Topping E2x2 Audio Interface Review

[Yoghourt]

• XLR in AC behavior, without phantom
  
  • Line-out is injected into XLR in through a load resistor per branch
  • Z+ = Vloaded/(Vopen-Vloaded) - (Rload + Zlo+) and so on
    Calculations below use Zlo+=53.2R, Zlo-=52.2R, Zlodiff = Zlo+ + Zlo-
  • Measure done through picoscope, without switching voltage range

Phantom	Freq	Input	R+	R-	V+pp	V-pp	Vdiff	Z+	Z-	Zdiff
0​	1kHz​	XLR	open​	open​	1,856​	1,645​	3,495​
0​	1kHz​	XLR	1604​	1606​	0,6034​	0,5353​	1,123​	798​	800​	1570​
0​	1kHz​	XLR	1604​	open​	0,6028​	1,643​	n/a​	797​	n/a​
0​	100Hz​	XLR	open​	open​	1,864​	1,652​	3,508​
0​	100Hz​	XLR	1604​	1606​	0,6097​	0,5405​	1,128​	806​	806​	1571​
0​	10kHz​	XLR	open​	open​	1,861​	1,65​	3,506​
0​	100Hz​	XLR	1604​	1606​	0,6042​	0,5355​	1,127​	797​	797​	1571​

• (Zxlr+ + Zxlr-) is consistent with Zdiff= 1.54k within 2%
• Spec impedance is to be understood as differential, and is met with < 5% accuracy
• There is no significant impact of frequency
• There is no visible impact of disconnecting a branch on the other branch ⇒ XLR in is a true symmetrical design.
• From frequency response below, -0.5dB @ 13Hz
  https://www.audiosciencereview.com/forum/index.php?attachments/topping-e2x2-audio-interface-dac-balanced-in-line-frequency-response-measurement-png.322564/
  Assuming 1st order low-pass → Flow_pass = 4.54Hz
  Assuming RC structure, AC cap s-e = 43.8µF
  43µF is effectively a capa value that can be found on market, even if not the most standard (39 or 47µF)
• XLR in AC behavior w/ phantom
  • Line-out is injected into XLR in though AC-coupling cap + load resistor per branch
  • Cap = Mundorf Ecap audio raw 470µF 63V → Z(1kHz) = -j*0.338R
  • My meter readings were not stable enough. I was reluctant to take any risk with my picoscope (20V max input), so below measures are from analog scope, also not recalibrated since epoch. Accuracy is, say, ~10%

Phantom	Inst	Input	R+	R-	V+pp	V-pp	Z+	Z-	Z+ + Z-
1​	0​	XLR	open	open	1,63​	1,45​
1​	0​	XLR	1604​	1606​	0,535​	0,474​	810​	805​	1615​

• Are we coping from a slight impedance change when phantom is switched on? Or it a bench artefact? Well, it’s easy to check : switch phantom power on/off, and check if it changes readings
  → nothing happened = XLR in impedances are fairly constant whether phantom supply is on or off

• Line in/inst in vs. phantom supply
  
  • No DC on line in/int in when phantom is switched on.
  • Line-out is no more involved here. In below table, DC voltages are measured when stabilised.

Phantom	Inst	Input	Load	Vleft+	Vleft-	Vright+	Vright-	Zph s-e	I se (mA)
0	0	XLR	open	0​	0​	0​	0​
1	0	XLR	open	46,9​	46,9​	46,9​	46,9​
1	0	XLR	10k se	27,9​	27,9​	27,9​	27,9​	6810​	2,79​
1	0	XLR	3x10k// se	15,4​	15,4​	15,4​	15,4​	6818​	4,62​

• Without external load on XLR, phantom set-up time = couple of seconds, power-down time > 1 min
  Now that’s interesting : not only we see effect of AC coupling cap in the signal path, but also we see a high-value discharge impedance…
• Let’s first consider the charge scenario, setting aside the “discharge resistor”.
  At the time we switch on, AC-coupling capa is discharged hence null voltage. Basic expectation is therefore to see the result of a resistor bridge between phantom-coupling resistor and AC impedance.
  • Let’s do the model and the math

• Assuming Rp=6.81k and Zin=800, we reach Uin = 67.1% E when t=T
• Let’s check that on picoscope, with 2,21MR between XLR in and scope probe 1MR for each branch. Stabilized result is E=14.52V, so Uin(t=T) = 9.74V

Here it is right channel, cold and hot points. Trace for left channel is virtually the same.
This leads to C = 0.964/(6.81k+800) = 123µF, hey quite close to standard E12 value!

This is much higher cap value than predicted by freq response, which could indicate AC cap dominated by another low-pass in further input stages.
The charge curve starts from 0, without the expected initial step E*Zin/(Rp+Zin)~1.5V
Hum… No idea there…

• Regarding the longer-lasting discharge, it can point to either a high-value supply discharge resistor, or a supply decoupling cap, or -more probably- both!
  The math is virtually the same except that target point is 4.84V, and we are looking for “Rp”.
  Here is the picoscope trace.

From which is found Zdischarge = 33.47s/123.3µF - 6.81kR = 20.3k
Like the 6.81k, this looks definitely a familiar value…

So here what xlr in looks like all in all

My apologize if any inconvenience and the verbosity. FYI, I will also take a similar look to line-in (instr off).

 

[yourmom210]

What is the AUX output on the OTG version? Does it have a separate output channel? I want to experiment with subs integration and am wondering if I can use the OTG version as a 3-channel DAC for this purpose.

edit:

found screenshots of the Topping software running with the OTG model, and it’s obvious that it is not a separate channel. But SPDIF output is. If I connect an cheap external DAC to the SPDIF output to get 2 more channels, would it degrade the quality of that DAC output in any way? By forcing resample for example. Will the clock be derived from SPDIF and delay will be constant, or will it drift over time?

Were you able to figure this out?

On macOS, the OTG version reads as an 8-channel configurable output device. So definitely lots of possibilities here.

I took some self-measurements of the aux (3.5mm) output on the OTG version and they seriously lagged behind the 1/4" outs (which I'm using as SE TS outs). I had to attenuate by around 7-8dB if I remember correctly (either digitally or around 1:30/2 o'clock on the analog pot) before getting a clean signal. Lots of distortion above that. Ran a a host of tests on the 3.5mm and they all came back dirty.

Edit: Here's a screenshot from midi options

 

[yourmom210]

I used the exact same everything with my Motu M4 which didn't produce this tone. I was meticulous as I was trying to hear differences between the preamps.

As mentioned above, the only thing I didn't test at the time was trying to figure out if the tone was captured by the microphone or it was an internal issue of the unit.

Maybe unrelated but the Topping E8x8 released later doesn't record up to 192k anymore, only to 96k.



True but nothing stops you to pitch down the sample even further with interpolation. See this video where I lower the ticking of a clock 8x (3 octaves).

I agree this is a niche issue that 99.999% of users won't bump into. For recording music, podcasts, etc I would rarely go higher than 48k.

Not sure if this is related, but my unit (OTG version) shows constant low-level noise from the disconnected inputs when set to 192kHz. Haven't seen it with other sample rates, but just noticed yesterday.

 

[John.D]

Can this product be used to learn some basics about testing your own gear like a DAC or something? Is there a guide already?
Thanks in advance

 

[lunaria]

Hi. Did somebody find out what this headers are for in the E2x2 pcb? UART? I2C? SWD? I2S? The “I2S_FRE” silkscreen text is intriguing. Could it have a “hidden” I2S input, maybe used for the OTG version?

The headers are located in a different place in the OTG version.

View attachment 520812

The XMOS chipset that the E2x2 uses would absolutely be capable of handling an additional I2S input in terms of hardware capabilities. However, even if there are unused accessible pins, how do you expect to access that additional channel without a custom firmware? As far as I can tell, a custom firmware would include a crazy amount of reverse engineering (practically almost impossible and the standard firmware doesn't include support for something like that).

The standard firmware almost certainly doesn't allow access to some hidden I2C channel.

(The Topping PA5 with its severe hardware issues was actually reverse engineered and I would be very happy for the same thing to happen to the EX2x2.)

 

[half_dog]

Not sure if this is related, but my unit (OTG version) shows constant low-level noise from the disconnected inputs when set to 192kHz. Haven't seen it with other sample rates, but just noticed yesterday.

This interface uses noise shaping on its inputs and it begins around 40kHz and becomes more severe in higher frequency. It seems the software dBFS level meter considers an average signal of the bandwidth, hence with the energy from the noise shaping, the total signal measured by the level when using this higher sampling rate displays a constant low level noise...

If you use RTA from REW (ASIO or WASAPI excluse mode), you will able to observe the noise shaping. I will try to post a screenshot when come home.

 

[half_dog]

@yourmom210 here!

 

[morillon]

What s input ? With gain? Open ? Shortcut? Etc

 

[half_dog]

What s input ? With gain? Open ? Shortcut? Etc

It is observed in both analog inputs, does not matter if it is XLR (48V on or off), Line either Instrument, minimum gain and open in this screenshot. When the gain knob is closed to its end, there is a low pass filter around 20kHz. It might reduces noise shaping, but I believe the noise background might be dominated by the gain stage.

 

[Guddu]

Can this product be used to learn some basics about testing your own gear like a DAC or something? Is there a guide already?
Thanks in advance

You can start with a quick search in this thread with keyword "measure" perhaps.... I just did the same.

 

[seanman]

Are there any other differences between the OTG and the regular e2x2 besides the main obvious features, either advertised or not?

I have no need for OTG, SPDIF, Aux Out, or more loopback options as it currently stands.

Only difference I can see in the specs on Topping site between is Line Input Impedance 6k (regular) and 9K (OTG version). Not super well versed in the spec stuff, so not sure how much of an impact this would make, if any.

Mainly using guitar and either a dynamic or condensor mic most days. Occasionally using a mini synth for line input. Curious if there's something else upgraded or "better" with internals or new measurements I'm not aware of and would be worth spending the extra cash for.

Thanks all!