[Clutz]

So I'm a bit confused - output caps on the end of a headphone amp are bad because they end up creating a high pass circuit - which cuts out some of the bass frequencies. So is the reason that a we can get away with a relatively small value for the output caps in this case because the input impedence of the amp is much much higher than that of headphones, so the high pass filter doesn't cut out the bass frequencies?

Cheers,
Clutz

 

[dsavitsk]

The input impedence of headphones is generally between 32 and 600 ohms. The input impedence of an amp is usually 50000 to 100000 ohms, though they sometimes dip down to 10000.

To determine the size of output cap you need, use

C = 1/(2 * pi * Hz * R)

where C is the capacitance and Hz is the 3dB corner frequency and R is the impedence the amp sees. Don't forget that there is a resistor from out to ground, so the cap sees this resistor in parallel with the input impedence. Also, for the corner frequency, there are phase distortions for frequencies about 10X higher than the 3db point, so people generally use 2Hz as a good point, though a bit higher is also fine. Also, some feel that it is better to use a higher qualty cap with a higher 3dB than a lower quality with a lower. So, if a 4uF MKP and a 1uF FKP are the same price, you might do better with the FKP.

On the output of the Millet amp, you'll notice that Pete uses a 1K resistor while most preamps use a 50K. This is because the 1K makes no difference to a 32 ohm load, but when paralleled with a 50K input impedence it would greatly influence the choice of cap.

So, your 10uF cap is a little large, but does not hurt anything. 4.7uF is a good safe value, though it is also generally larger than you need.

Last, paralleling caps can cause problems since the caps fire at different speeds so you get weird phase issues. It is sometimes worth it to use a very high quality cap for the small one, but if the small one is not a lot better, it probably does more harm than good. Your DAC will probably sound a bit better if you take out the .56uF cap and just use the 10uF (unless the .56 is one of the new solen teflon caps.)

 

[rickcr42]

Quote:

The input impedence of headphones is generally between 32 and 600 ohms. The input impedence of an amp is usually 50000 to 100000 ohms, though they sometimes dip down to 10000.

you can probably make your life easier if you drop 600 ohms entirely from your requirements list and limit it instead to 20-300 ohms.
I not not yet met the 600ohm can that could be properly driven by anything less than a full blown power amp.It will play but not even close to the way it should sound

 

[Alf]

My thoughts on output capacitors:
* The best electrolytic capacitors are better than mediocre/standard film capacitors.
* Good film capacitors are big and the best film capacitors are huge.
* The difference between best electrolytics and practical film capacitors is not that big.
* Adding film capacitors would mean increasing the board size.

If you would like use the best there is, nothing stops you from using your favourite film capacitors. Yes, you need a bit of creativity and a bigger enclosure but it is doable.

Also the DAC is not meant to be used completely standalone. There is always going to be an amp to connect to. Many amps have some kind of input DC protection circuit. If this is the case, you do not need CL/CR at all. You can either jumper them or connect output wires to the capacitors’ plus pads.

After a lot thinking on the subject I decided to leave electrolytics on the board. 47uF is overkill for the application but it is very common and does not hurt anyway. So what we have is a cheap and compact solution with very decent sound. This is a fallback strategy if nothing else is available.

 

[MASantos]

So what do you think will be the best solution, using the CL/Cr caps or using the MKT imput caps in a PIMETA amp?

Alf did you read in that TI text regarding the pad format(using a "rounder" approach to the cap) near the coupling caps? What do you think about applying that in the board? I was also intriged regarding mounting the caps vertically...

Manuel

 

[Alf]

Quote:

Originally Posted by MASantos So what do you think will be the best solution, using the CL/Cr caps or using the MKT imput caps in a PIMETA amp?

It depends on what you can get for your CL/CR and your personal preferences. Just choose the ones you like better.

Quote:

Originally Posted by MASantos Alf did you read in that TI text regarding the pad format(using a "rounder" approach to the cap) near the coupling caps? What do you think about applying that in the board? I was also intriged regarding mounting the caps vertically...

If you mean decoupling caps from the piece about proper decoupling, the way I understood it is a bit different. They suggest putting vias as close to the corresponding SMD pads as possible and ideally integrate them into the pads.

If you look at the layout pictures I posted previously, you can see that this very approach is used for all decoupling caps.

 

[Alf]

Last minute change:
* Moved traces connecting X1 and IC1 to the top.

Board v1.0 Rev 5 – All without bPlace

 

[MASantos]

Quote:

Originally Posted by Alf It depends on what you can get for your CL/CR and your personal preferences. Just choose the ones you like better. If you mean decoupling caps from the piece about proper decoupling, the way I understood it is a bit different. They suggest putting vias as close to the corresponding SMD pads as possible and ideally integrate them into the pads. If you look at the layout pictures I posted previously, you can see that this very approach is used for all decoupling caps.

I read it again and understadn it now! *Sits back and thinks to himself:looking at the pictures is not enough...

 

[drbig]

Alf, any chance of getting in on the next prototyping stage for this? I could provide feedback on the build process (from a novice viewpoint) and my impression of SQ ( for what it is worth).

 

[Alf]

I ordered 15 PCBs from www.eurocircuits.com today. I expect the boards to be delivered to me around 3rd of April.

As before, I would greatly appreciate it if you could help me with prototyping and provide feedback on building process and sound quality. Please remember that this is a prototype. It might not work at all. So if you decide to participate, you do it at your own risk. However, this is a great opportunity to lay your hands on the new design in about 3 weeks. The next batch of boards (group buy or another prototype) will not be around until mid May the earliest.

The price will be £6.20 per board. You can have max 2 boards at this stage. P&P will be £2.5 to all destinations. This now includes PayPal charges (I had to upgrade my account to accept CC payments).

If you are interested, please PM or email me specifying the number of boards you would like to order and your PayPal email.


EDIT: The PCBs are no longer available

 

[MASantos]

Quote:

Originally Posted by Alf I ordered 15 PCBs from www.eurocircuits.com today. I expect the boards to be delivered to me around 3rd of March.

Do you mean 3 of April?

I don't know if I will be able to participate in this second prototiping stage, could you reserve a board for me and I will let you know afterwards

Manuel

 

[Alf]

I updated the support web site:
* Overview
* Parts List

The new version of the DAC seems to be cheaper to build. If ordered from Farnell, the parts cost approx £16-£17 of which ICs cost £8.5. Of course, you need to add relevant taxes, the cost of PCB and enclosure.

 

[guzzler]

Put me down for another Alf, i'm on holiday just in time

I'll get some pictures of the build while I'm at it too for the construction guide

 

[MASantos]

ALf I was reading through note 4 in the parts list and you speak there about powering the dac directly through USB. Wouldn't this be really bad in terms of noise?

Do you think that there will be a audible difference between using an external PS and overclocking the chip to 5,4v and regulating the voltage to 4,5v using USB power?

I understand that this question is difficult to answer before assembly and listening tests, but from a teorical point of view what do you think?

Manuel

 

[Alf]

Quote:

Originally Posted by MASantos ALf I was reading through note 4 in the parts list and you speak there about powering the dac directly through USB. Wouldn't this be really bad in terms of noise?

Well, it is not direct. It is rather unregulated. There is CLCLC filter that takes care of some noise. In theory, you can select values of C1, C2 to tackle the noise from your computer in the most effective way and make it a non-issue. We will see how it performs in real life.


Quote:

Originally Posted by MASantos Do you think that there will be a audible difference between using an external PS and overclocking the chip to 5,4v and regulating the voltage to 4,5v using USB power? I understand that this question is difficult to answer before assembly and listening tests, but from a teorical point of view what do you think?

I expect RMAA to show < 2-3 db difference for the main parameters. Even the worst figures should be over 90. I cannot hear this difference. Someone might.

Important point. The DAC is going to be connected to an amp. The worst component of this combination dominates. You need to have a VERY good amp to make this DAC the weakest link. It may be that I cannot hear the difference because my PIMETA is just not up to the task.