• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Review and Measurements of New JDS Labs Atom Headphone Amp

jackenhack

Active Member
Forum Donor
Joined
Oct 25, 2018
Messages
203
Likes
497
Location
Stockholm, Sweden
I suspect they might be using recently released TI opamp especially designed to drive headphones - OPA1622. Maybe few in parallel to increase power.

I've designed a headphone amplifier using OPA1612 as a gain stage and using OPA1622 as output buffers. Each channel has 4 outputs summoned, thus lowering the already really low noise by around -6 dB. I use TPS7A4700 and TPS7A33 low noise DC to DC converters with 4 µV/16 µV noise. Only thin film 0.1% 10 PPM resistors in the audio path. I'm currently working on my last revision with further improvements in the power supply.

Here's a picture of my current test mule.
jacken-headphone-amplifier.jpg


Only problem? I only have a QuantAsylum QA401 audio analyzer, so I can't measure it. Not good enough. Unfortunately, the BOM cost of this amplifier is totally off the wall, and with the current state of audio devices on the market, It's hard to compete.

But I have a cool gang of beta-testers.

Here's a picture when I installed the amplifier in Rammstein's studio (Brag brag!). It's currently in the mixing chain, so if the next album sounds bad, you can partly blame me! :)

Yeah, the ugly bald guy on the left is me.

IMG_0963.jpg
 

AnalogSteph

Major Contributor
Joined
Nov 6, 2018
Messages
3,338
Likes
3,278
Location
.de
Is this D class amplifier?
Nope. Traditional AB.

Class D isn't popular in headphone amps due to EMI issues (can't do BTL output!) and lower power required. IC amps that need to conserve power are usually Class G (by whatever definition - depending on where you are, the definitions of G and H may be swapped).
 

derp1n

Senior Member
Joined
May 28, 2018
Messages
479
Likes
629
Sure but let's wait for the formal release of the product.
Can we have a look now it's out? The official picture of the PCB sucks.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,376
Likes
234,552
Location
Seattle Area
Can we have a look now it's out? The official picture of the PCB sucks.
Sure. Here you go. It is a high resolution picture so click on it to see it full size:

JDS Labs Atom Headphone Amplifier Teardown PCB.jpg
 

jackenhack

Active Member
Forum Donor
Joined
Oct 25, 2018
Messages
203
Likes
497
Location
Stockholm, Sweden
It looks like the JDS amp is DC coupled. So if something goes wrong, you could have a severe DC offset at the output. It's an easy way of lowering the distortion, but it comes with a price. Most modern devices have next to no output offset, but if anything bad happens with the source, the headphones will fry. Not likely, but it could be an issue.
 
  • Like
Reactions: trl

jackenhack

Active Member
Forum Donor
Joined
Oct 25, 2018
Messages
203
Likes
497
Location
Stockholm, Sweden
It is DC coupled and has a relay on the output. John's blog only says that relay is a delayed turn on. Wonder if it also detects DC and shuts off.

It doesn't look like there is any DC offset protection circuit.
I'm curious about the eight-pin SOIC at the AC input barrel. Can't read the chip id. Any chance of checking?
 

jackenhack

Active Member
Forum Donor
Joined
Oct 25, 2018
Messages
203
Likes
497
Location
Stockholm, Sweden
But I must say that I'm impressed with the performance of this unit! The limiting factor I can see on the board is the gain stage, where the amp is pushing the limit of the JRC op-amp. Very impressive!
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,376
Likes
234,552
Location
Seattle Area
I'm curious about the eight-pin SOIC at the AC input barrel. Can't read the chip id. Any chance of checking?
Sure, it is a 78L12a regulator:

1542329252086.png
 

AnalogSteph

Major Contributor
Joined
Nov 6, 2018
Messages
3,338
Likes
3,278
Location
.de
Only problem? I only have a QuantAsylum QA401 audio analyzer, so I can't measure it. Not good enough.
Nothing that couldn't be fixed with a bit of external help and some good ol' creativity and elbow grease, I bet.

The QA401 has balanced I/O, which is a good starting point. You will have to employ differential hookup as with a speaker amp.
Its limiting factors are:
Input noise floor is -113 dBV, somewhat higher than for very good headphone amps.
Distortion is down at -108.5 dBr, noticeably short of very good headphone amps again.

To tackle the distortion issue, use a passive attenuator to get input level down. The QA401 has -20 dB built right in, so trying that for starters seems like a no-brainer. Otherwise you can always build one for yourself:
Code:
(per channel)
out sig o--[   R1   ]--+----o in +
                       |
                       _
                      | |
                      |R|
                      |2|
                      |_|
                       |
                       +----o shield
                       |
                       _
                      | |
                      |R|
                      |2|
                      |_|
                       |
out gnd o--[   R1   ]--+----o in -
I'd try R1 = 470R, R2 = 47R or something. Increase R1 for greater attenuation if needed.

What about the noise then? Well, you wouldn't have a little portable mixer with two mic inputs floating around that could serve as a preamp, would you? A Behringer 8-10 channel job would do just fine. (Their main mix outputs are impedance-balanced, too. You'd have to make some adapter cables, of course.) Set channel volume and main mix at 0 dB, tweak frequency response flatness with tone controls if needed (my Q1002USB appreciated a nominal -1 dB on the mids), and calibrate overall gain to something well in the green like 40 dB on both channels. Set main mix to -10 dB and increase mic pre to 50 dB if you want to be really sure. Then subtract the same overall 40 dB for a good estimate of real noise floor.
Estimated EIN for my Behringer Q1002USB came out as about -126...-127 dBu for shorted input using two different methodologies, which while not SOTA, still beats -113 dBV (-111 dBu) by a long shot. (The amps in a Mackie 402 VLZ4 should be even better, and that still isn't super expensive.) Internal clipping level translated to 2 Vrms at around the -10 dB mark on the main mix fader.

All of this stuff should add less than 20% to the cost of the QA401 even if you have to buy it brand new, plus some elbow grease. Easily worth it, I say. Granted, it's not as comfortable or flexible as having an Audio Precision unit, but you can make it work well enough at a fraction of the price.

Incidentally, are there any series resistors in your input circuitry? If so, I would suggest shuffling things around so that the trace following the resistor becomes a whole lot shorter. This way the longest part of this rather long trace would see source output impedance only.

Speaking of inputs, you have prominent beta testers like that and no balanced inputs?
Unfortunately, the BOM cost of this amplifier is totally off the wall, and with the current state of audio devices on the market, It's hard to compete.
I'm not surprised. Your board looks like capacitor overkill. Tons of 2200µs, 10µ film caps, and are you sure you need no less than 8 of the violet caps? Among all of this excess, do you have automotive grade ceramic caps, too? (Those would be worthwile for longevity reasons, as they are much less prone to stress cracks, which have a tendency of eventually letting the cap fail short.)
The limiting factor I can see on the board is the gain stage, where the amp is pushing the limit of the JRC op-amp.
You think that's a JRC? Doesn't look like a '2068 to me if you compare it to the two present in the buffer stage... which seemingly isn't actually a plain buffer but does have some gain of its own.
Looks like they designed the output of the LM317/337 with a capacitance multipliers. Nice stuff and keeps the BOM price down.
Love me some cap multipliers. :) The cool thing is, you can easily limit their maximum output voltage (for protection) by including a single zener diode per rail, which will make them turn into an ordinary regulator once the threshold voltage is reached. Actually I can even see a little diode in the circuit, that could be it.
Is it just me smelling a whiff of NwAvGuy? :)
This is a supplier of the O2 amp you are talking about, no surprises there.
 
Last edited:

JeffB1961

Member
Joined
Nov 13, 2018
Messages
26
Likes
14
maybe someone with some metal fab skills and equipment will be inclined to make and sell a metal housing that would do the atom justice .... maybe even make a metal housing that will hold both the atom and the weisontex ....... maybe even make a box to hold both with a switch for the atom to choose input from the weisontex or the line in so we could still use another input source ........ but i'm a total newB at all this ........ ;)
 

trl

Major Contributor
King of Mods
Joined
Feb 28, 2018
Messages
1,967
Likes
2,523
Location
Iasi, RO
Top Bottom