The Best DAC is no DAC

Member
Joined 2007
Paid Member
Hi Ray,

As a sort of response to your question:

Through-hole parts aren't as scary to work with. Would like to hear your views - are SMD worth the extra effort sonically?

I don't think there's any single answer to this ... Depends on where the SMDs are used and what the application is. Resistor distortion e.g. are directly related to power rating and high power SMDs can be tricky to find. But in terms of decoupling an HF IC I would say SMD only (with all the potential reservations on "only"), and as close to the IC as possible. BTW much information on IC decoupling and digital layout on the internet.

Cheers,

Jesper
 
Yes, I was looking at that too; I think I'll stick with this one for now as I can use the high quality low noise 5V supplies already in place.

SN74HC74N - TEXAS INSTRUMENTS - LOGIC, DUAL D-TYPE FLIP-FLOP, 14DIP | Farnell element14

Ray
I use bog standard 74HC74 flip flop. Maximum Vcc is 7 volts. This allows me to use a 6V SLA battery (actual Vcc = 6.4V). This has 2 advantages:
1. SLA battery is already quiet. I hate building complex regulators when a SLA battery does the job already. I only use a regulator when I need a specific voltage.
2. I get a bigger voltage swing with higher Vcc. As you point out, the DSD output is lower than other sources, so the extra swing from the SLA helps.

For those who use mono flip flips (to improve separation) - do you hear, or even better, measure, any difference?
 
More generally, Using my JLSounds USB board with my muting arrangement and consistency of playback material (only ever DSD256) I am not experiencing any pops or clicks.

Ray
I'm not using any muting circuit, the only time I get any click is when I change sample rate in HQ Player.

I always play back DSD 256 so why am i changing sample rate?

For DSD or PCM files (44.1k/88.2k, 176.4k), DSD 256 = 11.2MHz.
For PCM files (48k/96k/192k), DSD 256 = 12.2 MHz.

Does this make a difference?

When I play back standard DSD/SACD file (DSD 64) at 11.2MHz, CPU load is around 43%. When I play same file at 12.2MHz, CPU load is 50%. So upsampling at integer values does reduce CPU load.
 
Yes, I was looking at that too; I think I'll stick with this one for now as I can use the high quality low noise 5V supplies already in place.

SN74HC74N - TEXAS INSTRUMENTS - LOGIC, DUAL D-TYPE FLIP-FLOP, 14DIP | Farnell element14

I've been reworking the PCB layout to reflect Jesper's feedback. One thing I keep thinking about is SMD components, I don't like hand soldering these as I'm a bit ham-fisted but you can keep everything very compact. Through-hole parts aren't as scary to work with. Would like to hear your views - are SMD worth the extra effort sonically?

Ray

when you are dealing with digital signals (or any steep edge high Di/Dt signal) then loop areas become critical as do parasitics, inductance being t6he killer...so for best signal integrity SMD and small loop areas are best, as is a contiguous (unbroken) ground plane to provide a return current path (broken copper pours are not ground planes).
 
Hi Ray,

As a sort of response to your question:



I don't think there's any single answer to this ... Depends on where the SMDs are used and what the application is. Resistor distortion e.g. are directly related to power rating and high power SMDs can be tricky to find. But in terms of decoupling an HF IC I would say SMD only (with all the potential reservations on "only"), and as close to the IC as possible. BTW much information on IC decoupling and digital layout on the internet.

Cheers,

Jesper


Most electronics are done with SMD these days including very sensitive analogue. Higher power based designs are also SMD because it does offer many advantages, including thermal engineering... For resistors always de-rate by a minimum of 50%, preferably more. There are many decent low ppm SMD resistors out there.
 
Member
Joined 2007
Paid Member
@marce:

Thanks marce for "chipping in" ;-) .. In relation to SMD components vs. other component types IMHO there's also the factor of "sound" which in my perspective may lead to SMDs not having similar qualities.

To this end please note that I'm not "opening" a discussion on this here - mainly sharing my perspective on this ...

Cheers & a Merry Christmas when/if you get to it ;-)

Jesper
 
I have SA9227 USB card. SA9227 (32BIT 384K) daughter card supports 32BIT 384K-in Amplifier from Consumer Electronics on Aliexpress.com | Alibaba Group
see last 3 images

HTB18CYvKpXXXXcbXXXXq6xXFXXXm.jpg



Does anyone know which pins to connect from card to get DSD. I do not have documentation and do not know where to get it.
 
Most electronics are done with SMD these days including very sensitive analogue. Higher power based designs are also SMD because it does offer many advantages, including thermal engineering... For resistors always de-rate by a minimum of 50%, preferably more. There are many decent low ppm SMD resistors out there.

Thanks marce et al for posting such useful information; this is a good learning experience - and I have a lot to learn!

So, I'll change to SMD components, at least for the flip-flop/filter elements.

Is there any mileage in progressing this as a group development/buy? I'm happy to keep updating and posting a PCB layout for the experts among us to critique. I'm also happy to handle a small distribution exercise.

If this is a possibility, I suggest some constraints to keep it manageable;
1. Keep the concept of it being a JLSounds piggy-back board (doesn't prevent it from being used in other situations.
2. Max 50mm square and two layers to keep the fabrication cost down
3. Retain as much flexibility for how it will be used as possible so people can try different filter approaches.

Ray
 
I can't see anything to suggest it has DSD capability?

It's DSD. First diode is DSD indicator. It lights when i play DSD. DSD playback http://hifimediy.com/download/SaviAudio-SA922x_DSD64 _DSD128_playback_on_foobar2000.pdf.

Datasheet http://download.esm-cn.com/ATTACHMENTS_201408_101635_20140829860_pdf/SA9227-v1.3a.pdf page 24 is solution to my problem. I think...

I'll try lpf r+cap. Any suggestion? Something like
An externally hosted image should be here but it was not working when we last tested it.

from AudioLinux - The audiophile realtime plug & play operative system
 
Smd is easy enough once you have solder paste, good tweezers and a small soldering iron tip. Trying without those will lead to sad times and frustration.

In layman's terms, what is the flipflop doing in this circuit? Acting as a buffer?

I heard Doede of DDDAC fame's direct dsd prototype earlier in the year and that used an lm7171 high speed opamp to very good effect. Is there a reason you're going for a flipflop and not considering something like that?
 
In layman's terms, what is the flipflop doing in this circuit? Acting as a buffer?

Flip-Flop is to derive balanced differential signals.
My pre-amp and poweer am p are all balanced so differntial out put is indeed the key driver for using a flip flop.

Depending on the flip flop you use and supply voltage, it can also act as a level shifter. My flip flop has 6.4V supply, so voltage swing is nearly double what is coming off the USB board (3.3V supply). Oh, and since the signal is now balanced, I assume that voltage is doubled again?