Living Room System

Friday, September 12, 2014

Toslink Upside Down

Nobody seems to talk about this, so I shall fearlessly proceed.

Toslink connectors are supposed to go in only one way.  However, in my experience, most Toslink connectors can get plugged in upside down.  Generally you have to apply unusual force to do this, and the instructions for Toslink connectors always say that you should not "force" the connectors.  So it is clearly wrong.

But sometimes very tempting.  When Toslink connectors are forced in, upside down and all the way in, they become very rigid.  Plugged in the usual way, Toslink connections can sometimes be very loose, sufficiently loose that with some equipment the signal can be unreliable or intermittent.  Though I can't recommend that others do this, I have chosen to plug in connectors like that upside down.  Currently this means the two Toslink connections to the first M-Audio CO2 in my kitchen.  If I plug those connections in "the right way" they become intermittent if I wiggle the cables.  That is not good for stuff that can get moved around on top of the kitchen table.

But I worry and wonder if plugging Toslink connections upside down damages either the cables or the equipment being plugged in.  It could in fact be that the reason my first CO2 requires cables to be plugged in upside down is that I did so sometime in the past, wearing out a certain part of the connection socket so that ever since it has been intermittent for cables plugged in the right way because it doesn't hold tight enough anymore.  Alternatively it could be that I damaged the connectors on the cables themselves in this way.

So I can't recommend that others plug in Toslink upside down, but for now, I am doing that for certain connections.

I am very glad that my new Schiit Modi DAC (OptiModi) has a Toslink connection that is not unreliable if the connector is wiggled, when plugged in the correct way which seems slightly loose.  I suspect that is the way they should be by design.  The connection should not have to be rigid to have a reliable connection.

But that doesn't seem to be the case with my CO2, and it also was not the case for an Inday TLDA22 Toslink splitter I bought, which introduced me to the idea of plugging in Toslink upside down, for it would barely work any other way.  It's possible that once I started plugging the cables into the TLDA22 upside down, it damaged the cable connectors so that now they won't go into the CO2 well enough the correct way.

I do have some Toslink cables with metal connectors and I think those would not be so easily damaged, though it might cause damage to the equipment.




Thursday, September 11, 2014

Good Schiit! (OptiModi)

I'm very pleased with my $99 Schiit Modi with Optical interface (OptiModi).  It looks, works, and sounds great, and is an incredible bargain!!!  Also made in USA, in a nice little metal box, audio jewelry on the cheap!!!

Unlike other devices I've had, the Toslink suffers no dropouts when I wiggle the connector.  It doesn't hold the Toslink connector as tight as I like, but no matter if it's completely insensitive to connector movement, as all Toslinks should be, but some aren't (my next post discusses Toslink connectors).

Loaded with top silicon, such as one of the latest converters from AKM, it may be the best sigma delta DAC out of a few dozen in the house.  I won't be testing it against my R2R 1704 DAC for the living room panels.  I am slightly afraid I might think this little job was better…

The night I first plugged the Modi into my Kitchen system, I was rocking out to the good sounds it made.

I plugged it into my Pioneer DVR-LX70, as it has always been intended to provide the analog audio that machine requires.  My living room receiver accepts digital inputs, so it seemed natural to provide it an optical input.  And so I have long used some kind of optical splitter on the Mac output to produce one Toslink for the receiver, and one Toslink for a DAC which produces the analog required for a DVR.  The DAC has always been the Emu 0404 USB I repurposed as a DAC for this use, temporarily, in 2011 or so.  I really intend the 0404 for use in making electronic and acoustical measurements.  It's been a pain in the neck because after each power outage, the 0404 has to be reset (since on power up, it goes to "OFF" on the main selector.  And then I can't remember if I need to do to get it working again.  (I strongly dislike DACs that forget their last state on power cycling.  My Aragon DAC was like that and I hated it.  Then it died.)

So the purchase of the Modi was really about streamlining and making my Kitchen system more robust, rather than "getting better sound."  I had every reason to believe the 0404 was top shelf, didn't really need replacement.  And further, that routing an analog converted signal through the LX70 and thence to my receiver in analog form would not, could not, be as good as sending the receiver the original digital.

But it was sounding so good as I was listening to Abby Road mixes, I was thinking my thinking had been wrong.  If the Yamaha 5790 is fed analog in direct mode, it never goes through the Yamaha's ADC and DAC.  It goes through in pure analog, "Pure Direct", which is made possible by having a parallel volume control that controls the DSP when doing digital processing, and an digitally controlled amplifier (an analog circuit that does not sample or quantize the input, but is controlled by a digital signal).  I tested and determined that the receiver does actually operate this way when I received it.  In "Pure Direct" you see no digital artifacts at the output if you provide an analog signal without digital artifacts (typically a waveform generator).

So perhaps the better sound (or at least heightened experience) resulted from the superiority of the Schiit Modi over the DAC built in to the Yamaha.  Of course Hydrogen Audio objectivists would say this is impossible, modern DACS and ADC's are completely transparent, and moreso than much analog circuitry.

Anyway, on Thursday I compared the two DAC's more directly bypassing the LX70 (which may be boosting the bass a bit--Legato Link???) by plugging the Modi straight in to the Yamaha.  While I initially thought there was a difference, as I flipped back and forth after the first comparison in critical listening I found there wasn't really any way to distinguish the two.  Finally I got very bored at the possibility of finding a difference, they were just too close.

So there you go.  At first I though there was a huge difference, the Modi was a "revelation", opening up music better than ever before, perhaps better even than my cherished 1704 dacs.

Later I find no reliable difference between the Modi and my receiver, which would exactly be an audio objectivist belief, all DACs sound the same.

It's easiest to have a strong belief if after doing A and B you then quit testing.  Going back to A again confuses things.  I have always found this.  But going back to A perhaps also reveals that the difference  experienced between A and B was not due to the stimulus being detectably different, but rather the state and expectations of the listener.

The Schiit Modi is perfect for what I bought it for, and perhaps more.





Left A'Diva Speaker Hung

Finally got around to installing the left speaker in 2nd bedroom.  Not hooked up yet because I plan to use flat wire that sticks to wall.  The wire I have is labeled Aurum but looks identical to Ghost Wire sold by Sewell.  Here are the instructions for running Ghost Wire.  My plan is not to use the big terminal blocks sold by Sewell on the grounds that they are big and ugly.  Instead, I plan to fold over just a bit of wire at the end (that way I can get behind the sticky part on top surface) and solder small 16 or 18 gauge wires to it.  At least up by the speaker.

Tuesday, September 9, 2014

What to buy next

Back in June when I attended THE Show in Newport, I decided that after my second vacation in July I'd get myself a DSPeaker since bass boom and/or lacking bass response is the single biggest problem in my living room system.  But then before my July vacation, I decided to buy a DVD-9000.  Issues related to the DVD-9000 continued through about mid August.  So then the window opened up again.  I'm home for the foreseeable future (no planned vacations for the next 10 months or so) and I can buy stuff now.  But by this time, I was no longer committed to buying the DSPeaker Dual Core, having found issues with it and other approaches which might be better (and, unusually, cost less).  What I really need to do is start making full system measurements using REW or something better.  In August, I picked up a nice $100 calibrated microphone with bass calibrated to 5 Hz.  I could get started on this at any time now.  I also did a system time alignment and tuning using Tact for measurements (but not correction).  But an independent fine resolution measurement with REW would be better.  I just have to get around to doing it.  Note: meanwhile I have been doing many other things and making great strides forward if not in the direction I planned in June.

So I'm now putting off major purchases until I do that.  Especially with regards to EQ or acoustical products.  I need to establish baseline measurements as well as get comfortable making those measurements.

After doing the measurements with REW, I may simply choose to adjust and add to the Parametric EQ's (PEQs) in my Behringer DCX crossover or DEQ equalizer.  Another alternative would be to get an OpenDSP product to implement filters designed using REW.  Existing commercial products Dirac and Accourate don't look like what I want--I might prefer the algorithms used by DSPeaker to those.  But in either case, I want to do the measurements first.  Then, audio things I could buy would include:

DSPeaker Dual Core 2.0  (after all)
OpenDSP-DI (an alternative that works with REW)
Bag End E-Trap (an alternative way of taming modes, and it might be best for problem spot I have)
RealTraps, GTK, etc: Mondotraps, Minitraps, etc

Those are specifically things to deal with bass boom and impact.  But meanwhile I have other audio projects going.  Just last winter I started the Turntable Project, which has been on hold since about April.  Things for that project include:

new 12" tonearm and arm cable for Lenco table
new extra phono cartridges: Dynavector 17D3 (I could use up to 2 more!)
second moving coil amp for living room
repair Technics EPA-100 tonearm
refurb Technics motor
dustcover, dustcover
Get LP12 fixed (requires new cartridge)
Get Sony PS-X800 fixed
New turntables ???

And then, that's hardly the beginning of audio projects I've been thinking about:

New R2R Dacs for super tweeters, subs, and master bedroom system (4 new Dacs in all)
New Behringer DEQ's to replace DCX's so I can use external DAC's
OR, modified Behringer DCX's (and DEQ's)

I'm sure there's more, but that's about all that's coming to mind now.  Meanwhile, I also hope to get some big home improvement projects done, starting before January, and on for the next 4 or so years I hope to get these things done:

New master bathtub and tile (high priority!)
Other bathroom remodeling (lower priority)
New driveway extension
New patio and patio cover
Garage/Gym redesign
Other kitchen and bath upgrades
Solar system

And, in addition to that, there are some other important expected purchases in next couple years or so:

New couches for living room back wall
New kitchen chair for 2nd person
New adjustable bed for Queen's Room

Finally, in about 2 more years, I'm going to need a new CAR, for which the least expensive model I'm considering is Nissan Leaf for about $35,000.

With all this on the buffet table, I'd better not scoop up too much audio bricks.  A sensible person might say I've already got enough audio stuff…

Analogmetric

Saturday, September 6, 2014

Toslink and SPDIF raw signals compared

At the September 2014 XCSSA meeting I brought a box of SPDIF Coax to Toslink converters, analog to digital converters (to generate digital audio signals), and an oscilloscope, all to see how the raw Coax/Toslink signals look in various conversion configurations, including ultimately 3x conversion.  Direct SPDIF coax is better than using Toslink conversion, as I expected.  But the differences are quite small, and it looks like many levels of conversion would likely be OK in a digital audio system.  I have been using 2x conversion (actually, 4 conversion elements) in my latest digital audio system at 96kHz, and even with other long cables, it has worked fine, and has sounded great.  Based on what I saw at the meeting, that is not surprising, though I also demonstrated that a new method of hooking up converters will likely work better than what I have been doing in the past--but I did not have enough equipment to test that directly.  I'll discuss what I mean by conversion and how it should be counted as I go along.  I also showed that a bandwidth measurement for digital audio transmission should likely go to 20-100 mHz.  The 2mHz oscillator I brought was simply too low frequency to give any digital audio transmission systems a meaningful test.  The actual 96kHz digital audio signal seemed to use pulses equivalent to a 6mHz square wave, from our Oscilloscope estimation (which I haven't verified).  But the risetimes suggested transmission bandwidths in the 35-100mHz range, even with multiple levels of optical conversion.

First off, I connected a Analog to Digital converter to a generic brand 12 foot video cable, and connected that to the 100 Mhz oscilloscope having a special 75 ohm terminator.  (I am so lucky to have one of those!)  The square pulses of the digital signal are almost perfectly clear and square.


 I replaced the 12 foot generic cable with a 6 foot Monster Cable Video 3  75 ohm cable.  This specific model was recommended in a Widescreen Review technical test as having good performance.  (The Monster Video 2 was only so-so.)  It seems to use a solid center conductor, and uses slotted center pin metal plug.  It had performance essentially indistinguishable from the 12 foot generic cable.


After doing that test, I cleaned up the picture slightly by reducing intensity and adjusting focus.  If I were trying to fool you, I'd delete the second picture and just say this is what the Video 3 cable response looked like.  And that's true.  But it wouldn't be honestly explaining what caused the difference.



Now the picture is so clean we're seeing high frequency ringing which may be occurring in the SPDIF generating circuitry.

As shown, I think that coax cables 12 feet or less in length actually make very little difference to SPDIF connections.  So I won't be paying attention to which coax cables I'm using in these tests, and in any case they were all Belden 75 ohm video cables of one kind or another, terminated with Canare 75 ohm RCA connectors, in lengths of 1 foot to 7 feet.  I also didn't pay any attention to the optical cables I was using since they were all short, 6 feet or less.  Long Toslink cables might have some ill effects, but I did not bring any long Toslink cables with me.  I have used 30 foot Toslink cables with good results, but 30 feet might be getting to the point where some degradation of the signal sets in.  But I won't know that from these tests.  What I believe makes the most difference are the Coax to Optical and Optical to Coax conversions.  And possibly if you used really bad cables or connections, which I didn't do.


But the third scope picture only looks so much cleaner because of scope adjustments.  BTW, adjustments in future pictures will vary, because I need to readjust intensity a lot when going to very high speed pictures like the one below.  Notice above that the rise time appears to be about one tick mark in a grid having 5 tick marks per 0.05 uS.  So one tick mark would be about 0.01uS rise time.  Bandwidth is calculated as 0.35/rt.  Funny that linked page does exactly the calculation I need, and the result is bandwidth of 35mHz.  Much of the bandwidth limitation is likely in the Analog to Digital converter, which doesn't need a 10Ghz buffer to drive SPDIF at 6mHz.  Some is also in the scope and cable.  The scope is only a 100 mHz scope and likely sufficiently out of calibration to be somewhat less than that.  To verify the 0.01uS rise time I pressed the 10x horizontal button.


Well, actually it's better than 0.01uS.  Each major division here is 0.005uS, and it seems that the rise time is about that.  So now it looks like 70mHz bandwidth.  We're possibly getting close to the limits of the scope here, so the actual coax bandwidth might be still higher.  Notice that the top squiggles seem to cycle in about 0.01uS, which mean they represent a frequency of 100mHz.

Now I do one set of optical conversion with two conversion elements.  I take coax from the ADC and run it through a M-Audio CO2 converter to convert the coax to Toslink.  I run a 1 meter Toslink cable to a second M-Audio CO2 to confer to Toslink back to coax, and then to the scope.



You have to look carefully to see any differences in the scope picture.  But the 100mHz squiggles down the whole top of the pulse are gone and there's a bit of rounding at the leading edge of the signal pulses.  Zooming in, we can see the rise time is now about 0.009uS, so the bandwidth has fallen from 70mHz to 39mHz.  One would doubt this would be a problem for any digital audio transmission at 96kHz.  (Heck, even very ugly looking digital signals work, and this looks nearly pristine.)



Next I added a second set of optical conversion, by inserting an active optical splitter in between the optical sender in one CO2 and the optical receiver in another CO2.  I've had good luck with this active optical splitter, an Inday TLDA1.  Don't bother with passive Toslink splitters, they usually don't work.



 After inserting the splitter this I had to readjust stability controls in the scope, and the resulting stable pulse is a different one than before, one surrounded by blank pulses and seems narrower--ignore that difference.  I also might have changed the time base a bit to get stability.  Otherwise, t looks about the same, though you can see some roughness in the top and bottom curves of the pulse, which looks like some kind of added vhf ringing.




As much as I've always thought the Inday Toslink splitter was a surplative product, here it appears like it adds some artifacts compared to using a pair of CO2 converters.  Though by necessity I can only test the Inday in combination with a pair of CO2's, so I can't really isolate their effects.  But still, I now believe CO2 is better because of cleaner signal and lack of stability issues on the scope.  (BTW, the M-Audio CO2 is no longer being made.  Isn't that the way things go.)

I tested a different model Inday Toslink splitter, TLDA22, one which gives two optical and one coax output from a single Toslink, a feature I thought was cool, but I had many problems with this unit (and for that reason I have gone to great lengths to duplicate the functionality of the TLDA22 using multiple CO2 devices).  Running just as the previous Inday splitter, using Toslink input and Toslink output, and forcing the Toslink cables in upside down as they are too loose the correct way, I got an essentially identical trace from the 96kHz digital.  I had previously believed this TLDA22 to have broken 96khz capability, but here it seems to work as well as needed, except perhaps for having similar stability issues as the TLDA1 on the scope.


Now I couldn't run all three CO2's I had brought at the same time because I was missing the 9v DC adapter for it (did I not read the eBay ad carefully?).  We couldn't find a suitable 9v DC adapter in the large box of AC adapters at 10bitWorks because the few we found didn't have room for the large center pin in the connector.  9V DC adapters are relatively rare, actually.

But what I could do was use a different SPDIF generator that has both SPDIF and Toslink outputs.  That was an Emu 0404 USB, running in standalone mode.  Selecting the analog inputs and connecting to the Coax output, it produces pulses with a slightly curved top, suggesting some HF damped resonance around 10mHz, but still a clean looking wave (and it was very stable, note this was a lower sampling rate, probably 44.1kHz, since I can't vary that on an 0404 in standalone mode).



Now I remembered to test one of the CO2's in coax-to-coax mode.  The CO2 has a selector switch which can select Bidirectional (the usual choice), Coax-to-Toslink, and Toslink-to-Coax.  In Coax-to-Toslink mode the Coax input is converted to Toslink, and, simultaneously, the Coax output operates as a coax pass-through (it's an active splitter, actually).  Inserting the a single Coax-to-Coax into the 0404 output chain only made the pulse look better--with a flat top.  The rise time (and even ringing, in a later test) appeared identical.  Thus the Coax-to-Coax looks better and cleaner than any kind of optical conversion (but I should have chosen the better test signal, arrg!, because on this uglier test signal you can't tell as much):



Next I did something almost opposite.  Starting with the Toslink output of the 0404, I ran that to a CO2 in Toslink-to-Coax mode.  But I connected from the Toslink output of that CO2 (remember both Toslink and Coax outputs are active in Toslink-to-Coax mode) to another CO2.  The second CO2 simply converted Toslink back to coax for the scope.  Now this chain actually involves 2 layers of conversion or more precisely 4 conversion elements and one buffer.

Layer 1:
internal signal to Toslink output inside 0404
Toslink to internal signal inside CO2 #1

Layer 2:
internal signal inside CO2 #1 to Toslink output
Toslink to internal signal inside CO2 #2

Buffer
internal signal inside CO2 #2 to Coax output


About the same, though it does look like the rise time has fallen about as much as it did in the first test using 4 optical conversion elements, though the digital audio generators are different and the tests are not exactly comparable.  Still, it looks perfectly fine.

By this time I was getting confused and decided to do three tests in a row.  Starting with direct from coax from the 0404 to the scope, then taking the optical output of the 0404 through a single CO2 to coax (one layer of conversion, including one element inside the 0404 itself), then the two layers of conversion (same as previous experiment above).  You can see some progressive deterioration but not much.  The CO2 converters are so clean I don't believe I messed with the scope stability controls at all during these three presentations, though perhaps a tad on the last one.






I tested running the Toslink output of the 0404 into the Inday TLDA1 splitter, then through a CO2 for conversion back to coax.  This is two layers of optical conversion again, same as in the last picture above, but with one of the layers occurring within the Inday TLDA1.  The result looks to have slightly higher rise time, perhaps, but slightly less clean than the double conversion plus buffer above using two CO2's.




Now for the most conversion I was able to do.  Triple layers with one buffer:

Layer 1:
internal signal in 0404 to Toslink output
Toslink to internal signal in Inday TDA1

Layer 2:
internal signal in Inday TDA1 to Toslink output
Toslink to internal signal in CO2 #1

Layer 3
internal signal in CO2 #1 to Toslink output
Toslink to internal signal in CO2 #2

Buffer
internal signal in CO2 #2 to coax output



Despite all the conversion, it still looks pristine, bandwidth at least 35mHz, etc.  The Inday splitter adds a tiny bit of ringing, hardly noticeable, but also seems to make the rise time even quicker than without it.

Using the other Inday splitter with coax output, it was hard to get good triggering with 2 layers of conversion.  It occurs to me that scope triggering is much like the triggering inside a digital receiver in that it focuses on the transition from very small negative voltage to very small positive voltage.  It doesn't look at the scope trace like we do.  What I could see beyond the lousy triggering was that the rise time seemed just fine, perhaps even better than the CO2's.  But the freedom from ringing makes me prefer the CO2 devices to the Inday devices.

My ultimate hookup for Mac to Living Room Stereo will use two CO2's.  The optical output from Mac computer will get converted to coax and optical (Optical->Coax) by CO2 #1.  The coax output of CO2#1 will connect to the second CO2, in Coax->Optical mode.  Then I will have two optical outputs, one from CO2 #1 and the other from CO2 #2, and one free Coax output, from CO2 #2.  The optical outputs go to devices in the kitchen and the Coax output goes through installed wiring to living room.



I had been concerned that the second Toslink outputs would see three layers of conversion, but now I see that is not true.  The Toslink output from CO2 #2 sees these 2 layers of optical conversion, plus some coax buffering (which I have determined to have little ill effect):

Conversion Layer 1
Mac internal signal to Toslink output
CO2 #1 Toslink input to internal signal

Buffering Layer 1
internal signal to Coax output
Coax input for CO2 #2 to internal signal

Conversion Layer 2
internal signal to CO2 #2 Toslink output
Toslink to internal signal in receiving device

So this is only two layers of CO2 conversion, which I tested in many ways above.  It is not as bad as three layers of optical conversion, which also looked good in testing.

Late on Saturday night I hooked up the two CO2 converters as described above, but the second optical output is not currently connected to anything yet because I will use it to install a Schiit Modi DAC.  The other outputs are working perfectly and sounding great, especially the living room connection which now sees only one layer of optical conversion while previously it saw two.