Sorry if I’m being obtuse, or overly concerned, but can someone please help me interpret these results? What is all the “noise” up in the 10kHz-20kHz band, and then for the Sansui in the 5kHz-10kHz band? Do I have something to be concerned about with the Sansui?
Here are three scenarios, all in the same physical space (my bench) using the same set of cables:
- QA-403 loopback test using barrel connectors in place of DUT
- Adcom GFP-555 preamp, line level input, volume control set for unity gain
- Sansui AU-919 integrated amp, separated preamp output, line level input, volume control set for unity gain
Hi @GlasgowGrip, thanks for the clear plots. This looks like 60 Hz harmonics ripping through the entire spectrum. Since you are seeing it in loopback, my guess would be leaky cables and/or environment. As a first step, check the QA403 noise floor with 0 or 50 or 75 ohm shorting blocks on the 4 inputs of the QA403. That should give you an RMS reading around -117 dBV (with 0 dBV full scale input) and there should be no powerline present in the spectrum.
Next, go back to your cables in loopback (the first picture you shared) and see if there’s any change in the 60 Hz and harmonics by re-positioning the cables about. If so, the cable ground/shield is probably not so good.
And then, take a look at how the 60 Hz harmonics change with QA403 location. If you move to a “quieter” location do the harmonic levels change? If the QA403 is vertical instead of horizontal? Upside down? If so, there’s likely a strong 60 Hz field nearby that will need to be addressed.
My favorite keyboard has backlighting, and if enabled there’s lots of hash in measurements as my keyboard USB cable runs under the QA403 on my home office. The levels of your 60 Hz are very small: -110 dBV is about 3 uVrms…so, it’s pretty easy to pick up unwanted emissions from nearby radiators. Cell phones too. I can sometimes see a text message is incoming (hash in the QA403 spectrum) in before I get the “ding” if the phone is on the desk.
In any case, once the loopback situation is cleared up, the other measurements will hopefully become clearer too.
I would not be concerned with plots like that on the Sansui or Adcom. It would not even pop up on my radar, so to speak. I would typically use a 6dBV input signal for preamps. Sometimes I have found that connecting a separate ground to the item under test will to lower the 60/120/180hz spikes…
Thanks guys, I was kind of convincing myself that everything was okay and this is just insignificant noise (in terms of home audio use). Every time I hook these preamps up to my scope (Keysight 100MHz 2 GSa/s DSO) and drive them with the QA403 outputs I can’t find anything except a clean sine wave, no oscillation or instability or anything nasty.
So a couple more quick experiments as inspired by @matt:
Self noise with 50 ohm BNC terminators on all four inputs, QA-403 still in the original physical position, outputs disabled and cables still connected to outputs but not connected to any DUT.
Same as #1 except cables connected from outputs to the Sansui amp. The amp is unplugged from the mains, so obviously is just acting like a big 60 Hz antenna.
I have noticed that the environment matters. For example, connecting my 403 to my laptop over on my test bench, where there are a multitude of Rigol and other test instruments, along with a large gigabit ethernet switch, I get lots of power cycle noise and some higher frequency garbage. Running the laptop with out the power supply helps, but does not eliminate the 60 Hz noise and harmonics. Moving to my desktop, away from the test bench equipment, improves the power line harmonics, and using a 4 ft. heavy duty USB cable and placing the 403 well away from the desktop practically eliminates the powerline noise.
I see what everyone is talking about. Through trial and error I have found several culprits, though clearly not all of them:
I have a small space heater near my bench that throws a bunch of hash into the 8kHz-12kHz range if it is running.
I have 1.5kVA isolation transformer sitting on the floor under my bench that apparently adds about 12-15 dB to the 60Hz spike (depending on the DUT) when it is plugged into the wall. It doesn’t matter if anything is plugged into the transformer, it apparently just puts out that much of a magnetic field if it is plugged in.
Thanks for the update @GlasgowGrip! Sometimes it’s hard to get everything cleaned up completely. Watch some of @VAR videos to see how much device/device variation you might expect and also channel to channel variation you might expect in vintage equipment. Very eye opening. Take a look at the crosstalk variation in an older McIntosh amp. The amp was overall generally very healthy. Still curious so much difference in this one area.
I appreciate referencing my little Youtube channel! I put out a video each week so that means I test a lot of gear, the majority of it “vintage”. >60dB crosstalk seems to be a decent # from what I have measured. Not sure why the difference between the two channel, but having a difference seems to be the norm so far- and it is not always the R-L better than the L-R. As far the the MC352, I have redone my load and test cable wiring since it was measured and think that the output impedance would be lower (larger damping factor, which met the spec). Maybe one day when I have someone over to help me lift it I will move it to my bench and remeasure it.
Subscribed! I really appreciate people that have the ability to put themselves in front of a camera and present well. Far from easy.
Thanks Dan, I appreciate your subscribing and appreciating the video. You would not believe how many takes I have to shoot to get it somewhat right, and I don’t know that I have any ability other than it is kinda fun putting everything together from the research to the testing and repair sometimes. Guess it is a labor of love and so I have I not run out of things to test or talk about…
I am trying to duplicate this test. I have looked at your screenshots and have set my QA403 to the same parameters, my results look nothing like yours, Should I be using a Automated Test in conjunction with these settings?
I think you should post your results otherwise it’s hard to help on specifics. I don’t think automation will change anything fundamental.
I have now realized that the cable from the Function Board was disconnected.
Here we go again with new measurements
Any feedback on these measurements?
Gain numbers appear very far out.
It looks to me like your 1khz signal has disappeared- something is not correct or your preamp is bad…or not set correctly
Adjusted to maximum volume:
This looks pretty much “normal” - or maybe I should say “typical”. Ideally it would be much cleaner.
You have a big 60Hz fundamental - most likely “pollution” from the AC mains in your environment plus possibly a contribution from the power supply in your preamp - and then you see its harmonics at 120Hz, 180Hz, 240Hz, etc.
Then you have the predominant signal at 1kHz - which is what you want to see - and then you see its harmonics at 2kHz, 3kHz, 4kHz, etc. I believe many of the smaller spikes in between those are still further harmonics from the 60Hz.
I would agree with GlasgowGip about what you are seeing. I like to show the output as rms volts instead of dBV since that is what most preamps provide as far as specs. Also, the standard preamp levels- what I have learned from Amir at Audio Science Review (who is the Michael Jordan of doing audio measurements IMHO), is to put 2vrms (+6dBV) into the input and adjust the volume to get 2Vrms out- 0dB gain. Your attenuator will change to probably 18dB- sometime 12dB will work.
Possible recommendations for reducing PS noise?