[FraGGleR]

Ok, maybe my search for a DAC might go more smoothly if I understood a little bit more about the chips that make them go. I don't need to understand (nor do I think I could) too much about the specifics, but then again I know nothing so feel free to educate me.

On ebay and other places, for the cheapie USB DAC (Go Vibe, Hotaudio, etc.) they usually list just a single DAC usually the PCM2702 or maybe PCM2705. I assumed that this took the digital signal from the computer via USB and then converted it to analog for the headphone jack. Larry (aka Headphoneaddict) pointed out to me that the hierarchy of quality isn't linear, as the PCM2702E is a better chip than the PCM2706. So even if I got the schematic right, the chip hierarchy still confuses me (anyone have a list?)

But then I see something like the Ibasso D10 that generally gets excellent reviews for its DAC, along with the Pico and the Headroom Micro DAC and they list out 3 different chips, all of which are some sort of d/a coverters to my puny brain. For instance the D10 lists out:

- PCM2906, converts USB into a S/PDIF signal
- CS8416, DA-receiver chip
- WM8740, D/A chip

Ok, they are all d/a chips, right? And from the numbers and general internet research are all top notch from their prospective manufacturers. But why all three? The simpler DACs just list a single chip. Is the signal really being converted 3 times? Is it the cumulative effect of these in series that produce the great sound? The Pico doesn't list all the stages, but does say they use the WM8740. I am assuming there is more magic going on there right? The Headamp Micro uses 3 sets up chips, but they are all from Cirrus including the CS8416 that the D10 uses. I have seen the exact same chips used in the Gigawork assembled board found on ebay, which goes for $200 less (perhaps a clone?)

And then there is the newest Headstage USB DAC in a cable which describes itself this way:

The cable uses Texas Instrument's 16-bit PCM2707 USB DAC receiver which sends the digital data to the Wolfson WM8740 high-end DAC. The ultra-low noise AD8656 is used to buffer the DAC.

This thing is $60 and uses the same Wolfson chip as the Pico and the D10. It is also looks to be the size of a large tootsie roll as opposed to a little box.

I absolutely understand that there is more to just the chips that make one DAC better than another, but I don't know much about any of it. Can I even guess how something sounds based on the chips used? Now the Pico is one beautiful piece of kit and I have certainly been known to pay a premium for something pretty (and of high quality craftmanship), but $60 vs $300 is very big to me.

Gigawork vs Headroom Micro is really curious to me because of the identical chip list.

Is there a relatively simple guide to sorting through all of this? Everyone hears differently so personal opinions only go so far. Is there a way to make maybe a more consistent comparison based on chipsets? I guess I can understand a little bit since I used to build computers. I can use the same top of the line processor, but if I use a cheapo motherboard or very slow ram, I can lose some performance, but at least in my experience there, I never noticed the gaps to be that big.

I have a pathological need to find the best deal I can, so I keep thinking I can outsmart somebody (not sure who even), by finding the good chips in cheap enclosures that will compete with commercial DAC's. If I had the time, steady hands, electrical engineering knowledge, I would totally build some Frankenstein of my own, but I have none of those. So I comb the forums looking for little nuggets, but I am kind of lost. The easy answer is to throw down $500-1000 and KNOW that I am getting something really good, but I think I could only ever be happy if I found something that sounded that good for $100-300.

Can someone throw me a bone or two and explain a little bit of this to me? Why is the D10 so much better than the Headstage? What should I look for besides good reviews when I am doing my research?

Thanks!

 

[kboe]

Magic.

 

[userlander]

Great post! I had some of the same questions looking at the pico dac/amp recently. I'd like to hear some guru level info on this topic, too. Subscribed.

 

[IPodPJ]

When using higher end DAC chips, the chips don't make as much of a difference (if any) as do the power supply, analog filters and output stage.

 

[mudhole]

I have the same questions. which stage will effect the SQ in the DACs?

 

[Chris19]

Hmm, lemme give a shot at an explanation.

First, the iBasso D10, the Headstage USB DAC, and the Pico DAC are all using the Wolfson chip for the D/A conversion. The WM8740 is the only chip converting from a digital signal to analog sound in all of these DAC's.

Second, what they're doing with that Texas Instruments chip, be it the PCM2707 or PCM2906, is using it purely to convert the signal from the USB format to either SPDIF or I2S. I'm not sure what format the digital sound signal is in when it's travelling over USB, but it's not something the Wolfson chip can decode.

When converting a signal, it isn't changing it to analog and back to digital. The digital stream has a particular format to it for SPDIF, or I2S, and the TI chip is just rearranging the bits. There shouldn't be a major difference between receiver chips used.

On the iBasso, it looks like they go to the SPDIF signal first. Taking a look at the Wolfson data sheet, it doesn't look like it can understand a signal in anything but I2S. So, that SPDIF signal then goes to the Cirrus CS8614 to be converted into I2S format. Why do this? Well, the iBasso can take inputs from coaxial and optical SPDIF too. The CS8416 is a multiplexer that can accept many inputs, and put out the signal that you select. So, that's all the CS8416 does -- select which input, and allow that input to be up to 192 Khz.
*** Please note that USB audio is pretty much limited to 48 Khz sample rates, except for special (expensive) configurations ***

With that little Headstage DAC-in-a-cable, they use the Texas Instruments chip to convert straight to I2S for the Wolfson chip. There aren't any other inputs to worry about, so you don't need anything else.

Finally, the Pico does something similar again. I don't know what chip Justin uses for the USB receiver, but it's probably one of the Texas Instruments ones. It really doesn't matter. He says it's converted straight to I2S and then upsampled, then fed to the Wolfson chip. Upsampling is a whole 'nother thing so I won't talk about it here.

As far as size differences go, most of that would be for power regulation, whether or not they have their own power source, and how many input jacks you've got.

I haven't heard any of these units, but I do have a Cambridge DACmagic which uses the same Wolfson WM8740 chip. I can say for certain that any of these solutions will sound better than whatever soundcard you have right now.

EDIT -- Forgot to mention that, yes, the PCM2707 is not just a USB receiver, but has a full-fledged D/A converter in it too. It's just not being used! The chip was designed to pass through the digital data after receiving it from the USB connection.

 

[some1x]

Quote:

Originally Posted by IPodPJ /img/forum/go_quote.gif When using higher end DAC chips, the chips don't make as much of a difference (if any) as do the power supply, analog filters and output stage.

Many really high end DACs don't use off the shelf DAC chips

 

[some1x]

edit: double post

 

[DarKu]

Quote:

Originally Posted by IPodPJ /img/forum/go_quote.gif When using higher end DAC chips, the chips don't make as much of a difference (if any) as do the power supply, analog filters and output stage.

101 % Agree with that

Also some companies even make custom DAC chips

For me the most important part is the Power supply and the ouput stage.
There are different types of DAC chips too, Delta Sigma, rail to rail, etc
I like the vintage Rail to rail DAC's more

 

[msahau]

Let me share my limited knowledge.

As a Computer head (or digital head), I had also fallen to the same trap of thinking about things in the same way (e.g. chipset etc) and bits are bits.

What happens in a basic DAC is:

1) A input of digital signal
2) A processing of that digital signal
3) That digital signal is fed into a DAC to get a Analog output
4) Since the Analog output will be square wave, some filter is applied
5) That Analog signal is slightly amplified to line level so it gets transmitted to the preamp at the right level

Actually, the chips you list out are used for 1) input handling, 2) digital processing and 3) conversion to Analog

Those chips are all off the shelf and you will see even the medium end or high end one will use the same chips, does that make them sound the same? Hell no.

Audio is not like computer, it is very sensitive to a lot of things.

A few points greatly differentiate the sound:

1) Any input signal, albeit digital, introduce timing differences, which affects the sound output. The more expensive solutions has something to reclock the data, or to sync its clock with the source of digital signal. This will lead to more accurate sound playback

2) Digital processing differs from simply using off the shelf oversampling to significant DSP algorithms. In addition, digital filter is also applied before sending them to the DAC.

3) After the Analog signal is output from the DAC, it gets amplified. As the output is at really really low level, any slight distortion is highly significant. There are the basic op am to class A circuit or even tubes to amplify this sound to line level which can significantly affect the sound.

4) The circuit design, which may include, channel separation, minimize cross talk, separate digital stage and analog stage, and good power supply to ensure good amplification of signal.

These are the general points and I am sure there are others including the selection of circuit material, caps and capacitors, symmetry and minimization of signal path.

Mostly, digital circuits and computing only need to ensure it sends a as loud signal as possible as a 1 and no signal as 0. In addition, there is also numerous things in error corrections to ensure that 1s are 1s and 0s are 0s in our computing device. When we save and open word files always retrieving the same data, it is hard to believe bits aren't bits, and believe audio products should be as cheap as computing devices.

Not so, Analog processing requires not only high and low, it is everything in between that must be kept intact by the circuit. Also, Audio playback is the ultimate real time application, your amplifier is continuously outputting signal to the speaker/headphone to your ears, involving no discrete stage like your CPUs getting data. Your CPU can be notified about error correction and wait for the next clock, your ears does not have that capacity to wait. Any timing difference is directly amplified and goes to your ear, no error correction and no waiting.

It is unfortunate that these circuit design stuff are trade secrets to the manufacturers and they are likely to only outline the technology they use in vague marketing terms (e.g. upsampling algorithm, DAC chip (seems emphasized by only the very low end markets), some basic circuit design) but they cannot tell you all the details, nor are they meaningful. Ultimately, it is your ears they are trying to please and so you should trust your ears and forget all about those.

In the end, I think buying audio equipment is not buying computer, you cannot select the chipset on an internet website and get more or less what you wanted. You must audition it and see how your ears and heart respond to it before buying.

I hope this answers some of your questions......

 

[IPodPJ]

Quote:

Originally Posted by some1x /img/forum/go_quote.gif Many really high end DACs don't use off the shelf DAC chips

Well, yes and no. For example, Chord Electronics uses Pulse Array DACs in their DAC64 and QBD76 but programs their own FPGA. Fabricating a chip from scratch is not cheap and many of these relatively small high-end audio companies couldn't afford to do that.

I think Wadia makes their own chips though because they upsample to the same resolution as DSD which is 2.8224 Mhz. I know they at least program them.

 

[IPodPJ]

Quote:

Originally Posted by DarKu /img/forum/go_quote.gif 101 % Agree with that Also some companies even make custom DAC chips For me the most important part is the Power supply and the ouput stage. There are different types of DAC chips too, Delta Sigma, rail to rail, etc I like the vintage Rail to rail DAC's more

What I forgot to also mention: what usually is of benefit is how many DAC chips are used in balanced configuration, i.e. 2 (obviously the most common), 4 (dual differential), or 8 (not sure what that's called, I think stacked dual differential).

 

[msahau]

I have a few more small thing to add.

1. As can be seen from my long answer, I think you can take that is a general "No" to your question of if you can distinguish good DACs from "chipset" alone.
I will give you a more specific example. I am now using PS Audio Digital Link III, which is listed at $699. If you perform a simple internet search, the "chipset" it uses and their price (if ordered in thousands) listed on the website of the manufacturer:

Input receiver CS8416 $2.56
Sample rate converter SRC4192 $6.55
DAC chip PCM1798 $3.25

The chipset gives a total of about $12.36, out of a $699 listed price. Do you think chipset means a lot? I am sure someone can make up a circuit with these component for less then $100, do you believe they will sound the same? Chipset is nothing because any audio grade product will be using these "top end" chipsets. If you look around the ads, you will notice only the lower end products would emphasis they are using whatever brand's top DACs PMxxx blah blah blah.

2. I am sure DIY computer guy think they can outsmart Dell or HP by building your own. I am sure if you are Mr Mark Levinson, you can come up with a DAC circuit design with pure material costs of $200 which will beat the DAC I am using now. I am also sure, you can buy some cheap Chinese DAC at $200 using the same chipset, and it is *possible* it can sound better then a $1,000 DAC. The problem is, chances are, you do not have the years of experience of Mark Levinson (the expertise they are selling you in their high priced products), and that you cannot tell which Chinese DAC is actually good when checking out on the internet. I am 100% sure you can find a good deal in an unknown brandname because of the "brandname" premium, but you also lose the "insurance" of getting something with at least performance you knew (from user comment or otherwise).

In the end, you need to trust your ears. Remember audio products involve your ears and emotions, not your eyes to look at specification nor your analytical mind in analyzing what ought to be good from circuit design.

With these, I have presented all I know of this, and anyone can disagree with me, and I am sure there are guru out there who disagree point 2 above as they have the extraordinary knowledge accumulated to get good bargains for their own taste.

For me, I try very hard to forget everything I know of products, brandname, designs, marketing terms, chipsets when I go out shopping for gears. I try to tell myself to get what my ear wants within the reach of my pocket, despite what other might say about it on the internet.

Of course I have not yet succeeded in 100% doing this as everyone is affected by others and what you have heard. At least that is a theory I try to keep in mind when I do my shopping.

 

[FraGGleR]

Wow, awesome responses guys. All of it was helpful, especially explaining the signal path and why there are so many chips involved. Of course, it also confirmed my fear that if I want a bargain, I have to be willing to dish out and possibly lose enough money that maybe I should have just purchased a known product in the first place.

There is a thread about a fascinating product from Maverick audio that I am following and contributing to because I love the price point, the looks, and am interested in the tube buffer they added to their unit. $200 seems like a bargain for what they claim to offer, but will it compete with things much more costly? Will it last beyond next Tuesday? Hmmm.

Seems like the best bargain is still to get someone like MisterX to build me a proven DIY design.

Thanks guys!

PS I can outsmart Dell with my build quality and bargain shopping skills

 

[linuxworks]

my take is that it is MOSTLY about the chips they use.

why do some vendors go outside OTS chips? marketing, for one. image and appearance to have 'done hard work' is another. perhaps someone wrote some dsp code that they think is 'trick' and will turn the world around, who knows

my money is on the big co's (wolfson, etc) and to use their research and development by using their chips and ref designs.

the current crop of dac and receiver (and upsampling) chips are so good I find it hard to justify effort BEYOND simply using them.

if you use the chip competantly and don't make gross mistakes, I think the results will be nearly identical with other designs using the same chip. it really is about being a 'user' of the chip in most cases.