Wow this is so cool. Congrats!
Wow this is so cool. Congrats!
here's the comparison:@IVX you have the meizu dongles too right? Any approximation of how much more power the E1DA needs to draw compared to a meizu dongle? Been wondering about that. Not that power is an issue but I suppose the 9038d will need to draw more than apple/meizu dongle
@amirm With about 8.9 mV and 300 Ohm the power is about 0.26 µW and (according to your original power vs. distortion graph above) the "good" channel should have about 65 dB THD+N. But your new measurement now says it's actually much better at 72 dB (just as good as the reference Topping DX3Pro Low).A quick note on this channel differential:
I was chasing a different problem and grabbed this dongle randomly. I was surprised to see one channel's noise shoot way up at lower levels. This had nothing to do with balanced or unbalanced measurements. But rather, was trigged by using of the "high performance" mode in my APx555. This feature uses dual ADCs, one for the signal itself and the other for the rest of the spectrum. The result is that the analyzer distortion is essentially removed from the equation. Due to historical reasons, I had NOT used this mode for the above graph. But recently went ahead and enabled it due to so many devices doing well in this graph.
This is the snapshot with the High performance mode on:
View attachment 122508
As you see, one channel in blue has gone crazy, raising the noise level. It causes SINAD to drop by 15 dB or so as a result. And the fact that balanced input is used (under Input Configuration).
See what happens when I turn OFF the high performance mode:
View attachment 122509
The problem goes away. I sent a bug report to AP but just got a standard message of checking if the hardware is good. The hardware is. I will either have to turn this mode off like I used to have it or watch for it. Have not decided which.
Note that this doesn't impact the dashboard where this mode is always on as the levels are high.
Anyway, the "good channel" is the trustworthy measurement here for E1DA.
It is a different measurement (load) system for headphone amps. I will do an article on it soon....In additon, I see "200 kohm" in "Termination" under "Input configuration".
This begs my question: is this new measurement for 300 Ohm load?
@IVX I'm very interested in the ultrasonic noise of the SMPS and how it affects the I/V stage.amirm and the dynamic range in your latest screenshot is near 122db unweighted, hence A-weighted(typically +2-3db) will match with my specs 125db. Good to see you didn't take my comments wrong and use balanced input now, please keep it up. I'm not against of somebody but I'm always on the side of the truth aka proved fact(a science as is).
Regarding AP, it is a kind of legacy issue, they did BW limitation ONLY after preamp and attenuator even in SYS1. Probably at the end of 80th it wasn't a big deal but today any dirty USB or smartphone with LTE On could seriously affect the precision of the test. I would add BW limit in the very first preamp stage to guaranty that my BW is truly limited and any HF noise content will be ignored.
Yeah it's just added as a freebie novelty itemYes I did And strangely, there was a spare PCB in the box of the SG3.
Jimbob54, 9038D isn't my idea at all, it was asked by our discord society like that: please give us the same 9038S DAC but 3.5mm unbalanced. So I did as requested
2.5mm>3.5mm headphone adapter, right? To use balanced cable with 9038d? Just to avoid unfortunate conclusions(or just carry a 2.5-3.5mm adapter with me at all times ;-)
2.5mm>3.5mm headphone adapter, right? To use balanced cable with 9038d? Just to avoid unfortunate conclusions
cursed "adapter" wording haha
capslock, regarding the gift-PCB, it is the old version of 9038S which we ordered more than we needed, and I can not throw it out because nice looking and gold-plated, so I pushed Sunny to put one in the giftbox.
Sounds like a potential youtube topicAs a side thing, people keep thinking that my tests are "simple" sine wave and that music waveforms are more complex. Here is the time domain (i.e. what the samples look like) of the multitone signal:
View attachment 119203