Neville D

Do you want to run RT code or develop RT code? Neville.

QUOTE (mdoan @ Aug 14 2008, 07:16 AM) Towards the end, shortly before they pulled the plug on it, they were bundling HiQ as part of the LV Pro Dev (or was it Full Dev?) package. You could try NI Developer zone or NI Labs had a version of the 3d plotting capability for earlier versions of LV (8.2 or so, I believe). Neville.

QUOTE (bbean @ Aug 7 2008, 05:30 AM) I forgot to mention I was ftp'ing to a PXI 8187 controller and LV-RT frm a Windows machine, and 100MB/s connection. I have a Gig-E network, but the 8187 only accepts 100MB/s. And yes I have Filezilla that uses default passive, so I would guess command line uses passive too. Neville.

QUOTE (Neville D @ Aug 6 2008, 01:08 PM) Just tried it. http://lavag.org/old_files/post-2680-1218062985.vi'>Download File:post-2680-1218062985.vi Making the Active boolean to false =79ms (using Internet Toolkit); your command line VI=171ms. Neville.

QUOTE (Draqo85 @ Aug 6 2008, 01:28 PM) Sometimes a search of the LAVA archives pulls up lots of http://forums.lavag.org/TCP-Listener-Wait-On-Close-Open-t3735.html' target="_blank">useful stuff. Download Jack's example from his website. It pretty much does what you want. Neville.

QUOTE (Val Brown @ Aug 6 2008, 01:38 PM) Was it a DVD-sized download? It seems the dvd includes everything and the kitchen sink. I usually wait for the package in the mail since LV-RT is never available for download. Neville.

QUOTE (Dan Bookwalter @ Aug 6 2008, 12:34 PM) Have you tried tieing all other unused channels to ground? Using shielded cables ? Do you get the noise on all the channels? Try plugging the amp into a different power line? There are devices that can isolate the power line for a couple of plugs (around $300 or so). Maybe "measure" the noise on a AI channel and then subtract out the channel in software. Noise issues are always difficult to fix and usually isolation is the only thing that always works. I think there are some NI DAQ cards that have channel isolation built into them (for a future upgrade).. Neville.

QUOTE (bbean @ Aug 6 2008, 10:06 AM) The FTP toolkit is really showing its age. Lack of documentation, disorganised pallets, lack of good examples etc. What you can do is make sure the Active? boolean is set to false (I think it tries to open a fresh TCP connection if its false). The dir listing VI also seems to be writing to property nodes and doing all sorts of parsing (though that shouldn't slow it down too much unless you have very complicated directory structures). You might try to use the FTP Command.vi (vi.lib\addons\internet\ftp\ftp2.llb\FTP Command.vi) to experiment sending the bare minimum commands to speed it up. I like your command line trick and it is indeed fast. Neville.

What usually works best is to use a card with isolated input channels. In the past I have used NI SCC signal conditioning accessories to isolate channels entering the DAQ card. Works great, but can get expensive fairly quickly. In the end, if you have more than 5 or so channels requiring isolation, or measuring extremely low voltages, it is better to use SCXI systems. Hopefully you are using a differential AI channel referenced to ground? PS I have never used your card before. Neville.

QUOTE (Pollux @ Aug 6 2008, 11:57 AM) Using it for Arrays will affect performance, since every local variable of an array will make a separate copy of the array. Also, reading and writing locals switches to the user interface thread. Scalar variables might still cause race conditions when a write might accidentally trample a read operation. Overall: use wires where you can. Its easier to read, will give best performance, and not cause race conditions. Locals are OK, if for example you are reading vals from an ini file and initializing a bunch of controls on the front panel once. If doing anything in a loop, avoid them. Also don't try to get around that by reading or writing to the "value" property of a control or indicator. That is even slower than using locals. Again a "one off" use is fine. Neville. Neville.

QUOTE (Pollux @ Aug 6 2008, 05:27 AM) Reading and writing to arrays as local variable copies will affect performance, readability and introduce race condition bugs. Use shift registers or stretch the wires out.. Neville.

Hi everyone, I just found the NI Vision 8.6 Development Module Readme The new optimizations for multi-core and improvements to overlays seem particularly timely. Also enhanced font support for real-time targets, and remote monitor support for dual-core PXI targets. Good stuff! Can't wait to get my hands on it. Neville.

QUOTE (george seifert @ Jul 31 2008, 10:13 AM) Individual lines are always more flexible. You have more control, and don't have to remember the states of the other lines. Two caveats: 1 You have some logic that requires you to remember prev line states.. e.g: Line1 can only come ON if Line2 and Line3 are OFF..or something like that. 2 You have to carry around 16 separate task ID's. Neville.

Might depend on the hardware you are using.. Low cost DAQ's may not allow the lines to be treated individually.. Why not try an example with two lines and tasks and if it works, go from there? Neville.

QUOTE (Gary Rubin @ Jul 24 2008, 09:35 AM) Fair enough.. you could build an exe, you know.. Neville.

QUOTE (JRTritsch @ Jul 7 2008, 07:26 AM) Check the sample rate, and whether you are waiting to acquire all the samples before displaying.. (I don't have LV at the moment to look at your example). Neville.

Won't be able to make it this year.. But have fun at the LAVA bbq! :beer: :beer: :beer: :beer: Neville.

I thought John Lokanis had written a utility that did just that.. check out the LAVA CR. Neville.

Your attachments don't show. Dialog boxes won't pop up when viewing your front panel with a browser. Change your Front panel to show messages in a text display with buttons that are appropriately displayed/hidden. Neville.

Download a utility from the NI website that helps you run some diagnostics on your card. In particular check the ground traces on the breakout board connected at the end of the cable from the DAQ card. If there is any damage, the breakout board will be the first to be affected. Reboot your PC, or power down, and put the DAQ card in another slot. Neville.

QUOTE (Viswanath @ Jul 23 2008, 08:12 AM) You probably can't do that with hardware timing. These cheap DAQ cards have only a single DMA channel (I'm not exactly sure how the USB card works..) and FIFO but the long and short of it is: you can do only 1 hardware timed operation at a time. Instead generate BOTH AO's with the higher frequency, and then downsample the output of the lower frequency one in software. There are VI's in the pallets to downsample either the entire waveform or pt-by-pt in a loop. Another approach is if the frequencies are low enough, output both waveforms with software timing in two separate loops. You'll have to experiment to see if this works for you. Neville.

QUOTE (Xrockyman @ Jul 23 2008, 08:29 AM) I think your probably not looking for the execution priority setting. Changing that setting, for a VI in series with user events, will do nothing till the user has actually finished choosing stuff. If you want your User Interface to be more "dynamic", handle user events with an Event structure, and have a second parallel loop that acts on the event instructions generated by the Event structure. Look for examples or a template for a Queued State Machine architecture under LV help. Neville.

QUOTE (kelvin @ Jul 21 2008, 04:47 PM) I'm no Gig-E expert, but have you gone in under device manager in Windows and checked which driver the camera is using? Maybe you need to use the NI Gig-E driver? Are jumbo packets enabled? That is quite critical the one time I played with a Gig-E camera. Contact IPX and see if they can offer any help with using their device with LV.. My local Prosilica tech support walked me through using their camera with LV, and I finally got the demo camera to work. Neville.

QUOTE (urh8 @ Jul 22 2008, 03:33 AM) What do you mean? Are you connecting the AO channel to an AI channel and observing the AI channel on a plot? If you are just looking at the AO samples being sent to the card, they will of course always be perfect. Its been a while since I did any DAQ, but change the RSE (referenced single-ended) to Differential, and see if that makes a difference. I seem to remember you can't measure negative voltages in single-ended mode (I may be wrong here..) Check grounding connections for differential carefully (see manual). Read the pdf manual that comes with your card (don't know if they ship that any more for the newer cards). It talks about how connections are to be made to the card and what the differences are between RSE, NRSE, and Differential. Grounding is extremely important, or else you will get all sorts of problems. Ground AI chans not being used (as a start) to see that those are not causing any issues. What sort of breakout board are you using? If in doubt, check that the ground pins on the breakout are not fried. (Turn the board over and eyeball the traces or check continuity between the screw terminal and the connector pin on the other side). Ground the AI chan connected and see that it reads 0V in LV and with the scope. Now connect a battery and see that it reads a reasonable constant noise-free static value. Check with scope. Make sure your scope is on a 'reasonable' V/div setting Now output a constant voltage with your AO chan, connect to the AI chan and repeat above to check. Then generate a low freq waveform and confirm above. Finally generate your required high freq waveform and check. Report results of the above for further help. Finally some of the E-series cards have a self-test utility somewhere on the NI site. Download and run (if applicable to your card) to check if there is a card malfunction (I doubt it, in all my years of DAQ, I have seen only a single "bad" card). You can use NI-MAX for debugging or your own code, but MAX is a good place to start. Another thought occured to me: if you are using a "low-cost" card, they usually have a single DMA channel, in which case it can be either used for high speed AI or AO, but not both. Try slowing down your sample rate to see if the problem disappears with that. What card are you using? Neville.

QUOTE (Draqo85 @ Jul 18 2008, 12:49 PM) Check out the examples TCP client-server.. there are some examples that do what you need (multiple clients). Also, you can avoid having to deal with communication if you use LabVIEW Shared variables. You can define a buffer for each variable so as to not loose data. Again, see some examples of communication using shared variables. For simple cases, it will save you a lot of time. For high performance, you might have to use TCP or UDP. Neville.