PDA

View Full Version : Some Qs on the output configuration



occam
2005-08-10, 05:54
1.The pcm1748 is capable of digital attenuation of -64db in steps of .5db, for a total of 128 settings. The SB2 implements this feature with 40 steps. Is there any practical way of expanding the configuration of the SB2 to access this finer granularity, such as reprogramming EEPROM, if its controlled by such?

2. The output opamp is configured for a gain of about 2. Is it inverting or non inverting? Is that NJM2041 configured as straight amp or as a lowpass filter? If so, how many poles and what is the pole frequency? Is the NJM2041 stable at Av<=1? Can I adjust the gain by changing Ri & Rf?

Many, many thanks in advance.

dean
2005-08-10, 07:16
On Aug 10, 2005, at 5:54 AM, occam wrote:
> 1.The pcm1748 is capable of digital attenuation of -64db in steps of
> .5db, for a total of 128 settings. The SB2 implements this feature
> with
> 40 steps. Is there any practical way of expanding the configuration of
> the SB2 to access this finer granularity, such as reprogramming
> EEPROM,
> if its controlled by such?
The Squeezebox2 does the attenuation in software, processing the
samples with 24 bit accuracy before they get to the DAC, but with
only 8 bits of resolution in the multiplier. We expect to expand
that range for the next major firmware release.

> 2. The output opamp is configured for a gain of about 2. Is it
> inverting or non inverting? Is that NJM2041 configured as straight amp
> or as a lowpass filter? If so, how many poles and what is the pole
> frequency? Is the NJM2041 stable at Av<=1? Can I adjust the gain by
> changing Ri & Rf?
Will let Sean reply to that one.

-dean

seanadams
2005-08-10, 20:04
We are not using the TI's built-in volume function. It is done in software before the DA converter (so we can apply it to either the s/pdif or the analog or both).

Here's a schematic of the output. Note the AK4384 is not populated.

http://www.seanadams.com/dac.pdf

You can tweak the gain a little but you can't go much above 6vpp with this circuit (9V supply biased at 5) - what do you want to do?

occam
2005-08-10, 22:58
The output opamp question was with regards to dropping the output gain, basically making it a voltage follower. And yes, I see why you're not using Vcom for biasing as you're not using the 5v as your rail for the opamp.

Many thanks.

Robin Bowes
2005-08-11, 02:57
seanadams wrote:
> We are not using the TI's built-in volume function. It is done in
> software before the DA converter (so we can apply it to either the
> s/pdif or the analog or both).
>
> Here's a schematic of the output. Note the AK4384 is not populated.
>
> http://www.seanadams.com/dac.pdf
>
> You can tweak the gain a little but you can't go much above 6vpp with
> this circuit (9V supply biased at 5) - what do you want to do?

Hmmm, I'm a little confused. I obviously don't understand analogue
electronics as well as I thought I did as I don't understand how the
NJM2041M is being used.

You seem to be feeding the audio output from the PCM1748 directly to the
Left and Right outputs, but also to the non-inverting input of the
NJM2041M. The NJM2041M output then feeds J3 & J4 (what are these?).

What configuration is this?

More pertinently, I'd like to reduce the output level by, say, half. Is
which components would I need to change, and is there a simple formula
that I can use to calculate the relevant values?

Thanks,

R.

--
http://robinbowes.com

If a man speaks in a forest,
and his wife's not there,
is he still wrong?

Patrick Dixon
2005-08-11, 04:55
Hmm, I not sure it's entirely fair to ask Sean for detailed instructions on how to hack his hardware design, but just to point you in the right direction:

The left and right channels from the dac also feed the headphone amp.

The gain of the output stage is 8K25/4K7. I'm sure with a bit of googling you'll figure out what the 330pF does.

I think you'll find J3 & J4 are the phono output sockets.

Robin Bowes
2005-08-11, 05:25
Patrick Dixon wrote:
> Hmm, I not sure it's entirely fair to ask Sean for detailed instructions
> on how to hack his hardware design, but just to point you in the right
> direction:
>
> The left and right channels from the dac also feed the headphone amp.

Ah, that makes sense. My brain obviously wasn't working when I looked
athis earlier this morning.

> The gain of the output stage is 8K25/4K7.

Yeah, a standard inverting amp configuration. So, replace the 4k7 with
8k25 to give unity gain.

> I'm sure with a bit of
> googling you'll figure out what the 330pF does.

Well, without googling I can say it's a single-pole low pass filter.
With a bit of googling, I can tell you that the roll-off is 20dB /
decade and f0 is 1/(2pRC) = 64.3khz.

> I think you'll find J3 & J4 are the phono output sockets.

That's what threw me! Well, that and a bit of brain fade!

R.

--
http://robinbowes.com

If a man speaks in a forest,
and his wife's not there,
is he still wrong?

Davey
2005-08-11, 05:33
Gents,

The 330pf shunts the 8.25k at high frequencies (above the audio range) to reduce the circuit gain so noise performance and stability are improved. The circuit presently has a gain of 1.75 or 4.9db. If you reduce the value of the 8.25k resistor you'll lower the gain of the circuit. (4.7k would yield unity gain.) I don't believe this op-amp will have any kind of stability problem with that.

I'm not quite sure why you'd want to lower the gain anyway. The analog output is already a tad lower than what you'd see from a typical player. (I measured 1.86Vrms when playing a 0dbFS recorded track.)

Sorry, Robin answered the question while I was typing. :)

Cheers,

Davey.

Robin Bowes
2005-08-11, 05:42
Davey wrote:
> reduce the value of the 8.25k resistor you'll lower the gain of the
> circuit. (4.7k would yield unity gain.) I don't believe this op-amp
> will have any kind of stability problem with that.

Ah, but that would affect the frequency response of the circuit too.
Increasing the 4.7k value to 8.25k would also yield unity gain but
without affecting the filter response.

R.

--
http://robinbowes.com

If a man speaks in a forest,
and his wife's not there,
is he still wrong?

Robin Bowes
2005-08-11, 05:48
Robin Bowes wrote:
>
> Yeah, a standard inverting amp configuration. So, replace the 4k7 with
> 8k25 to give unity gain.

Hmmm, does that mean that the analogue output of the SB2 is inverted
from the "input" (i.e. the original digitised source) ?

R.
--
http://robinbowes.com

If a man speaks in a forest,
and his wife's not there,
is he still wrong?

Patrick Dixon
2005-08-11, 06:34
I shouldn't think so because the dac has a flag for normal or inverting phase, and I'm confident that Sean will have programmed the 'right' option. Vinne's Red Wine Audio mods might be though, since he's dispensing with the inverting op-amp stage altogether!

Davey
2005-08-11, 06:43
Yeah, that's a good point. The cutoff frequency is already pretty high and reducing Rf would raise it higher. I guess my thinking was that it would be easier to tack a resistor on top of the existing feedback resistor vice changing out the other one.

Regarding absolute polarity inversion....it would be fairly easy to check. You'd need an impulsive signal for a source and then have to monitor with a scope. Anyway, it really doesn't matter since there's no way to know which way is "right" or "wrong" on the source material and there's much discussion whether this effect is even audible or meaningful.

Cheers,

Davey.

Robin Bowes
2005-08-11, 06:45
Patrick Dixon wrote:
> I shouldn't think so because the dac has a flag for normal or inverting
> phase, and I'm confident that Sean will have programmed the 'right'
> option.

Sean - can you confirm?

> Vinne's Red Wine Audio mods might be though, since he's
> dispensing with the inverting op-amp stage altogether!

Heh. I can see the headlines now "Audiophile in 180 out-of-phase shock"

:)

R.

--
http://robinbowes.com

If a man speaks in a forest,
and his wife's not there,
is he still wrong?

Robin Bowes
2005-08-11, 06:50
Davey wrote:
> Regarding absolute polarity inversion....it would be fairly easy to
> check. You'd need an impulsive signal for a source and then have to
> monitor with a scope. Anyway, it really doesn't matter since there's
> no way to know which way is "right" or "wrong" on the source material
> and there's much discussion whether this effect is even audible or
> meaningful.

It's been absolutely proven that absolute phase is not detectible. But
it's generally good practice to preserve phase in audio equipment.

R.

--
http://robinbowes.com

If a man speaks in a forest,
and his wife's not there,
is he still wrong?

Vinnie R.
2005-08-11, 06:57
Patrick Dixon wrote:
> I shouldn't think so because the dac has a flag for normal or inverting
> phase, and I'm confident that Sean will have programmed the 'right'
> option.

Sean - can you confirm?

> Vinne's Red Wine Audio mods might be though, since he's
> dispensing with the inverting op-amp stage altogether!

Heh. I can see the headlines now "Audiophile in 180 out-of-phase shock"

:)

R.

--
http://robinbowes.com

If a man speaks in a forest,
and his wife's not there,
is he still wrong?

IMO, it really doesn't matter if the opamp inverts or not... just listen!

However, you do want both channels to be in phase or there will be serious "Audiophile in 180 out-of-phase shock." :-)

Regards,

Vinnie

seanadams
2005-08-11, 07:00
Sean - can you confirm?

> Vinne's Red Wine Audio mods might be though, since he's
> dispensing with the inverting op-amp stage altogether!

Heh. I can see the headlines now "Audiophile in 180 out-of-phase shock"
Well, I agreee one or the other is inverted. :)

I'll double check it later today. One thing I know for sure is correct (that's easy to screw up) is phase between channels - it's easy to be off by one sample if you have an inverted LRCK.

Robin Bowes
2005-08-11, 07:12
Vinnie R. wrote:
> IMO, it really doesn't matter if the opamp inverts or not... just
> listen!

Agreed.

> However, you do want both channels to be in phase or there will be
> serious "Audiophile in 180 out-of-phase shock." :-)

:)

R.
--
http://robinbowes.com

If a man speaks in a forest,
and his wife's not there,
is he still wrong?

Davey
2005-08-11, 07:56
Davey wrote:[color=blue]

It's been absolutely proven that absolute phase is not detectible. But
it's generally good practice to preserve phase in audio equipment.

R.


:) Many golden-eared audiophiles would disagree with you on that.

Cheers,

Davey.

Robin Bowes
2005-08-11, 09:43
Davey wrote:
> Robin Bowes Wrote:
>
>>Davey wrote:[color=blue]
>>
>>It's been absolutely proven that absolute phase is not detectible. But
>>
>>it's generally good practice to preserve phase in audio equipment.
>>
>>R.
>>
>
>
> :) Many golden-eared audiophiles would disagree with you on that.

;)

R.

--
http://robinbowes.com

If a man speaks in a forest,
and his wife's not there,
is he still wrong?

seanadams
2005-08-11, 19:00
I'll double check it later today.
Sorry, didn't get around to this today. Will have a look ASAP though.

seanadams
2005-08-12, 14:29
D'oh - it's inverted. Firmware fix coming.

Robin Bowes
2005-08-14, 14:50
seanadams wrote:
> D'oh - it's inverted. Firmware fix coming.

Sean,

Any chance this can be a configurable option?

For example, if I (or others - hi Vinnie!) decide to bypass the op-amp
output stage then we would require the output of the DAC to be "correct"
phase (actually, "correct polarity" would probably be a better
description) rather than the "inverted" polarity that would be required
to feed the inverting output stage.

R.
--
http://robinbowes.com

If a man speaks in a forest,
and his wife's not there,
is he still wrong?

dean
2005-08-14, 16:50
Yes, that's possible. Please file an enhancement request on
bugs.slimdevices.com.

-dean

On Aug 14, 2005, at 2:50 PM, Robin Bowes wrote:

> seanadams wrote:
>
>> D'oh - it's inverted. Firmware fix coming.
>>
>
> Sean,
>
> Any chance this can be a configurable option?
>
> For example, if I (or others - hi Vinnie!) decide to bypass the op-
> amp output stage then we would require the output of the DAC to be
> "correct" phase (actually, "correct polarity" would probably be a
> better description) rather than the "inverted" polarity that would
> be required to feed the inverting output stage.
>
> R.
> --
> http://robinbowes.com
>
> If a man speaks in a forest,
> and his wife's not there,
> is he still wrong?
>
>

robinbowes
2005-08-15, 01:31
[QUOTE=dean]Yes, that's possible. Please file an enhancement request on
bugs.slimdevices.com.

http://bugs.slimdevices.com/show_bug.cgi?id=1976

occam
2005-08-16, 17:06
Sean,

Your diagram of the ps -



AC -> 5V 2A -----> Dual ----> 3v3 (IO, wireless, logic)
SMPS | Switcher ----> 1v2 (CPU core)
|
|----> Linear reg ----> 3v3 (headphone)
|
|
|----> Switcher ----> Multiplier -----> 11VAC VFD ef
| |
| |
14VDC | |-------> 55VDC VFD vdd2
----------RC filter<-----|
|
|
|------> Linear reg ---> 9vdc (op-amps)
|
|------> Linear reg ---> 5vdc (DAC)



Which 3.3v regulator is providing the digital V+ for the dac? Is it the supply from the dual switcher, or the 3.3v linear regulator that is also feeding the headphone amp? What is the Part# of the 3.3v linear regulator?

Thanks in advance,
Paul

seanadams
2005-08-17, 10:19
Which 3.3v regulator is providing the digital V+ for the dac? Is it the supply from the dual switcher, or the 3.3v linear regulator that is also feeding the headphone amp? What is the Part# of the 3.3v linear regulator?

Thanks in advance,
Paul

Digital side of the DAC is powered by the digital 3v3 rail from the switcher. You could move it to the headphone amp if you want.

void
2005-08-21, 09:36
Heh. I can see the headlines now "Audiophile in 180 out-of-phase shock" :)
Most audio systems mess up the phase linearity so badly, that you won't hear an improvement when switching the absolute phase. Even on many very expensive high-end systems, the difference is often subtle.

But on those few systems which have accurate phase linearity, I think few people would have a problem to hear the improvement when absolute phase is right (less dull, more realistic, tangible, richer sound). A good single driver speaker and a good source help a lot.
Some CD's are out-of-phase, on some CD's an individual instrument is recorded out-of-phase, when you switch phase, this instrument will sound better but the rest will sound worse.

It would be nice if the absolute phase could be selected on the SB2. And it would be great if that could be implemented on Digital out (my best DAC is inverting).

Patrick Dixon
2005-08-21, 09:57
And it would be great if that could be implemented on Digital out (my best DAC is inverting).If you think it makes a difference, just switch your + & - speaker wires at one end.

radish
2005-08-21, 10:05
Most audio systems mess up the phase linearity so badly, that you won't hear an improvement when switching the absolute phase. Even on many very expensive high-end systems, the difference is often subtle.

But on those few systems which have accurate phase linearity, I think few people would have a problem to hear the improvement when absolute phase is right (less dull, more realistic, tangible, richer sound). A good single driver speaker and a good source help a lot.
Some CD's are out-of-phase, on some CD's an individual instrument is recorded out-of-phase, when you switch phase, this instrument will sound better but the rest will sound worse.

It would be nice if the absolute phase could be selected on the SB2. And it would be great if that could be implemented on Digital out (my best DAC is inverting).


This page goes into more detail : http://home.att.net/~Lemon.J.Russell/audphase.htm. Not sure I'm entirely convinced, but the problem strikes me as more one of speaker design. A cone could easily behave differently in a push vs a pull due to mechanical design, and it's that difference which causes any audible change. Fix that (electrostatics?) and you're golden. As for human voices having a "large unidirectional impulse", I'm not sure how that could arise given the nature of vocal cords as, well, cords. Percussion I can much more easily believe - the first impact of a stick on a drum skin will always be of higher amplitude than any subsequent vibration, and will always be in the same direction.

styx
2005-08-24, 05:34
If you think it makes a difference, just switch your + & - speaker wires at one end.

That's not particularly practical, is it? I'm voting for a software toggle! There's actually a quite old bug in the bugzilla requesting this as well.

John Stimson
2005-08-31, 12:52
It seems pretty practical to me. Way easier than modifying the software to do it. How often do you change between inverted and noninverted?