Schiit Modius E And Cinemag CMLI-15 Level Input Transformers: Uh Oh It's Magic

I'm putting this under the mods and tweaks section despite the mention of the Modius E. @JK47 brought the Modius E to my attention when he mentioned the use of this DAC with the CMLI-15 Line Input Transformers. There had been a lot of talk about these transformers back in the day when people were trying to get better clarity out of their Yggdrasil A2/OG for feeding the single-ended inputs of their amps. The single-ended outputs of the Schiit A2 DACs were said to be more grey sounding or veiled compared to the balanced outputs. (This is probably something that may be worth exploring in a later thread).


Photo of Cinemag CMLI-15 Line Input Transformers stolen from someone's For Sale post.

Anyway, back to what I was going to say. The Modius E was just another delta-sigma DAC that I measured, discounted, and threw into my headphone measurement rig on the account of its excellent measurements. (It replaced an AKM Modius). I didn't really think about it until @JK47 mentioned the Modius E in the context of using it with the Cinemag transformers.

The Modius E w/Cinemag combo has a wow factor when it comes to faster paced music with dynamic gusto compared to the lazier more romantic MM2. The MM2 has a thicker, liquid density in the mids that gets lost in fast complicated sections of certain songs, but that dense tone is also addicting on some tracks.​

In different thread, another reader mentioned that perhaps we should revisit the Modius E. I guess I will be doing this partially here. I do have to admit "R2R / ladder DAC stubbornness" on my part as for the reason I never seriously considered the Modius E. Even more so because I had heard another rig in town where the MMB1 and MMB2 were replaced by the Modius E. That person said the Modius E sounded way better! Of course I erased this from my mind because of my stubbornness. I only revisited the Modius E while trying to find an appropriately priced pairing with the NITSCH Sound Pietus Maximus headamp (https://www.superbestaudiofriends.o...maximus-optimus-prime-review-the-touch.13957/). @CEE TEE also mentioned that the Pietus Maximus sounded great with the built-in ESS card.

And dang, was I surprised at its performance! The Modius E wasn't bad. It wasn't OK. It wasn't good. It was very good. Now since I had been moving my A2 Schiit DACs around, I happened to have my XLR-SE converters with the Cinemag CLMI-15 line input transformers sitting around. So what the heck? I just tried them for fun on the Modius E. Hot damn. This combination is exemplary! (Not that I ever doubted @JK47).

I will confirm @JK47's observations and add a few more.

• The Modius E + CMLI XLR->SE generally has better clarity than the Modius E's SE outputs. I suspect part of this is use of the super duper balanced opamps used for the Modius E's balanced outputs.
• The Modius E + CMLI XLR->SE imparts a dose of inner warmth compared to the Modius E's SE outputs
• The Modius E + CMLI XLR->SE has more energy and oomph than the Modius E's SE outputs. It happens sometimes. The thing about wires with coils is that they are current storage devices. It could also be added distortion in the bass.
  
• The Modius E + CMLI XLR->SE has a more vintagey kind of sound. By this, I harken back to the days of cassette tape, mixing consoles and microphones with transformer couplings, etc. It's called saturation. Where higher level signals gradually distort, but in a pleasant way (as opposed to digital clipping which is bad).
• The Modius E + CMLI XLR->SE sprinkles a little bit of magical pixie dust onto the music. I have no better way to explain this. It's like how a lot audiophiles really like the sound of a good transformer volume control as a passive preamp.

Of course I was most curious from a measurement standpoint. How much distortion, at what frequencies, and at what levels (again, the saturation thing) did the Cinemag CMLI-15s impart on the Modius E's output (which we know to be almost perfect from a measurement standpoint). Inspired by @atomicbob's distortion surfaces, I knew this was the time to explore.

On to science!

 

Schiit Modius E
XLR-SE using Cinemag CMLI-15 Line Input Transformer
Distortion Surfaces

• 100k-ohm load
• 90kHz bandwidth
• APx555 "cheat" HPSA mode on
• Note: I have included H2, H3, H4, and H5. The latter two replaces H4-10 as with other plots. I did this here because there is interesting stuff going on.

H2 - second harmonic distortion product


H3 - third harmonic distortion product


H4 - fourth harmonic distortion product


H5 - fifth harmonic distortion product


Thoughts:

• With distortion measured in db, we can naturally expect the value to steadily decrease (unless distortion is actually rising) as we increase the output level. This because there will be more dynamic range (remember, decibels are ratios not absolute numbers).
• What's interesting is that H2, H4, and H5 follow generally follow this expectation, with the twist that the distortion tends to be a higher in the lows, with a marked increased below 70Hz.
• Third order distortion in absolute terms seems to gradually increase as the output is increased. This especially so below 125Hz. I suspect this explains the dose of inner warmth.

Now are these results gonna make the Pink Panther shit in his pants? It depends. If you like numbers for the sake of numbers, then this is not for you.

For reasonable people, it's more complex. There's reason why I made the color scale the way I did*. There's definitely a saturation effect going on at all harmonics H2 to H5 in the lows, a little more in the odd than the even, and overall for H2 and H3. For sure this distortion is audible. There's some warmth, even some fuzziness that can be heard in the lows at loud volumes (almost any pop with modern beats). However for the rest of the audio band from upper lows to top octave, distortion is too low (remember, 16-bit only only has 96db of dynamic range).

At the end of the day, I conducted with simple a Radio Shack (RIP) switch a blind test the following: Topping E50 single ended outputs, Modius E single ended outputs, and Modius E with XLR-Cinemag->SE into the Pietus Maximus headamp. The latter won with respect to preference. (It was also the most obviously different sounding from the first two.)

*Red means broken, yellow probably ain't so good but we can probably get by, green is audible but still good. No, I didn't pull these outta my ass. They are based on over a decade of measuring and listening to gear.

 

Both. Increased sample points and tessellated the square further via interpolation. I didn't like how the colors were coming out before. I plan to stick with this.

Exactly you said. I broke it up for a reason. To see D3>D2 and D5>=D4.

There's some fuzziness going on the lows in pop music with modern beats. However, mids still sound more clear with the Cinemags compared to the single-ended outputs. Measurements don't explain everything. If they did, everybody would buy their audio gear, food, and cars based on measurements.

And everybody would be using DACs with ESS chips with OPA1612 for IV and buffer, and amps with the TPA6120 and OPA1612, or at the very least a discrete opamp with 125db+ of gain and corresponding amount of negative feedback, globally, and for each nested stage. Something about the National Fastpitch Coaches Association, I mean Nested Composite Feedback Amplifier.

 

It's the CMLI-15. After gear gets to Mjolnir 3 level and higher, that is SBAF Class B- or higher Recommended Components we can hear the CMLI-15 steal plankton. The bigger and more expensive CMLI-600 don't steal as much plankton, but they are much harder to drive (they will cause A1/GS XLR outputs to distort).

 

Regular way. If they weren't each successive harmonic would be shifted right. This is why H5 takes a dump near 20kHz. 5x20kHz > 90kHz <-- measurement bandwidth.

Anyway...

 

For reference, comparison between [XLR] and [XLR--CMLI15-->SE] outputs from Modius E. To see the effect of the line input transformers. What's the DAC doing and what's the transformer doing to the original DAC signal. Won't be running the entire gamut of harmonics, just H2 and H3 will suffice and give us the idea.

Schiit Modius E
H2 - second harmonic distortion product
[XLR--CMLI15-->SE] -vs- [XLR}


Schiit Modius E
H3 - third harmonic distortion product
[XLR--CMLI15-->SE] -vs- [XLR]

 

I would expect the magnetization would go away after playing music, say Daft Punk's Random Access Memories, for an extended period of time?

 

Here is the measured frequency response/bandwidth of the CMLI-15, straight into a 100kohm load. We can expect that the frequency response/bandwidth would change into a different load. As an aside, it's probably best that the output be loaded with a 10k-ohm resistor) for flattest frequency response, although at my age, the 2db rise at 20kHz may be preferred.

 

I'm trying it play it down. Audiophiles can be a neurotic bunch. The last thing I want are SBAF guys with these boxes, and there are quite a few, playing 25dBu 20Hz tones to demagnetize their line input transformers once a week.

 

Obviously extreme cases, but gives you an idea of the resultant waveform.

-10dbFS @ 10Hz


-5dbFS @ 10Hz


-1.5dbFS @ 10Hz


-1dbFS @10Hz


To put things in perspective, you are gonna get a lot worse with transducers in the lows at moderate volumes.