[joneeboi]

A guy takes a year and half hiatus and you go ahead and change everything on him?
Presenting the Carrie Amp v3.
 


 
Yes, I'm back, and I've been silently toiling away on a new design for the past few weeks. Wow, it's great to see so many new and exciting projects here on head-fi; the DoodleBug seems relevant. Carrie Amp v2 was a great success for me, seeing where a little ingenuity and a lot of hard work could go. Now I'm back with another product, a new approach to the same problem.
 
This product essentially boils down to what I'd like to use at work. The original Carrie was made for my days at university, studying in a loud area with my laptop's broken headphone out jack. Now, it's being used for my cubicle-dwelling job as an engineer, where I sit all day and listen to music. (Office Space is starting to lose its humour as it approaches my reality.) I'm not paying tons of attention to what I'm listening to, so it's not the worst thing in the world if there's no audible difference. The world of computer ground noise is deep and mysterious, so we'll see how this bad boy performs out in the wild.
 
Some highlights of the design:
 
- Separate power supplies and grounds as much as possible. With the potentiometer and output jack/termination style, I don't really have much of a choice here. Let's see where the limit of the noise and intereference is, especially with a full-on solution like the DoodleBug. This will allow some side by side listening tests with a Carrie Amp v2 with its Mini^3 amplifier circuit. This duality also doubles the component cost, but a person could conceivably populate one power supply and jumper it to the other channel.
- Dual single CMoy design with rollable opamps. Carrie Amp v3 is my learning vehicle (school bus?) in power supply and PCB design. I want to have a great sound for my 650s and SR60s, but I'm not sure yet which opamps will get me there.
- Through-hole components where possible for easier build.
- Lowest noise components where possible, eg. TLE2426 SO8 package, LP2985-10 VREG with bypass capacitor.
- Still the same board size for compatibility with the DACs you know and love, ie. bantamDAC, GrubDAC, and SkeletonDAC.
 
You may notice some changes/oddities about the design as well.
 
- Two single-channels instead of three. I'll admit, I got caught up in the O2 Amp hullabaloo, and it got me thinking about the circuit design. In the discussion, it came up that crosstalk could only be mitigated by fully separate and bipolar power supplies and proper grounding. Any non-flamey input on this matter would be greatly appreciated.
- Not symmetrical design. This could lead to weird imbalances between channels in impedances and interference. Will it sound different? Let's find out.
- Connected input ground of U1 with the output ground. I know this seems like taking two steps forward and anywhere from one to ten steps back in terms of noise mitigation. This seems necessary to creating a +/- 5 V power supply, though I haven't tested if the output voltage would float away otherwise. Yay for prototyping.
 
Still to do:
 
- Fix TLE bypass cap connection
- Add back VALUE layer for U1 and U3
- Add SO8 opamp underneath U4 a la PIMETA v2. Requires rerouting VR- to C10R-.
- Reposition components to not cut off ground plane
- Add LED to the board
- Move output jack and potentiometer before final FPE plate design
- Finalize BOM
 
This is still in the prototyping stage, so I need nine volunteers to help me test this sucker. These are the problems I hope to answer in the prototyping stage:
 
- Will the output ground provide adequate current? I plan on manufacturing the boards with 4 oz. copper, but just from the looks of the layout, it seems grossly imbalanced towards the right channel in terms of signal pollution. Do I rip it all up and go with a star ground?
- What kind of noise can we expect from the DAC's output ground? Does a person need to simply stick a DoodleBug in the signal path and simplify the rest?
- What is the difference in audible and measurable noise between using one power supply and two?
- See how difficult it is to solder the SMD portion is around U1's output portion.
- Find which opamps work best with 650s and SR60s
 
Discussion
 
In my design, I see two competing ideologies: "just listening" and aiming for purity of electrical design. The classic "fun vs. accurate" dynamic, convenient vs. objective. For the past year and a half, I've been playing in the fun side of the arena.
 
Where have I been all this time? Just listening to music. I'm horrible for relistening to the same song ad infinitum (Clarity by Zedd is on repeat one in my car), so I spent the time exploring new styles and services. I have iTunes Match and Rdio, which have made music discovery much easier. I used to listen to WAVs, and now I stream 320 kbps. It's not the gold standard of audiophilia, but I've stopped listening to the equipment in favour of the music. I got my 650s almost two years ago, and they've enabled me to stop imagining audible changes and listen. My modded SR60s balance out the audio equation with the help of my PortaPros for my walk to and from work. I equalize my car stereo for extra bass boost, and I love the sound of my EarPods on my iPhone. I bought a Pioneer A4 for AirPlaying from the iPhones, iPads, and MacBooks in our home (dat subwoofer), and my two bookshelf speakers in the home theater cover the other end of the home when AirPlaying to the Apple TV 2 or XBMC on the Raspberry Pi. I live in a 500 sq. ft. studio, so it's not hard to reach our music.
 
Sounds pretty rosy, eh? So why am I back? I'm bouncing back and forth between fun and accurate because I learned that each leads to the other. I did the portable battery powered amp via line out dock thing, both on the Creative Zen Vision:M and iPod ecosystems, and I did the diyMod thing. I organized meets, rolled some opamps in higher fidelity amplifiers I built, and openly scorned low quality audio online and to friends. After a while, I kinda got sick of it all, and then I just listened to my music. Now I'm bored of that, and the pendulum is swinging back towards accuracy.
 
The design isn't perfect. I'm still learning, and this is still the prototype stage. I'd love to get some feedback from all the much more talented designers and builders here at head-fi. (That's you!) It's a hypothesis, so please let me know if there are any glaring oversights or areas of improvement.
 
tl;dr New design. Need nine prototypers.

 

[Avro_Arrow]

I'll test one for you...

 

[Avro_Arrow]

Why not use use the DCP010512DBP and get rid of the TLE?

 

[Forte]

Welcome back.
Love my Carrie Amp, great to see this being developed further.

 

[joneeboi]

Excellent question, Avro. That's a vestige of the OPA690's power supply maxing out at +/- 6.5 VDC, hence why I kept the voltage low in v2. Let me get rid of the TLE, but now I need to reconsider the VREG. Hm. Great feedback.

 

[Avro_Arrow]

Here is something to consider...

 
DCP010512DBP for power conversion (these can be stacked for more power)

 
317/337 regulators with the voltage set by zener diode (much quieter than standard implementation)

 
LME49990 followed be OPA633 buffer

 
That will give you two ground centered channels with 100mA output ability.

 

[Avro_Arrow]

  That will give you two ground centered channels with 100mA output ability.

 
I guess 100mA is pretty optimistic.
 
The DCP010512DBP will run out long before that...
 

 

[MisterX]

Replace the TLEs with this---->
 
http://cds.linear.com/docs/en/datasheet/3260fa.pdf
 
You would no longer need an isolated boost converter but the thermal pad needs to be soldered to the PCB.

 

[Avro_Arrow]

  Replace the TLEs with this---->

 

 
 
That looks like a nice part but I see two problems.
 
1. Needs an oven to solder.
2. Does not boost voltage, only inverts.
    with only 5 volts in, you will have +- 5 volts less LDO.

 

[MisterX]

1. Needs an oven to solder.

 
Extend the pad on the PCB out past the edge of the IC and it can be soldered with an iron.
PITA to remove with an iron though.  

 

[joneeboi]

That's an interesting chip, MisterX. From what I understand, I could get +/- 4.5 V roughly from the LTC3260. However, now that I'm ripping up the PSU altogether, I'm thinking we could go with +/- 15 V, which is a category of DC/DC converters that doesn't have a deep bench. For now, I'd like to keep the option of isolated grounds on the table because I'm not certain a simple cut down the middle of the ground plane will be enough for noise and/or headphone return current. Getting rid of the TLE will give me considerably more room, so that's quite exciting already. The DCP010515DBP looks to fit the bill, but I'll take a look at some others.
 
Based on my headphone collection, I figure the most current I'd need for my SR60s is 22.3 mA. How's my math?
 

SR60i sensitivity is 98 dB, and worst case, let's go with 110 dB max output.
 
dB = dB_1mW + 10 log₁₀(P)
10^{(110 - 98)/10} = P
P = 10^1.2
P = 15.85 mW
P = I²*R
I = (0.01585 W/32 Ohm)^1/2 = 22.3mA

 
The TI DCP 15 dual output chip can supply 33 mA, which gives 50% more headroom than the worst case listening volume. The LME49990 can output 26 mA into 600 Ohm at +/- 18 V, so that would be a good candidate for what I'm looking to do. How it pairs with the 60s, 650s and EarPods will be up to the old earballs, but I still have plenty more research to do on this front.

 

[MisterX]

That's an interesting chip, MisterX. From what I understand, I could get +/- 4.5 V roughly from the LTC3260.

Yeah, if you were to power the IC directly from a +5 volt supply but that is not how I suggested using it.

 

[00940]

I would suggest a visit to murata website. They have quite a few interesting dc-dc converters providing bipolar suplies.

I would also suggest scrapping the regulators. They are not doing anything useful. Rather put decent LC filters after the switching converters.

 

[Avro_Arrow]

I'm not so sure I would use LME49990 to drive headphones directly.
I've used LT1363 to good effect in battery powered CMoy application
and am pretty happy with the sound.
 
I'm with 00940 as well, the regulators may just be wasting voltage.
We want to squeeze every last bit of power out of the DC/DC converter.
 
In the end, it's your design and you have to be the one who is happy
with it.

 

[joneeboi]

Thanks for the feedback, guys. Back to the drawing board for me. I thought the regulators would be necessary because of variation in voltage vs. load. Why wouldn't that apply here?