Good suggestion!
The first plot is the QA401's internal generator, for reference. The second plot is a 1KHz Victor oscillator adjusted for that same -16dBFS, the third a 10KHz Victor oscillator, and the 4th a 10KHz on the left channel and 11KHz on the right. The 5th plot is back to the internal generators in the QA401 wired in loopback, with one set for 10KHz at -16dBFS and the other 11KHz at -16dBFS.
Those plots show another cool feature of the QA software. Click near any peak and it will post the frequency and amplitude in purple. Just click it again to remove from the list.
Those were all good suggestions from Richeem! He really knew his stuff.
Attachments
Last edited:
Ok, so back to my earlier opamp question - so if I wanted to measure THD with it configured as a buffer, would I terminate the - input, connect the + input to the output of the DUT and then terminate the - output and connect the + output to my circuit's input? Just looking for basic connection and what tests/settings make sense. I am not sure where to start.
Depends on the gain of your test stage. Getting the best results will also depend on finding the optimum grounding, never easy or obvious. Attenuate the input to your dut to the target level with an external attenuator if you need to lower the input a lot. Play with it. The best results take experiment.
Sent from my SGH-M919 using Tapatalk
Sent from my SGH-M919 using Tapatalk
Here goes...
* The first two are the 1KHz internal QA401 generator with averaging set to 5 and then to 20. The 3rd capture is back to 1KHz and averaging = 5, but I've turned on the QA401's internal attenuator. It is activated by that green button in the lower right of the control panel and produces a relay click inside the box.
* The next 2 are one internal generator set for 10KHz and the other 11KHz, with averaging set to 5 and then 20. Then the next capture is the 10KHz + 11KHz at avg=20 but with the attenuator now turned on.
* The final two are Victor's 1KHz oscillator with averaging set for 5 and then 20.
Remember that clicking on the lower left of these will zoom them up.
Attachments
Last edited:
1audio:
thanks for the encouragement. Here's what I tried. I shorted the negative input and looped in my device
I can't seem to get the control-click working through VMWare, so adjusting the generator output level is pretty tedious, using my laptop trackpad. It would be nice if there was another way to invoke this menu.
So, what should I tweak / try next? This is a little discrete opamp circuit configured as a buffer. I could configure it for some gain and also try and use three as an instrumentation amplifier to test the differential input. Ultimately I'll use them as building blocks for S-K filters, so FR will become an important test.
thanks for the encouragement. Here's what I tried. I shorted the negative input and looped in my device
I can't seem to get the control-click working through VMWare, so adjusting the generator output level is pretty tedious, using my laptop trackpad. It would be nice if there was another way to invoke this menu.
So, what should I tweak / try next? This is a little discrete opamp circuit configured as a buffer. I could configure it for some gain and also try and use three as an instrumentation amplifier to test the differential input. Ultimately I'll use them as building blocks for S-K filters, so FR will become an important test.
Attachments
Does anybody know if the designers ever took in consideration to build in a function to analyze RIAA response?
Given that the board has a HW attenuator that can be engaged for low frequencies and the ADC is 10bit, they could sense the input signal with internal loop measurement and provide inverse RIAA curve signal accurate enough to perform a quick fairly accurate RIAA check.
Obviously it wouldn't be expected to hit +/-0.01dB accuracy that anybody could hit with a bench meter and signal generator and an excel table, but even if this method was accurate to +/-0.1-0.2dB, it would be a very quick and neat way to check RIAA work whether on initial design stage or other manufacturer's products.
Given that the board has a HW attenuator that can be engaged for low frequencies and the ADC is 10bit, they could sense the input signal with internal loop measurement and provide inverse RIAA curve signal accurate enough to perform a quick fairly accurate RIAA check.
Obviously it wouldn't be expected to hit +/-0.01dB accuracy that anybody could hit with a bench meter and signal generator and an excel table, but even if this method was accurate to +/-0.1-0.2dB, it would be a very quick and neat way to check RIAA work whether on initial design stage or other manufacturer's products.
To those who know more than I do, I have two questions:
1) Why is the bottom of the sine wave so narrow from the internal oscillator and wider on the external oscillator? (looking at the -100 to -130+ dB range on the 1K signal)
2) Why is there some 60Hz noise on the images with the external oscillator. I'm assuming cable routing or some AC device close by (or maybe agdr's cat from earlier pictures!)
I was about to guess "phase noise" but after a moment's thought, the internal oscillator is surely synced to the A/D sampling clock, which makes all the oscillator-related FFT bins come out really nice and sharp. There's math behind this, and if I knew it it would surely be too complicated to explain here.
Power mains signal is ubiquitous, and apparently just having an external cable is enough to pick it up. As I recall, this is Victor's oscillator which is battery powered. I'm thinking a shorter cable should lower the 60Hz and its harmonics, or maybe put the whole setup in a Faraday cage.
Cat-generated noise is always a possibilty!
I was wondering about both myself. As for #1, I have no idea. benb's thoughts make a lot of sense.
As for #2, that one I worked on a bit when I saw it. The Victor boards are being powered from my Victor Board Power Supply project here and I checked for 60Hz & powerline harmonics at the time and nothing significant. This particular one is using the Hammond 1598REBK fully shielded box, so it is effectively a Faraday cage. The end panels are aluminum and contact the inner shielded coating (Hammond has the inside of the plastic box sprayed with conductive matierial).
The Victor power supply board can use either the AC section from a wall transformer or a bunch of 9V batteries in series. I was using AC initially so my first thought was EMI from the AC section. I unplugged the (24Vac) wall transformer from the case, inserted the six 9V batteries, put the box back together, switched it over to batteries and... same 60Hz! In fact there seemed to be even more powerline harmonics (about 2x amplitude more) with the batteries, which seemed odd. So back to AC.
I haven't messed with it anymore since but my best guess in thinking about it today is earth grounding may help. I'm thinking that may be why the AC was slightly better than batteries. With the wall wart cord plugged in there may have been some EMI conduction through the transformer windings to the neutral side of the AC line, in effect partially earth grounding it.
The 60Hz could be coming in via the USB cable to the QA401 and laptop, which (the laptop) was running on AC for this test, but I don't think I saw the 60Hz line when using the QA401 internal generator.
The BNC cables are fairly good Pomona cables, that shouldn't be the leak, and the QA401 is solid metal. I do have some slots in the back panel for ventilation. I should try putting conductive tape over those and see what happens.
The way things are wired here is the "ground" of each Victor board is tied together via mounting on that front panel. The Victor board ground is the mid-point between the two TL431's on his board, which of course is different from the "ground" of what is feeding the boards. So in this case either side of the batteries, or either side of the AC supply if AC is being used, is "floating" with respect to the front/rear panels and the Victor board ground. The front panel, rear panel, case internal shielding, BNC grounds and all the Victor board grounds are tied together. And in this case each Victor board is being fed via its own constant current sink.
Attachments
Last edited:
I am curious about using the balanced inputs. This should be a big improvement over the QA400. To my understanding this would reduce/eliminate any mains noise picked up by the interconnect cables. I think I understand how to utilize the balanced inputs connected to my 8 ohm dummy load when testing amplifiers. I am not sure how to do it connected to Victor's excellent oscillator.
I am impressed with what has been demonstrated so far with this analyzer. Maybe it is time for me to order one.
I am impressed with what has been demonstrated so far with this analyzer. Maybe it is time for me to order one.
The 60Hz noise is most likely due to magnetic induction. A coaxial cable in a magnetic field will have voltage induced in the cable. Any nearby 60Hz power cable (including the fixed wiring in walls, ceiling or under the floor) will produce a magnetic field and therefore induce a voltage in the coaxial cable plugged into the analyser.
Coaxial cable provides less immunity from magnetic fields than tight twisting of two insulated conductors. The larger the loop area enclosed by the conductors the larger the induced voltage.
regards,
Rick
At the -100+ dB levels even a minute amount of power line current leakage will show. Let me show how- If the connecting cable as 1 Ohm of resistance and the 60 Hz artifact is -120 dBV that would be 1 microvolt. For 1 Ohm of resistance you need 1 microamp of leakage. If you start with 120V 1 microamp comes from 120 MegOhms of series resistance. 20 pF of stray capacitance is enough to get there. I have been through this and eventually you figure the best grounding that keeps the leakage out of your signal but its never the same twice. Even with the isolated power and a battery powered source putting the source on the wrong box of electronics will still degrade things.
The widening at the base of the fundamental has two sources. The dominant one is the phase noise of the analog oscillator. A digital source is much better. Using an independent digital source will confirm this. The second source is the windowing, which looks like the passband of a filter in this context. You see that on the digital source.
Here is a very good analog oscillator + QA401 at -2 dBFS and the same source into the Shibasoku 725 showing its actual residuals (the top line is -90 dB, the marker is references to -90 dB). You can see some of the effects I mentioned above. In this case when the source ground was floating there was a lot more AC noise residual around everything.
On my unit the analyzer has the residual distortions you see below. The source side has a lot of 2nd harmonic (no pictures for now) about -105 dB. I'll try to plot it in more detail later. If your DUT has distortion well below -100 dB the QA401 may not be your best bet for accurate readings. However for most real world stuff it should be adequate.
It has an internal attenuator for the DAC to scale the output. At higher attenuations the noise will be an issue. You could create a script to measure RIAA with high accuracy by doing a self calibration to confirm the attenuation accuracy and then a passive external attenuator to scale it for your device. Set the level at 20 KHz (the largest signal) and run the sweep. I'[m sure someone here has the skills to create this. Matt has published the necessary info to start. An Excel spreadsheet setup to import would be the remaining task.
What would be really neat is a series of Excel macros to run the QA401.
The widening at the base of the fundamental has two sources. The dominant one is the phase noise of the analog oscillator. A digital source is much better. Using an independent digital source will confirm this. The second source is the windowing, which looks like the passband of a filter in this context. You see that on the digital source.
Here is a very good analog oscillator + QA401 at -2 dBFS and the same source into the Shibasoku 725 showing its actual residuals (the top line is -90 dB, the marker is references to -90 dB). You can see some of the effects I mentioned above. In this case when the source ground was floating there was a lot more AC noise residual around everything.
On my unit the analyzer has the residual distortions you see below. The source side has a lot of 2nd harmonic (no pictures for now) about -105 dB. I'll try to plot it in more detail later. If your DUT has distortion well below -100 dB the QA401 may not be your best bet for accurate readings. However for most real world stuff it should be adequate.
It has an internal attenuator for the DAC to scale the output. At higher attenuations the noise will be an issue. You could create a script to measure RIAA with high accuracy by doing a self calibration to confirm the attenuation accuracy and then a passive external attenuator to scale it for your device. Set the level at 20 KHz (the largest signal) and run the sweep. I'[m sure someone here has the skills to create this. Matt has published the necessary info to start. An Excel spreadsheet setup to import would be the remaining task.
What would be really neat is a series of Excel macros to run the QA401.
Attachments
Last edited: