can you derive i2s output from the emu1212 as well? read that the digital and analog card are connected by the i2s interface, i look at the analog card it has cs4398 and ak4394 as d/a and a/d chips. what signals do i have to look in order to trace the i2s pins?

Wow, I had not checked this thread for a while since starting it. I'm glad to see there is still some interest.

First off, I am driving a P-3A with the Juli@ I2S signals directly from the header on the digital board (picture). I am now driving 4 separate RG-174/U cables ( 50 ohm small flexible coax ) with the 4 signals. The shields are all tied together to ground on the Juli@ and are tied to the metal housing of the mini-DIN connector (which looks awful but works fine ). These cables are 15 feet (yes feet ) long to span from my desktop across my office to the P-3A. I can lock on with the P-3A up to 96kHz ( upsampling with foobar2000 and SRC ).

The I2S signals have a lot of ringing on them. They are 3.3V signals but there is about 1.0V of ringing on the ground and VDD side. The MCLK signal looks more like a sine wave. But the P-3A is not smart enough to know that it should not work.

I think the transceiver chip (DS26C32AT according to an email from Peter Madnick) is the secret. It is a differential receiver with sensitivity of 200mV. I have tried the Juli@ with an Audio Alchemy DTI Pro32 - it does not work. The DTI is looking for TTL input levels, and I think these signals have too much noise.

I swapped out the P-3A opamps for LME49710 that I got as samples from Nat Semi. These are the single versions of the LM4562 quads that people were raving about.

The sound is fantastic. The system consists of a modified Heathkit AA-151 and Klipsch Tangent 500 speakers with rebuilt crossovers and titanium diaphragms. I have musicians as neighbors who are amazed by this budget system.

I am not proud of this "design" as very little design went into it, it was just a kluge to see if it would work, and it did. I would like to build a board with perhaps a 74LS244 quad driver or discrete transistor buffers, and to add some additional supply voltage filter caps.

Question for audioengr - Any advice on what type of driver circuit to use?

Quote:



BTW, that differential receiver on the P-3A is really slow so it adds jitter.

The driver on the Juli@t probably cannot drive a 50 ohm parallel terminator, so I would recommend a series resistor on each line of 30 ohms to start with. Put them as close to the driver as possible, in series with the trace for each signal. You probably have to cut the traces to insert the resistors. Use SMT resistors. Then remove the 100 or 110 ohm parallel resistors at the P-3A receiver. Then, I recommend establishing a new voltage reference for the 26C32. These are balanced or differential inputs, so lift the 4 legs on the chip for the negative inputs and tie them all together. Then drive this with either a voltage divider or a voltage reference.

Look at the signal first at the chip pins with a really short stub-ground on your scope probe (less than .5" long). Determine the best place to sample the signal rising edge - probably around 2V. Then get a voltage reference or do a resistor divider to generate this DC voltage and tie it to the four "minus" input pins. Then you will have a much lower jitter signal at the P-3A.

You can also add a buffer at the Juli@t, but this will be a kludge and needs to be done right, including power, decoupling etc.. I would try the series terminator first.

Steve N.
Empirical Audio

wackyterbacky, thanks for the pics. So the sound is just fantastic or is that a comparison to the SPDIF connection?

audioengr, to connect the M192 to Juli@ Can I just connect the 4 different grounds to one of the VSS on the Juli@?

Nopietns:

I vastly prefer the I2S. I have not performed frequency sweeps or measured square wave responses, so I find it hard to quantify. There is more detail and wider soundstage with I2S. I think instruments like cymbals sound more realistic, but honestly I have not done a blind A-B test. In fact, I like the sound of the 44.1kHz as opposed to the 96kHz upsampled. Again, hard to quantify, it just sounds more natural. Perhaps there is more jitter at the higher clock rate due to my cabling system.

Steve N. - Thanks so much for your detailed reply. I really had not considered modding the P-3A receiver, although I know you and others performed extensive upgrades on this DAC. One more question if you don't mind?

If I stayed with the Juli@ unbuffered outputs driving a cable, is there any advantage to changing cables to a CAT 6 or CAT 7 shielded twisted pairs? Would that be a less reactive load?

Quote:



Yes, but close to the driver if possible. Best at the driver VSS pin.

Steve N.
Empirical Audio

Quote:



The Cat5 or 6 cable has much lower bandwidth and higher loss than the coax, assuming that the coax you are using has Teflon dielectric. The impedance of the CAT cable is closer to 100 ohms, but it is balanced and you are not using the receiver in a balanced mode anyway. I would stick with the coax for such a long run.

If you series terminate at the Juli@t and remove the parallel resistors in the P-3A, this is the preferred way to do it. You could start by putting radial-lead resistors at the start of the coax cables and see how this sounds - you must remove the resistors at the P-3A though.

Removing the SMT resistors on the P-3A board is simple. If you dont have 2 soldering irons, then use a hot tip and alternate sides melting the solder until you can push the resistor off the pads. Then use solder wick to clean-up the pads.

Steve N.
Empirical Audio

a littlebit offtopic but
i want to power the juli( only digital section) via a riser card externaly.
do you know which voltage is needed. 3,3 , 5 or 12 volt or all three.
thanks
alfred

Quote:



Which order are these in on the RJ45 connector (easier if you compare to coloring of CAT5 wires (Brown, Green, Blue, Orange).

Of course if anybody else simply knows the pinout of the NortStar I2S connector, that would be very helpful.

Thanks

audioeng has kindly shared the pinout for the NorthStar m192 I2S RJ45 input and I have made a pair of cables to plug-and-play with the ESI Juli@ (cut a cat5 cable in half, and attached the ESI Juli@ connectors to it (audio board was ruined after some failed modding anyway). If anybody wants an already prepared Juli@ I2S cable that works with their NorthStar m192, PM me and I'll ship it for free. This cable is not likely to work with any other I2S input so if you ask, please make sure you have an m192. Cable is about 22 inches long so the computer has to be 18 inches from the DAC for it to reach.

Tuning in (so to speak). I like the looks of this. DIYI2S! Neat! I suppose I should get my hands on a capable DAC first, though...still a good thread.

Steve Nugent gave some helpful tips which I finally had time to implement.

I removed the I2S input resistors from the P-3A board. I have little experience with SMT components, but two soldering irons make it easy. The picture shows the metal shrouded PCB connector with the 4 sets of pads behind it. The resistors were in between those pads.

I put a 10 ohm resistor in series with each coax cable at the connector on the Juli@. I changed the ground point for the shields to the screw hole just adjacent to the connector. The screw hole is surrounded by feedthrough holes which connect to the ground plane. I flowed a little solder into each hole. I soldered a small spade connector on my ground wire and screwed the spade tightly against the feedthrough holes. This may make no sense unless you have a Juli@.

Also, I added a 220uF electrolytic bypassed by 0.1uF to the 3.3V VDD.

I had to repair the Mini-Din connector so I reinforced it with several chunks of heavy shrink tube.

The waveforms on the P-3A are much cleaner than before. There is still some ringing but it is attenuated from the prior set up.

I am convinced that this helped the sound. I am noticing more separation between instruments and vocalists. I like Norah Jones "New York City" and use the title track to compare different tweaks. There is a harmony during the chorus, and I now hear the voices coming from different focal points. Perhaps there is less jitter.

Or perhaps I am just imagining Norah Jones and her long lost twin sisiter in my lonely listening room?

Quote:



The waveform is a bit bumpy due to the impedance discontinuities and only 3V P-P, but otherwise probably an improvement in edge-rates and therefore jitter. If you are using 50 ohm coax, 10 ohms is probably a bit low, depending on the driver impedance.

Steve N.

Thanks Steve.

I know you recommended 30 ohms. I had 10 ohm and 100 ohm handy, so went with 10. I will get other values, since this did make a noticeable difference.

I noted that the DAC receiver chip uses a 1.15 Volt reference for the I2S inputs. This does give noise immunity even for my noisy signals. The sharper rise and fall times definitely help the sound quality.

My Audio Alchemy DTIPro32 still does not sync on those signals.

Also, I thought I read one of your posts saying the Perpetual DAC sounds better without the case. That is how I am running it now, so I could probe the PCB. Do you have a mechanism for that?

Dear Wackyterbacky

Thanks for taking time to put your juli@ mods here as I have been able to tap the i2s for my TDA1541 dac.

Dear Steve

Thanks for detailing the i2s cabling info and sharing your excellent knowledge here.



Could I ask for some help in solving my little problem please.

I have i2s signals exiting juli@ to my TDA1541a dac & can get a lock on 16/44. Each signal is going via a star quad cable (Gotham) with 2 wires for signal & 2 for ground. The grounds are tied together in juli@ at vss (pin 5 header 7) & in the dac at analog ground & master clock ground. Overall sound & presentation is amazing. At present I have 390R damping resistors just before the dac (as part of original dac design - Arcam delta 70.2)

The problem I am having is a crackling sound every minute or so. The sound remains whether I choose internal reclocking in dac or not. As this is not a continuous crackle, I am not sure it can still be ground or is it a clock/timing issue.

Many thanks

Regards

Fib