Zähl H1 Review & Measurements

 

Introduction

The Zähl H1 is a headphone amplifier built by Michael Zähl, who is most well known for his mixing desks used by artists such as The Chemical Brothers, Nils Frahm, and Aphex Twin. It follows the HM1, using the exact same amplifier as its bigger brother, but now designed as a pure headphone amp without the ‘extras’ that many users might not need or want.

Design

The H1’s design just like its bigger brother follows two main principles; everything analog and linear, and no compromises.

The H1 uses a pure class-A amplifier design, with no switching components anywhere, the ability to run without any negative feedback error correction if desired (and outperforming all other zero-feedback amps I am aware of whilst doing so), and delivers what to my ear is the most stunning example of transparency and a genuine reference sound in a headphone amp.

Besides being slightly slimmer than the HM1, the H1 also now has the power supply transformer internal within the unit, rather than in an external box. This was made possible because the H1 does not need to include all of the quite complex preamplifier, routing and EQ circuitry that is packed within the HM1, and despite an internal transformer usually meaning a higher likelihood of mains noise pickup, the MU-metal shielding seems to be extremely effective as mains leakage is kept entirely below -150dB. This transformer also has three independent windings, one for the left channel, one for the right, and a third dedicated to the front LEDs and logic circuitry so that none of it is sharing a supply with the audio signal path itself.

The rectification, filtering and voltage regulation for each channel is also completely separate, to ensure that sudden power demands from one channel do not impact the potential amount of energy available to the other channel.

Just like the HM1, the chassis is comprised of a beautifully finished pair of matte-black endplates, and a space-grey coated main body that has a slight shimmer to the finish. Pictures don’t quite do it justice, but the overall level of premium feel on this amplifier without going towards any sort of flashy aesthetic is fantastic.

The H1’s key difference from the HM1 is as mentioned the streamlining of the design into a dedicated headphone amplifier without the extras its bigger brother offers. This means that there is only one input on the rear, and no preamp or ‘A-Thru’ outputs.

The H1 also adds a -10dB gain setting, which in combination with the roughly 5dB lower noise floor for very low level outputs, could make this amp quite appealing for listeners using extremely sensitive IEMs. The amplifier left/right inputs are reversed compared to most amps, so check to make sure that you’re hooking it up correctly!

On the front are the controls, including the typical power and volume controls, but still retaining the left/right balance adjustment, the amplifier feedback toggle, and much to my own delight; the stereo-base adjustment. This was a favourite feature of mine on the HM1 and so I’m very glad to see it has made its way onto the H1!

Another ‘extra mile’ design element is hidden within the volume controls themselves. When either volume knob is turned down to minimum, a relay shuts off that input entirely to prevent any noise or residual signal level coming through and keep things absolutely silent, regardless of whether the input is actually enabled or not.

To the left of the volume controls is a knob with two options; ‘Class A’, and ‘Class A + Servo’. The H1 like the HM1 is a full class A amplifier, and does not use an AB design, nor any form of sliding bias. This means that it is effectively running at full tilt all the time, and any energy not used to drive the outputs is converted to heat. As such, fairly significant cooling is needed as the HM1 does pull around 50 Watts from the wall at all times.

Whilst many would say that Class AB done properly is just as good as Class A but without the drawbacks and challenges associated with the extra power consumption and heat output, others feel that Class A is still the way to go and won’t let a little extra on their power bill get in the way of their music.

Given as the H1 follows that ‘no compromise’ approach, Class A was the only way to go here. But what’s interesting is that whilst most amplifiers will use negative feedback (a method of error correction in amplifiers) to achieve higher performance, and this is usually inherent to the design and cannot be disabled, the H1 just like the HM1 gives the user the choice.

Selecting ‘Class A’ takes a purist approach. Full Class A, and absolutely no feedback or error correction. Whereas ‘Class A + Servo’ still keeps the amplifier in Class A, but enables feedback for better objective performance.

Whilst some may find that one of these modes is their outright favorite, personally the reason I love it so much is simply that different headphones sound their best on different modes. And I’ve found myself changing this setting depending on which headphone I’m using.

Hifiman Susvara? Servo all the way. But HD800 or ZMF Atrium, no feedback gave me a better experience. And having the ability to switch is fantastic.

The H1 is impressive in just how well it performs without any feedback. Because getting good performance without it is quite tricky, there aren’t many other zero-feedback amplifiers to compare to, but the Enleum AMP-23R and KG CFA3 ZF both get around 60-80dB of SINAD, falling far short of the H1’s over 100dB. In fact there are plenty of amplifiers that DO have inherent feedback which don’t touch that.

Turning the feedback on however bumps it up to truly top notch performance, again now outperforming most other Class-A headphone amps I’ve been able to test such as the Luxman P100 Centennial. The only other Class-A amplifier I’ve tested which got to similarly low levels of distortion was the Holo Bliss, and that achieves that result through quite a bit more complex of a signal path, though it is also still one of my all time favourite amplifiers.

Then we come to the feature I was worried may have had to be left out on the H1 and I’m very happy to see it was retained; the stereo-base adjustment. This can be described simply as a ‘soundstage knob’.

Many devices have a ‘crossfeed’ feature, which feeds some of the signal from the left channel into the right (and vice versa), but with a slight delay. This is done to try to emulate the effect of a pair of speakers in front of you, where each ear can hear each speaker but with different amplitudes and delays, rather than headphones where the left ear only hears the left channel and the right ear only hears the right channel.

This usually has the effect of pushing things out in front of you a bit and can certainly help with tracks where elements are hard-panned to the left or right, but I wouldn’t call it an ‘increase’ in soundstage and usually isn’t something I find myself utilizing unless I’m listening to a track with hard-panned elements, usually older recordings.

But this knob is not just crossfeed, it’s quite a bit more clever than that, and it does a multi-step process to alter soundstage.

Firstly, it performs a ‘mid/side’ comparison, which looks at what content from the Left and Right channels is similar vs different, and makes adjustments to only the content that is different in each channel.

The result is that rather than having things pushed out in front of you like crossfeed, the soundstage is expanded or contracted in an extremely convincing manner. It separates elements in the mix more than they already are and the effect is honestly stunning. Centered elements remain untouched and don’t get smeared or blurred, and elements in a mix that are off at an angle don’t get brought out in front loads like with crossfeed, but simply placed further away, with a slight change in angle depending on how much wider/narrower you’re making things. I did not realize how badly I needed this until I had it, and now find myself listening to almost everything with this set to the +1 setting.

There are two levels of increase available, and two levels of decrease. Allowing you to expand or shrink the stage to suit the mix or keep it in the middle to pass things through untouched. You can also turn it all the way down to output mono.

Whilst I do tend to find myself using the +1 setting for most of my listening now, it’s quite interesting in that this feature really is not one with a ‘best’ setting, and is incredibly track dependent. Very well mastered and produced tracks tend to sound their best with no adjustment at all, and can get a bit overstretched if you push it up too much. Some tracks seem to have somewhat exaggerated soundstage in the first place, and do sound better when you bring things in a bit.

I find myself changing this setting quite frequently and nearly on a track by track basis. It’s brilliant to have this at your fingertips and works VASTLY better than typical crossfeed only implementations. And once again, this is ALL being done entirely in the analog domain, no DSP whatsoever.

If you’d like to hear my full thoughts on the sound of the H1, check out the video review! But for now, lets see how it measures.

Measurements Include:

• THD+N (SINAD)
• Bandwidth
• THD+N vs Frequency
• Power
• Output Impedance
• CMRR (Common Mode Rejection Ratio)
• Power On/Off Behaviour (Safety Test)
• Volume Matching vs Level of Attenuation
• Noise (20hz-20khz, 20hz-96khz and 20hz-1Mhz)
• Intermodulation Distortion (IMD)
• Dynamic Intermodulation Distortion (DIM)
• Crosstalk
• Multitone
• Square Wave Output

Additional measurements and test information available in the full report

Tests conducted with standard 4V line level XLR input, and tested at 4V output, 700mV (Headphone Level) output, and 50mV (IEM Level) Output.

Test Setup

• Audio Precision APx555 B-Series analyzer
• Measurement setup and device under test are running on regulated 230V power from a Furman SPR-16-Ei
• H1 was warmed up for 6 hours prior to testing
• Balanced input and XLR-4 output used unless otherwise specified
• Exact analyzer/filter configurations for each measurement are detailed in the full reports
• CH1 (Blue) = Left, CH2 (Red) = Right

Full Measurement Reports

32 Ohm (Class A + Servo)

300 Ohm (Class A + Servo)

300 Ohm (Class A No Feedback)

THD+N / SINAD

1khz Sine, 4V input, 4V output (300Ω Load, Class A + Servo):

Just shy of 114dB THD+N here, one of the best performing class-A amplifiers I’ve measured. This is nearly identical to the performance and overall behaviour/profile on the HM1, which is no surprise given as both are using the same amplifier circuitry.

1khz Sine, 4V input, 4V output (300Ω Load, Class A):

With feedback disabled, we still get over 100dB SINAD, and our profile changes to a more 2nd order dominant profile, which subjectively tends to provide a warmer sound.

1khz Sine, 4V input, 700mV output (Headphone Level, 32Ω Load, Class A+Servo):

At medium/typical output levels for moderately sensitive headphones like a Hifiman Arya, the H1 comfortably exceeds the dynamic range of 16 bit music.

1khz Sine, 4V input, 50mV output (IEM Level, 12Ω Load, Class A+Servo):

This is where we do see a difference from the HM1. Due to no longer having any additional preamplifier and other additional circuitry in the signal path, the noise floor is reduced by around 5dB. This will almost certainly make little or no difference for the majority of headphones, but if you’re an IEM user, some very sensitive IEMs can do with every last dB in terms of pushing down noise, so this as well as the new -10dB gain option make this quite a compelling option for IEM users.

Bandwidth

The H1 has an extremely wide bandwidth. Zähl says that the -3dB point is at 500khz! I unfortunately don’t have a sine signal generator which can go that high so cannot test this directly, but can show that up to 200khz it’s completely flat.

THD+N vs Frequency (96khz capture bandwidth)

(96khz bandwidth used on the analyzer. Don’t compare this directly to standalone audible band THD+N measurements as the measurement setup is not the same.)

Power

32Ω: 3.5W

300Ω: 0.4W

THD vs Output Level in dBv, 0dBv=1V

The H1 seems to show almost no change in distortion with more difficult loads, just as on the HM1. Meaning it will perform the same even with very current-hungry headphones like the original Hifiman Susvara.

Maximum output is limited by voltage at 10.6Vrms not current.

Additionally, due to the 10.6Vrms output limit, and the +6dB gain at the headphone outs, if you have a DAC with an output level of anything up to 5.3V, you can run it with the HM1, completely max the amp volume and be certain that the amplifier is not reaching any limits or struggling to drive your headphones no matter what they are. But the 10.6V output limit does mean it’s probably not the best choice specifically for the ModHouse Tungsten.

THD vs Output Level in Watts

The H1 and HM1 are quite good examples of why the maximum power spec of an amplifier alone does not necessarily give you much information about how well it will power difficult to drive headphones.

A ‘10 Watt’ amplifier may be worse for driving difficult headphones than a ‘2 Watt’ one.

How can this be? Well let’s explain:

The maximum power of an amplifier is given as the max power it can output before reaching 1% THD+N. It does not mean it will behave or perform the same all the way up to that level. Usually even if an amplifier can supply a lot of power such as 10 Watts, it’s often showing noticeable increases in distortion at lower levels like 0.5W.

Here’s a demonstration; below is the HM1’s THD vs Output level graph in dBV, and I’ve added the THD vs Output level results for another class A amplifier that has a higher maximum output power than the H1.

As you can see in the graph, despite the fact that the other amplifier can reach a higher MAXIMUM output, it begins rising heavily in distortion far earlier. So it might have more maximum power, but driving difficult loads at higher levels, the H1/HM1 is actually going to perform better even though the max power specs might have suggested otherwise. The power spec of an amp is a hard maximum where it will either shut off or have huge levels of distortion, a little deeper inspection is required to see how well it is likely to drive difficult to power headphones.

As to why this is shown with dBV instead of Watts on the X-axis, it’s because if Wattage itself is used as the X-axis unit, then the traces shift left/right depending on BOTH output level and impedance, not just output level alone, making it extremely difficult to see how an amplifier’s performance changes for the same output but a different load unless you have a calculator handy. It gets quite messy.

The same graph as above, but plotted using Watts on the X-axis would look like this for example:

With dBV it’s also incredibly easy to see how an amp will perform for a given volume with YOUR headphones.

For example if you have HD600, that has a specced impedance of 300Ohm, and sensitivity of 105dB(1Vrms).

1Vrms = 0dBV, so just look on the 300 Ohm line, at 0dBV on the X-axis, that's the distortion your amp will have driving the headphones at 105dB SPL.

If you listen at 95dB SPL, just look at -10dBV instead. Easy!

Output Impedance

Class A: 0.8Ω

Class A+Servo: 0.04Ω

CMRR

100hz: 76dB

1khz: 75dB

10khz: 76dB

Power On/Off Behaviour

This test shows the output of the device when turned on/off, to check whether the output may put connected headphones at risk. Regardless of what this test may show, please ALWAYS disconnect your headphones when turning any amplifier on/off unless the manufacturer explicitly instructs you to do otherwise.

Turning on (from cold, 300Ω load):

Turning off (from warm, 300Ω load):

DC Offset is low, and no transient spikes above 1mV when turning off or on. So connected headphones are safe. Always good practice to disconnect headphones before turning any amplifier on/off though.

Noise

256k FFT, 3 Averages

1M FFT, 3 Averages

No squelch/idle-mute circuit detected

IMD

Dynamic IMD (DIM)

Crosstalk

Crosstalk is not as low as some other amps such as the Holo Bliss, though this is likely unavoidable given the stereo base adjustment feature which requires channel comparison.

Multitone

Square Wave Output

Thanks to the extremely wide bandwidth, square wave output is extremely clean.