Review and Measurements of New JDS Labs Atom Headphone Amp

[Veri]

I know with my headphones I only use about a max of 200 mVrms, so it looks by using High Gain I can get a 5dB improvement on this metric?

And you'd think that additional amplification beyond unity gain (eg High Gain) would worsen distortion, but it seems the opposite here....so what's going on here?

Not an audible issue. Low gain might give you a more enjoyable volume pot range. You shouldn't up the gain for a 5dB theoretical performance gain, not if you need to use lower pot range to make up for it. Just use what makes sense. That's what I do

 

[MRC01]

...
And you'd think that additional amplification beyond unity gain (eg High Gain) would worsen distortion, but it seems the opposite here....so what's going on here?

It's hard to believe it has higher distortion in low gain. Swapped labels on graph?

 

[Jimbob54]

Not an audible issue. Low gain might give you a more enjoyable volume pot range. You shouldn't up the gain for a 5dB theoretical performance gain, not if you need to use lower pot range to make up for it. Just use what makes sense. That's what I do

And the lower down the Vol knob is, the worse any channel imbalance would be.

 

[Swtoby]

And the lower down the Vol knob is, the worse any channel imbalance would be.

On their current crop of amps, JDS has done a nice job of minimizing this. I hear almost no shift at very low volume with my sensitive headphones.

 

[Robbo99999]

Not an audible issue. Low gain might give you a more enjoyable volume pot range. You shouldn't up the gain for a 5dB theoretical performance gain, not if you need to use lower pot range to make up for it. Just use what makes sense. That's what I do

It's true it lowers the useable volume pot range, on High Gain I don't use it past 9:30am on my harder to drive headphones. It does seem to sound cleaner, but it's so hard to compare these things accurately. If you look at the graph the actual output of the majority of the music is gonna be around 50mVrms where you're talking about a difference between -87dB to -94dB. My K702 operates at worst case on the 3rd line up from the bottom on the following graph which is only 112mv, and that's related to peak RMS of the bass when EQ'd, so most of the music is below that and also after taking into account that tracks are not at 0dBFS the whole time:

So I'm theorising we can probably look at around the 50 mVrms section of the JDS Labs Atom distortion graph, which is where it's a difference between the -87dB to -94dB:

About the volume pot, it runs from around 7 o'clock to around 4 o'clock on the dial as it's max movement, which is about "9 hours" in total. If I run it at 9 o'clock that's about 22% of it's total travel and judging from Amir's following graph on channel mismatch vs volume pot position we can see it's only falling apart from 16 seconds through to the end of travel at 20 seconds:

That's the last 4 seconds of travel where it become unpredictable, which is the last 25%, which roughly equates to the first 25% of the volume dial.....so yes that equates up to about 9:30am on the volume dial as being the more unreliable area. So yeah, I might be giving up to about 0.5dB in channel balance accuracy by running High Gain. Marginal gains overall.....subjectively I seem to like High Gain better. EDIT: I normally use it Low Gain historically, just tried it in High Gain recently.

It's hard to believe it has higher distortion in low gain. Swapped labels on graph?

@amirm , is it possible the labels were accidentally swapped on the following graph, where it's showing higher distortion in Low Gain than High Gain?

 

[Jimbob54]

It's true it lowers the useable volume pot range, on High Gain I don't use it past 9:30am on my harder to drive headphones. It does seem to sound cleaner, but it's so hard to compare these things accurately. If you look at the graph View attachment 148292 the actual output of the majority of the music is gonna be around 50mVrms where you're talking about a difference between -87dB to -94dB. My K702 operates at worst case on the 3rd line up from the bottom on the following graph which is only 112mv, and that's related to peak RMS of the bass when EQ'd, so most of the music is below that and also after taking into account that tracks are not at 0dBFS the whole time:
View attachment 148289
So I'm theorising we can probably look at around the 50 mVrms section of the JDS Labs Atom distortion graph, which is where it's a difference between the -87dB to -94dB:
View attachment 148292
About the volume pot, it runs from around 7 o'clock to around 4 o'clock on the dial as it's max movement, which is about "9 hours" in total. If I run it at 9 o'clock that's about 22% of it's total travel and judging from Amir's following graph on channel mismatch vs volume pot position we can see it's only falling apart from 16 seconds through to the end of travel at 20 seconds:
View attachment 148293
That's the last 4 seconds of travel where it become unpredictable, which is the last 25%, which roughly equates to the first 25% of the volume dial.....so yes that equates up to about 9:30am on the volume dial as being the more unreliable area. So yeah, I might be giving up to about 0.5dB in channel balance accuracy by running High Gain. Marginal gains overall.....subjectively I seem to like High Gain better. EDIT: I normally use it Low Gain historically, just tried it in High Gain recently.

@amirm , is it possible the labels were accidentally swapped on the following graph, where it's showing higher distortion in Low Gain than High Gain?
View attachment 148292

Also, I think the noise floor will be higher in high gain so if you were using sensitive cans/ IEMs (you wont be apart from the NAD?)

I'm assuming harmonic distortion isnt an issue on amps like the Atom regardless of gain setting

 

[Robbo99999]

Also, I think the noise floor will be higher in high gain so if you were using sensitive cans/ IEMs (you wont be apart from the NAD?)

I'm assuming harmonic distortion isnt an issue on amps like the Atom regardless of gain setting

I don't know about that, but yeah the NAD HP50 is my most sensitive headphone, haven't used it for months.....at the moment I'm only using my K702 & HD560s and sometimes the HE4XX.

Regarding using more sensitive headphones though, there is a way to use say the sensitive HP50 on High Gain mode on the Atom Amp - instead I could lower the volume digitally. Effectively I'd be combing the following G6 DAC distortion graph with the exact same graph on the Atom Amp . I'm betting I could work out theoretically what is the best combination of reduced digital volume vs the supposed positive tradeoff of being able to use High Gain on the Atom Amp. I could probably also use the same G6 graph to see where I'd be at if I reduced the volume digitally enough to be around half pot on the Atom Amp - thereby negating the channel balance problems. I could probably find some online calculator of how to combine the respective figures from the graphs to see where the sweet spot is. EDIT: I think I remember something called the "10dB difference rule" when comparing SINADS in an audio chain, so I could try to make sure the G6 is always 10dB better than the amp. It looks like I could lower the G6 to -20dBFS and still be at -100dB, so I probably wouldn't want to be below that given the around -90dB I was talking about on the Atom Amp at 50mVrms. EDIT #2: -20dBFS is a Windows Slider Volume of only 26%, so there's a lot of scope for lowering the volume.....but I have to take into account negative preamps used in EqualiserAPO for EQ purposes too.

 

[MRC01]

What doesn't make sense is that in amps like the Atom that have a switch for high / low gain, that switch usually changes the resistors in the gain-feedback loop. Low gain has higher negative feedback, high gain has lower negative feedback. All else equal, higher negative feedback means: less gain, less noise, less distortion, wider bandwidth. Put simply, "the more gain, the more pain". So it's a bit unusual and suspicious that the graph shows low gain having higher distortion.

 

[Robbo99999]

What doesn't make sense is that in amps like the Atom that have a switch for high / low gain, that switch usually changes the resistors in the gain-feedback loop. Low gain has higher negative feedback, high gain has lower negative feedback. All else equal, higher negative feedback means: less gain, less noise, less distortion, wider bandwidth. Put simply, "the more gain, the more pain". So it's a bit unusual and suspicious that the graph shows low gain having higher distortion.

I thought it was strange too, which is why I originally asked how it could be that higher gain = lower distortion for the Atom Amp, I tagged Amir in one of my responses today, so hopefully he'll see it and be able to at least confirm to us that the Low Gain / High Gain labels on the graph aren't mixed up, and perhaps a theory as to why High Gain has lower distortion on the Atom Amp. I'm just curious to optimise my G6 DAC output levels in conjunction with the gain settings on the JDS Labs Atom (by comparing the IMD graphs I showed earlier) whilst hopefully being around mid pot on the Atom to have best channel matching for the various headphones I own. High Gain seems to sound a bit cleaner though, hard to compare, and certainly not level matched (roughly subjectively level matched yes).

EDIT: I did just find this just now, Amir has tested the Atom Amp in another thread to compare Low Gain vs High Gain:
https://www.audiosciencereview.com/...s/study-headphone-amp-gains-low-or-high.5495/
He found High Gain reduced Sinad by 10dB, and that was by raising the noise floor.......but there's IMD and noise floor, they're seperate things right, I'm not an expert on distortion. So which is more important for sound quality: noise floor or IMD?? And it's IMD that seems improved through High Gain on the Atom Amp.

 

[Jimbob54]

I thought it was strange too, which is why I originally asked how it could be that higher gain = lower distortion for the Atom Amp, I tagged Amir in one of my responses today, so hopefully he'll see it and be able to at least confirm to us that the Low Gain / High Gain labels on the graph aren't mixed up, and perhaps a theory as to why High Gain has lower distortion on the Atom Amp. I'm just curious to optimise my G6 DAC output levels in conjunction with the gain settings on the JDS Labs Atom (by comparing the IMD graphs I showed earlier) whilst hopefully being around mid pot on the Atom to have best channel matching for the various headphones I own. High Gain seems to sound a bit cleaner though, hard to compare, and certainly not level matched (roughly subjectively level matched yes).

EDIT: I did just find this just now, Amir has tested the Atom Amp in another thread to compare Low Gain vs High Gain:
https://www.audiosciencereview.com/...s/study-headphone-amp-gains-low-or-high.5495/
He found High Gain reduced Sinad by 10dB, and that was by raising the noise floor.......but there's IMD and noise floor, they're seperate things right, I'm not an expert on distortion. So which is more important for sound quality: noise floor or IMD?? And it's IMD that seems improved through High Gain on the Atom Amp.

You won't hear the noise floor on the Atom regardless of gain unless you're really really loud on really sensitive headphones /iem.

Whether the imd reduction is audible in listening, who knows. It's pretty low regardless.

 

[JohnYang1997]

This may have been mentioned in this thread before, but shouldn't we be using this headphone amp in High Gain Mode to get the least amount of distortion?
View attachment 148279
I know with my headphones I only use about a max of 200 mVrms, so it looks by using High Gain I can get a 5dB improvement on this metric?

And you'd think that additional amplification beyond unity gain (eg High Gain) would worsen distortion, but it seems the opposite here....so what's going on here?

This is distortion vs generator level. This means that the output voltage will be much higher at the same input voltage. Should've used output level instead. What's shown is the noise floor of AP.
You can also see the actual distortion is starting to take over after 500mV input voltage at high gain. Low gain stays clean. But again with the same input level, the output levels are different. It can be a different graph when measuring output voltage.

 

[Robbo99999]

You won't hear the noise floor on the Atom regardless of gain unless you're really really loud on really sensitive headphones /iem.

Whether the imd reduction is audible in listening, who knows. It's pretty low regardless.

I've certainly never heard the noise floor with my Atom Amp, as the background is dead silent, no hiss or any audible noise.......so in that case maybe there is an element of sense to concentrate on the measured IMD aspect rather than the SINAD aspect.....which might lead some credence to my endeavours to keep IMD at the lowest levels based on the G6 DAC IMD graph and the Atom Amp IMD graph whilst being in around a good mid point on the pot of the Atom Amp to avoid the larger channel imbalance in perhaps the first 25% of the volume pot.

 

[Robbo99999]

This is distortion vs generator level. This means that the output voltage will be much higher at the same input voltage. Should've used output level instead. What's shown is the noise floor of AP.
You can also see the actual distortion is starting to take over after 500mV input voltage at high gain. Low gain stays clean. But again with the same input level, the output levels are different. It can be a different graph when measuring output voltage.

Damn, I think I see what you mean, which means I can't use those IMD graphs in the way that I am currently....because I've been assuming any given point on that x-axis was the output voltage to the headphone, which was an incorrect assumption based on what you're saying there. Well, if that's the case then there are zero measurements to back up my subjective listening opinion that High Gain sounded a bit cleaner than Low Gain. If this is true though, then what's the point of the following graph, if it's not showing output voltage, wouldn't it be more useful if it was IMD distortion vs output voltage?

 

[JohnYang1997]

Damn, I think I see what you mean, which means I can't use those IMD graphs in the way that I am currently....because I've been assuming any given point on that x-axis was the output voltage to the headphone, which was an incorrect assumption based on what you're saying there. Well, if that's the case then there are zero measurements to back up my subjective listening opinion that High Gain sounded a bit cleaner than Low Gain. If this is true though, then what's the point of the following graph, View attachment 148351 if it's not showing output voltage, wouldn't it be more useful if it was IMD distortion vs output voltage?

You can think about shifting(remapping) the curves with high gain to the right. (500mV to 2.5V, 2V to 10V something like this) But as the input voltage is still limited to 2V for low gain there's little to compare. What you will see is just a bit higher noise when using high gain and higher output level.

In real world usage, you would hardly go pass 1V, so there really should be no difference other than a bit higher noise with high gain which you might be able to hear with ultra sensitive IEMs.

 

[Jimbob54]

Damn, I think I see what you mean, which means I can't use those IMD graphs in the way that I am currently....because I've been assuming any given point on that x-axis was the output voltage to the headphone, which was an incorrect assumption based on what you're saying there. Well, if that's the case then there are zero measurements to back up my subjective listening opinion that High Gain sounded a bit cleaner than Low Gain. If this is true though, then what's the point of the following graph, View attachment 148351 if it's not showing output voltage, wouldn't it be more useful if it was IMD distortion vs output voltage?

The graph as it is highlights issues like the liquid spark, or if IMD shot up in one gain mode on the Atom. Less useful for comparing 2 well behaved units or the same unit in 2 different modes. I think for the Atom, the old adage of "use low gain until you need more" is probably the right one.

 

[Robbo99999]

You can think about shifting(remapping) the curves with high gain to the right. (500mV to 2.5V, 2V to 10V something like this) But as the input voltage is still limited to 2V for low gain there's little to compare. What you will see is just a bit higher noise when using high gain and higher output level.

In real world usage, you would hardly go pass 1V, so there really should be no difference other than a bit higher noise with high gain which you might be able to hear with ultra sensitive IEMs.

Ah, I totally see what you mean now there. High Gain on Atom Amp is actually 4.5x though so I worked out the result of where the High Gain line would sit if equalised to the same output voltage, so I shifted the High Gain Curve to the right by a factor of 4.5 and the following is what I got where I moved a 22mv output to 100mv (a 4.5x factor), it's basically 7dB worse IMD for any given output voltage vs Low Gain:

Well, in that case, my perception of increased sound quality whilst using High Gain was either the fact that it's difficult to accurately level match when doing comparisons, and probably it was a bit louder when I was listening to my music tracks on High Gain, and also combined with a bit of psychological expectation bias. Needless to say I'll go back to using Low Gain then, as I can't put faith in my listening tests vs what we've learned here re the measurements of High Gain vs Low Gain.....that's gotta be more weighty.

 

[magicscreen]

Use 1Kohm potentiometer instead of the 10k one. Decreased thermal noise at any potentiometer position. And it decreased the input impedance of the output stage.

Will it be louder if I use 1Kohm potentiometer?

 

[Caenhorn]

Any way I could clean the pot in Atom Amp? It's not just scratchy. Right channel gets muted after adjustment quite often, and I need to wiggle it to get it back

 

[jseaber]

@Caenhorn - Hm, that's not right. Please reach out with your order number so that we can swap the potentiometer for you.

 

[Caenhorn]

@jseaber Appreciate it! Unfortunately, I'm not from US and sending it back is way too pricey. I wonder if I could do something about it myself, unless exchanging it is the only way.