Stereo AD Converters

I know of no audio A-D products that do not use an A-D converter per channel all timed from the same sampling clock. This will mean that there is no delay (at audio timescales) between the channels in a device.

Differences in performance between models are due to a multitude of things from the design of the audio driver and clock circuits, crosstalk minimisation, the power supply and power distribution and decoupling to the quality of the converter chips themseves.

It's probably not appreciated in the practical audio world what really happened when chip designers made advances from the early audio A-D converters that used successive-approximation techniques. These are the types that went into the early digital audio products and, along with the companion D-A converters, were largely responsible for the "harsh" sound that characterised the first available CDs and CD players.

Since the advent of "over-sampling" devices, a lot of the early problems have been reduced, but not without causing trouble elsewhere. Over-sampling ADCs and DACs need high-quality clocks locked to a multiple of the audio sampling rate, which means in practice using phase-locked loops to generate a clock at 64, 128 or even 256 times the basic 44.1KHz (or 48KHz etc for non-CD units), so attention focuses on the quality of the clock multipliers used. A rock-solid 44.1KHz sampling clock can be completely degraded by sloppy chip design in the multiplier circuits or by poor implementation of PCB layout and power decoupling in this area. It's the attention to this sort of thing in addition to specifying the better-quality parts that marks out good designs (products) from poor ones.

Multi-channel (i.e. more than 2) A-D converters usually impart the same sonic characteristics on to all their tracks. If one thinks of the sound of a cheap condenser microphone, it may be possible to bury any unpleasantness from it in a mix of tracks where the others are recorded with better mics, but a project studio that multi-tracks everything using the one cheap mic risks multiplying up the audio deficiencies and producing a mix that offends the ear. Spread this example out from serial tracking to parallel operation by taking a low-cost multi-channel A-D converter unit, and it becomes clear why some designs have a sonic signature that is audible all the way through to the 2-track mix. Products that do not suffer from effects such as this are usually the higher-cost ones that have had the necessary attention applied to the detail of the design.

From the technical pov it is not possible to get 1 channel in an 8-channel Pre-amp offsynch, other then by design and on purpose.
If you lock 2 separate but identical Pre-amps to one clock the difference in timing between the 2 is not audible, shouldn't even exist, unless something's
gone haywire or you use real crappy equipment from the dark ages ( in Alsihad some Plugs caused single sided delay due to a bad or non-existing latency compensation, though).

I worked with all kinds of good and bad, old and new converters. I never saw anything like that. The digital signal send off to the recorder or PC, card is multiplexed ( optical & wire). If there
is a fault in the mux chip the whole thing goes wild or some channels are missing. The mux technology like in MADI or ADAT link is quite old, safe and explored. It it is no techical challenge and no budget question, anymore.

The Sound of AD/DA converters is often depending on quality an prize. The synch capabilty is technically on the same problem-level like the rubber feet of its case, prolly less.

Yeah, I think it's mostly a non-issue at this point. I've never encountered any problems like that. As long as you're working from one master clock (again, inside the converter box), you should be fine.

The only sync issues I ever worry about is if we're locking a variety of machines together for biggg projects; like DAWs, Video playback, digital consoles, etc. But I rarely have to deal with that; I'm just not that kind of (big) production company.

And THOSE guys (big NYC or LA studios) have "people" who handle it. ;-)

Well just when you think you've got lol. I'm obviously not be explaining this correctly and understanding the part in converters that effects what I think is being effected. I'm an Auditory learner so its always been challenging, processing technical info from reading. It takes me longer to get it this way, but I do eventually as I go over and over things. My hearing is great and probably why I love working with sound and choose it as a profession. I'm going to post something once I get permission. I'm hoping you will all help me understand this better.

Lavry Engineering gave permission to post this in whole. It's what got me wondering about stereo imaging and how the clocking or ? could be effecting this like delay does. I have somewhat cross referenced this to non constant latency and wondered if this is one reason to invest in high quality converters with very good power supplies and dump firewire. The difference between my FF800 and the AD10 is pretty convincing. I found quoted below very interesting.

Hi

Below is a copy of a post from my Lavry forum:

When I hear the term “latency” I think of some guitar player trying to monitor in real time their performance via headphones, or some stage performers listening to themselves via “spot” monitoring speakers.

In such cases, the question becomes, how long is too long? I heard many different answers to that question, but they mostly range in the 5msec-10msec. How long is long? Here is the “reference”: Sound travels (acoustically) at about 1msec per foot. If you want say 2 msec “round trip”, even if you place the 1 foot away and the speaker 1 foot away, there is no “time budget” left for the electronics. The lowest time possible (a big overkill) will be when the monitoring is done in direct analog, and by use of headphones (no acoustic distance on playback).

When it comes to recording, the issues are different then “self monitoring”. An orchestra can span distances of a whole stage, and at sound travel of 1 msec per foot, it can sound just fine. One needs to draw a distinction between “self monitoring” and other cases, and not throw everything into one pot, ending with a confusion between “latency” and conversion quality. It is not.

There has been a lot of talk about latency, mostly due to “endless problems” with fire wire based devices. Aside from dropout and clicks, many problems were due to latency that changed from power up to power up. The wrong conclusion would be to assume that one needs as low latency as possible. When problems occur due to non constant latency, the solution is not to reduce latency; the solution is to make it constant.

A couple of month ago I gave a lecture about it in Argentina AES on surround. I just came back from India where I pointed the same issues out. The issue to watch for is about phasing. Eliminating phasing is NOT about reducing delays to zero (which is an impossibility). Eliminating phasing is about having THE SAME DELAY for “SHARED SIGNALS”. When you look at you signal paths (between the mic and the MIXER), you can view the signals as “SHARED SIGNALS” or “UN-SHARED SIGNALS”. If say mic A picks up ONLY a flute, and mic B picks up ONLY a guitar, there is no shared signal, and there will be no phasing. But if Mic a picks up the guitar and say 30% of the flute (mic B still picks the flute), you have 30% shared signal. When that portion of the music appears on multiple channels (more then one), AND the delay between channels is NOT THE SAME, you will have phasing of the shared portion.

Clearly, you can eliminate phasing even with equipment that has huge delay (latency), be it a second or an hour, as long as the channels you mix are subject to EQUAL delay.
Also, you can eliminate phasing if you make sure that each mic pickup field pattern covers a single sound source (be it a singer or an instrument). Of course avoiding all shared signals is not always possible or desirable, thus one needs to be aware of how to set things up, and how to do it right. For example, most vocalists get their own mic, while for large orchestra one has limited number of channels…

So low latency is not the issue for eliminating phasing. EQUAL LATENCY is. How close equal? Very close. For mixing, we are not dealing with acoustic delays. We are dealing with ELECTRIC signal propagation. Take a sine wave of say 20KHz. Take another identical 20KHz signal but with a 25usec delay DIFFERENCE. When you add the 2 channels, you end up with a total cancellation! With only 25usec, the amplitude vs frequency is no longer flat. It is a comb filter with one notch at 20KHz.
Take say 100usec delay, which is a lot less delay then any sigma delta audio converter, and you frequency response is a mess – it has a comb filter with a few notches…

Electrically, the signals travel very fast on wires and though much of the analog circuits. But when in the digital world, it is not uncommon to accumulate many milliseconds of delay not only in conversion, but also using DSP. For example a long FIR filter can add many msec. I already pointed out that it does not take much delay to have phasing problems. Yes, 1usec will not hurt you, but 25usec will kill your 20Khz an 100usec will be much worse, not to mention a 1msec DIFFERENCE.

Thus the responsibility and the solutions for phasing is a lot less about converter absolute delay then one may think. Adding more delay difference adds “teeth” to the comb filter at lower frequencies. Taking care of keeping relative delays very small (a few usec) is important when dealing with “shared signals”.

The term “shared signals” is not common. I actually created the phrase “shared signals”, to help explaining the issue. So a search for “shared signals” may not be as productive as you wish. I do hope my explanation is clear enough given the short time I can allocate right now.

My advice is to keep your ears on the ball, and not fall for a contemporary slogan “low latency”. Latency is not a key indicator for converter quality. BTW, in the case of fire wire and USB devices is the SUM OF the converter delay PLUS the DRIVER. When someone talks about low latency, first figure out if it matters in your case. If you think it does find out the NUMBER (in msec). Lately I see the phrase “low latency” bouncing around out of context, and often not quantified!

Regards
Dan Lavry

A little longish...
Is this an older statement?

Good microphone technique and good converters can spare you the mentioned problems.
I have never considered any USB/FW devices acceptable for studio work, although it is better, now, then it was.
Runtime, crosstalk and phasing problems before the acoustic signal hits the mics is not a converter problem and
converters that give you problems like this, you better give a kick.

I flawlessly record and play back several tracks at once at 0.7 millisecs and then there still is direkt listening, too.
Btw, I consider low latency processing well as a property of a good converter ( and driver)... one property.
As a last resort I recommend [[url=http://[/URL]="http://www.uaudio.c…"]Little Labs® IBP Phase Alignment Tool Plug-In | Universal Audio[/]="http://www.uaudio.c…"]Little Labs® IBP Phase Alignment Tool Plug-In | Universal Audio[/] or the hardware version of it.
When doing multiple mic recordings I use a 70 cent Knackfrosch to record its sharp click for the mic/room runtimes, just in case I needed it later..

I suppose its a few years back now.
Remember, I'm thinking these are pretty sweet for a remote rig including the 2-bus out of my mastering chain. It has the USB there as an option but I would use AES EBU personally but I'm sure others that don't have this luxury can still get happening.

It kicks the FF800 out of the arena for certain as a 2-bus AD and pre.

AES/EBU is the way to go.... and can transport the higher sampling rates, too, compared to (ADAT format) opticals, anyway.

Big K, post: 366641 wrote: AES/EBU is the way to go.... and can transport the higher sampling rates, too, compared to (ADAT format) opticals, anyway.

Does this include BNC?

I'm assuming you are including the other digital interconnect options other than firewire itself, yes?

I would differentiate between ADAT, COAX MADI, AES/EBU, optical digital audio and USB, FW, etc.
The latter are computer interface links. RME uses its own FW protocol to the PC cards, though.

Of all digital audio connections the ADAT is the weakest as to samplingrates.
AES/EBU is great, but means a lot of plugs and cable ( multicores).
MADI is a comparatively cheap single coax connection for digital audio (64 channels over 100 m ( optical for 2000 m) , MIDI and RS 232.

There are a range of boxes for connecting and converting ADAT and AES/EBU signals to and from MADI.
For all the possibilities work through RME MADI info pages. It is all well described there from the guys that make 'em boxes.
:-)

Now, I'd like to know, if you need to cover large distances in your studio. If it all sits in one rack or side by side, MADI might be an expensive option, after all, because you either need all I/O units to be MADI or you need an additional converter box to connect non-MADI Mic-pres, etc.

As it is, I am only going 3 meters

But I may use one converter in the acoustic room instead of running all that cable.

I'm looking at two systems. I really have the stereo one complete. I love these Blacks and am interested in the AD11 now. They are compact and sound awesome. In my other thread, I am going to upgrade to a new 16 IO . I'm done with FW now. Its was a starting point for me while I gathered my hybrid system.