HI Niels,
Thanks.
Oh apparently I didn’t put the noise value into the list of values. Whoops thanks for catching that.
Mark
The fix will be in the next release. And it also applies to the THD vs Amplitude test.
Hi Mark
The calculation of power is wrong. Left channel shows 6,3V, right channel 5,8, but the wattage is the same for both channels.
Regards
Niels
Thanks, you’re absolutely right.
I’m refactoring the entire top level of Spectrum (along with the rest of the tests), so it’s already fixed and probably replaced with different bugs. That’s the last step to part A of getting load/save working. I should have that done in a day or two and that’ll be 1.1.4.
I expect 1.1.4 to be an interesting discussion here. I’m adding the ability to load/save tests (results and inputs), but it’ll be gradual with the end point being able to do charts with multiple datasets on them (at a minimum 2 so I can a-b compare).
After that: fixing bugs and finishing the QA402 USB support - all of which I hope to finish by the end of May.
1 Like
I can’t find it offhand but I know someone not in the US asked about the power markers for 50Hz devices.
The answer is that I currently do an eval at 50Hz & 60Hz and whichever has the highest magnitude I decide is your power line frequency.
This was never intended to be the final algorithm and given power supplies double or triple this, for 1.1.4 I’ve updated it to testing 50,100,150 and 60,120,180 and whichever harmonic has the worst distortion determines the power line frequency. That is almost certainly not the ‘final’ algorithm but seems good for now. I hate to have you specify your power line since this should be findable.
Mark
1 Like
Might cepstrum methods be a possibility?
Wow, I have never heard of using a Cepstrum for anything. I’m impressed and obviously clueless.
It looks like the math is trivial and so I could support it. I saw a little bit of literature about using it but let me read up on a bit and see. Is this a thing you rely on?
Mark
Version 1.1.4 release announcement
This version is a major upgrade for a few reasons. The primary reason is that the arithmetic code for the spectral tests (Spectrum, Intermodulation, and Scope) is now thought to be correct and complete. Please send bug reports.
REST Support - To get around hardware issues, this version adds back the option of using the REST interface rather than the direct to USB. Simply click the checkbox in Settings to revert to REST. This does require a running QA40x program to provide that REST interface. It will run more slowly.
Load and Save
This is a work in progress. Currently there are three buttons implemented on each of the three completed tabs (Spectrum, Intermodulation, and Scope).
Save → this provides the option of saving the entire tab (settings and data) to a compressed file
Load → this loads a saved tab, both settings and data - so you can see how the test was performed as well as look at the data.
Get → this loads a saved tab but does not make it the settings or data. Instead it puts the data into an “Other” set of data that can be displayed using the Other Left and Other Right checkboxes. This data will display slightly transparent and underneath the test data. The idea is to allow comparison with prior tests.
I’m very open to suggestions on the load/save functionality. Note that “Other” is implemented as a list currently limited to one item.
Mark
Also there is a new option labelled “Tab Info” which brings up a small window with the ability to see/type a name and description for the tab. Those strings are saved and loaded and the Name is shown in the plot title.
Finally, I changed the baseline for dBFS display since it isn’t clear what Full Scale is. Now it is 32 dBV to match the QA403 input limit.
1.1.4 Fixed Bugs for 1.1.5
a) the shadow colors for Other Left and Other Right plots were hard to see. improved.
b) the intermod data summary tables were both channel lefts. fixed.
1 Like
Contents
Routine Settings 3
Saving Settings 3
Loading Settings 4
Saving the Screen Image 5
Markers and Cursors 5
Minimizing Hum and Noise in Measurements 6
Measuring High-Power Amplifiers 7
Shorting Blocks 8
Loopback Performance 8
Analyzer Noise vs FSI Setting 10
Analyzer THD vs FSI Margin 10
Gain Measurement 11
Frequency Response 13
Noise 14
User-defined Noise Weighting 17
Signal-to-Noise Ratio (SNR) 18
Noise Density 21
THD and THD+N 22
THD vs Output Power into 4 Ω and 8 Ω loads 23
Burst Power into 4-Ω and 2-Ω loads 24
THD vs Frequency 26
THD vs Frequency vs Output Power 28
THD Visualizers 29
GenEx Burst Waveform 31
CCIF 19 + 20 kHz 32
SMPTE IM 60 and 7000 Hz 35
Multitone IM 36
Crosstalk 40
Output Impedance and Damping Factor 41
Automated Measurements 44
Power Output - THD vs Power at 1 kHz 46
THD vs Power at 16 kHz 50
THD vs Power with 4-Ω Load 53
THD versus Frequency versus Input Level 54
Frequency Response 57
Frequency Response by Chirp 59
Output Impedance and Damping Factor 61
Crosstalk vs Frequency 63
Intermodulation Distortion vs dBV Input 65
Intermodulation Distortion vs Power 67
Hi Mark
I have compared SNR for your app and QA analyzer, and SNR seems to be calculated very different between the two apps:
1 watt
5 watts
1 watt
5 watts
It this a bug?
Also included is a screenshot of the powermarkers in a 50 Hz country
Left Other and Right Other, what is this new stuff? It does not seem to do much
Regards
Niels
Hi Niels,
- I think I’m just going to add a 50Hz power option in Settings…
-
The other left and other right will display a previous saved test result translucently on top of your current test results. Simply use the Get button on the bottom left of the screen to ‘get’ the saved file and “Other Left” and “Other Right” will magically show those gotten values as if they were just swept in addition to your current data.
-
It’s very hard to compare SNR without knowing that the two devices are using exactly the same sweeps. I do a few things different from possible factory specification. First, I’m not using A weighting, it’s just noise from 20-20KHz. Second, I don’t use the AES2015 spec, which I think is too broad. Instead I’m using a 1/5 octave fundamental bandwidth (which should make my numbers ‘better’ but more accurate).
Still, here’s what I see with 1.1.5 when I compare the two with the exact same amplifier output. It’s pretty close…
Mark
Great, thanks for the explanation
Niels
1 Like
When running your program, the QA403 makes a repetitive noise. Not relay activation or release, but it almost sounds like a reed relay or one of the internal relay drive coils being modulated somewhat.
It’s kinda disconcerting. Relays like to be either on or off…
If the relay clicks each time a sweep happens, that’s generally not a good sign and indicates that you’re receiving more voltage than the attenuator is set to allow.
The way the program currently switches the output relay through a test →
- When you click Start :
- Every time the device sends out a wave the output is scanned for optimal output settings and that is set. If the output level changes (such as during the THD vs Amplitude test) then it will possibly click each time that happens.
- During autoranging the generator level is set once or twice and this may cause a relay click.
The way the program currently switches the input relay through a test →
- When you click Start :
- If you are auto-ranging (all tests except Scope, Spectrum, and Intermod) then the relay will click once or twice as it first sets it to max, determines gain, then sets attenuation to optimum. This also applies to Scope, Spectrum and IMD if you use auto-ranging there.
- Next the relay may click before/after the noise test as the attenuation is lowered if not auto-ranged.
- At this point it’s done clicking until it finishes the test, at which point the relay will click one last time to put the device back into 42 attenuation.
** Exception: the THD vs Amplitude test, which will usually increase the attenuation as the test proceeds.
Possible problems:
If the device you are testing has a very non-linear gain curve it is possible that some of the gain tests are underestimating the gain. I think I’ve been very careful but you never know.
If the right channel has more gain than the left channel and again is not at all linear then potentially there could be an overload issue.
Finally:
If the relay clicks on every sweep (not when you click start, but those tests which run continuously) then you have an overload issue - which I wish to fix - and please report exactly the steps you followed and what you are testing. Best of all send me a saved copy of the test+data.
Thanks,
Mark
As much as I’m not overjoyed with the ‘clicking’ either - note that modern day reed relays are rated at tens of millions of operations.
Postscript – that may have been a little easygoing. I looked up what I think are the Panasonic relays being used and they’re spec’ed at a minimum of 50,000 operations electrically (although 50MM mechanically). I’ll see what I can do to minimize relay usage.
I think you’re misunderstanding what I mean.
Start the program.
Short your inputs, connect nothing.
Press start.
Listen to the QA403.
After the intial relay operation, listen.
It is a very fast partial relay coil drive/interruption that makes a noise in time with the “run” light flashes. It sounds like a traditional reed relay (that’s how quiet it is as compared the normal relay sounds from the QA403). But the QA403 has no reed relays. So the program is pulsing the drive coil/s so fast during that LED blink they don’t even move. Why?
1 Like
Huh,… I had never noticed this but I think you’re right. If I run the Spectrum test the the start of every sweep there’s a very slight noise. I have to glue my ear to the chassis to hear it but it’s there.
I did some investigation and I think I’ve stopped that behavior by doing a few things different with the USB control. It will be 1.1.7 (a few days at most) so please download that version and give it a shot.
Thanks, yet another issue I would never have noticed.
Mark
1 Like
QA40xPlot v1.1.8 is now feature complete for the first Beta.
The Get operation now lets you compare multiple saved tests and there are a few other minor updates.
Thank you all for your help getting to this point. The next phase is just bug fixes for a while, although I’m collecting feature requests.
Mark
1 Like
Great program, thank you!
Feature requests:
- add color selection, instead of “MintCream” I’d like “AliceBlue”
- add option for inversed colors i.e. black display background
- allow to save window size and position to the default settings.
1 Like
You’re welcome.
Thanks for the feature requests.
The third (keep window) I view as a current bug. The only reason I haven’t done it yet is that I know it’s very tricky to get right with dual monitors.
I’m currently using almost all system colors - except the MintCream : ) - in the probably remote hope that theming the ui will be easy.
Mark