Introduction to the Schiit Audio Midgard review
First of all, I absolutely refuse to do Schiit jokes. All-in-all it is a proud German name, host to a long line of audio engineers. We should show some respect. Also, it is kind of cliché at this point. Maybe next time. I’ll admit, I can’t recall the last time I was so excited about a review. Yes, we do play with the coolest toys on the market, but hear me out. It’s a $219 amplifier made in the US of A. Metal enclosure, discrete circuit, pre-out. I’ve seen dudes spend more on strippers. Duh, I’ve seen a dude spend more on ONE stripper. We’re talking wire strippers, obviously.
So, for what it is, Midgard is priced more than fairly. At least on paper. I love to see great sound quality getting more affordable. It excites me more than the new expensive stuff. And the other thing, that is a cherry on top: it does something different. It’s hard to get me curious about headphone amplifiers, as I’ve seen hundreds at this point and even built a couple.
I’m speaking about the Halo topology being used for the first time. In short, it is a mixed-feedback loop that includes headphones. Possibly improving sound quality. Why possibly? Because it’s very hard to measure the effect on the headphone’s output. More on that in a dedicated paragraph. So to sum things up: Schiit promises state-of-the-art performance for the price of a steak dinner in a big, European city. Steak, and a few glasses of wine, but can you do one without another?
They did something really smart here. The box, is a regular shipping cardboard, the one that you know its colour, but it’s weirdly hard to name it. Or is it just me? Inside the device is secured by a special, custom-made cardboard insert that holds it physically at a distance from the box sides for secure shipping. Smart, cheap and works. Inside you’ll find a quick start manual, some stickers and that’s basically it. Gotta keep the price down, but honestly? It’s secure, keeps the price low and arrives in one piece. It ain’t stupid if it works.
Design and Build Quality
Out of the box, I knew one thing. It’s Schiit Audio quality. Full metal chassis, nothing to complain really. We have to appreciate the metal volume knob and Neutrik XLR connectors, something not entirely obvious at this price range. The logo on the top is stamped and then filled with paint, while other markings are made with screen printing. The only thing I don’t like is the two switches on the front. They feel kind of wobbly, but then I looked at the price and forgot about them. It’s an aesthetic little box. Sadly, the 4-pin XLR headphone connector can bend inside when plugging headphones. Some bracing could help that. A shame. Besides that, it’s all good.
Tech inside the reviewed Schiit Audio Midgard
Now we’re getting to the good stuff. Reviewed Schiit Audio Midgard is not exactly a balanced amp. The inputs are balanced, as is the volume control. The input of the amplifier is balanced too, you can see that by the input filter components. The output of the amplifier is single-ended though. So the conversion from a symmetric signal path occurs here. It means we get all of the benefits of balanced inputs, namely Common Mode Rejection Ratio, which means interference resistance.
We don’t get extra output power when using the XLR output, so keep that in mind. Not that you’re going to miss it, since it delivers 5,5W @ 16 Ohm, 3,2W @ 50 Ohm and 375mW @ 600 Ohm. It drives a Susvara to a very satisfying result. But yeah, I basically touched the end of the potentiometer wiper on low gain. The manufacturer described the amp topology as open-loop linear, with low-loop feedback. I get all tingly on the inside when I read stuff like that. But, what does it mean in English? So, amplifiers use global feedback to achieve low distortion and low output impedance. Of course, there are some “no feedback” designs, like tube amps, or those crazy amps from Dartzeel.
They do use feedback in fact. It’s not global, it’s local. You can’t build an amp with absolutely zero feedback of any sort. What Schiit meant, was that they have a design that uses a lot of local feedback and some fancy engineering to create an amplifier that is pretty linear in open-loop (no global feedback applied), so it doesn’t require too much global feedback to achieve stellar performance. On the other hand, most manufacturers prioritise very high global feedback to achieve absurdly good measurements. Think NFCA circuits from Topping.
But that means, that their amplifiers are not very inherently linear. Topping prioritises fast design and deployment of their product and using huge amounts of global feedback makes it possible. To build an inherently linear amplifier it takes a long time and lots of tweaking. We know Jason likes to take time with his stuff. There is a whole book about this type of design: Audio Power Amplifiers, towards inherently linear amplifiers by Dr. Arto Kalinummi. It’s an essence from his PhD study, which ended in him building a couple of no-global feedback amplifiers. If you’re a geek like me, this book is a must-have. It was distributed by Linear Audio, and I think it still is available on Amazon.
Okay, what about the Halo topology? I have a separate paragraph about that, down below. The power supply uses a 16VAC wall-wart brick and a stacked power supply design giving a couple output voltages, but the most important part runs on bipolar +/-30V and +/- 20V rails. It gives the amp great performance into high-impedance headphones. Headroom is king. It is by measurements the best SE amplifier Schiit has made. Also, the most powerful. The preamp outputs can’t be turned off, that is the biggest con of this amplifier, one we shouldn’t have to bother with. There is space on the PCB and on the front panel for that. Shame it wasn’t done.
So the Halo topology is based on a 40-year-old paper (which was not disclosed by Schiit) that described using mixed feedback to improve the speaker’s performance. According to Schiit, the original paper mentioned a 6dB improvement in acoustic measurements. Sadly, I couldn’t find any paper that fits the description. The Quad current dumping amplifier seems to be on a decent track, but don’t quote me on that. The idea is, that Halo output has sense resistors (hence 2Ohm output impedance), that allow for mixed-mode feedback (voltage and current feedback).
I suspect that mixed mode feedback allows for a partially current drive, of the driver, which should result in the driver being powered “better”. It is also assumed (which I agree with), that this system will have the biggest impact on headphones with impedance peaking. That makes perfect sense to me, as the “floppier” the driver, the more error it will create, giving the Halo more to work with.
That said, all of that wasn’t supported by measurements. Jason said, that they don’t want to spend resources and time. They invited people with measurement gear to get reviewed Schiit Audio Midgard for free and to measure it to heart’s content. This is how I got mine, even though I only have access to electrical measurements, not acoustical. I told them, but still got it. Thanks for this chance, I appreciate it. As usual, I wasn’t asked to say anything positive and they have no influence on this review. They’ll see it at the same time, as you do. As usual with Ear-Fidelity.
For starters, I would like to thank Remco Stoujesdijk from It’s Only Audio for providing those measurements. His help was crucial in understanding the Halo as much as we were able without excessive reverse engineering. Measurements were done with Audio Precision APx585. Measurements were taken with XLR inputs and both headphone outputs. As you can see, Midgard is a fantastically measuring amplifier with impressive specs. Please note the scale of the attached measurements. They have been scaled up, so we can actually see something.
Frequency response, please note the scale. It’s completely flat in the audio band better than (-0,125dB).
THD+N versus measured output, which is astonishingly low. You can see it’s optimised for around 2V.
SNR @ 10VRMS input, max gain, over 22kHz bandwidth, brick wall filtered. With bandwidth up to 96kHz the result is 6dB worse @ 123dB indicating the white noise performance of the amplifier (flat noise density across the bandwidth, i.e. thermal noise).
Distortion products @ 1VRMS
Spectrum @ 1VRMS @ 1kHz
IMD 19kHz + 20kHz
Proof of mixed feedback. Loading was 100Ohm resistor (mildly inductive) + 66nF capacitance. If pure current feedback was used, the -3 dB point would be near 24 kHz, instead, it is 71 kHz so current feedback does not dominate.
In a sweet shot with a thermal camera, you can tell where the transistor’s die is. Notice a really hot, 85-degree C resistor up top.