Tractor with QA403

Was considering exploring the functionality of Tractor, but all references are for the QA401 and there does not appear to have been any development or discussion of this recently. Does Tractor support the QA403 and is it recommended for new applications?

Hi @Robbn, not yet, but the work was started. Let me check on it. I think whatever cleanup was left could be pretty quick. Thanks!

Hi @RobbN, I checked on this and I think a release can be out tomorrow for first try. Thanks, Matt

Hi @RobbN, the release with support for QA40x is at the link below.

When setting up your test, you need to make sure configuration (See Settings menu, then Setup Group) has QA40x. This will be changing. Right now, you need to specify a group of test equipment you want to use. But shortly there will be the ability to specify an audio analyzer, a programmable load, a voltmeter, a current meter, etc). And it won’t require pre-configured groups anymore.

But please try with QA40x and let me know what you think. I did see an issue this morning related to setting Y axis scale for measurements. This really only impacts the bitmaps that are captured by Tractor and stored in the testlogs directory.

Thank you! I’ll check it out ASAP. Unfortunately, time is precious and I’m still just experimenting with your systems with a limited knowledge of making the most of the products. I still rely on other systems for our test and measurement needs. It will likely take a while for me to have any useful feedback.

We are checking Tractor with QA403 at the moment. Both great tools.
We found 3 issues in Tractor:

  • the selectable values for the “Analyzer Input Range” is 6dB or 26dB as in QA401, so the only usable setting with QA403 is 6dB at the moment
  • if you do a frequency response measurement (Test “FreqResponseA01”) with a mask applied the return value to Tractor is always “Pass” even if the QA40x Analyzer Software shows “Fail”
  • as matt already quoted the settings for Y-axis are not as defined

Hi @ChrisWN, I have confirmed the first, and we’ll get a fix out by tomorrow. We’ll have more info on the other items then too.

Hi @ChrisWN, and new Tractor release is up (version 1.101) that will allow more input levels based on the hardware you have selected. The mask issue you highlighted is a bug in the QA40x software, and that will be fixed in a release coming early next week. Link to Tractor releases on Github is below.

Thanks very much for reporting!

Hi matt,
thanks for the new releases. Unfortunatly I cannot test the bugfixes as Tractor (1.101) shows the message “The test could not be started: Unable to connect to the QA40x. Ist the QA40x application running? Is it version 1,154 or later?” if I try to start a test in Tractor with the new QA40x Software 1.160B installed. As soon as I go back to 1.154 the test starts again.
Can you check that please?

Thanks chrisWN

Hi @ChrisWN, sorry about that. The version name of the last QA40x release was 1.160B. The ‘B’ was added because of a build tool problem where the new visualizers weren’t showing up. So, that was fixed and the release was renamed. But rather than going to a new number, the B was added. But when Tractor tried to connect, it got that release number (1.160B) and tried to convert to a double to verify the version was suitable. And that caused an exception.

It’s been fixed in the 1.161 release. Once you are able to run the tools together, please verify the freq response mask issue is working as you expect. Remember that by default the left and right will be tested, and if you are only using left channel, the test will fail. So, make sure to turn off the right channel in the Tractor test if not needed. That caused me some confusion at first.

I’ve finally gotten back to trying Tractor. Does it only work with 48K sampling?

I worked through your tutorial and made a noise test as described, then added in other tests, but I’d like to test the broadband response of an amplifier to 80KHz with 192K sampling and the chirp test always fails. When I go digging into the program, the RMS limits are always set back to 20KHz. I thought it may be because of the noise test that preceded it, as it does not seem to allow me to set the upper frequency to 80K. Deleting that test from the program, and running the tractor test set again still results in a fail because measurement frequency is set back to 20KHz. To reset the analyzer manually again, I seem to have to fiddle with the buttons, switching back and forth with the sampling and generators before I can change the measurement frequency. It seems like kind of a mess. Or am I missing something?

Hi @RobbN,

Yes, it only works with 48K sampling rate.

Every test starts off by setting the analyzer back to the defaults. Without this, the state of the analyzer couldn’t be known from run to run. So, changes you make the analyzer will be undone on every test run. Tractor is mostly to enable factory tests, and there’s a balance between options available and simplicity. Otherwise, you’d have hundreds of options to configure for each run (eg weighting, window types, etc). The sample rate could be configured, but then you’d need to require the user to specify the measurement start/stop frequencies.

So, for most 95% of production tests I think the recommendation would be to keep it simple, and find ways to test around particular issues.

For the remaining 4% of production tests, the source is available to Tractor and you can write your own tests that whatever you need. That test can then plug into Tractor. For example, you could write a test called FreqResponseA01-192k and then in that particular test set the sample rate to whatever you like, and set the other parameters for testing start/stop to whatever you’d like. Ditto with window types and other parameters.

And for the last 1% of testing with really esoteric needs (engineering validation type of stuff), you probably would just want to write a test in Python or DotNet. There are some great environments, like Jupyter, that allow an interactive approach, and they have buit-in graphing capability.

Below is a plot of a simple test in a Jupyter notebook. I’m not much of a Python guy, but you can see what is going on there. The test swept the DUT input from -30 to 0 dBV and measured the THD. It took about 3.3 seconds to run, which is very reasonable for an interactive environment.

Is modifying Tractor an option?

Hi @RobbN,

Here’s a post showing how you could modify Tractor to allow you to make a 96K sample rate RMS measurement by making a custom test. But hopefully it’s also a starting point to see the potential for controlling other things with custom tests. Let me know if you have any questions on this.

Matt, I just want to thank you for all of your attention to this. I know that I’m capable of these things, as I’ve done similar code modification before without really knowing languages, but I’ll spend a big chunk of my day doing this, so the thought of diving in is daunting when compared to other tasks that will be deferred. My impression of the QA403 thus far is great. However, every time I use it, I come to the same conclusion that the software and software documentation could be vastly improved for those of us that don’t code. I want to use it, but I need to make a deep dive into learning about controlling it or perhaps contract with somebody who can help.

If anyone is reading this that would be interested in writing some tests for us, please feel free to make contact.