Author Topic: New Keithley DMM6500 and now DAQ6510  (Read 296891 times)

0 Members and 1 Guest are viewing this topic.

Online HighVoltage

  • Super Contributor
  • ***
  • Posts: 5453
  • Country: de
Re: New Keithley DMM6500 and now DAQ6510
« Reply #275 on: December 04, 2018, 03:22:55 pm »
I ordered a DMM6500 today, might receive it before Christmas.
It is confirmed to come with a free Kickstarter license (Promotion until March 2019)
There are 3 kinds of people in this world, those who can count and those who can not.
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 14080
  • Country: de
Re: New Keithley DMM6500 and now DAQ6510
« Reply #276 on: December 04, 2018, 04:59:01 pm »
What exactly are you trying to do?  That's a lot of info you're asking for...  Like for DCV Bandwidth: it varies depending on a handful of settings and isn't spec'd, so if you tell me why you want it or how you'll use the info, I can estimate a figure for you.  For something like a block diagram, what are you looking to find? 
I am an engineer old school.
I used to understand the device with which I work. It is very difficult for me to work with a black box which I do not understand.

For example, I measure the noise of a low-noise power source. And I want to know the band of the device as the measured noise level depends on it. And I want to have this answer in my head and not ask the engineer Keithley every time.

Or I do not want to hear the switching of relyushek in double dimensions. For example, I crossed out this mode for myself from the applicable. And it turns out that under certain conditions I can use it.

Those. I really want to work with the device myself without distracting Keythley engineers from work.

Moreover, the programmers received detailed instructions for working with the device for 1000 pages. And engineers have nothing but a very modest verbal description which is difficult to understand :(

With the modern DMMs it is not common anymore to get a detailed description on how they work, that was in old days, like 30 years ago. In part this is to make is more difficult to copy the instrument in China. The other part is that the operations get more complicated with many more modes. So it would be quite a thick book to describe the operation.  Another problem is that details might even change over revisions.
So I don't think there is much hope - at least for a more normal, lower cost instruments. One might get a little more information for a metrological grade instrument - because one would need it to get the best and reliable performance.

Still it would be nice to have at least a little more information, so one does not need to start with a kind of reverse engineering / testing to get information such  as the actual data rate / time spacing, actual aperture and measurement sequence in the Ohms modes. Also for the digital filtering used in the meter, it would be nice to know about the exact filter function, as this effects noise estimates from the readings. With filtering use the simple std. deviation values calculated have to be taken with a grain of salt - as the raw data may be correlated.

Sometimes it makes a difference if the reading is a continuous integration of e.g. 500 ms or the average of 5 readings at 100 ms. This also effects things like the time from trigger to the actual reading. Also when using the scanner the timing of the reading can be important.
So at least for the more serious meters such information should be in a good manual.

Not knowing such details might lead to problems like not always waiting long enough for settling. This could lead to problems like the odd out-layers once every 160 readings seen in some cases - though in this case more like a problem for the DMM internal software.
 

Offline Brad O

  • Regular Contributor
  • *
  • Posts: 85
  • Country: us
  • Keithley Apps Engineer
    • Keithley homepage at Tektronix
Re: New Keithley DMM6500 and now DAQ6510
« Reply #277 on: December 07, 2018, 09:00:48 pm »
I am an engineer old school.
Oh well in that case: I gathered some info from the design engineers.  We're really quite happy to share details, but it helps us if you're specific in what you want to know and why you want to know it.  I think I addressed all the questions below:

To know when the relay will click, and when not.
The relay positions aren't documented right now so I just found these by trial and error.  I'll see about making a cohesive table for this later, I know reading it in text form is complicated.  If you go into or out of the following functions/ranges, you will hear a relay actuate.  These are only for Voltage, current, and ohms.  The other functions reuse these relays. 

For the DMM6500: Relays are used for the AC I/V functions, ACV 10V and higher, DCV 1000V, 10 Ohms and lower (both 2W and 4W), 100mA range and 1A and higher ranges both ACI and DCI.  The DC/ACI >=1A relay is the same as the <1000DCV relay, so switching between those will not have a relay click.  Digitize functions use the same relays as their DC counterparts. 

For the DMM7510, assume it's the same as the DMM6500, but with these differences: No relay for low ohms but instead has a relay for 10MOhm and above.  DCV has no 1000V relay.  DigiV has a relay for 10V and below (>10V shares the relay with DCV).  I don't think I missed any...

What is the input bandwidth for DCV.
From a purely hardware perspective up to but not including the ADC: up to hundreds of kHz to support the DCV digitizer function for 10V and less ranges. 100V and 1000V ranges a lot less than that due to internal 10Meg divider.  Generally similar to digitzer specs.

Is it possible to somehow find out the exact values of current shunts. The device itself must know them?
The Nominal values are copied here from the specifications: 10 μA-10 kΩ; 100 μA-1 kΩ; 1 mA-100 Ω; 10 mA-10 Ω; 100 mA-1 Ω; 1 A-100 mΩ; 3 A-100 mΩ; 10 A-5 mΩ.  These have a 1% tolerance.  The exact values are never measurable by the DMM or otherwise because of the path resistances and connections inherent to the DMM (unless you open the box and cut the resistors out).

How are the processors, trigger connected inside ...
And any other information on the device. Scheme. Block diagram Sketches.
The processors/trigger/measurement relationship is too complicated to document here and changes depending on the specifics of how you're measuring.  Is there some case you had in mind?  We want to try and answer your question, but we could make hundreds of different block diagrams about the box.

the actual data rate / time spacing
Do you mean the flow of data out of the box?  Or the spacing of measurements?  Typical reading rates out of the box for the best case scenarios are given in the datasheet, they'll go down from there.  Spacing of measurements are controlled by things like autozero, multiphase measurements (so like 4-wire ohms, DCV ratio, things where one measurement really takes more than one measurement), line sync, aperture, autoranging, autodelay, etc.  I don't think it's practical or possible to list all the possible combinations and the spacing for each function.

actual aperture and measurement sequence in the Ohms modes
The aperture for ohms modes is settable like in any other function, unless you meant something else? The measurement sequence is different depending on what range you have selected (the 10M and 100M range use a ratiometric method, described starting on page 204/5-59 of the manual) and certain measure settings like offset compensation and open lead detection add extra phases before the measurements take place. Depending on your particular settings, the measurement sequence gets up to 7 phases long for 4-wire ohms.

Also for the digital filtering used in the meter, it would be nice to know about the exact filter function, as this effects noise estimates from the readings. With filtering use the simple std. deviation values calculated have to be taken with a grain of salt - as the raw data may be correlated. 
"Filter" Filtering is applied by the digital process after calibration is applied to the readings.  There are two different types of filtering that are detailed on page 198/5-53 of the manual, plus Hybrid filtering which is in the 1.0.01f firmware release notes or I wrote about it earlier in this thread (https://www.eevblog.com/forum/testgear/new-keithley-dmm6500/msg1768589/#msg1768589).  All this "post-processing" follows an order of operations on page 215/5-70, I attached an image of it too (which is actually a bit misleading, going to the display and the buffer actually happen simultaneously, I put a bug report in for that).

Not knowing such details might lead to problems like not always waiting long enough for settling. This could lead to problems like the odd out-layers once every 160 readings seen in some cases - though in this case more like a problem for the DMM internal software.
You can definitely configure the box so that the DMM won't return settled measurements.  The power up default enables Autodelay which ensures settling in most cases.  The times added by autodelay are given starting on page 189/5-44 of the manual.  If you have an atypical setup, like really long cables or circuits, you may need to add manual delays.
 

Offline drummerdimitri

  • Frequent Contributor
  • **
  • Posts: 525
  • Country: lb
Re: New Keithley DMM6500 and now DAQ6510
« Reply #278 on: December 08, 2018, 12:11:11 am »
Since the DMM6500 can take simultaneous readings of current and voltage is there some script to measure power other than using a shunt resistor?
The device cannot measure current and voltage simultaneously. Voltage only or current only. At the same time spending a relay resource.

How was I able to measure the DC voltage using the rear ports and DC current as a secondary measurement then?

There were no relays switching between the two modes  :-//
MegaVolt is right, the DMM6500 can't measure Voltage and Current simultaneously, but you might not always hear relays clicking.  There are a couple range combinations that don't require a relay switch, typically because the the signal paths use the same range resistors.  10V and 1A is an example combo, but there are a few others.  This behavior was the same for the 7510 as far as I know, but the range combinations might be different.

I noticed that but for the ranges where a simultaneous non-relay switching measurement, is there a script that can multiply the current and voltage and give me a power reading instead of using a shunt resistor and the voltage ratio method?
 

Offline TiN

  • Super Contributor
  • ***
  • Posts: 4543
  • Country: ua
    • xDevs.com
Re: New Keithley DMM6500 and now DAQ6510
« Reply #279 on: December 08, 2018, 12:58:11 pm »
Are there are plans to release calibration/service manual for DMM6500? I'd be interested to see the calibration points for ACV.

MegaVolt

Some of the internal pics for DMM6500 you can find here. Perhaps they can shed some more light on how it's designed and working.

YouTube | Metrology IRC Chat room | Let's share T&M documentation? Upload! No upload limits for firmwares, photos, files.
 

Online MegaVolt

  • Frequent Contributor
  • **
  • Posts: 910
  • Country: by
Re: New Keithley DMM6500 and now DAQ6510
« Reply #280 on: December 08, 2018, 10:33:20 pm »
Oh well in that case: I gathered some info from the design engineers.  We're really quite happy to share details, but it helps us if you're specific in what you want to know and why you want to know it.  I think I addressed all the questions below:
Many thanks to Brad O for the information gathered.

Then if you do not mind, I will continue to ask questions.


Quote
What is the input bandwidth for DCV.
From a purely hardware perspective up to but not including the ADC: up to hundreds of kHz to support the DCV digitizer function for 10V and less ranges. 100V and 1000V ranges a lot less than that due to internal 10Meg divider.  Generally similar to digitzer specs.
Does this mean that I must include an external anti-aliasing filter to eliminate the effect of spectrum overlap. And at the same time reduce the input noise level?

Quote
How are the processors, trigger connected inside ...
And any other information on the device. Scheme. Block diagram Sketches.
The processors/trigger/measurement relationship is too complicated to document here and changes depending on the specifics of how you're measuring.  Is there some case you had in mind?  We want to try and answer your question, but we could make hundreds of different block diagrams about the box.
For example, I absolutely do not understand how I can get all the signal samples through the remote interface. Not one buffer. Namely, all data without gaps. I do not understand how the data pass. How does screen information or network communication affect information gathering? How do I communicate on the network so as not to interfere with the measurements. I do not understand how I should set up triggers and buffers so that the data is not lost and at the same time I had time to read them over the network.
 

Online MegaVolt

  • Frequent Contributor
  • **
  • Posts: 910
  • Country: by
Re: New Keithley DMM6500 and now DAQ6510
« Reply #281 on: December 08, 2018, 10:40:19 pm »
Some of the internal pics for DMM6500 you can find here. Perhaps they can shed some more light on how it's designed and working.
Thanks for the link. But I still do not know how to reverse engineer in photography :)

Although I am very interested in the type of resistor 10 MΩ and the method of its connection to the input.

I asked about the information that can not be obtained even from the scheme. Some logical block diagram of how the device is arranged.

For example, above Brad O spoke about the set of processors that have access to data with different priority. They still communicate with each other. And it would be good to understand what cubes are, how they are connected and what interesting of them can be folded.
 

Offline Brad O

  • Regular Contributor
  • *
  • Posts: 85
  • Country: us
  • Keithley Apps Engineer
    • Keithley homepage at Tektronix
Re: New Keithley DMM6500 and now DAQ6510
« Reply #282 on: December 10, 2018, 09:09:49 pm »
I noticed that but for the ranges where a simultaneous non-relay switching measurement, is there a script that can multiply the current and voltage and give me a power reading instead of using a shunt resistor and the voltage ratio method?
Not that I know of right now, but there could be!  I will see about making you a simple version in a little while that adapts the shunt method script.

Are there are plans to release calibration/service manual for DMM6500? I'd be interested to see the calibration points for ACV.
The calibration values are listed in the back of Reference Manual, Appendix C.  There is also a full calibration manual though over at https://www.tek.com/tektronix-and-keithley-digital-multimeter/dmm6500-manual/model-dmm6500-6-1-2-digit-multimeter-0 that describes the process to actually do the adjustments.  It's in general very similar to the DMM7510 procedure.

Oh well in that case: I gathered some info from the design engineers.  We're really quite happy to share details, but it helps us if you're specific in what you want to know and why you want to know it.  I think I addressed all the questions below:
Many thanks to Brad O for the information gathered.

Then if you do not mind, I will continue to ask questions.
Of course!

Quote
What is the input bandwidth for DCV.
From a purely hardware perspective up to but not including the ADC: up to hundreds of kHz to support the DCV digitizer function for 10V and less ranges. 100V and 1000V ranges a lot less than that due to internal 10Meg divider.  Generally similar to digitzer specs.
Does this mean that I must include an external anti-aliasing filter to eliminate the effect of spectrum overlap. And at the same time reduce the input noise level?
For digitize functions, there is a high frequency low pass filter ahead of the ADC but it is not sufficient (with respect to the noise floor) to eliminate all alias effects for a signal with content >500kHz. Whether or not that matters to your measurement depends on what the signal looks like (FFT-wise). 

If you have >500kHz content that is significant, yes, you should add additional filtering to avoid low frequency aliasing.  Generally if the > 500kHz content is small compared to the signal in band content, other specs swamp those errors and so the effect may not be noticed.

Quote
How are the processors, trigger connected inside ...
And any other information on the device. Scheme. Block diagram Sketches.
The processors/trigger/measurement relationship is too complicated to document here and changes depending on the specifics of how you're measuring.  Is there some case you had in mind?  We want to try and answer your question, but we could make hundreds of different block diagrams about the box.
For example, I absolutely do not understand how I can get all the signal samples through the remote interface. Not one buffer. Namely, all data without gaps. I do not understand how the data pass. How does screen information or network communication affect information gathering? How do I communicate on the network so as not to interfere with the measurements. I do not understand how I should set up triggers and buffers so that the data is not lost and at the same time I had time to read them over the network.
It is definitely possible for the instrument to take data faster than it can transmit.  No remote interface has an actual "streaming" option (that is possible with KickStart, though it still isn't magically faster than the bus).  What interface are you trying to use? (LAN/USB/GPIB?)  What measurement function are you using?  What kind of sample rate are you trying to use? 

Although I am very interested in the type of resistor 10 MΩ and the method of its connection to the input.
The 10MOhm resistor is custom and isn't available on the market.  It's designed to support the temperature coefficient and time drift of the DC specs for the 100V and 1000V ranges.  It connects from HI to LO and is switched on with a low leakage analog switch when required.  Input measurements are divided by 100 through it.

Because of the large resistance and value and stray capacitance, it has a very limited signal bandwidth.  The large value also has significant thermal (johnson) noise so it should only be used on the 1V and higher ranges if possible.  The firmware default is configured to be the least confusing, not necessarily the best measurement.  Impedance is forced to 10M by default to prevent apparent errors like "My DMM is measuring 10V with nothing connected", which our support engineers used to get many, many calls about.  We recommend switching to Auto input impedance for the best measurement. 
 
The following users thanked this post: thm_w, kj7e, Octane

Online MegaVolt

  • Frequent Contributor
  • **
  • Posts: 910
  • Country: by
Re: New Keithley DMM6500 and now DAQ6510
« Reply #283 on: December 11, 2018, 08:44:38 am »
For digitize functions, there is a high frequency low pass filter ahead of the ADC but it is not sufficient (with respect to the noise floor) to eliminate all alias effects for a signal with content >500kHz. Whether or not that matters to your measurement depends on what the signal looks like (FFT-wise). 
Can you show the graph of the frequency response of the filter? So that I can make decisions about the need for additional filtering?

It became clear about the digitizing V mode.

Can you tell us about the filter in DCV mode? Does DCV use the same filter with a cut-off frequency of 500 kHz?

After all, for DCV mode should the input band be already?

Quote
It is definitely possible for the instrument to take data faster than it can transmit.  No remote interface has an actual "streaming" option (that is possible with KickStart, though it still isn't magically faster than the bus).  What interface are you trying to use? (LAN/USB/GPIB?)  What measurement function are you using?  What kind of sample rate are you trying to use?

 For example, take the digitization rate of 100,000 samples / s
For 8 bytes per count it is 800,000 bytes. This is very little for any interface.

For example, it is 10% of the LAN. Or 2% of USB.

How should I set up the device so that after filling in one buffer, it will without a break begin to fill the next buffer?

---------------------------------------
And one more question:

I observe some pause between measurements in DCV. Its values are approximately 970 μs.

1NPLC = 0.02 ms + 970 μs = 50 Hz 47 Hz
0.1 NPLC = 0.002 ms + 970 μs = 500 Hz 336 Hz
0.01 NPLC = 0.0002 ms + 970 μs = 5000 Hz 854 Hz

What makes this delay?
Is it possible to somehow disable it?
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 14080
  • Country: de
Re: New Keithley DMM6500 and now DAQ6510
« Reply #284 on: December 11, 2018, 12:38:44 pm »
The extra delay between readings can have different sources. If the DMM uses a more conventional multi-slope ADC, there is some extra time for the run-down an reset phase. However nearly 1 ms would be relatively long.  There may be some extra adjustment measurements (e.g. AC scale, temperature etc. ) going on too. For a more sigma delta like ADC there can be some time for the "soft" start of the aperture window. In principle an SD like ADC could get away without an extra delay in an non AZ mode, if slight overlap / correlation is accepted.

Another possible reason maybe time for data transfer from the ADC to the display / output part of the meter. 

So it is very unlikely on could remove that extra delay.
Are these directly read delays (e.g. from the time stamp in the data file) or are these number only calculated back from the data rate ? Depending on the mode, the data may not come in at a constant rate but could include some extra delays (e.g. for auto-zero) from time to time.
 

Online MegaVolt

  • Frequent Contributor
  • **
  • Posts: 910
  • Country: by
Re: New Keithley DMM6500 and now DAQ6510
« Reply #285 on: December 11, 2018, 02:20:51 pm »
However nearly 1 ms would be relatively long.
I think so too.
Quote
There may be some extra adjustment measurements (e.g. AC scale, temperature etc. ) going on too. For a more sigma delta like ADC there can be some time for the "soft" start of the aperture window.
This is not AZ. This delay is after each measurement. But I think it's worth trying without AZ.
Quote
In principle an SD like ADC could get away without an extra delay in an non AZ mode, if slight overlap / correlation is accepted.
How can I do that?
Quote
Another possible reason maybe time for data transfer from the ADC to the display / output part of the meter. 
Then the maximum operating frequency would be 1 kHz. And the device works without problems at 1 MHz.
Quote
Are these directly read delays (e.g. from the time stamp in the data file) or are these number only calculated back from the data rate ?
time stamp in the data file
 

Online MegaVolt

  • Frequent Contributor
  • **
  • Posts: 910
  • Country: by
Re: New Keithley DMM6500 and now DAQ6510
« Reply #286 on: December 11, 2018, 02:43:54 pm »
Here is the table.
It is curious that this delay is not very stable. It may be in the region of 970 μs or 936 μs.

It looks like I can not build FFT according to such data.

I understand that there are some other ways to get data. Tell me who knows them?
 

Offline TiN

  • Super Contributor
  • ***
  • Posts: 4543
  • Country: ua
    • xDevs.com
Re: New Keithley DMM6500 and now DAQ6510
« Reply #287 on: December 11, 2018, 03:03:50 pm »
Calibration manual indeed suggests that there is temperature compensation procedures in place during normal operation.
Page 3-7 clear about it:

Quote
Disable temperature correction
Before you start your adjustment, you must turn off temperature correction. Run the following commands to turn off temperature correction.
cal.adjust.step.setup("TC_EN")
cal.adjust.step.execute("TC_EN", 0)

Brad O.
Thanks for manual, interesting read! We volt-nuts always like to know what more equipment we need to buy for calibrations  :)
I missed it before, because Tek site ain't too much friendly for old-school folk. Clicked DMM6500 -> Manuals -> Service = nothing, so I assumed it's not published yet.

Quote
Impedance is forced to 10M by default to prevent apparent errors like "My DMM is measuring 10V with nothing connected", which our support engineers used to get many, many calls about. 

Hear, hear, we get thread like that every few weeks "my meter is broken, heeelp" :).
YouTube | Metrology IRC Chat room | Let's share T&M documentation? Upload! No upload limits for firmwares, photos, files.
 

Online MegaVolt

  • Frequent Contributor
  • **
  • Posts: 910
  • Country: by
Re: New Keithley DMM6500 and now DAQ6510
« Reply #288 on: December 12, 2018, 01:32:20 pm »
I calculated the time for one countdown for digitalization mode V.
1 KS / s - 924 Hz
10 KS / s - 5506 Hz
100 KS / s - 11096 Hz
1 MS / s - 13199 Hz

Tell me who knows what a set of settings should be to see 1 MS / s?
 

Offline MrFox

  • Contributor
  • Posts: 35
  • Country: ca
Re: New Keithley DMM6500 and now DAQ6510
« Reply #289 on: December 12, 2018, 07:05:59 pm »
I calculated the time for one countdown for digitalization mode V.
1 KS / s - 924 Hz
10 KS / s - 5506 Hz
100 KS / s - 11096 Hz
1 MS / s - 13199 Hz

Tell me who knows what a set of settings should be to see 1 MS / s?
I haven't seen any problem filling a 7M buffer in exactly 7 seconds.

You can't capture in real-time at that speed through any external link. You have to use a large buffer and dump it afterward.
 

Online MegaVolt

  • Frequent Contributor
  • **
  • Posts: 910
  • Country: by
Re: New Keithley DMM6500 and now DAQ6510
« Reply #290 on: December 12, 2018, 07:39:22 pm »
I haven't seen any problem filling a 7M buffer in exactly 7 seconds.

You can't capture in real-time at that speed through any external link. You have to use a large buffer and dump it afterward.
Please tell me your settings? I made measurements in the internal buffer.
 

Online MegaVolt

  • Frequent Contributor
  • **
  • Posts: 910
  • Country: by
Re: New Keithley DMM6500 and now DAQ6510
« Reply #291 on: December 12, 2018, 08:50:51 pm »
I understood!!!! "Count" should be put as much as possible.

This is a very unexpected behavior.
Does the device measure or display ?? At the same time can not?
 

Offline MrFox

  • Contributor
  • Posts: 35
  • Country: ca
Re: New Keithley DMM6500 and now DAQ6510
« Reply #292 on: December 12, 2018, 09:31:24 pm »
Yes, I always use a single manual trigger with a large count.

I guess it starts making sense when you have a trigger model running instead of auto triggering, because between each capture operation there's a lot of code to run. I find it logical the samples per seconds are stable only within the capture count.

It's a bit like a scope's segmented memory, the rate of triggering is not the same as the ADC rate.

 I really love trigger editor it's amazingly powerful, it takes a while to learn though.
« Last Edit: December 12, 2018, 09:39:40 pm by MrFox »
 

Online MegaVolt

  • Frequent Contributor
  • **
  • Posts: 910
  • Country: by
Re: New Keithley DMM6500 and now DAQ6510
« Reply #293 on: December 12, 2018, 09:39:14 pm »
I guess it starts making sense when you have a trigger model running instead of auto triggering, because between each capture operation there's a lot of code to run. I find it logical the samples per seconds are stable only within the capture count.

It's a bit like a scope's segmented memory, the rate of triggering is not the same as the ADC rate.
Unfortunately, I still do not quite understand where the trigger is connected and how it works.

But I thought that in the 21st century, working at 1 MHz is possible without any delays at the same time as the screen, calculating statistics and other actions.
 

Offline MrFox

  • Contributor
  • Posts: 35
  • Country: ca
Re: New Keithley DMM6500 and now DAQ6510
« Reply #294 on: December 12, 2018, 09:44:03 pm »
It's because it cannot run the trigger script a million times per second. To get a guaranteed timing there's no way around setting the size of your segments.
 

Online MegaVolt

  • Frequent Contributor
  • **
  • Posts: 910
  • Country: by
Re: New Keithley DMM6500 and now DAQ6510
« Reply #295 on: December 12, 2018, 10:50:50 pm »
It's because it cannot run the trigger script a million times per second. To get a guaranteed timing there's no way around setting the size of your segments.
For the digitizing V mode, is it possible to somehow disable the trigger and possibly the screen.

Are there any other ways to get continuous data in the buffer? Particularly interested in DCV mode.

Is there somewhere a tutorial on setting up triggers and writing scripts?
 

Offline MrFox

  • Contributor
  • Posts: 35
  • Country: ca
Re: New Keithley DMM6500 and now DAQ6510
« Reply #296 on: December 13, 2018, 01:56:00 am »
It's because it cannot run the trigger script a million times per second. To get a guaranteed timing there's no way around setting the size of your segments.
For the digitizing V mode, is it possible to somehow disable the trigger and possibly the screen.

Are there any other ways to get continuous data in the buffer? Particularly interested in DCV mode.

Is there somewhere a tutorial on setting up triggers and writing scripts?
You mean continuously overwriting itself? It doesn't look like it, not at 1MS/s.

Not sure how it works exactly, but the buffer needs to be processed in software after each acquisition (it says "processing backlog" if it cannot do it realtime). So in very high speed digitizing modes it cannot be a free running loop regardless of the size. In fact when I set a massive buffer it automatically disable the continuous trigger, which is fine considering this is a pretty big operation to process millions of samples, calculating averages and std deviation, etc... But yeah it would have been nice to sample continuously and process only when stopping.

The trigger interface is like a visual block programming directly on the device, relatively easy when you know what you're trying to accomplish with it, much more powerful than the hard coded things I'm used to. I haven't read the manual yet I'm still playing around.
« Last Edit: December 13, 2018, 02:31:31 am by MrFox »
 

Online MegaVolt

  • Frequent Contributor
  • **
  • Posts: 910
  • Country: by
Re: New Keithley DMM6500 and now DAQ6510
« Reply #297 on: December 13, 2018, 08:30:13 am »
You mean continuously overwriting itself? It doesn't look like it, not at 1MS/s.
Not necessarily yourself. One solution I see is the alternate filling of two buffers. But I am afraid that it is impossible to configure the device so that it does not lose data when switching from one buffer to another.
Quote
Not sure how it works exactly, but the buffer needs to be processed in software after each acquisition (it says "processing backlog" if it cannot do it realtime). So in very high speed digitizing modes it cannot be a free running loop regardless of the size. In fact when I set a massive buffer it automatically disable the continuous trigger, which is fine considering this is a pretty big operation to process millions of samples, calculating averages and std deviation, etc... But yeah it would have been nice to sample continuously and process only when stopping.
It amazes me that the work of calculating the mean and standard deviation is not assigned to the FPGA. As well as other mathematical operations.

It is very simple and very fast. I have a board on my desk where I calculate these numbers in real time for the signal coming from the ADC 1 GHz.
Quote
The trigger interface is like a visual block programming directly on the device, relatively easy when you know what you're trying to accomplish with it, much more powerful than the hard coded things I'm used to. I haven't read the manual yet I'm still playing around.
I'm trying to understand what he can. And until I found a textbook with pictures for dummies :)
 

Offline MrFox

  • Contributor
  • Posts: 35
  • Country: ca
Re: New Keithley DMM6500 and now DAQ6510
« Reply #298 on: December 13, 2018, 04:35:50 pm »
Ah ok, Brad already explained it in a previous post...
The fastest is actually to use a trigger model, because there's a separate processor for it. I am curious why the default CONT mode isn't just that trigger model.

MikeP:
Probably you set the digitize Count to 1 and went into continuous trigger mode?  When you do that, triggering is handled by the display processor, so when that processor gets busy it will stop triggering and catch up with whatever else it's being told to do (like update the graph image).  That's where those gaps in data are coming from.  A couple ways around this come to mind:
  • Set a higher count to capture all the data you need.  Triggering within a Count set is handled by a separate processor that won't get caught up in display stuff, you also won't see the lines connecting separate groups of data that may be a little confusing.  The graph's smart scaling in x will by default show you the latest group of readings.
  • Use a trigger model.  Triggering from a trigger model is also handled by a separate processor and a very simple model can allow continuous data capture.  I'm attaching a script (change the .txt to .tsp to use it) where the last lines set up a trigger model that starts a digitize voltage measurement (that continues infinitely) and then stops the measurement when the TRIGGER key is pressed.  The display might lag behind slightly depending on your other settings, but there won't be any gaps in the data.  If you digitize at a really high rate then you may see a pop-up like "Processing reading backlog...".  That message means the display processor needs to catch up with the data buffer and it will stop other activities until it catches up, usually no more than a second or two.
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 14080
  • Country: de
Re: New Keithley DMM6500 and now DAQ6510
« Reply #299 on: December 13, 2018, 05:40:41 pm »
There can be more calculations done than just mean and std. deviation.  Other tasks may be corrections of temperature effects. Scaling from raw data to voltage (or amps) and also filtering zero readings / offsets.

The processor at the display part is way slower than an normal PC. I would expect is to be comparable to maybe an Rasberry at most.  This processor also likely has to handle other stuff, like the display. So I kind of understand that this might not be enough to do this in real time.  Especially some filtering might be difficult in real time as it involved forward looking data.

The trigger models are a little confusing, but this the price for an instrument with so many options.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf