Ok so I’ll have a full measurement post shortly but here’s a few tidbits
First thing is some good news and bad news.
The bad news?
The specs for the amp are wrong.
The good news?
It performs quite a bit better than the spec says.
For example Spec says that when outputting 2W @ 30 Ohm it gets about -105dB THD+N, I’m getting about -114dB
Doing a 4V in 4V unity gain test THD+N is 111dB or so. This amp does not seem to care what you load it with, it shrugs off high impedance loads even at high output as shown above. And the limitation in performance is usually just what voltage you’re actually outputting.
Swapping to ‘Class A’ / No Feedback mode, we see some pretty amazing performance for a zero-feedback design, and a nice, 2nd order dominant, descending harmonic (almost single-ended-triode like) structure.
Though worth noting that in no-feedback mode, distortion will increase more with more demanding loads than when feedback is enabled.
EQ:
The EQ settings are as shown below, shelves with upto +/- 5dB
I quite like that the bass shelf does not adjust stuff above 100hz much, I personally find that some bass shelf EQs extend too far and bloat things when you use them, I don’t get that here:
Left/Right Balance:
The Left/Right balance adjustment adjusts balance by 0.4dB each time. Every notch to the left will increase L-Channel volume by 0.2dB and reduce R-Channel volume by 0.2dB for example. This is nice because it’s really fine grained. If you have some enormous channel imbalance it’s not going to fix it, but this wasn’t meant to. The controls on this amp are meant as small, fine tuning adjustments to make a mix perfect, not to correct larger inherent issues with your setup or hearing.
Stereo Base Adjustment:
The stereo base adjustment is the most interesting thing about this amp for sure.
For context, many amps (and some DACs/software) have a feature called ‘crossfeed’. This feeds some of the left channel audio into the right channel at a lower level and with a slight delay. This is aiming to sort of emulate the effect of listening to a pair of speakers.
This is also done with a frequency dependent curve. So lower frequencies are fed into the other channel louder than higher frequencies for example.
Secondly, in production and mastering there is a technique called mid/side processing, which looks at what content of the signal in the left and right channels are different, and what is similar.
Similar content is treated as a ‘mid’ channel, with content only existing in the left or right channels alone being treated as the left and right ‘side’ channels.
The width of the stereo image can be adjusted by increasing the amplitude of only the differences. This keeps centred stuff as is but widens the perceived stereo image.
I assumed that the HM1 was doing this. It’d be pretty impressive on its own given as it’s done entirely in the analog domain with no digital processing…but it turns out the implementation is even more clever than that.
The HM1 is actually combining mid/side processing and crossfeed in a very unique way.
It first performs mid/side separation to separate content that is similar vs different. Then adjusts the amplitude of the differences as with normal stereo image adjustment, but ALSO applies some crossfeed, though only to the differences!
I’ve not seen a crossfeed implementation that follows a mid/side comparator like this. And the effect of this is that you have both an increase in width AND depth of the soundstage, but without any smearing or messing with centred content.
No wonder this adjustment sounds so good…
We can actually show this with a specially crafted test.
First, I created an audio file where the right channel is playing a 100hz and 1khz tone, and the left channel is playing a 1khz and 5khz tone.
So the 1khz tone is playing in both channels, but the other two are playing only in the left or right.
Left channel:
Right channel:
You can see some small amount of the 3rd tone in each just cause crosstalk is around -90dB or so (sort of to be expected when there’s a channel comparison circuit like this in use, still very low and almost at the point where a 16 bit file couldn’t show that content anyway). So don’t worry about that.
But now, let’s look at what happens when we use the stereo base adjustment knob.
Unaltered:
Stereo base adjustment turned up:
Note the following:
-
the 100hz and 5khz tones are now a couple dB louder than the 1khz tones. Meaning the differences have been amplified but similar/centred content has been left untouched.
-
Some of the 100hz tone from the right channel has been fed through to the left and some of the 5khz tone from the left channel has been fed through to the right.
-
Because the 1khz tone has not increased in level at ALL (checked with a more precise method than just looking at an FFT), we can also conclude that not only has the 1khz content not been affected by the mid/side level adjustment, but it also has not been affected by crossfeed as this would result in a small increase in level due to the addition of the crossfed signal.
This combination of using mid/side stereo adjustment, AND then crossfeed but only on the side-signals is really cool and if you get a chance to listen to the HM1, I strongly recommend doing so. This is the best ‘spatial adjustment’ I’ve heard, and makes normal crossfeed implementations sound like outright gimmicks.
