Increasing the scan rate for Keithley 2700 f AC Voltage measurement

[oap3952]

I am trying to use the sample example for Keithley 2700 series for AC voltage measurement. The data is stored in the form of Excel File. Currently, the code scans only one reading per second. How Do I increase the scan rate? I have attached the picture of code as well as the file. It shows some "header error".

[RTSLVU]

Doh... nevermind... 😛

I just looked at your code but I don't have the Keithley drivers installed so I do not see any way to increase the measurement rate unless there is something inside the Keithley vi's that could be adjusted.

If that unit is anything like an HP/Agilent/Keysite 3497xA data logger there might be a few things you can do.

1. Set the range on all channels (autoranging adds time to every measurement)
2. Group similar measurement modes together by channel (switching measure mode between channels takes extra time)
3. Check for any measurement delay or interchannel delay settings
4. Decrease resolution. (higher resolution takes more time)

Also the 3497xA has two ways of taking measurements.

1. Direct, you basically call out the channel and the measurement. This mode is slow as the instrument has to be programmed to take a measurement every time for every measurement and each request only returns the one measurement you requested
2. Program every channel at once and a scan list. Then initiate a scan every time you want to take a measurement. This is the fast way as the instrument is pre-programmed and takes all the measurements you setup in one scan and returns an array of all the measurements.

[oap3952]

I have attached the code file.

[RTSLVU]

Also how fast do you want? 

Remember these instruments are not high speed DAQ devices. 

But their website says is is capable of scan rates of up to 500 channels/second so it sounds like you are not configuring it correctly 

My experience with the 3479xA is that you have to do it the "hard way" all of the vi's to make it scan and take fast measurements are under "Advanced Configuration", all of the "EZ Config" vi's were slow.

[oap3952]

Scanning rate of 100 readings per second is good enough for me. Any thing above 20 readings per second works for me well.

[cstorey]

There's an example packaged with the driver called "Keithley 27XX Performance.vi" which shows you how to get faster measurements.  (Very similar advice to RSTLUV's)

- reduce the resolution to 3.5 digits

- disable Auto-ranging

- set the number of power line cycles over which the instrument averages a reading to minimum

- turn off filtering, auto-zeroing, display, etc..

- disable or reduce any delays after switching

You'll need to find the trade-offs between speed and acceptable data in terms of noise and drift.  I leave auto-zeroing on, otherwise you end up with drift due to instrument temperature fluctuations over time.  Disabling auto-ranging is usually a big time saver if you know roughly what value you are going to measure and that value doesn't vary by orders of magnitude over time.  Turning the display off is another big time saver.

You'll need to play with all the settings to find the balance.

Now is your goal to measure all channels nearly simultaneously?  If so look at buffering all readings then reading it all back in one chunk, which will add a delay between repeat measurements, but all channels will be scanned as quick as possible. 

One issue with really speeding things up is it looks like you are running this over the serial bus at 9600 baud.  Every iteration of your while loop is sending/receiving at least 2 command strings and then reading a large chunk of data.  That's not a speedy process.  I'm not sure how much of an increase you could get from switching to GPIB or Ethernet, but they both have better throughput than RS232.

Craig

[RTSLVU]

---

@cstorey wrote:

One issue with really speeding things up is it looks like you are running this over the serial bus at 9600 baud.  Every iteration of your while loop is sending/receiving at least 2 command strings and then reading a large chunk of data.  That's not a speedy process.  I'm not sure how much of an increase you could get from switching to GPIB or Ethernet, but they both have better throughput than RS232.

---

Oh my yes 9600 baud is way too slow. I have several Yokogawa power analyzers in the lab that only have 9600 baud serial communications and if I am taking more than one single measurement an update rate of 2-3 seconds is as fast as I can pull the data over the serial port at 9600 baud. 

Besides all the other advice you need to increase the serial baud rate or use a faster interface like GPIB, USB or LXI.

[oap3952]

I am just reading one channel 101 for AC Voltage. And the baud rate I select is 19.2K.

[oap3952]

I tried using "Keithley 27xx Performance.vi" but it shows me an error "221: Settings Conflict" I didn't make any changes in standard program but it shows me this error.

[cstorey]

 You might want to elaborate on your cards, measurement type, etc.. 

The provided examples might not work with your cards/setup so you will have to do some trial and error work.  Use highlight execution (light bulb in the block diagram editor) to see what VI in the Performance example give you errors.  Read about the settings for your cards and measurement type, then change the inputs to that VI to match your card/setup.   

Craig