new RIGOL DS1000Z and... MSO4000

[Phili76]

1.6V still ok, above 1.7V it starts clipping.
around 1.68V is the last change in peak voltage.

If you need any other test, no problem :-)
philipp

[sync]

I tested the clipping on different vertical scales. I used a 16 div. pp triangle signal on each scale to make comparison easier. Stopped the scope, reduced the scale and make the screen shot. The values in the file names are the scale settings of the measurement.

[Phili76]

just tried the other channels, with only 500MS/s complete different behavior ...
philipp

[marmad]

Thanks, guys!

It appears that the DS1000Z is mapping even less of the ADC to the display than the DS2000. In sync's images, whenever the scale is increased by 2x, you can see that the waveform occupies ~5.5 divs peak to peak - which means it clips at ~11 divs full scale.

That means that the DS4000 does 250 to the 400 vertical pixels of the waveform display area; the DS2000 does 200, but it seems the DS1000Z is something like 184.

[Phili76]

@sync: does it change when you disable the second channel?
philipp

[sync]

The clipping varies with the vertical scale, numbers of channel used and maybe other things. For this measurement I used the Vpp measuring function on a reduced scale to measure the clipping.

[marmad]

The clipping varies with the vertical scale, numbers of channel used and maybe other things. For this measurement I used the Vpp measuring function on a reduced scale to measure the clipping.

Wow... that's a total mess. There seems to be no standard - which would make recreating the display on the PC or another device a major pain-in-the-ass (unlike the other UltraVision DSOs).

But even worse: the fact that you sometimes see less than half the scale of the ADC on the DSO screen (and often between ~50 - 70%) kind of sucks.

[sync]

Wow... that's a total mess. There seems to be no standard - which would make recreating the display on the PC or another device a major pain-in-the-ass (unlike the other UltraVision DSOs).

Doesn't the DSOs scale/normalize the samples before sending to a computer? I would expect that.

[marmad]

Wow... that's a total mess. There seems to be no standard - which would make recreating the display on the PC or another device a major pain-in-the-ass (unlike the other UltraVision DSOs).

Doesn't the DSOs scale/normalize the samples before sending to a computer? I would expect that.

Perhaps - and I need to find that out. On the DS2000/DS4000, display memory reads always return values between 0 - 255 (equating to ADC values). I need to know how those values scale to the vertical display area in order to make the waveform look correct when displaying it real-time. As mentioned, on the DS2000, it's always 1 step = 2 pixels / on the DS4000, it's always 1 step = 1.6 pixels.

If you've got the NI VISA drivers installed, you can test what the DSO is outputting from the display memory using the attached small program. Just start it, then select channel(s) (it doesn't matter if the channel is ON or OFF at the DSO), and click read.

You'll see a decimal list of the bytes from the display memory for that channel, including a summary at the bottom of the lowest/highest value in the list. If the channel is OFF (at the DSO), you would normally get all 127s - ADC half-scale (flat line).

Adjust a signal so that the top and bottom are close to the top and bottom of the full display area, then STOP the DSO, and use the vertical position knob to make the waveform exactly touch the upper or bottom grid line - then do a read. That will give you the max (or min) at that edge - which is what I need to know. You can also check that the same values (at display edge) are returned regardless of scale or number of channels on.

On the DS2000, a waveform touching the bottom edge of the display area always produces the value = 28 (min) and the top edge produces the value = 227 (max).

EDIT: BTW, USB connection is fastest and automatically detected - but if you're using LAN, you'll have to enter your IP address when the program starts using the following format (copy and paste, but replace the shown IP with your correct one):
TCPIP::192.168.1.200::INSTR

[sync]

I can't run your program. I'm a Linux user. But I captured a few waveforms via USB. The values are like the DS2000 (:WAVeform:FORMAT BYTE). The values are clearly from the display memory not the ADC/ sample memory. Are you sure that the DS2000 gives the ADC values?

I had problems with :WAVeform:MODE MAX and :WAVeform:MODE RAW. They gave wrong or no data (error message on the scope). I don't know if this a problem with the scope firmware or my setup.

[marmad]

I can't run your program. I'm a Linux user. But I captured a few waveforms via USB. The values are like the DS2000 (:WAVeform:FORMAT BYTE). The values are clearly from the display memory not the ADC/ sample memory. Are you sure that the DS2000 gives the ADC values?

I'm not looking for the actual ADC values - those are easy to get and not a problem - but you can't do it real-time while the DSO is running. OTOH, the display memory is a series of vectors derived from interpolation and peak-to-peak decimation which are EQUATABLE to ADC values - giving back a value between 0 - 255 that are ALWAYS mapped (on the DS2000) - regardless of scales or number of channels - precisely the same way to the display. If you clip the signal on the top or bottom, the returned value for that part will always end up at 0 or 255.

I need to find someone who can run the small program and discover what the scale if for the DS1000Z - and if it's fixed or not.

[sync]

On the DS1000Z it's 0-255. 127 in the middle. 25 values per division. So 201 values for the displayed 8 divs.

[marmad]

On the DS1000Z it's 0-255. 127 in the middle. 25 values per division. So 201 values for the displayed 8 divs.

But this doesn't match up with your previous data. I'm not sure how both this and the chart you published before could be accurate. The display memory is a reflection of what's been captured. If you stop the DSO with part of the waveform off the screen, it will be either clipped or not clipped if you reduce the scale. If it's not clipped, it means the data has been captured by the ADC and was already present in the display memory.

EDIT: Ahh.... I just realized that the DS1000Z must be doing magnification of the ADC values!

[marmad]

I just checked on the DS2000:

When running at the small scales (<= 5mv/div) EXCEPT for 500uV (which we already learned over in the other thread is a magnification of 1mv/div), the clipping always happens at 10 divs - meaning the full ADC is mapped to 10 divs. But at 500uV/div, the clipping happens at 20 divs - which shows that it's actually 1mv doubled.

If the clipping is happening at a larger number of divisions than 10 - depending on different settings of the DS1000Z - you have to imagine that it's actually using magnification of the ADC values to produce those values.

Depending on the magnification value, if Dots display is turned on, you can sometimes see streaks in the displayed pattern.

[Phili76]

just tested, 1.4Vpp 200mV/div, 39 min 215 max
rescaled to 178mV, 29 min 227 max
so should be like DS2000

philipp

[Hydrawerk]

(...) DS2000: (...)
EXCEPT for 500uV (which we already learned over in the other thread is a magnification of 1mv/div)

I think that this is not mentioned in the datasheet. 

[marmad]

just tested, 1.4Vpp 200mV/div, 39 min 215 max
rescaled to 178mV, 29 min 227 max
so should be like DS2000

philipp

Thanks, Philipp. I had assumed this was correct.   We just got side-tracked by the fact that the DS1000Z appears to use magnification of ADC values depending on the number of channels ON and the scale. On the DS2000 this is a constant - unless you are on the 500uV/div or vernier setting.

[Dave Turner]

Marmad

I have a DS1074Z-S software version 00.02.00.SP1, which I'm still learning to use effectively. I'd be happy to try tests which do not screw the warranty or potentially brick it as I can't afford a replacement. If I can help as a control please give me idiot proof destructions.

Dave 

[sync]

EDIT: Ahh.... I just realized that the DS1000Z must be doing magnification of the ADC values!

Yes, it scales the ADC values to the display values. And :WAVeform:DATA? returns the display values.

[marmad]

EDIT: Ahh.... I just realized that the DS1000Z must be doing magnification of the ADC values!

Yes, it scales the ADC values to the display values. And :WAVeform:DATA? returns the display values.

But as mentioned, it shouldn't really be doing that unless it's specified in the datasheet - which it isn't (just as the 500uV magnification isn't specified in the DS2000 datasheet). And it appears to be doing it a lot and at many different settings. When are you actually seeing > 75% of the ADC's full scale?

[sync]

But as mentioned, it shouldn't really be doing that unless it's specified in the datasheet - which it isn't (just as the 500uV magnification isn't specified in the DS2000 datasheet). And it appears to be doing it a lot and at many different settings. When are you actually seeing > 75% of the ADC's full scale?

Simpler (low cost) analog front end. The DS1000Z is not a high fidelity scope.

Edit: the programming guide is clear about the :WAVeform:DATA? values:

:WAVeform:YREFerence?
Query the vertical reference position of the specified channel source in the Y direction. The YREFerence is fixed at 127 (the bottom of the screen is 0 and the top is 255).

The middle of the display = 127.
0 and 255 for bottom and top of the screen is obviously wrong.

:WAVeform:YINCrement?
Query the waveform increment of the specified channel source in the Y direction. The unit is the same as the current amplitude unit.
Explanation YINCrement= VerticalScale/25

25 values per division.
Bottom gratitude line: 27 = 127 - 4 * 25
Top gratitude line: 227 = 127 + 4 * 25

[marmad]

Simpler (low cost) analog front end. The DS1000Z is not a high fidelity scope.

This has nothing to do with it. Rigol has been pretty good about being fairly honest and following established procedures for their documentation, but in this regard they're not (look at other DSO manufacturers specifications). If scales are actually magnified ADC values (beyond what is necessary for translation to pixels), it should be noted. 256 values are already a small amount - if they're only showing you 50% of that, you should know.

Edit: the programming guide is clear about the :WAVeform:DATA? values:

I have never been unclear about the :WAVEform:DATA values - but I think you have.   I only needed to know how the values were mapped to the display - as noted many times already, it varies between the DS2000 and DS4000.

25 values per division.
Bottom gratitude line: 27 = 127 - 4 * 25
Top gratitude line: 227 = 127 + 4 * 25

As mentioned many posts ago, bottom line is 28 - top is 227. It's 200 points inclusive.

[sync]

As mentioned many posts ago, bottom line is 28 - top is 227. It's 200 points inclusive.

Yes, you are right. I measured a screen shot. The distance between top and bottom of a clipped waveform are 398 pixels. It uses vertical pixel doubling. So 200 points inclusive.

[marmad]

Yes, you are right.

I should hope so - I've been writing software that reads and displays the Rigol display memory on screen for over a year now

It uses vertical pixel doubling. So 200 points inclusive.

But that's only doubling to the screen - which any 8-bit DSO has to do to some degree with higher resolution screens. But your DSO appears to be multiplying smaller subsets of the real ADC values in order get faster waveform update rates. If I was a DS1000Z owner, I would want to know from Rigol exactly WHEN and HOW MUCH of that is going on.

[sync]

which any 8-bit DSO has to do to some extent with higher resolution screens. But your DSO appears to be multiplying smaller subsets of the real ADC values in order get faster waveform update rates.

I don't think it's about faster waveform rates. It's more computing for the additional scaling (magnification) than a simple 1:1 mapping. I think it's a cheaper analog front end. The analog gain and offset doesn't need to be accurate set. The correction is done in software.