Tír Na HiFi

rickmcinnis wrote:Those isolators will not be needed with the SDTrans and I will want to bypass them.

Nothing more than remove them and install a wire from in to out?

There could not be any reason to retain them, could there?

If you have a good low-jitter and low-noise source, I can't imagine they provide benefit.

OTOH, in situations where they might, the Si8620 devices used are likely not the best. Check for comments by or attributed to John Swenson, mostly on CA, where he says these devices add a good bit of jitter (and maybe noise) and that the NVE GMR-based ones are better.

On removal, you are correct, just in-to-out, making sure you have a common ground.

Of course, being a cautious type, I'd first wire in-to-out first to confirm operation, then remove them and re-wire.

Greg in Mississippi

EDIT: P.S. When you have a good low-jitter/low-noise source, I believe there is likely EVERY reason to bypass or take the isolators out. They are adding jitter to the signal, which the reclocking then tries to remove. IMHO, it is better to start with as good of a source as possible and then only isolate and reclock if there is a good reason to do so (such as coming from a somewhat noisy source or one that the isolation is needed to protect the source, such as with a BeagleBone Black as a computer player).

Greg,

Yes those things might be necessary in many applications, BUT ... One should eliminate the need!

Any idea why they are not all going in the same direction?

Are they these: https://www.silabs.com/Support%20Docume ... 61x-2x.pdf

My assumption is they are the single channel versions?

When you look at them in HIFIDUINOs good picture of the board one can only see a trace going to BCLK - which means nothing I know, but it is a little puzzling.

rickmcinnis wrote:Greg,

Yes those things might be necessary in many applications, BUT ... One should eliminate the need!

Any idea why they are not all going in the same direction?

Are they these: https://www.silabs.com/Support%20Docume ... 61x-2x.pdf

My assumption is they are the single channel versions?

When you look at them in HIFIDUINOs good picture of the board one can only see a trace going to BCLK - which means nothing I know, but it is a little puzzling.

Rick,

Yup, those are the devices. I'm not sure which of the Si8620 variants are used, but all are 2-input/2-output, so with 3 devices, there are 6 isolated channels. The isolators used only go in one direction, but there are 7 isolated data connections with two configured as outputs, so 2 channels are configured with inputs coming from the FPGA and outputs going to the connection points (either by using a different chip or how the board is configured).

How I'll do it (I'll at least replace mine at some point with the equivalent ILxxx GMR device) is to:

1. Find continuity between the appropriate input (like I2S BCLK IN) and the corresponding device input pin.

2. Use the datasheet layout to determine which is the corresponding device output pin.

3. Wire from input to output.

And that 7th channel? I suspect it is the 'I2S FSEL IN' which is either 'high' or 'low' and that can go through the device labeled 'D3'.

I hope that helps. If not, let us know.

Greg in Mississippi

Just thought that there are probably channels for the grounds, also.

No doubt, tracing will be what is required.

Thanks,

rickmcinnis wrote:Just thought that there are probably channels for the grounds, also.

No doubt, tracing will be what is required.

Thanks,

Rick,

There shouldn't be any channels for ground, just the signals. The input signals reference to the input grounds, the output signals to the output grounds, but no feed-through of the grounds.

Greg in Mississippi

these isolators are doing far more harm than i thought
got a good lift in sq
a must do mod for battery systems

THANKS for that!

Now I wonder if we should remove that middle one just to have it out of the way?

Another thought - why not bring the I2S leads directly to the right side pads of the chips and then the grounds to the STAR ground? I am using the HIROSE coax and figure that has to be better that those board traces? Or is that not necessarily the case?

With my board on the kitchen table I am hoping I will replace those FPGA capacitors this weekend. I keep telling myself it couldn't be that hard. So far, I have not been able to convince myself of this.

Did you use solder paste? I am hoping there is enough solder remaining after removal and just flux will be required (along with a tweezer) to hold them in place for soldering.

Dithering between the heat gun and lots of tape and Randy's two irons approach fro removal.

used hot air and solder paste mostly because it was the first thing i came across
if using an iron id be thinking about using this sort of stuff
http://ie.farnell.com/chip-quik/smd1/re ... Aub48P8HAQ

was a little easier to remove than the output opamps

i didnt bother figuring out what the middle isolator was for so left it there for now

dread to think what sq could be possible without an fpga chip

It is a shame to have a chip that does nothing more than tell the chip to do nothing with the NOS "fllter".

No comment about the direct connection from player to right side pads?

That is what I am going to do unless you say NOT to.

Paul Jacobsen/moreDAMfilters used that stuff and really liked it until it splattered and made a mess of his, I think, FPGA chip. Plus, I would have to wait for it.

Heat gun is hard to beat for removal. My heat gun is not made for installation, just a utility gun with a small bore nozzle I made from copper sheet. Stuck using an iron with a tiny tip. As my Mother would have said, "it builds character".

nige2000 wrote: <SNIP>
these isolators are doing far more harm than i thought
got a good lift in sq
a must do mod for battery systems

There you go again Nige, advancing the art of what is possible out of this DAC! MANY THANKS to al of the tweaks you've tried and confirmed here... you are the other 'DAM DAC god' in my book along with Paul for his filters!

rickmcinnis wrote:THANKS for that!

Now I wonder if we should remove that middle one just to have it out of the way?

Another thought - why not bring the I2S leads directly to the right side pads of the chips and then the grounds to the STAR ground? I am using the HIROSE coax and figure that has to be better that those board traces? Or is that not necessarily the case?

With my board on the kitchen table I am hoping I will replace those FPGA capacitors this weekend. I keep telling myself it couldn't be that hard. So far, I have not been able to convince myself of this.

Did you use solder paste? I am hoping there is enough solder remaining after removal and just flux will be required (along with a tweezer) to hold them in place for soldering.

Dithering between the heat gun and lots of tape and Randy's two irons approach fro removal.

Rick, the middle isolator is probably the one for the isolated RS-232 serial connection. If you aren't planning to use that, removing it should be ok.

Your idea on connecting directly to the right-side (output side) pads of the isolator chips with the Hirose coax is probably a very good idea. That coax should be a lot better for I2S signal integrity than traces and jumpers.

On the FPGA caps, I plan to use the 2-iron method. I do have a small, inexpensive SMD re-work heat gun and while it is MUCH better for things like the isolator chips (with lots of Kapton tape, like you said), resistors and capacitors are easier for me with 2 irons. After removal, I use small solder-wick to clean the pads, then apply a tiny amount of thin Wonder Solder to the pads, position the part, hold it down with a small screwdriver tip, tack done one side, then the other, then followup with a quick resolder of each side with the Wonder Solder.

Finally, how I understand it is that the FPGA does a LOT more than just the filter selection, such as the clocking, muting, configuring the output signals to the shift registers, selecting the inputs, volume control, and lots more. The FPGA is 80% of what makes the DAC work!

Greg in Mississippi