will it be possible to use the different versions side by side?

Sven:

her is non-intuitive at least for me. Settings under submenues wtih no show on the main pane what is set with no instant setting change possibility in the main pane. Start and toggle buttons same image without sign what it does. Yes, I know with mouse over button shows it. Is there still the IDLE mode like at the previous version for continues generator output? Would be great that you review the design of the user interface for easy usability and clear show what is set for measurement. Sorry to say but the Q

Hi @Sven, the purpose the 0.1 is to test subsystems. The UI is very far from what will be final, it’s just pieces right now. And each piece needs to be tested on Win/Mac/Linux to confirm they work.

In the final app, there will be a play/pause button for PC and a play/pause button for the analyzer (they are there today, just in different places. You can see the “running man” for automated tests at the top of the app. The PC and analyzer run/stop buttons will end up there most likely). The PC and analyzer will each follow the generator settings. So, whether chirp or sine, the you have one place to set the values and the PC and analyzer will both follow (you will need to specify a mapping from abs to rel). This means if you want to send an expo chirp out of your DAC, you set the system for chirp, hit the PC play button, and you’ll hear it from your PC. And then capture it on the analyzer and you’ll see the freq response of the the PC DAC.

Don’t fret, there’s a lot of change coming. And yes, Qa40xPlot is very good! There’s a chance that Qa40x_v2 will be too complicated for some, and that is OK too. There’s QA40x_v1 and also Qa40xPlot for fallback.

The translations will be switched to DeepL in the near future, and they translate press pool as “pool de presse” and “trace pool” as “pool de traces”

But we can add exceptions, and so if something is confusing it can be corrected by native speakers and that will be important for top-level app concepts.

1. Yes, everything has been built to enable it, but it will probably be one of the last things worked on before the 1.0 release.

2. Yes, in the input range select you’ll see two additional buttons: One is to optimize for THDN (noise) the other is to optimize for THD (distortion). This will cause additional measurement to be made to optimize for your preference, giving the autoranging.
   

   

.
These additional measurements for autoranging occur up front very early in the pipeline, so you might hear some clicking, and you’ll see [AR] reported in the lower left acq status. But the benefit of it happening early in the pipeline is that downstream measurements aren’t aware it’s happening, and so for the automated tests you no longer need to specify Input Level Range Adder. You just decide if you want to optimize for noise or distortion and everything just works. Once autoranged, it won’t run again until input level changes by +/- 1 dB.

Also in the automated tests you can override global settings. For example, in the Power THD versus Freq test seen below, we have overridden the buffer size and autoranging. But we have accepted the global default for Window. This should make it a lot easier to have a bunch of canned tests. And down the road, imagine the ability to chain together a bunch of automated tests and a final pass/fail. That will effectively pull Tractor into the main app.

4. Yes, below you’ll see the dac routing block. This isn’t done yet, but you can click on each of the switches you see and watch the change as you click. That changes the routing for the system.

If I click on the switch before the PC DAC in the lower right you can see I’ve selected the ADC monitor. The gain block will let you pick 0/-20/-40 dB. And so, whatever is going into the audio analyzer can be monitored on your PCs speakers (with a few hundred ms of delay)

Also, it’s a bit confusing now since there’s some QA404 stuff mixed in. On the QA404 (when it’s ready isn’t known yet, not in the next 6 months for sure) you can select either XLR input (BAL) or BNC (single ended). You also have the ability to ground the outputs (the drawing is a bit confusing right now as it looks like you are grounding the DAC). But grounding the outputs means you can leave your DUT connected to the analyzer output and make grounded input measurements on your DUT. This is really needed for some of the new Class D amps like Purifi

This first release has a cross platform audio subsystem called PortAudio (whcih I’m hoping more can test above). This is really the enabler for the tighter integration with the QA40x hardware. So, longer term this means flexible switching so you can source analysis from a sound card too. That’s not part of the 1.0 release. But with the routing flexibility and non-bursted mode, should be straightforward to see how it can come together.

Lots more work to do…but again, this 0.1 release was to ensure some of the bigger parts (cross platform audio) are working on lots of different systems.

As I am in the middle of a project and can’t get into BETA land I had a few questions on the exciting release.

1: Do we have the ability to have a steady state oscillator during FFT analysis?

2: Does the system now have auto range input?

3: Can we monitor the input on the PC or MAC?

Cheers and thanks Matt a bunch for your tireless improvements in this platform. Looks very cool..

I would translate it to French as “Liste des traces” instead.

“piscine” is really the French for “swimming pool” which has little to do with a QA403.

I’m pretty code illiterate, unless everything is explicitly specified, step-by-step. I have been using my QA403 on a Windows laptop I purchased on the used market just for this product and am getting familiar with it. But I REALLY appreciate work on a Mac native app! I had not used Windows in many years and did not miss it! Thanks.

Please see this thread: Qa40x_V2 Version 0.1 - #6 by matt

This is a first release of the app that doesn’t require mono to run cross platform. The Qa402 support will be added in the next few days. I’d be grateful if you could try and report your platform and what you are seeing. Thanks!

Hi @ScottRichardson, yes, agree. It mostly for Linux folks at this point since they have such a broad array of platform options/choices.

matt:

• Confirm localization appears
• Confirm file system persistence
• Confirm local tone generation
• Confirm USB operation (should be nearly rock solid)

Hi @matt. I ran a quick test to confirm what you asked. Operating system: Windows 10. Everything you asked about has been confirmed. I haven’t tested anything else. I’ll test the same things on Windows 11 as soon as possible.

EDIT I can confirm that I didn’t encounter any issues with the same tests on Windows 11 either

1. Download and Unzip

Download the zip and unzip. There’s no installer.

2. Running

For Windows, you can go into the unzipped directory and run qa40x.exe directly.

For Linux/Mac, you will need the dotnet 8.0 runtime installed. On Ubuntu, for example, that looks like this:

sudo apt-get install -y dotnet-runtime-8.0

With the dotnet runtime installed, you can switch to the unzipped directory and run the app with:

dotnet Qa40x.dll

You might get a Skia compat issue (The version of the native libSkiaSharp library (88.1) is incompatible with this version of SkiaSharp.). If so, let me know.

Next, watch for an issue with libusb. If you see:

libusb-1.0 was not found

install the Linux system package:

Ubuntu / Debian

sudo apt install libusb-1.0-0

Fedora

sudo dnf install libusb1

Arch

sudo pacman -S libusb

Unversioned Symlink Issue

LibUsbDotNet may look for libusb-1.0.so while the OS only provides libusb-1.0.so.0.

Check what exists:

find /lib /usr/lib -name 'libusb-1.0.so*' 2>/dev/null

If libusb-1.0.so is missing, either install the development package:

sudo apt install libusb-1.0-0-dev

or create the symlink manually, for example on Ubuntu x64:

cd /lib/x86_64-linux-gnu
sudo ln -s libusb-1.0.so.0 libusb-1.0.so
sudo ldconfig

3. Linux: USB Device Not Found

If the log shows:

Usb.Open failed: Usb.Open(): Device Not Found.

then the native USB library is loading correctly, but Qa40x cannot find the analyzer.

The current code looks for this USB VID/PID:

16c0:4e39

Check detection with:

lsusb
lsusb -d 16c0:4e39

If nothing is returned, check:

• The QA403 is connected
• The device is powered
• The cable is good
• The device is not attached to another host, VM, or container
• The firmware is exposing the expected VID/PID

4. Linux: USB Permissions

If lsusb -d 16c0:4e39 shows the device but the app still cannot use it, add a udev rule:

sudo tee /etc/udev/rules.d/99-qa403.rules >/dev/null <<'EOF'
SUBSYSTEM=="usb", ATTR{idVendor}=="16c0", ATTR{idProduct}=="4e39", MODE="0666"
EOF
sudo udevadm control --reload-rules
sudo udevadm trigger

Then unplug and reconnect the device.

5. Linux: Useful Diagnostics

Check whether the USB device is visible:

lsusb -d 16c0:4e39

Check libusb sonames:

find /lib /usr/lib -name 'libusb-1.0.so*' 2>/dev/null

Inspect the deployed Skia native library:

find . -name 'libSkiaSharp.so'

Run the app:

dotnet Qa40x.dll

Running the App

With versions and usb issues addressed, the app should come up in one of nine languages. English (en), Deutsch (de), Espanol (es), Francais (fr), Italiano (it), Japanese (ja), Korean (ko), Portugues (Brazil) (pt-BR), Chinese (Simplified) (zh-Hans). You can force a language via the command line. For example,

dotnet Qa40x.dll --lang fr

will force French.

The app should come up looking as below:

TASK:

Confirm app will show the selected language. The language may be spotty as localization is mechanical and not all the way up to date. The key is if you see SOME strings rendering.

Trace Activity

In the two lower windows, you’ll should see activity. The lower two windows are connected to a C# trace generator that runs and updates every 250 mS or so. This can be turned off by locating the C# Trace Generator tile, and clicking the gear.

That will open the generator properties:

Uncheck the Loop Generation and hit Apply. The updates should stop. Close the app and restart it.

TASK:

Confirm on re-start that the generator is no longer autogenerating. This means that the app has persisted settings successfully.

PC Audio Generation

Click the global settings gear:

Select the PC Audio Tab, select your DAC output (ADC isn’t yet connected) device and click the test button. You should hear a one second tone.

TASK

Confirm you can generate a tone on your local hardware.

Connecting the QA403 Hardware

Next, plug in the QA403 if not already. The lower left should show Idle (it currently shows Idle if the QA403 is recognized or not).

near the top of the app you’ll see a series of play/pause buttons:

Press the rightmost button. This will start bursted activity with the analyzer and you’ll see time domain display in the upper left and freq domain displayed in the upper right. To change input levels or experiment with autoranging, do the following: Stop the analyzer USB activity. Right click global settings gear, click on Generator tab, set the sine generation to whatever you want (greater than -20 dBV). The select the Opt THD+N or Opt THD. Connect and disconnect the loopback cable to initiate with the analyzer running, and also start and stop the acq. Depending on the button you select the system will automatically optimize for best THD or THDN as incoming levels change.

That’s it for this phase. The tasks were:

• Confirm localization appears
• Confirm file system persistence
• Confirm local tone generation
• Confirm USB operation (should be nearly rock solid)

Please leave your results here (include platform (Win/Mac/Linux + OS ver) or send mail to support at quantasylum.

Thanks, Matt

thank you for your response. My spreadsheet results are the following:

The spreadsheet is using SI units for calculation.

The input values for fs and fsm were taken from the attached measurement report.

It seems that the Cms calculation in SW 1.222 has a bug. If I use the HTML report to calculate the Cms value manually, the result is 1.3718 m/N (1371.8 um/N) - very close to my spreadsheet. I guess this is easy to fix.

Thank you an best regards

Andy

I have an amp made by crown that is multimode AB & B when testing the amp, I noticed a strange dip in THD+N on the channels. It consistently occurs at the 1.4kHz and recovers at the 3.3kHz. Is this graph normal? It kind of looks like I am hitting a bandpass filter but I’m not certain.

I am using the QA403 and running the output of the amp through a 200W 8Ohm load resistor. There is a Hall effect sensor that is measuring the wattage going to the load resistor and the analyzer probes are connected to the amp output. The software version is 1.215.

Any pointers will help,

Thank you.

Thanks for your message

Maybe I will try

Walter

However, in practice, running the Windows version under Mono will remain unstable until QA40x is redesigned especially for Linux/macOS. Hello World is manageable for mono. Complex graphical user interfaces are another matter.

Hi @Allen14, there will be a test release of the QA40x_V2 in the coming weeks that runs on Linux/Mac using just dotnet framework (no mono). The purpose of the release is to get the kinks out with local PC tone generation, USB hardware, languages, etc. I’ve discussed before that it’s based on Avalonia, which relies on Skia for underlying rendering. This means 2D primitives will get rendered using your 3D hardware, and so things are very fast.

The screenshot below shows where things are headed. There are lots of windows available (from single graph to a 4x3 array of graphs). Each graph can pop-out for inspection/editing, has cursors, measurements, markers, can export in PNG or SVG. Left, bottom, right axis are config (in the plots below you can see right axis is dBR tracking while left axis is absolute dBV). You can drag/drop to re-arrange all graphs to suit your needs.

All graphs source their data from a trace pool (on the left). There is a hardware source (which is the QA403/QA402 analyzer), but you can also add a WAV source or a C# source. The benefit of the latter is if you wanted to code up a test signal that was 0 dBV in amplitude with 3H at -40 dBc and 7H at -80 dBc with 10mV of uniform noise, you could. Each trace source can have transformers applied. A transformer is a block of code that does something interesting. Which means you could add a 60 Hz powerline notch, A or C weighting, or a Farina chirp transformation. As you add transformers, more endpoints emerge in the tile (small boxes inside the tile). These are the attach points for your graph. So, the upper left graph can be looking at a C weighted while the upper right could be looking at A weighted.

Each graph is specified as either frequency domain or time domain, and in freq domain you have the option for showing either spectrum or transfer function. For spectrum, it’s like the QA40x today. For transfer function, it will show gain/phase relative to the right channel. This fixes a deficiency in the main screen of the QA40x app where you couldn’t show phase (and had to use the automated test to get that).

Why 12 windows? The aim here is to allow larger, automated tests to show you a full view of your problem in one shot. That is, graph one might be captured time domain, graph 2 might be spectrum with THD/THDN meaurements, graph 3 is bargraph of spectrum, and graph 4 is a plot of THD versus power.

It will take some months to get to parity withe the QA40x app, but once that’s reached, it will very quickly exceed it.

Localization to 7 languages will be in the initial release. These will be machine translated, so a native speaker might be able to help with improvements. The v1 app usually needed 100 to 200Mbytes of RAM while running, with a super tiny 5Mbyte distro. The v2 will probably sit and 300-400Mbytes of RAM while running, with a larger 100-200Mbyte distro. There’s been a lot of work done memory use, and so the consumption variability of the v2 app should be quite a bit better than the v1 up.

EDIT:

Here’s two SVG graphs copied from Qa40x_v2 app and pasted directly into Word. Full vector is preserved converting from Word to PDF. Very nice for publishing.

Mark

Cheers

Ian

one strange question

But ARTA sw can run with 403?

Walter

First does the input bandwidth follow the sample rate. In other words at 192KHz sample rate can I measure signal up to 96KHz and display them or is the bandwidth limited to 20KHz no matter what the sample rate.

Secondly, I see all the example displays show levels in dBV. I am measuring pro audio equipment and we measure everything in dBu. Can the display be switched to show dBu?

Cheers

Ian

Sorry for the late reply. I understand all that you said. Thanks for the explanation. Much appreciated.

Michael

today I was using the QA403 (SW release 1.222) and the QA461 transducer driver to measure the Thiele Small Parameters of an Audax HM210G6 woofer chassis (new chassis, no burn-in). I was using the SPKR Impedance automated test.

According to the datasheet (see https://www.audax-speaker.de) the Vas value of the chassis is 85 liters. With the SPKR Impedance automated test (added mass method), I measured the Vas at 10.279 liters. Although TSP measurements of loudspeaker chassis usually show quite some deviations from the datasheet, this result seems wrong. I have used a spreadsheet to verify the Vas calculation.

Formulas used are:

(fs and fsm: resonance frequency w/o and with added mass, Sd: effective area of cone, c: speed of sound, Rho: density of air)

The result of my Vas calculation is 97.08 liters. This seems within the typical error range of such measurements. I re-verfied the Vas calculation using the ls_param spreadsheet from the Elliott Sound Products website ( Measuring Loudspeaker Driver Parameters ) - Vas calculated as 98.09 liters.

It seems that the SPKR Impedance automated test has a bug in the Vas calculation. Does anyone else have experience with this automated test?

Best regards

Andy

The tool you want is here:

Set the transition times to zero. I like a half second pre-conditioning at lower level, then maybe a half second at the higher level (+20dB), then a couple seconds back at the lower level again, because release times are always longer than attack times. Make sure each duration is long enough that the gain stabilizes at the final value before the next transition.

“De-skew the input” offsets the time axis in case there is a delay through the processor. It’s common to see a spike in the reported gain right at the transition if the IN/OUT are not exactly synchronized. Gain is only measured once per cycle, so use a high frequency for high time resolution.

As in all things measurement, you have to know ahead of time what the measured result will be to correctly set up your measurement, so expect a series of successive approximations, especially if you want a publishable result.

Dale

There is a one minute demo showing a file load and short explanation here (you may want to raise the volume):

YouTube

I have exactly one more major feature on my TBD List: improvements to the generator to allow constant output.

Mark

The link below is a good starting point. Some other suggestions from the algorithm as in 2) and 3)

1. Measuring Compressor Response - #4 by amp-haus

2. Pulsating oscillator - #18 by matt

3. The issues with QA403 and 1.183 in terms of usage - #18 by matt

When you start `mono QA40x.exe`, Mono must use **libgdiplus** to imitate the Windows GDI+ stack, which is where the `GenericError` and X11 `BadDrawable/BadAlloc` often originate. QA40x.exe is a Windows-targeted.NET/GTK# application. Xeyes functioning just indicates that X11 is still in use, not that GDI+ rendering is.

WSL exacerbates this as well. A single weak connection will cause the GUI to malfunction. WSL + X11 + Mono + GTK# is a stack of shims.

If testing QA40x under Mono is something you really want to do:
Use a genuine Ubuntu virtual machine (not WSL). * Install gtk-sharp-2.0, libgdiplus, and mono-complete. Use `MONO_LOG_LEVEL=debug mono QA40x.exe` to execute it.

However, in practice, running the Windows version under Mono will remain unstable until QA40x is redesigned especially for Linux/macOS. Hello World is manageable for mono. Complex graphical user interfaces are another matter.