Soekris Dam Dac

[nige2000]

Nige,
What did you change apart from removing the logic shifter bypass caps (which I don't think it is a good idea)?

Firstly this was only an experiment

Second I've got two identical builds( except for one still has these caps) so comparisons can be made directly and easily

I always preferred the sound of my pci e USB card clock mod with no bypass cap and direct from the cell

Also Preferred the sound of this DAC providing power direct from cell to the FPGA clock pad (bypassing the bypass cap)

On each shift register there is a 1uf and a 100nf cap I removed both and the cap that was on the old vref circuit

Thing to do next is to remove all the 1 uf caps on the other DAC and see how they compare

[rickmcinnis]

Nigel,

So you are saying that if one wants to use a 26650 for the input you have to replace the 1.2V reg?

That's puts me back a week.

Was hoping I could save that for later.

Just want to be sure so I can start dreading the process in earnest.

Would have thought a sm reg would need almost no dropout - learn something everyday ...

[jkeny]

Sure, Nige, I know it's an experiment (but a bit of a PITA to solder the caps back) but normally bypass caps on digital chips that are switching at high speed are needed because the inductance of any traces or wires carrying the power slows down the high speed current pulses - taking away those caps increases the inductance to that of the copper tracks & wires from the battery to the chip. Each chip has a different length of track running to the power pin.

Bypass caps also allow the inevitable self-generated noise on the chip's power pins to travel a short route to ground via the caps - the best configuration. Removing this short link to ground allows the spread of this noise to unwanted areas.

It also means that the power pins of all these 595 chips will not have any isolation from one another with the possibility that this noise will cross pollute between chips.

I'm surprised that the removal of those caps improved the sound - maybe removing the 1uF caps & leaving the 100nF in place is not a bad idea, I'm not sure. You will be able to check this with your A/B comparisons

[rickmcinnis]

jkeny,

My instincts are that might not be such a good idea, too.

Would not think those caps would have an effect on sound but could likely affect how the chips operate.

I think Nigel's smd skills are getting damned good. I would NOT look forward to replacing those things!

[rickmcinnis]

Nigel,

I see NOW that you are using the 1.2 volt "test points" for the new reg.

I initially thought you were replacing the SM for the LDO on the board.

That will be easy.

Thanks for the photos.

[nige2000]

Jeezzz guys I know what bypass caps are for
It was just a surprise it actually improved
Extra resolution decay and echo is not usually what I expect from extra noise
Next experiment is other build with only 1uf caps removed
Will report back when I get it tested no caps on shift registers vs 100nf ceramic
Getting great use out of the hot air
I can flick off the caps with the tweezers in seconds
Pity I didn't relalise this for the first dac
If it sounds better with 100nf I'll put on the film ones ;)

Jk any reason for them caps to be between the FPGA and shift registers ?
Thinking of bridging them

[jkeny]

What hot air gun ar you using?

One reason for those caps, I reckon, is to block DC which might either damage the input pins of the 595 or the output pins of the FPGA.
Another reason would be to stop DC causing false data input signals
You could test for DC on either side of those caps
The 595 chips can handle something like 2V DC on its data input pin but don't know about its other input pins
Don't know what the FPGA output pins can tolerate?

[nige2000]

What hot air gun ar you using?

One reason for those caps, I reckon, is to block DC which might either damage the input pins of the 595 or the output pins of the FPGA.
Another reason would be to stop DC causing false data input signals
You could test for DC on either side of those caps
The 595 chips can handle something like 2V DC on its data input pin but don't know about its other input pins
Don't know what the FPGA output pins can tolerate?

http://www.circuitspecialists.eu/solder ... -csi-825a/

im sure you asked me that before

bit like the oscilloscope dont use em very often but very very handy now and again

got all the 1 uf caps off the other build just after checking over it for shorts etc

did a very very quick a/b/a test
before any cap removal i was happy more or less these two builds sounded more or less identical
and unfortunately im preferring the no cap version
its not really logical, but i prefer to test then ask questions later

ill retest tomorrow but doubt ill change my mind, with bypass capsi found the sound slightly restrictive, maybe try the film caps

got a feeling the batteries swamp out the noise

then move to the fpgs caps to test for voltage etc
the irony is i dont like seeing caps and resistors in the signal path digital or analog :)

[Sligolad]

its not really logical, but i prefer to test then ask questions later

Some of the best discoveries of the 20th Century came out of that approach!!

[jkeny]

http://www.circuitspecialists.eu/solder ... -csi-825a/

im sure you asked me that before

bit like the oscilloscope dont use em very often but very very handy now and again

Thanks - looks like a good one

got all the 1 uf caps off the other build just after checking over it for shorts etc

did a very very quick a/b/a test
before any cap removal i was happy more or less these two builds sounded more or less identical
and unfortunately im preferring the no cap version
its not really logical, but i prefer to test then ask questions later

ill retest tomorrow but doubt ill change my mind, with bypass capsi found the sound slightly restrictive, maybe try the film caps

got a feeling the batteries swamp out the noise

then move to the fpgs caps to test for voltage etc
the irony is i dont like seeing caps and resistors in the signal path digital or analog :)

I agree that the usual large value, reservoir caps used on PSes are detrimental to the LiFePo4 battery PS - they seem to lack the low impedance of these batteries & slow down current delivery. So for high speed current pulses feeding digital chips, they seem to be a bad idea.

On the other hand small value, bypass caps serve a completely different function - which I know you know, Nige.

The issue with taking off these small 100nF caps which are right next to the PS pin of each 595 chip is that it forms the bridge to ground for the self-generated noise coming from the chip - by removing them you remove this electrical channel for that noise - it now has to find an alternative, longer route back to ground. There is no way of knowing this route without studying the layout of the Soekris 4 layer board - it's pretty certain that the next layer down from the top layer is the ground layer so any noise problems may well be mitigated because of this configuration?

I still think that it's generally advisable to use a bypass cap on the 595 chips - if the current ones are affecting the sound then better ones should be used