Crappy USB solution?
Could You explain a bit further? I would really like to know about your preferred choice.
And hope it's better than <10psec jitter (measured directly on the output)- practically XO performance (not VCXO)
~5kHz-200MHz SPDIF bandwith,
24bit/192kHZ bit perfect output.
If You have a better solution for less than 100 Euro, I'm game!
Could You explain a bit further? I would really like to know about your preferred choice.
And hope it's better than <10psec jitter (measured directly on the output)- practically XO performance (not VCXO)
~5kHz-200MHz SPDIF bandwith,
24bit/192kHZ bit perfect output.
If You have a better solution for less than 100 Euro, I'm game!
That would be off topic, lets just say it doesn't involve a PC laptop and USB devices with kluges. You're a smart guy make your own choice> if you really need 24 bit/192KHz? Foobar player et al.
Last edited:
Now this would be really off topic;
No, I'm not that HD files buff at all. But by using sox or other resamplers it's a very nice possibility to have the hard calculation done in the PC with as much precision as you wish, and by switching to the max input frequency for the dac, effectively bypassing it's internal, less quality digital filters.
And the possibility to switch on the fly between filter types and listen..
Ciao, George
No, I'm not that HD files buff at all. But by using sox or other resamplers it's a very nice possibility to have the hard calculation done in the PC with as much precision as you wish, and by switching to the max input frequency for the dac, effectively bypassing it's internal, less quality digital filters.
And the possibility to switch on the fly between filter types and listen..
Ciao, George
Hi
What kind things are you doing to use 24bit/192KHz playback?
and dont you need more support from the vendor for BluRay?
I think your guess is wrong - the on-board pulse XMFR are Murata 78604/2C 2:1 - not a penny pinching option (& not necessarily worse than the SC ones). The Hiface is also made in BNC or RCA variants - it's the buyer's choice which they buy.
I don't think your analysis is accurate or fair to M2tech! In fact, Joe K's measure of <10ps at the output of the Hiface tells a different tale!
Last edited:
Well if the Murata is so great why do you think you need the attenuator then? That's not the one TI recommends infact if you look again at the white paper I linked to its overshoot looks alot like Brand X. Just when we get to the bottom of the story you want us to go in circles again. LOL
BTW ...I will Q any measurement I haven't seen the block diagram of> of what exactly equipment used and the test conditions >>and then the data interpretation. I'm an Engineer but if you want to take a single number and run with it that's your choice.
BTW ...I will Q any measurement I haven't seen the block diagram of> of what exactly equipment used and the test conditions >>and then the data interpretation. I'm an Engineer but if you want to take a single number and run with it that's your choice.
Last edited:
RF Attenuators = Jitter Reducers
So we can mark this one as not demonstrated except maybe in the case of the Hiface?
You guys have signally failed to make your case for the generality, in particular, rkeny your insistence that receive impedance could not be determined with a DMM is debunked, and you, Joseph K, have retreated from your insistence that your plots showed anything of significance and are now groping to retrieve some credibility in the case of one device, the Hiface. Your original posts mentioned the Hiface only in passing, leaving the conclusion to be drawn that the results were generally applicable. You insisted, moreover, that reflections frequently resulted, even in the case where the termination was correct, from inaccuracies in the cable impedance.
Anybody who feels the need to do so can look up the specifications of co-ax cables.
There is no doubt in my mind that if I, or somebody else, had not robustly rejected your claims, that you would have happily continued to present your plots as evidence in the general case.
Why not have the good grace to admit that you both let your enthusiasm run away with you, and apologise? If not to me, then to all the people with less expertise who you might potentially have misled.
If this post had been titled Good Cable = Jitter Reducer it might have turned out to be one of the few cable threads with some justification. As it is, up until now it's just been some kind of folie à deux (or trois, if Jocko Homo is included). At least now I have some hope it has been prevented from becoming a folie à plusieurs like so many urban myths in audio.
w
infinia, if you'll take my advice, you'll stop treating these guys as though they can be reasoned with or appeased. You have to simply oppose them.
So we can mark this one as not demonstrated except maybe in the case of the Hiface?
You guys have signally failed to make your case for the generality, in particular, rkeny your insistence that receive impedance could not be determined with a DMM is debunked, and you, Joseph K, have retreated from your insistence that your plots showed anything of significance and are now groping to retrieve some credibility in the case of one device, the Hiface. Your original posts mentioned the Hiface only in passing, leaving the conclusion to be drawn that the results were generally applicable. You insisted, moreover, that reflections frequently resulted, even in the case where the termination was correct, from inaccuracies in the cable impedance.
Anybody who feels the need to do so can look up the specifications of co-ax cables.
There is no doubt in my mind that if I, or somebody else, had not robustly rejected your claims, that you would have happily continued to present your plots as evidence in the general case.
Why not have the good grace to admit that you both let your enthusiasm run away with you, and apologise? If not to me, then to all the people with less expertise who you might potentially have misled.
If this post had been titled Good Cable = Jitter Reducer it might have turned out to be one of the few cable threads with some justification. As it is, up until now it's just been some kind of folie à deux (or trois, if Jocko Homo is included). At least now I have some hope it has been prevented from becoming a folie à plusieurs like so many urban myths in audio.
w
infinia, if you'll take my advice, you'll stop treating these guys as though they can be reasoned with or appeased. You have to simply oppose them.
The attenuator deals with 2 things - the knocking down of the high SPDIF signal (~2Vpp) & as a by-product, the reduction in reflections.
I'm not sure how the attenuators will help with the overshoot?
If using the SC XFMR are you saying that these issues would not exist? I believe the high signal would & the reflections are due to mis-terminations so the SC XFMR would not help here. Maybe the overshoot but are you sure this is from the XFMR or the SPDIF driver?
BUT you may be correct that the Murata XFMR is pants?
I agree that no values should be taken at face value (but I trust Joe K knows what he's doing as I've seen his measurements & test set-up in the past!). I also agree that a single jitter figure is usually meaningless but when it's this low the spectrum analysis of the jitter is also going to be fairly impressive, don't you think?
Last edited:
Jeez, waki - why not read the first post & stop making up your own interpretation of what this thread is about. The rest of your post is best ignored as it starts out with this false statement!
You know it was highlighted in bold then but you are determined to ignore it so I'll quote it again here:
DO YOU UNDERSTAND WHAT I'M SAYING THEN & NOW?
Last edited:
Maybe the only way to find out out about the SC part is to spend the extra $12 here and let your ears tell you. LOL
Well let's just say that by only showing time plots of NOT random data proving a case for clock recovery improvement does not help my confidence in someones testing ability. If he showed a phase noise plot with a block diagram I'd be really be impressed. EDIT> Alas but this will never happen because this won't help YOUR cause one iota . Before I was curious but now I'm getting sarcastic in my responses. Must be getting near the end of this thread.
Last edited:
No I've already got a Newava 1:1 XFMR 🙂
As he said already - got a Wavecrest to loan him?
Disagree. Low interwinding capacitance=wrong approach.
Connect low-interwinding-capacitance-high-leakage-inductance-low-bandwith-pulse-transformer to TDR.
yes wrong approach for return loss, but quite right approach for noisy PCs SMPS, USB power on Tx clocks and using singled ended connections. Get a spectrum analyzer or BER tester.
@wakibaki
bellow is digital input of actual product. What can be measured with DMM?
75R resistor? Is everything OK or something wrong with this circuit?
Please teach me, how to measure this with DMM.
My DMM will go only to 10, maybe your DMM can go to 11?
Attachments
Last edited:
That IMO will unlikely be an improvement,
the TI chip mostly won't like driving that low of impedance also making Rds (a variable) more of an influence on Zo, also increases CMOS ground bounce it's the whole reason for using a XFMR in the first place (CMRR ie reducing ground noise).
Why would you expect this approach to be better than the TI reference design?
Nothing wrong here stormsonic, just measure with the DMM. The capacitor isolates the rest of the circuit at DC and the 75R will swamp everything else. You put one probe of the DMM on the centre conductor and the other on the outer. If your DMM only goes to 10, time to get a new one.
w
Ask me a hard one...
w
Ask me a hard one...
Sure I understand what you're saying.
'Hence they can only be used on a cable where the SPDIF signal is higher than normal & can bear some reduction & not drop below SPDIF spec.'
But this would still only make them useful in the case where reflections had been demonstrated to be degrading the reception.
Recall that transformers are not normally employed in consumer SPDIF equipment and that co-ax cables are commonly specified to +-3 ohms or better.
Let's suppose that the terminating resistor at the receive end is a 1% with an unfavourable coincidence with the cable; i.e the cable is 72 ohms (to get the worst case at the transmit end) and the resistor 76 ohms. The coefficient of reflection is (76-72)/(76+72) (just taking the absolute value) or 4/148 = 0.027027.
Now suppose the termination at the transmitting end, taking your own value for Rds(on) of 18 ohms, plus a 1% resistor of 303R, thru a 2:1 transformer is 80.25ohms. This gives a coefficient of reflection of (80.25-72)/(80.25+72) = 0.0541.
So that by the time the reflection has returned to the receiving end its value is 0.00146 of the amplitude of the original incident edge, or 1.5 thousandths worst case, to put it in round numbers.
Of course this could still be causing jitter. Without measurement of the received clock jitter both in the presence of this level of reflection and without, it is impossible to say. On the other hand, it seems extremely unlikely, and against that we have the assertions of a guy who didn't know that you could measure the receive termination with a DMM.
So we can mark this one as not demonstrated except maybe in the case of the Hiface?
Perhaps you know some other equipments where the SPDIF signal is higher than normal?
w
As I told Joseph K, it's not necessary to SHOUT and doing so does nothing for your credibility.
'Hence they can only be used on a cable where the SPDIF signal is higher than normal & can bear some reduction & not drop below SPDIF spec.'
But this would still only make them useful in the case where reflections had been demonstrated to be degrading the reception.
Recall that transformers are not normally employed in consumer SPDIF equipment and that co-ax cables are commonly specified to +-3 ohms or better.
Let's suppose that the terminating resistor at the receive end is a 1% with an unfavourable coincidence with the cable; i.e the cable is 72 ohms (to get the worst case at the transmit end) and the resistor 76 ohms. The coefficient of reflection is (76-72)/(76+72) (just taking the absolute value) or 4/148 = 0.027027.
Now suppose the termination at the transmitting end, taking your own value for Rds(on) of 18 ohms, plus a 1% resistor of 303R, thru a 2:1 transformer is 80.25ohms. This gives a coefficient of reflection of (80.25-72)/(80.25+72) = 0.0541.
So that by the time the reflection has returned to the receiving end its value is 0.00146 of the amplitude of the original incident edge, or 1.5 thousandths worst case, to put it in round numbers.
Of course this could still be causing jitter. Without measurement of the received clock jitter both in the presence of this level of reflection and without, it is impossible to say. On the other hand, it seems extremely unlikely, and against that we have the assertions of a guy who didn't know that you could measure the receive termination with a DMM.
So we can mark this one as not demonstrated except maybe in the case of the Hiface?
Perhaps you know some other equipments where the SPDIF signal is higher than normal?
w
As I told Joseph K, it's not necessary to SHOUT and doing so does nothing for your credibility.
I think this WAS a hard one 😉
No offense, but I believe that the 100 ohms series resistor is effectively in parallel with the 75 ohms. The inverter creates a sort of virtual earth at it's input so the 100 ohms is effectively terminated at a virtual gnd.....
No?
jd
Ah, a breakthrough - you've opened your mind up to the possibility - this is a great leap forward
Oops, I sopke too soon - that crack of light only shone for a millisecond - ah well! keep exercising those muscles that open minds & you'll get there
QED, waki - go back to the books! Throw away your shovel now - the hole is getting too deep!
Did I ever say anything else?
Did I say I did?
Last edited:
- Status
- Not open for further replies.