Small form factor, budget DACs?

Bombaywalla,
I also did some thinking about your comparison of the SN DAC and the Wadia. I'm thinking that some of the differences you were hearing had a lot to do with the output impedance of the Wadia (51 ohms) verses the SN DAC ( guessing maybe 3000 ohms). The input of your Preamp is 50K ohms I think?

The Wadia should sound more dynamic and linear..I'm guessing. The differences you described between the units. I also noticed when using or not using a buffer between my DAC and integrated. The buffer which I believe your player has built in, gives more presence and makes the musical lines easier to follow.

This is why in IMHO..unless you have a cd player or DAC with analog outputs of the Wadia's caliber or a Preamp with a high input impedance ..say 100K ohms. A buffer is mandatory to get the most out of the DAC or CD players analog outputs.

I also believe this is one of the reasons many love PC audio. Some of the sound cards used have an output impedance of 50 Ohms.

>> 04-02-06: Gmood1
>> Bombaywalla,
>> .......I'm thinking that some of the differences you
>> were hearing had a lot to do with the output impedance
>> of the Wadia (51 ohms) verses the SN DAC ( guessing
>> maybe 3000 ohms).

Gmood1, I'm having a hard time believing this. Off the top of my head, I don't know what the input impedance of my preamp is, but I think that your guess of 50K is pretty damn good one. I'll have to look in the user's manual where it is stated.
AFAIK, if the input impedance to the next stage is 10X higher than that of the prev stage, the input imp gets defined by the prev/driving stage (in this case the DAC output). Thus, both 50 Ohms & 3K Ohms are small enough for an input imp of 50K.
The SN Saru DAC+ uses Burr-Brown OPA627 buffers. I briefly looked at the TDA1543 DAC spec page & if I read it correctly, it's a current o/p DAC. So, these OPA627 buffers must be doing a dual job of current->voltage conversion + buffering. There has simply got to be feedback around these OPA627 opamps (in the wcs, it's being operated as a unity gain buffer) in which case, the opamp's buffer o/p impedance gets divided by the OPA627's open loop gain. This DC gain is usually very high implying that the (closed loop) o/p impedance must be very small (less than 1 Ohm).
A long way of saying that I don't believe that o/p imp has anything to do w/ the sound difference.

>> The buffer which I believe your player has built in,
>> gives more presence and makes the musical lines easier
>> to follow.
This makes sense - the TDA1543 DAC does not have the capability to drive the interconnect cable + preamp input in terms of creating enough voltage swing at the preamp input. It just wasn't designed for that! Hence, the need for a buffer. The component values in feedback network for the buffer need to be carefully selected so that they do not load the TDA1543 o/p. Additionally, overall thermal noise from resistors also needs to be considered.

>> I also believe this is one of the reasons many love PC
>> audio. Some of the sound cards used have an output
>> impedance of 50 Ohms.
I don't know much about PC sound cards. Somebody w/ more experience can confirm or not whether the o/p impedance is 50 Ohms or not.

However, unless I see a good reason to contradict, I believe that Hudsonhawk is on the right track w/ his hypothesis of the sound diff - the clock jitter.
As I wrote in my prev post - the Crystal Semi 8412/8414 locks onto the recovered clock embedded in the digital data stream using an on-chip digital PLL. The o/p clock from the 8412/8414 cannot be any cleaner (jitter-wise) than what is fed into it. Hence, the clock to the TDA1543 sample & hold ckt is a jittery clock (esp for badly recorded CDs). This will certainly create D->A errors resulting in "digital" sound. The more I think about this issue, the more I'm convinced that this is the issue. If there is someone out there that thinks I'm wrong, please correct me.

One thing that could be done to alleviate this issue (& higher-priced DACs like Audio Note, etc might be doing) is to create a very low jitter clock ref for the DAC (say, using the Tent XO module or something similar). It can be 44.1KHz or 48KHz or 88.2KHz or 96KHz. Then, using the Crystal Semi 8412/8414 to lock onto the embedded clock in the data stream, dump the incoming data into a FIFO at the CD transport clock rate. Then, using the low-jitter DAC clock, clock the data out from the FIFO into the DAC. This separates the CD transport noisy & jittery clock from the DAC clock. The sound o/p must improve dramatically.

Look at a sound card - I think that you'll see a clock/crystal on that PCB! it is clocking the data into its buffers from the PC hard-drive using that clock & non OS DAC is locking onto that clean clock. Hence, the sound o/p is much better. Bet you, that's what happening!

Yes Bombaywalla the 10x factor works OK. I'm not sure this is written in stone though. After listening to the differences of the average CD player or DAC output impedance (which is around 3000 ohms)verses using a separate buffer(100k ohm input and 16 ohms output at the moment) to alleviate the load of the interconnect and the amplifier. There is more dynamics,deeper tighter bass and the images are more defined. It's almost like some one took the governor off and let the engine run without it being held back.

This is using a 25K passive volume control in the loop.I should have mentioned passive volume controls. You can hear the differences very easily.

I'm not as technically oriented as you are my friend. In layman terms, it's like using a CD player that is designed to run directly into an amplifier. Then running it through a linestage/buffer before the amplifier. Most would prefer the linestage/buffer to just the straight connection. Even though there's an additional component in the loop..it sure sounds better with that buffer in between.

One means to measure output impedance of a CD player, with close results, is to have a test CD with a 1Khz signal, play it and measure the open unloaded, output signal on a good AC voltmeter. Then add a variable resistance across it, adjust it until the value is half of the open measurement, remove this resistance and measure its resistance with a standard ohmmeter. That value should be very close to the source impedance, at least at 1Khz.

I've tried the BVaudio SR10 buffer unit in the past.This was done using analog outputs not as high grade as your Wadia. The difference was noticeable. When I moved to a more substantial buffer. The difference was unbelievable! Maybe this website can explain it better than I can BVaudio . By the way my TDA1543 based DAC doesn't use Op amps at all.

Howdy mates...I also have a non os dac- the Scott Nixon Tube + ...Love it..not bad for $500, seperate power supply-a buck and quarter...Old School :)

congrats Hudsonhawk

Gmood1,
yes, 10K input imp of the next stage vs. the driving stage is not written in stone. Usually the rule of thumb is 5X-8X. However, I've found that is not high enough & that 10X works 99% of the time successfully.

OK, from your 2nd lengthier post I think I understand better where you are coming from. Let me see if I can summarize: Hudsonhawk & I were musing why these non OS DACs based on TDA1543 DACs sounded (really) bad on bad recordings. He & I were hypothesizing that it's the lousy jitter performance from the recovered clock.

NOTE: both Hudsonhawk & I have non OS DACs that use a Burr-Brown opamp buffer that drives the RCA outputs. From the spec sheet of this opamp, it has very low harmonic distortion over 20Hz-20KHz + it can drive large capacitative loads - we are talking 5nF & still have a gain-bandwidth product greater than 1MHz + it has very good settle time. So, it seems that this buffer does the job of the BVAudio SR10 & similar after-mkt buffers. Thus, I see very little advantage in further attaching an after-mkt buffer. If there is an improvement in these 2 specific non OS DACs, I surmise that it is most likely to be very little. I wish I had one on hand to give your theory a try (I would be double buffering).

Now, in your Audio Sector Premium non OS DAC that uses passive buffering, the ball-game is entirely different. As an aside, if you look @ the TDA1543 data sheet, you'll see (in Fig 1) that they have suggested the use of buffer opamps that have some bandwidth limiting (that parallel cap in the feedback). It is not the only way to "terminate" the TDA1543 output i.e. one could also use passive buffering. However, the passive buffering will rely on the TDA1543 to drive the interconnect parasitic C + the preamp input. It'll do the job (as your ears have discovered) but you know that the sound could be better (again, as you have discovered). In your particular case, the BVAudio SR10 & similar products work & show the difference since the passive buffered TDA1543 has to drive 100K & very little parasitic capacitance of the active circuitry, a much easier load. I do not think that it'll be quite the same for my SN DAC or the DAC-AH.
There is nothing magical about the 50Ohms output impedance of the BVAudio SR10. It is a standard impedance used in test & measurement equipment & by the RF engineers. It is low enough where it'll work w/ 99.9% of the equipment in the market-place no questions asked. For that matter, 600 Ohms would have worked just as well (would have been an easier load actually) as it would have been low enough to work w/ all the preamps out there.
So, you have to pay the Piper - now (opamp buffer as part of overall DAC in the same chassis) or later (use after-mkt buffer).
So, in your particular case, it appears you have 2 issues affecting the sound: insufficient drive from the Audio Sector DAC & poor jitter performance from badly recorded CDs. IMHO.