Experience with this DIY DAC ?

[sendler]

These Big Dac Boards are a very flexible test sled with everything socketed which will allow me to plug in a Twisted Pear Opus when I get some connectors. They are also a lot of fun in that you never know what combination of plug in modules will actually work with each other. Apparently not many people have tried the DIR9001. If you had there would be more raves about the sound quality. It is much better than the CS8416 even through the upsampler and/ or being fed upsampled or direct data from my SRC2496, making the 8416 sound a bit hazy in comparison. There would also be more complaints about the function of the 9001 module. I have two boards and doubles of all of the modules with the exception of never having tried the 4398 yet. One of my DIR9001 modules will only lock to the optical regardless of the input selection jumper settings and the other will only lock to the coax but only if the jumpers are set to optical? I will eventually remove the electronic switch IC under the module to just hard wire them both to coax. One DIR will work with the upsampler and one won't whereas both Big boards will allow the upsampler to work with either 8416. The upsampler is great and offers a more fleshed out, non digital sound except that the CS43122 has a low level whine when the upsampler is used. Anyone listening to 16/44 digital without upsampling it somewhere is missing out.
.
The transformer I/V PCM1798 doesn't whine with the upsampler but I can't get rid a low level buzz. I have tried all of the various ground and loading schemes possible. 40 ohms (plus 20 ohms trans primary) from each leg to ground with the single ended output starred to audio ground under the module, or ground lifted. 80 ohms between the legs (input center tap and resistors ground lifted) which surprisingly still played just the same. No resistors on the legs, loaded only 4-7k at the amps back through the secondary. This tends to be more critical to resistor matching for stereo balance as any deviation between the resistors is multiplied by 64X. This will play only if the input center tap is grounded and seems to sound quite similar although I didn't bother to go back and forth comparing the before trans/ after trans loading schemes yet. I also tried matching the load on both sides with 100R across the primary and 6400R at the amps instead of the usual 20k which interestingly doesn't double the listening level of the I/V. Every possible combination still hums on both boards even with 12v battery power. Maybe I butchered the module board while removing the opamps? I have a couple empty modules to try with PCM1794a later. Listening over the hum, the 43122 sounds better anyway although it still has a slight hiss/ noise floor issue which means I wouldn't want to commit to this set up for an upcoming pc active cross project either. Now that I know the importance of clocks thanks to the DIR9001, I should get my Oettle clock installed with the 4395 in the DCX before I give up on that chip altogether to hear if it is just the receiver that is giving the Big Dac Board most of it's advantage now.

[Bill Fuss]

You will have to show us the post. Mine is clearly set up to upsample to 192. It's not that hard to trace the pinouts, I did it, and I'm old and half blind.

[sendler]

All of my clocks on two new V10 Big boards upsamplers and two DIR9001 just received in the last month are still 24.576.

[Bill Fuss]

Originally Posted by sendler

These Big Dac Boards are a very flexible test sled with everything socketed which will allow me to plug in a Twisted Pear Opus when I get some connectors. They are also a lot of fun in that you never know what combination of plug in modules will actually work with each other. Apparently not many people have tried the DIR9001. If you had there would be more raves about the sound quality. It is much better than the CS8416 even through the upsampler and/ or being fed upsampled or direct data from my SRC2496, making the 8416 sound a bit hazy in comparison. There would also be more complaints about the function of the 9001 module. I have two boards and doubles of all of the modules with the exception of never having tried the 4398 yet. One of my DIR9001 modules will only lock to the optical regardless of the input selection jumper settings and the other will only lock to the coax but only if the jumpers are set to optical? I will eventually remove the electronic switch IC under the module to just hard wire them both to coax. One DIR will work with the upsampler and one won't whereas both Big boards will allow the upsampler to work with either 8416. The upsampler is great and offers a more fleshed out, non digital sound except that the CS43122 has a low level whine when the upsampler is used. Anyone listening to 16/44 digital without upsampling it somewhere is missing out.
.
The transformer I/V PCM1798 doesn't whine with the upsampler but I can't get rid a low level buzz. I have tried all of the various ground and loading schemes possible. 40 ohms (plus 20 ohms trans primary) from each leg to ground with the single ended output starred to audio ground under the module, or ground lifted. 80 ohms between the legs (input center tap and resistors ground lifted) which surprisingly still played just the same. No resistors on the legs, loaded only 4-7k at the amps back through the secondary. This tends to be more critical to resistor matching for stereo balance as any deviation between the resistors is multiplied by 64X. This will play only if the input center tap is grounded and seems to sound quite similar although I didn't bother to go back and forth comparing the before trans/ after trans loading schemes yet. I also tried matching the load on both sides with 100R across the primary and 6400R at the amps instead of the usual 20k which interestingly doesn't double the listening level of the I/V. Every possible combination still hums on both boards even with 12v battery power. Maybe I butchered the module board while removing the opamps? I have a couple empty modules to try with PCM1794a later. Listening over the hum, the 43122 sounds better anyway although it still has a slight hiss/ noise floor issue which means I wouldn't want to commit to this set up for an upcoming pc active cross project either. Now that I know the importance of clocks thanks to the DIR9001, I should get my Oettle clock installed with the 4395 in the DCX before I give up on that chip altogether to hear if it is just the receiver that is giving the Big Dac Board most of it's advantage now.

Great post Scott,

Wait til you try it direct without any receiver!

The 4398 is their flagship chip, hows come you haven't tried it yet?

I have yet to experiment with a current out dac, but do you think the 1794 might be having internal problems with the relatively high load impedance? Maybe trying 5 or 10 ohms load could be the solution.

[sendler]

Originally Posted by Bill Fuss

Great post Scott,

Wait til you try it direct without any receiver!

The 4398 is their flagship chip, hows come you haven't tried it yet?

I have yet to experiment with a current out dac, but do you think the 1798 might be having internal problems with the relatively high load impedance? Maybe trying 5 or 10 ohms load could be the solution.

Snow tells me that the 43122 sounds better than the 4398 and I believe her judging from what I am hearing. The only things keeping me from declaring a rave are the slight hiss and the whine when using the upsampler. The 43122 doesn't seem to like something about 192k even though the spec sheet says it can handle it. Is there hardware control on the upsampler to chose different frequencies? Maybe it would like 176K better. As is, it is much better than the CS8420/ AK 4395 in my Behringer.
.
The resistor/ transformer I/V on the PCM1798 requires at least that much resistance even though I am using 1:8 transformers in order to play loud enough. The 1794 will need only 1/2 of that and mono mode will be 1/2 again, taking the required resistance all the way down to 10-15R to get somewhere around 0dbu which is all I need and will be right in the sweet spot for load. I had 160R on each leg of the PCM1798 for test at one point. It was very loud but very distorted.

[Bill Fuss]

Hiss and whine, sounds like me when i'm in one of my moods.

I'm still under the impression that the 43122 was a design excercise predating the 4398. They may have had consistancy problems with them because they were using new techniques that have supposedly been perfected now in the 4398. Anyway, I tend to discount vendor's opinions, especially when they are selling something as an upgrade.

I had it in mind that you were using the 1794 for some reason. The 1798 would seem to require a higher trafo ratio than 1/8

[Ryssen]

I decided to take a risk this evening..
I plugged in the PCM1794 modules coupled to the Lundahls trafos.no smoke...
but it was sound.
Have just been listening a short while with audio from my satelite reciever,
I think it sounds darker than the 4398,but it is still very detailed.
And the sound level is about the same with the 120 Ohms resistor as I/V.
Will listen more closley later,I have to finnish my subamp so I get some bass 2.

[sendler]

Anyone in the USA have a 4398 module for sale?

[McGyver]

Originally Posted by sendler

Anyone in the USA have a 4398 module for sale?

But why bother? In my opinion CS4397 sounds better than CS4398. And CS43122 is probably essentially the same as CS4397.

And when I tried PCM1798 with passive conversion (150R Riken) and tube gain stage (16x gain) ...
Now I'm absolutely sure there's no return to Cirrus.

[sendler]

Originally Posted by McGyver

But why bother? In my opinion CS4397 sounds better than CS4398. And CS43122 is probably essentially the same as CS4397.

And when I tried PCM1798 with passive conversion (150R Riken) and tube gain stage (16x gain) ...
Now I'm absolutely sure there's no return to Cirrus.

Wow! so much gain. 150R per leg? And 16X. That's 18 db more than I am using with my PCM1798. Have you tried smaller resistors? When I had 160R per leg to ground, the sound was horribly distorted. The lower the resistance the better right? Most active I/V circuits aim for a virtual 15R or lower. Did you have tubes following the 4397?