Page 2 of 2

Re: Integrated Lab Measuring and Controlling (Labview, Labjack, ...)

Posted: Sat Jan 19, 2019 11:56 pm
by Richard Hull
Chris has it right. USB is still forming and might never stop evolving the RS serials and GPIB are so dead, they are set in stone. Slow yet reliable and easily worked with to this day. You pay for the late and great super fasts which may or may not survive a new operating system or connectivity rage. Technology marches on as we need to grab data at lightning speeds.

Richard Hull

Re: Integrated Lab Measuring and Controlling (Labview, Labjack, ...)

Posted: Sun Jan 20, 2019 8:10 am
by Harald_Consul
Thanks Chris.

Ok. I have understood now, that USB is in some directions pretty suboptimal for future-proof PC based high frequency measuring.

On the other hand serial port (aka RS-232) and parallel port (aka GPIB or IEC-625-Bus) are too slow to measure high frequency (> 3 MHZ) directly. And the measuring instruments of that decade do not have a digital storage to cache the high-frequency data mass.

I even haven't heard of digital scopes with an ethernet port a lot.

Thus, there is no future-proof intgrated high-frequency measuring solution yet, right?

Re: Integrated Lab Measuring and Controlling (Labview, Labjack, ...)

Posted: Sun Jan 20, 2019 9:54 am
by Chris Mullins
Well, "future-proof" and "PC-based" are two terms that don't really go together in general, unless you're willing to lock down the PC once things are working (no OS upgrades, no changing PC hardware, not networked, etc.) Speaking generally for long-lived systems, it's best to avoid consumer hardware/software - the new technology treadmill cycle is much too fast in that world. Or, freeze time by never changing anything in the system once it works.

Scopes with Ethernet ports have been around for a while, and are dramatically cheaper than in the past. The Siglent SDS1202X-E in my system has a 200MHz bandwidth and a 14Msample buffer length (per channel, with 2 channels), along with an Ethernet port. It has a waveform storage mode with timestamps, and can capture 400K waveforms/second. Likely there is a lot of fine print, and implementation bugs/shortcuts limiting actual performance compared to a real LeCroy/Tektronix/Agilent scope, but that's not bad for $379. Rigol has similar scopes too.

I'm sure there are Ethernet-based high speed data acquisition systems out there. Buying something like that from an established company is about as "future-proof" as you could get. Here's a USB-based system that gives you up to 12 MHz sampling on 2 channels: https://www.mccdaq.com/usb-data-acquisi ... -2020.aspx, assuming your PC can keep up with the storage. Naturally it requires their own USB drivers and software. Whether you can get that to work with a new Windows PC from 20 years in the future - who knows?

Re: Integrated Lab Measuring and Controlling (Labview, Labjack, ...)

Posted: Sun Jan 20, 2019 12:29 pm
by Rich Feldman
It gets easier when you have deep pockets. At work I sometimes use a late-model 30+ GHz oscilloscope without setting foot in the lab. It cost more than I make in a year. It's got an Ethernet cable connection.

Remotely log in from this computer to the oscilloscope's Windows PC, after sometimes asking a previous user to log off.
Then the Infiniium program (why should we start calling it an app?) takes over the screen, or a sub window, with familiar digital oscilloscope display. When there's need to enter the lab of noise and cold air, I can pick up my session at the oscilloscope's console.

I can save waveforms directly to files on 'scope's drive, or on a server, or on this PC. Generally as text files -- we are no longer locked into binary wfm formats. They can get a bit unwieldy beyond a few million lines. If the network connection is down, files can be saved or copied to a USB flash drive.

It's because of embedded industrial computers that Microsoft supports some operating systems long after the "consumer" and "pro" versions have gone into the dustbin.

Reminds me of a national security concern. We know utility grids and pipelines are remotely controlled by computer. Are they allowed to use connections through "the world wide Internet" for that? Sounds like profit-driven foolishness.

Re: Integrated Lab Measuring and Controlling (Labview, Labjack, ...)

Posted: Sun Jan 20, 2019 3:21 pm
by John Myers
industrial Ethernet has been gaining popularity. It offers fast real-time/deterministic communication. POWERLINK and EtherCat are two protocol examples.

Re: Integrated Lab Measuring and Controlling (Labview, Labjack, ...)

Posted: Mon Jan 21, 2019 7:10 am
by Harald_Consul
Very interesting.

I just did a quick Ebay research. New 200 MHZ oscilloscopes with Ethernet-Port roughly start at 1500 USD. That's more or less for the deep pocket.

On the other hand 2-Channel 200MHZ oscilloscopes with SD-card slot start at 330 USD. (Do not say, that the low price segment is good quality as well.)

Is it possible to use a sophisticated analog signal generator signal (by BNC cable) to precisely time-synchronize the measuring of two or more digital scopes (or other HF-measuring devices)?

I mean, the new signal generators (often attached within a scope) allow to program flexible wave patterns. Is it possible to program a sophisticated analog HF wave pattern, that is unique enough per time frame, that it would allow use it as index and to SQL-join the measuring of two (or more) different scopes?

Re: Integrated Lab Measuring and Controlling (Labview, Labjack, ...)

Posted: Mon Jan 21, 2019 1:42 pm
by Rich Feldman
Harald, it sounds like your time sync question is about matching the starting times.

Do you, or other readers, also care about finely matching the time scale factors? What sort of time accuracy do the low cost instruments come with?

Each instrument in a setup has a different reference for what appears as 1 microsecond or 1 MHz. Timebase accuracy tolerance ranges from dozens of ppm (e.g. from a computer clock or embedded XO) to very small numbers. It's common, in time and frequency instruments claiming 1 ppm or better, to support frequency locking to other instruments via BNC connector on back panel. Generally 10.0 MHz with some master/slave connection topology.

Here's an example where a timebase speed difference is right in your face. Single-channel scope acquisition of a serial data waveform to study jitter. In a chart of edge position errors vs. bit number we see them gradually ramping down, so bits around 100,000 are 1/4 bit early with respect to bit 1. No big deal, it just means the data source is running 2.5 parts per million faster than the oscilloscope.

Tangential story-telling time: -------------------------------------------

Among the standards discussed in this thread, old RS-232 can tolerate mismatches measured in whole percents, because it re-synchronizes at the beginning of each character. Who else remembers electromechanical teletype machines and 110 baud? Or even knows what a UART is?

Many modern multi-gigabit standards tolerate 100 ppm differences, to minimize cost of the local frequency reference. Data scrambling or encoding ensures that receiver always gets plenty of transitions, from which Rx clock can be recovered.

The Cassini-Huygens deep space mission to Saturn and Titan barely escaped a major comm link failure in 2005.
https://www.evaluationengineering.com/h ... ns-mission
After the spacecraft (C and H) separated, H probed the atmosphere of Titan while C entered orbit around Saturn. Data from H had to be received and recorded by C, for later transmission to Earth.
Long story short: the radio receiver designers accounted for predicted Doppler shifting of the carrier frequency, but overlooked the fact that serial baud rate would be shifted by the same Doppler ratio. :-)

Re: Integrated Lab Measuring and Controlling (Labview, Labjack, ...)

Posted: Tue Jan 22, 2019 2:46 pm
by Harald_Consul
Richard, if these guys eventually provided a 1 Ghz, 16 bit, 5 GSample/s and 1 TSample buffer oscilloscope for 500 USD, it would truly be about time synchronisation, only.

But for now, as the effective measured time period of recent oscilloscopes is unbelievable small (due to small buffer size), it is about about (start) time synchronisation AND remotely triggering the start of the record from my point of view (as you also mentioned).

Re: Integrated Lab Measuring and Controlling (Labview, Labjack, ...)

Posted: Fri Jan 25, 2019 10:22 am
by Harald_Consul
Ok, I still owe you a summarizing synopsis, which comes now. I guess, it will contain some errors (as I am terribly stupid ;-)).

Integrated lab measuring consists out of
  1. Measuring
    • Generally future-proof realtime measuring requires:
      • A standardized communication interface to the peripheral meter (e.g. USB, Ethernet, Wlan, Bluetooth, Serial Port, Parallel Port).
      • A standardized protocol to access the data (over the com interface). This can either be a file transfer protocol or a streaming protocol.
      • (Maybe in the old serial and parallel ports the communication protocol and data protocol are integrated in one? I am not familiar with these ones.)
    • Afaik only Raspberry and Arduino/Atmel measurement data logging can provide standardized communication interface and standardized data protocol.
    • Arduino/Atmel maximum measurement frequency typically is up to 10 kHz, extension kits up to 1MHZ are available.
    • Future-proof integrated high frequency measuring is very problematic.
    • Running a virtualized old Windows to make use of an old USB-driver is a random game.
    • High-end oscilloscopes indeed offer Ethernet/Lan interface. But the data access is still proprietary, which means the driver will be antiquated in 10 years, as well.
    • Offline-measuring by a low cost oscilloscope with an USB-stick or SD-card interface may be an attractive workaround to achieve a future-proof measuring equipment. However, a turn key time synchronisation for this approach is not known.
  2. Control
    • Remote switching on and off is a peace of cake. Most measuring I/O adapters also offer output, which using a relais panel can be adopted to any voltage and current.
    • For remote controlling a variable voltage a variable voltage supply can used best to muliply the voltage of the I/O adapter. E.g. from Spellman.