D/A input bit depth

[Phasenoise]

Hi there,

I was wondering if there was a way I could test to see if my dac will pass bit depths other than 16 bits.

The d/a in question is the Micromega Duo Pro.2, I have had this dac for a while and it's stunning even after all these years, I've never been able to get in to it as the previous owner / whoever has stripped the head on one of the torx screws, I did ask the previous owner and he said he was just trying to have a look but never managed it.

From what I remember the D/A should be either the Philips SAA7323 (16-Bit) or the 7350 (16 - 20 Bit). Most Micromega d/a from this period sport the CS8412 as an input receiver. I have a Meridian 518 so can select various ouput bit depths etc...

Is there any real way to test what bit depths it will pass before truncating the signal?

Thanks in advance, if you have any advice about getting that stripped screw out with out wrecking the lovely case please let me know.

[poynton]

Use a dremmel to cut a slot in the torx for a standard driver or araldite a torx bit in place for a week .
either way use a drop of WD40 and tighten slightly before trying to undo.

Andy

[Kurt von Kubik]

I cannot find the Duo Pro on any DAC list at all.
Only Dialog and Microdac which features TDA 1547 and SAA7321GP – YM3623B.
Micromega mostly featured Philips or Analog Devices DAC chips. Recievers were at that time either Yamaha or Crystal.
Normally the reciever will not be limited in bit depth, but in sampling frequency. Thus the CS 8412 is a 48KHz max chip, later came CS8414 which was intended to be used up to around 100 KHz and then later on the 8416 which accepts 192 KHz or in reality 210KHz or so.
Normally it is in the digital filter where the bit depth is at first limited. The digital filter was mostly set-up to recieve 16 -20 bit data @ 30-50 KHz sampling freq. and to oversample and do interpolation often up to 4-8 times.
Excessive data will at this point be truncated, otherwise this will happen in the DAC itself.
I.e. if the filter can do 20 bit, and the DAC only 18 bit, the 2 least significant bits will be truncated by the DAC chip.
So if you feed your DAC with i.e. 24 bit 48KHz (where ever you will find this) you´ll listen to 48KHz 24 bit less the truncated LSB´s.
If you try to feed the DAC with higher sample frequencies, you´ll experience problems with locking to the signal.

[Phasenoise]

Thanks for the replies,

Micromega in most of their early stuff like the Like the DUO.** Trio.** were all based on the Yamaha input receiver and the SAA7321GP D/A in the case of the Trio there is two D/A in differential mode. The later stuff and the mk1 T-Dac used the CS8412 input receiver still using the 7321GP D/A.

The .2 Stuff in DUO used the SAA7323 or the SAA7350 guessing the same CS8412 as the other Gear was using it.

On the concept series ( STAGE, DAC ) For Stage 4, 5, 6, DAC 1, and DAC 2 all use the TDA1305T as D/A with the DAC 1 and 2 the input receiver is the Philips TDA1315H

All the early stuff like the Logic and optic are basically a DUO.** grafted on to a philips machine they just soldered on to the SAA7220 to get a filtered signal, simple and cool really.

I have tried running a dither tone into the dac and the noise level does drop when input it set to 18 bits or higher, wasn't sure if I could play some test tones into it and measure the output and get some idea if truncation was happening. Not sure what tones or test to do.

I love micromega gear, used to sell it years back and I've had pretty much all of the gear they made apart from the data/dialog which I would love to get hold of.

[Phasenoise]

Sorry forgot to add, i'm only really bothered about the bit depth rather than sample rate, I feed the dac directly into a pair of modded Meridian M60 actives so high the bit depth the less the dither noise added from the Meridian 518 which is acting as the volume control.

[Mooly]

Try heating the head of the screw with a soldering iron.
Old cutters are great too for gripping damaged heads, and you get plenty of "leverage"

[Kurt von Kubik]

Originally Posted by Phasenoise

Sorry forgot to add, i'm only really bothered about the bit depth rather than sample rate, I feed the dac directly into a pair of modded Meridian M60 actives so high the bit depth the less the dither noise added from the Meridian 518 which is acting as the volume control.

It seems impossible to find any data on the 7321 Chip, instead the 7322/23 is popping up.
If they are alike, which I think they are to some degree, the chip is a 16 bit chip, accepting I2S input @ 44,1KHz. The I2S standard accepts different wordlengts, so no problem should occur feeding the DAC chip with 24 bit.
But the excessive bits are truncated at arival.

[Phasenoise]

Whats the best way to measure the noise floor of the d/a? surely increasing the input signal bit depth should result in a drop in the noise floor? so a drop at 18 bits then 20 bits then say a 24 bit signal does reproduce a drop then I would know either it can't be resolved or it's truncating. I thought maybe use sox to create some test signals and then pump them in to the dac via an spdif from an old pc then measure the dac output with the a/d of my PowerBook. Not the best way to measure but can't think of anything else really. Are there any specific patterns or tones that work best to check for differences?

Thanks

[guido]

Originally Posted by Kurt von Kubik

It seems impossible to find any data on the 7321 Chip, instead the 7322/23 is popping up.
If they are alike, which I think they are to some degree, the chip is a 16 bit chip, accepting I2S input @ 44,1KHz. The I2S standard accepts different wordlengts, so no problem should occur feeding the DAC chip with 24 bit.
But the excessive bits are truncated at arival.

Some notes i made long ago:
7320: 1bit dac+dig filter with flaws
7321: 2* 1bit dac+dig filter
7323: improved 7320, full spec
7322: as 7323, but relaxed spec.

Not sure i meant with that.. 2 stereo 1bit dacs to use differential?

Originally Posted by Phasenoise

Thanks for the replies,

Micromega in most of their early stuff like the Like the DUO.** Trio.** were all based on the Yamaha input receiver and the SAA7321GP D/A in the case of the Trio there is two D/A in differential mode.

I have tried running a dither tone into the dac and the noise level does drop when input it set to 18 bits or higher, wasn't sure if I could play some test tones into it and measure the output and get some idea if truncation was happening. Not sure what tones or test to do.

Mmm, two D/A in one chip??

Adding dither does drop the noise floor in the audio band. And it increases it outside the band (which is then filtered out). You can see this without a test tone (just playing "silence") using a spectrum analyser (in my case a pmd100/tda1541 and a 7L5 SA). All with 16 bit.

[guido]

Originally Posted by Phasenoise

From what I remember the D/A should be either the Philips SAA7323 (16-Bit) or the 7350 (16 - 20 Bit).

"The new Duo Pro 2 DAC is based on the true time-aligned differential mode 7323 chips of its predecessor and features a dual mono circuit configuration with multiple power supplies as did the Duo Pro."

Micromega Duo CD3.1 & Duo Pro 2 DAC | Malcolm Steward: audio journalist