"The Wire" Ultra-High Performance Headphone Amplifier - PCB's

TNT said:

Maybe volfsin youtube link was aimed to me maybe I need it. I get -15,3 Volt but only + 1,3 ... out of my PSU V2. Sighhhhh.

//

Click to expand...

TNT said:

I use the BOM values. Thanks for the tip!! Yes, I assume it's a short. The question is - where?

Click to expand...

TNT said:

Yes I would say they are a bit crocked. Does it matter if the component reach outside the pad?

Click to expand...

agdr said:

Page 12 of the voltage regulator datasheet under "capacitor recomdendations" says the noise reduction and feedforward caps need to be low ESR and recommends (at least) X7R or X5R MLCCs. 1uF MLCCs are not available in COG, so X7R is the best you can do there. At 0.01uF the X7R and COG cost the same these days, about $0.35 USD each for 5% tolerance 50V at Mouser (leaded), so might was well go with COG which won't have termperature and voltage variances to worry about.

Also on page 12 TI recommends the 1uF for noise reduction cap and the 0.01uF for feedforward and makes the statement "the solution shown delivers minimum noise levels... and PSRR levels..." I always go under the assumption that data sheet recommendations are the best possible for any given chip, given the enormous resources the chip companies have to devote to testing, some really smart people on board, and a financial stake in getting the recommends right to sell more chips.

Click to expand...

TNT said:

How does one verify the integrity of an regulator "in circuit"? Is it possible? Out of circuit?

Are they sensitive? What could cause its life to end "in circuit"?

//

Click to expand...

TNT said:

Thanks... but what to measure for? Sound reassuring that its hard to break.

Click to expand...

jdkJake said:

So, R13/R14 and R17/R18 form a simple voltage divider. With no pot in the circuit, the gain from the L-pad is (R2/(R1+R2)) or (10k/(1k+10k)) = 0.909
If you simply insert a 50k pot in series, the equation becomes (10k/(51k+10k)) = 0.164
To get the same gain characteristic, you need to increase the value of R2. Once you do this, the relatively low default value of the R1 resistor (1k) becomes inconsequential when added in series with the 50k pot. So, using a 1M resistor for R2 yields (1000k/(51k+1000k)) = 0.951. A 500k resistor in R2 will yield 0.907 or close enough.
In any case, you get the idea. Just swap R14/R18 with a 500k to 1M resistor and bump R15/R19 to 833R (2X gain) and see what you think. I would have 435R (3X) and 295R (4X) on hand in case you need more gain. Personally, I would not go beyond that, but, that is always a matter of perspective and taste.

Click to expand...

jdkJake said:

You are going about this somewhat in the wrong way. You need to resolve the attenuation situation first and then worry about the gain of the amp. You do not want to solve gain through manipulating attenuation. You want to allow as much signal as possible through the attenuation circuit when the attenuator is fully reduced (no attenuation). Then, if still not loud enough, you want to amplify that signal with gain from the amp.
The size of the attenuator should be chosen such that it presents a load your source can effectively drive. Most modern sources can drive a 10k load, but, other sources may struggle. Hence the desire to bump up to a 25k or 50k pot. The noise level goes up a little, but, the load is easier for the source to drive. Once you select the pot based upon your source, you calibrate the L-pad to optimize the selected pot. The goal is to optimize the amount of signal allowed through the attenuator circuit. Remember, you can only subtract at this step, not add to the signal level. This is important to ensure you do not waste precious signal through excess attenuation before it hits the amp.
Note that most L-Pad circuits in this position/application are 1:10 ratio. Play with the math a bit and you will see why this works best. Again, the actual value of the pot drives the resistor values of the L-Pad. In the circuit for “The Wire”, the R13/R17 resistor can be represented by the pot and the pot alone or you can leave a smaller resistor in series, just remember they sum when used like that. The R14/R18 resistor should be sized to maintain the 1:10 ratio with the resistance in front of it. The R14/R18 resistor also provides a path to ground in the event no signal is connected (or in this case, the pot or pot wiring fails). So, for a 10k pot, the R14/R18 resistor should be 100k. For a 50k pot, the R14/R18 resistor should be 500k. Some would just put a 1M there and be done with it as that accommodates up to a 100k pot (and everything below it). Be advised, it you ever remove or change the pot, this circuit need to be revisited and possibly corrected with different resistor values applicable to the new pot (or complete lack thereof).
Once you have optimized your attenuation solution, then you can worry about the overall amp gain. I would be surprised if you require more than 2X out of “The Wire” for the vast majority of sources and phones out there. But, you can always bump it to 3 or 4 if you want more oomph. “The Wire” can certainly provide it.
Just my two cents…

Click to expand...

opc said:

Hi Guys,
jdkJake is exactly correct in what he's saying above. You don't really want to replace your pot, you just want to increase the input impedance of The Wire so your pot works correctly with it.
To do this, you just increase the value of R14 and R18 to 500k or 1M ohm and short out R13 and R17. Job done. You don't even need to short out R13/17, but I would suggest it since it's free and easy, and makes for one less series component.
As Jake said, you can look at R13/14 and R17/18 as fixed value pots on the input with a ratio of 0.909. The wire was originally intended to be directly driven by a low impedance source without a pot, so the input impedance was optimized for that exact scenario. The relatively low input impedance is easily driven by the output of most modern DAC circuits, and the low input impedance helps keep noise as low as possible when no source is connected. If you want to use a 50k pot, just increase the input impedance to work best with that. The beauty of the SE-SE and BAL-SE versions of The Wire is that you can make the input impedance whatever you want, and as long as your source impedance is low, there's not really any noise penalty to pay.
To summarize, the 50k pot you have now basically takes the place of R13/14 and R17/18 to form the input voltage divider of the amp. The only difference is that now you can dynamically change the two resistor values by moving the pot, which gives you volume control. The pot's impedance becomes the amp's new input impedance, and you're off to the races. Keeping R14 in the circuit and changing its value to 1M ohm simply gives you a safety bypass in the event your pot fails. If that happens, the 1M resistor in R14 will ensure the amp always has a GND reference at the input which will prevent DC on the output.
Cheers,
Owen

Click to expand...

neeviour said:

I am thinking of integrating the Wire SE-SE with a MCU controlled remote volume module.
The module is pretty much a pot, but by MCU controlled relays and two sets of fixed resistors. I could not reprograme the MCU so it has to be, vol = 0 (MCU logic), Rpot = 0ohm, Vol = 100, Rpot = Rmax (10Kohm or 20Kohm, etc. depends on how I configure it).
The first solution is to use this module to parallel R14/18. I should install R13/17 as 10K or so, so when vol = 0, Rin = 10K, source is not overloaded. As mentioned before, to make sure if the wiring fails there is still a signal route, I should probaly place R14/18 = 1M or so, then the pot just parallel to R14/18. Fly wire between R13/14, I think. Rmax = 100K.
Another solution is to parallel the pot with the whole Wire input (H20/23), and put a 10K resistor before them in series. Rmax = 100K, R13/17 not important. R14/18 = 1M. This is essentially the same if R13/17 = 0, a good thing is no fly wires from pads.
Which one is a better plan? Any better idea? Advices please. Thanks!

Click to expand...

Fsatsil said:

I'm putting together a digikey order for an SE-SE build and I have a question about pot selection. My main sources will be an iPhone 4s line out or an iPad Air+Nuforce iDo DAC and headphones will mainly be Grado SR225i's and Sennheiser Hd598. What value pot would work best? Also do I need to change any other resistor values if I'm using a pot? I've been reading through the thread and the wiki but I haven't been able to find the info I'm looking for. Any help would be greatly appreciated.

Click to expand...

Fsatsil said:

Thanks Neevior. I think I got it now. If I want to use a 10k pot, the pot takes the place of r13/r17 so I would jumper r13/17 and increase r14/18 from 10k to 100k. Is this correct?

Click to expand...