ns/div. Persistence off 57500 wfm/s and persitence 1s selected 61400 wfm/s or somethink like this (10k memory) 19500-->20600; 19300-->20400 wfm/s

If two channels are in use then user can not select between 12 and 8 bit ADC mode.
As long as it can use 500MSa/s for both channels it use ADC in double channel 8bit mode and if samplerate is 250MSa/s it use ADC in double channel 12bit mode.  More low samplerates are only decimation from true samplerate.

True raw samplerates are, until different evidences:
Single channel 8bit mode 1GSa/s, Single channel 12bit mode 500MSa/s
Double channel 8bit mode 500MSa/s and double channel 12bit mode 250MSa/s.

[Hydrawerk]

Well, for a 100MHz scope in 2015, I want 1GS/s per each channel. Anyway, what does the intensity gradation  (aka DPO) look like?

[rf-loop]

Anyway, what does the intensity gradation  (aka DPO) look like?

Its existence is nowhere mentioned.  So question is bit weird. If someone have claimed that there is intensity grading, then can ask what it looks like.
Usually I do not know what it looks like what do not exist.  As also at this time.

[Hydrawerk]

So it is worse than Rigol DS1000Z in this aspect. https://youtu.be/580izt-2jWk?t=1m46s

[Hydrawerk]

I wonder what that CINCH-like connector is for. Well, it is a video output.
http://www.tomtronix.pl/owon/xds.aspx
Have you seen this manual?
http://www.owon.com.hk/probook/XDS3000_Series_Oscilloscopes_QUICK_GUIDE.pdf
http://www.tomtronix.pl/pub/d_owon/karty/XDS_oscyloskopy_kk_ang.pdf

[rf-loop]

About 8bit and 12bit ADC mode.

Single shots using 8bit and 12bit ADC mode.
Both single shots zoomed horizontally 10x and vertically 10x
12bit frequency response is full of course, it is not at all like ERES or poor box car "hi-res".

In 8bit zoomed image can see clearly that vertically one div is 25 levels. (2.5 in zoomed)
Owon vertical 10div display have 250 8bit ADC levels from 256 full scale.
12bit ADC mode true 12bit res staircase figure can not naturally reach due to sum of noises etc. Also fit to display resolution may produce  some efect.

[Orange]

Hi RF-LOOP,

Can you repeat these tests with a time base setting of 10 or 100nS/div ?

[rf-loop]

Hi RF-LOOP,

Can you repeat these tests with a time base setting of 10 or 100nS/div ?

Here 100ns/div

 
8bit
first normal run - still image (it need note that because it is still image, live visual effect is different -   more or less.)
then single shot
after then this single shot zoomed vertically 10x using V/div
after then zoomed  also horizontally 10x using t/div

12bit
normal run
single shot
then this single shot zoomed vertically 10x using V/div
after then zoomed  also horizontally 10x using t/div


Then 10ns/div
8bit
normal run
single shot
then this single shot zoomed vertically 10x using V/div
after then zoomed  also horizontally 5x using t/div   (min is 2ns/div)


12bit
normal run
single shot
then this single shot zoomed vertically 10x using V/div
after then zoomed  also horizontally 5x using t/div

Signal, levels etc selected so that fixed Sin(x)/x affect as little as possible. This can not turn off.
In these images this can see clearly in images fx and fy. (note this effect can not see in run mode)
I hope Owon develop FW so that user can turn Sinc on and off when ever he want, also in stop mode and also so that scope then show only true sample datapoints and then also line (vector) mode where just only these true sampled points are connected usin straight line. Now user is out from reality.
Also when do Sinc, it must never  draw trace what do not go through true real raw sampled datapoints as some toys may do.
It looks like Owon follow this rule number 1 but it is difficult to confirm because it do not show what are real true raw datapoints from ADC.. Also in some situations vector and or real dot mode is useful. Owon dot mode is not pure real. Scope FW add some intermadiate points betwwen real points but it do not mark real points example highlighting them.. )

With some cheap 8 bit scopes it is (perhaps) difficult to get visible quite clean 8bit ADC staircase...

[Orange]

Looks like a real 12bit system then.

Thanks for all your testing rf-loop, this gives a good insight of the DAC of the scope.

[Performa01]

Here 100ns/div

The 12-bit results look very good indeed.
But what the heck are these extra transitions/spikes in the 3rd image (bx-100n-8b-single-Vzoom10) ?
That doesn't look right for my tasting...