Lavry DA10 Inputs...which is better XLR/Optical...and best way to get it there from pc

[Currawong]

Quote:

Originally Posted by Dan Lavry /img/forum/go_quote.gif I am responding to your post, Gregorio, and this post is even more technical then the previous one. I still wonder if my posting here is too technical for most. I do not want to break the flow for the majority here. Regards Dan Lavry

Dan, some concrete technical info is welcome. I understand it's hard to explain a complex issue in a simple manner. In many threads, there's discussion of jitter, almost all by people with only elementary knowledge (such as myself). Having people around who actually know what they are talking about is very useful. Your posts help educate us and are ones we can reference later.

 

[gregorio]

Quote:

Originally Posted by Dan Lavry /img/forum/go_quote.gif Both converter types tend to be impacted more by jitter when the slew rate is higher (such as a full scale 20KHz).

Thanks for the info Dan, very interesting. I brings up a question though, from a mastering standpoint. If I were to apply noise shaped dither, as part of the mastering process when reducing a 24bit master to a 16bit distribution version, and I use a distribution algortihm, which redistributes the vast majority of the quantisation distortion (noise) to the high frequency band, is it possible this additional HF energy could negatively impact the jitter in a DAC? Would this be worse in an oversampling (delta-sigma) DAC or is the amount of additional energy we are talking about, (probably 3 or so LSBs worth at a guess) insignificant?

Cheers, G

 

[Dan Lavry]

Quote:

Originally Posted by gregorio /img/forum/go_quote.gif Thanks for the info Dan, very interesting. I brings up a question though, from a mastering standpoint. If I were to apply noise shaped dither, as part of the mastering process when reducing a 24bit master to a 16bit distribution version, and I use a distribution algortihm, which redistributes the vast majority of the quantisation distortion (noise) to the high frequency band, is it possible this additional HF energy could negatively impact the jitter in a DAC? Would this be worse in an oversampling (delta-sigma) DAC or is the amount of additional energy we are talking about, (probably 3 or so LSBs worth at a guess) insignificant? Cheers, G

I would say no and no.

As I mentioned in the earlier post, jitter has little impact on signals of very low frequency and amplitude. It has more impact on signals with high frequency and amplitude. Let me state it a little clearer - signals with BOTH high frequency and amplitude. A high slew rate (fast changing slope of voltage over time) calls for BOTH high amplitude and high frequency.

Say your signal is a 20KHz sine wave, and it is 10V peak (20V peak to peak). Then the maximum slop is around the zero crossing, and it is 1.256V per microsecond. If your jitter is late by sat 1nsec, the error is 1.256mV (milivolts). That is a lot! It costs you 4 LSB's error on a 16 bit recording. It costs you 64 LSB's on a 20 bit...

Say your signal is a 20KHz sine wave, and it is .1V peak (.2V peak to peak). The slew rate is 12.56 milivolts. It costs you .04 LSB's error on a 16 bit recording. It costs you .64 LSB's on a 20 bit...

Say the signal is 10V amplitude but the frequency is 200Hz. Again, the slew rate is 12.56 milivolts. It costs you .04 LSB's error on a 16 bit recording. It costs you .64 LSB's on a 20 bit...

Noise shaping dither:

Dither, even a noise shaping dither, is only a very small amplitude signal. It may be higher at around 20KHz, but it is still only a few LSB's so the slew rate contribution due to noise shaping dither is very small.

Oversampling DAC:

Oversampling does not alter the slew rate of a signal. It only adds intermediate sample values, but the slop of the signal is the same. Say you have a slop that is 1V per usec, and say you sample every usec. The first sample is at 0V, the second sample is at 1V, the third sample is at 2V... The slope between the first and second samples is clearly 1V per usec. If we up sample by X2, we leave the original samples as they are, but we insert additional samples every 1/2 usec. So now the first sample is at 0V, the second is .5V at 1/2 usec, the third is still at 1V at 1usec... The slope stays the same. If you up sample by say X4, both the change in voltage and the time between samples (on the slope) is halved again, so the slope is still the same.

Up sampling “fills in” more “dots” but the signal envelope is the same. The final analog filter removes the “steps”, making for a smooth slop. The analog filter has an easier life filtering many small steps (compared to few large steps), and that is the main advantage of up sampling, (contrary to much marketing hype such as "more bits").


So the jitter is not going to harm the signal due to up sampling. At least not in principle (no conceptually). If you observe such a thing, I would look for another implementation specific reason.

And also worth noting, a multi-bit sigma delta is less jitter sensitive then the older 1 bit sigma delta (as in DSD which is the engine of "supper audio CD" format). A 5 bit multi-bit modulator is less jitter sensitive then say a 3 bit. I am not advocating a 5 bit over a 3 bit, it is just one of the very many considerations.

Regards
Dan Lavry
Lavry Engineering

 

[gregorio]

Thanks Dan, that's great information. I've found it very difficult to get this sort of information in terms I can understand, ie., without the need for a PhD.

I appreciate you sharing your time and knowledge. Perhaps in the not too distant future I will be able to afford some of your top of the line profession converters. I'm currently specifying equipment for a new re-recording theatre but my budget is very tight at the moment. Many years ago I heard a dBTech Gold (or was it a Blue?), it remains in my mind the sweetest ADC I've ever heard.

Cheers, G.

 

[Dan Lavry]

Quote:

Originally Posted by gregorio /img/forum/go_quote.gif Thanks Dan, that's great information. I've found it very difficult to get this sort of information in terms I can understand, ie., without the need for a PhD. I appreciate you sharing your time and knowledge. Perhaps in the not too distant future I will be able to afford some of your top of the line profession converters. I'm currently specifying equipment for a new re-recording theatre but my budget is very tight at the moment. Many years ago I heard a dBTech Gold (or was it a Blue?), it remains in my mind the sweetest ADC I've ever heard. Cheers, G.

Thank you for your comments.

We changed the company name to Lavry Engineering around 8 years ago (there is another audio company in Europe that makes speakers and other audio gear in Europe with the name dB Technologies). What you heard is the same basic architecture and sonics but since we have upgraded the gold and blue further. The black series is the latest
DA's, MicPre and AD).

Regards
Dan Lavry

 

[doctorcilantro]

Qsp - Kmixer is gone amigo, and wasn't an issue with ASIO on XP.

I've used Lavry DA10 and AD10 in the past and I found them to be great units, but I had to sell them for economical reasons and have been considering adding an external clock to my 1616M. I'm feeding an external ESS Sabre DAC chip and my question is if I add a BLA Microclock with low jitter and use a patch cable 6" (BNC>RCA) as a source for the 1616M is there really a risk of ending up with more jitter?

BLA claims 1ps clock.

The 1616M is listed as about 500ps and I'm not promoting the use of an external clock or whether or not 1ns of jitter can be "observed", just curious if it is easy to end up with say, 1ns of jitter, when you hook up a master clock in such a way. Dan mentioned terminations, EMI, etc. and I guess it doesn't take much gievn these very small units of time that jitter is calculated with.

DC

Quote:

Originally Posted by Bones13 /img/forum/go_quote.gif I think doing something other than motherboard SPDIF out is a good idea. Most motherboard SPDIF is limited to 48 out, whereas the CD is encoded and therefore ripped at 44.1. There will certainly be resampling by the kmixer from 44.1 to 48 in the innards of your PC. Probably in the middle of the CPU cycles. Of course some better motherboard cards may do better, most do not. The USB to SPDIF or AES should provide a bitperfect data stream to the DAC, that should improve the sound. I doubt the small jitter element will make it worse. And as always you can spend just about as much as you can on a box that does this. Most people are happy with either a good sound card making bitperfect SPDIF, (even the lowly XFi cards can do this.) or a USB - SPDIF/AES bit of gear. Eliminate the kmixer, and output bitperfect to your dac however you can, you will notice a difference. Of course if you had a Mac, you would have bitperfect optical out already.

 

[bergman2]

after just receiving my trends audio ud-10.1 and hooking up the aes-ebu to the lavry da10 i can unequivocally state that it is superior to both rca unbalanced and optical -- much wider and natural soundstage with greater pop and extension

 

[DoYouRight]

can someone compare the DA10 or DA11 to a Buffalo32?

 

[midnight59lp]

Quote:

Originally Posted by Dan Lavry /img/forum/go_quote.gif Hi bergman A. XLR: 1. It has the advantage of being balanced which is a good method to overcome common mode noise. Let me explain: both pin 2 and pin 3 wires (the signal wires) occupy the same space (they are near each other and are parallel). So they both pick up nearly the same environmental electromagnetic noise interference. The receiver at the end of the cable "looks at" the difference between the voltages of pin 2 wires and pin 3 wires. Since the noise pickup is the same, the difference is zero so the noise is canceled. But the signal imposed by the driver side is not the same on both wires. It is a "forced" voltage difference... 2. XLR signals are transformer isolated (AES specification). The transformer helps isolating the driver and receiver units from ground loops, which is an unwanted current flow. 3. The AES standard is based on a few volts signal, which is a good thing; it helps defeat weaker unwanted interference. The higher the voltage means more power. The cable and load are 110 Ohms so a 2V signal is really 18mW power. A 3V signal is 41mW power. 4. The XLR has 3 pins. Pin 1 is use for a grounded shield. It is best to have the shield connected only at the driver side. Connecting at both sides will provide a possible path for ground currents between the driver and receiver chassis. B. RCA: RCA is a single ended signal (un balanced), thus no common mode rejection. There is typically no transformer isolation, and the signal is relatively weak. 400mV into 75 Ohm amounts to 1.07mW power. That is a lot less then the XLR signal. Also, a typical RCA cable does not have a separate shield. That is why XLR can be used for very long distances (hundreds of feet) and in a harsh environments. The RCA is restricted to around 15 feet if I recall correctly. But for short distances, say 6 feet for example, the RCA is just fine. Most consumer and hi fi gear does not offer XLR, but if it does I would use it. RCA is just fine for reasonably short distances. The XLR was designed for pro gear where distances can be very long. C. Optical: Optical has some advantage; it provides the best electrical isolation between units, and an opaque sleeve certainly blocks external light interference. But optical has some issues as well. The limitations depend on the type of optical transmitter, optical receiver and the light pipe itself. It would take a very long post to explain optical. For the most part, 15 feet or less works fine, better then RCA, and not as good as XLR. I like optical at short distances, it is very robust. Additional comments: People often get confused between the hardware and the format. What I said above is about the hardware, not the format. It is true that traditionally, RCA and optical (Toslink) were invented to support SPDIF format and XLR for AES format. But the lines have been blurred. One can send an AES format over RCA or optical, and SPDIF over XLR. One of the reasons for the "blurred line" is that while the various format features are different, the part of the data that contains the music is the same. So one can use the same digital audio transmitter and receiver IC's for both formats. I hope that helps. And yes, one more comment: In all cases, short cable or optical link is always better then longer one. Whenever possible, go for short. I do not mean that one has to overdo it, but if you can use a 6 foot length, do not use a 15 foot. It may help, and if it does not help, it will certainly not hurt. Regards Dan Lavry

Dan,
I have the AD10 and DA 10. I have been running it toslink into my mac pro desktop. Do you believe is would be a better setup running the xlr spidif out?
thanks,
mike

 

[doctorcilantro]

Quote:

Originally Posted by DoYouRight /img/forum/go_quote.gif can someone compare the DA10 or DA11 to a Buffalo32?

Ask Jude, he likes the DA11, not sure if has heard Buffalo 32.

I may pick one up to demo against my Eastern Electric ESS 9018 based DAC some time.

Toslink is immune to EMI and RFI, but AES has good noise rejection and you don't have to do a conversion at the interface of each device as with Toslink.

DC