Hello spzzzzkt!
Many thanks for your answer and for all your informations.
This is an important help for me!
I ask you only an ulterior detail:
When you talk about the DAC chip supplies from the +/-18V, you say:
" I'd highly recommend that you apply the bypass caps to Radj on the LM337+LM317 regulators as these are required to get best performance from the LM3x7's ".
What does "Radj" means?
Escuse me, but my english is not good, and often the technical words assume different terms and abbreviations in the different languages, so I am not able to translate "Radj".
Many thanks again. Best regards.
Davide.
Many thanks for your answer and for all your informations.
This is an important help for me!
I ask you only an ulterior detail:
When you talk about the DAC chip supplies from the +/-18V, you say:
" I'd highly recommend that you apply the bypass caps to Radj on the LM337+LM317 regulators as these are required to get best performance from the LM3x7's ".
What does "Radj" means?
Escuse me, but my english is not good, and often the technical words assume different terms and abbreviations in the different languages, so I am not able to translate "Radj".
Many thanks again. Best regards.
Davide.
regal: assemblage DAC 2.0 with parts upgrade kit.
spencer: i've possibly damaged one of the smd cap or resistors? i done some basic trouble shooting but could find the cause.
t.s.: take a look at http://www.acoustica.org.uk/t/3pin_reg_notes1.html Radj is the the adjust resistor R2 in the description.
spencer: i've possibly damaged one of the smd cap or resistors? i done some basic trouble shooting but could find the cause.
t.s.: take a look at http://www.acoustica.org.uk/t/3pin_reg_notes1.html Radj is the the adjust resistor R2 in the description.
Reg volt adjus pin
Often better results to use "string" of LEDs (in series) to replace this R2.
There are many simple plans about improving regulator performance on the net - take your time and try out some of the simpler circuits without getting into the complicated feedback I.C. systems - and get a better understanding about what works for you.
It's quite surprising what a "cleaner" regulator does for both digital and (seperate) analogue circuits.
Often better results to use "string" of LEDs (in series) to replace this R2.
There are many simple plans about improving regulator performance on the net - take your time and try out some of the simpler circuits without getting into the complicated feedback I.C. systems - and get a better understanding about what works for you.
It's quite surprising what a "cleaner" regulator does for both digital and (seperate) analogue circuits.
t.s. i forgot to say that adjust bypass caps are C2 or Cadj. It might be worth trying a low-ESR 100uF in this position.
I'm using paralleled zeners plus diode as suggested in the link and discussed in the digital forum as the "raygulator". The LED's would probably be better but the zener/diode units barely squeeze into the available board space so fitting LED's might be a bit of a challenge.
I've finished off the Tangent LNMP so should get a chance to make up a test rig to see what effect the different mods actually have on noise levels.
I'm using paralleled zeners plus diode as suggested in the link and discussed in the digital forum as the "raygulator". The LED's would probably be better but the zener/diode units barely squeeze into the available board space so fitting LED's might be a bit of a challenge.
I've finished off the Tangent LNMP so should get a chance to make up a test rig to see what effect the different mods actually have on noise levels.
as long as you use a low esr cap. personally I'd go for a 100uF panasonic FC/FM or a rubycon ZA in this position.
the pmd100 doesn't work with either cs8412 recovered clock or with the vcxo clock - both of which are 256*fs. I've pulled the resistors off the pmd board and I'll replace them as they look like they've now lost the end caps.
i made the mistake of trying to add the extra reg to the vcxo board and had trouble soldering to the tarnished copper. It looks like I short circuited the center pin to output and ended up pumping 8v into the pll bias and digital chip supplies. This appears to have killed both the 4046 and 4020. The vcxo is still putting out a clock but I'm not sure if I've dumped 8V into the adjust voltage - if that has happened I've also cooked the vcxo.
Anyway I'm now stuck listening to the recovered clock from the CS8412 until I get replacement chip and it is a definite step backwards, especially from the last (working) revision on the VCXO. The imaging is less defined, and the sense of empty space around instruments is far less apparent.
Before the "disaster" I discovered that the output of the VCXO was 90degree out of phase with the original clock. I had thought this might be an issue but rereading the PLL section of the "Audio DAC" pages it seems that the sm5842 is designed to cope with a LCRI that is up to 3/8 of clock period different from the data clock. If I understand correctly this is the equivalent of 120degrees out of phase so the difference is acceptable but does mean that the sm5842 needs to run in "jitter free" mode.
the pmd100 doesn't work with either cs8412 recovered clock or with the vcxo clock - both of which are 256*fs. I've pulled the resistors off the pmd board and I'll replace them as they look like they've now lost the end caps.
i made the mistake of trying to add the extra reg to the vcxo board and had trouble soldering to the tarnished copper. It looks like I short circuited the center pin to output and ended up pumping 8v into the pll bias and digital chip supplies. This appears to have killed both the 4046 and 4020. The vcxo is still putting out a clock but I'm not sure if I've dumped 8V into the adjust voltage - if that has happened I've also cooked the vcxo.
Anyway I'm now stuck listening to the recovered clock from the CS8412 until I get replacement chip and it is a definite step backwards, especially from the last (working) revision on the VCXO. The imaging is less defined, and the sense of empty space around instruments is far less apparent.
Before the "disaster" I discovered that the output of the VCXO was 90degree out of phase with the original clock. I had thought this might be an issue but rereading the PLL section of the "Audio DAC" pages it seems that the sm5842 is designed to cope with a LCRI that is up to 3/8 of clock period different from the data clock. If I understand correctly this is the equivalent of 120degrees out of phase so the difference is acceptable but does mean that the sm5842 needs to run in "jitter free" mode.
Obviously messing about with perf board and adhesive copper sheet isn't a real solution, so I've dug out the PLL board I'd started on. I'll get a proto made once I'm happy with the layout - still quite a way to go.
I've stuck with thru hole but it looks like I'll have to use smd for mlcc decoupling caps and ferrite beads to decouple the chips "guido style". I'll possibly have to use SMD versions of the 4046 and 4020 too, due to space constraints.
I'm trying to keep the boards under 2"x 1.375" to allow them to panelize neatly on the small 5.5" x 4" proto board offered by custompcb.com, which would allow me to sell boards for about $8.00US if anyone was interested. The caveat here is that there are no guarantees that it will work with anything except the CS8412/4 and SM5842 as laid out.
attached is a "life size" pic of layout as it currently stands, I'm still in the process of parts shuffling so it's only partially routed.
I've stuck with thru hole but it looks like I'll have to use smd for mlcc decoupling caps and ferrite beads to decouple the chips "guido style". I'll possibly have to use SMD versions of the 4046 and 4020 too, due to space constraints.
I'm trying to keep the boards under 2"x 1.375" to allow them to panelize neatly on the small 5.5" x 4" proto board offered by custompcb.com, which would allow me to sell boards for about $8.00US if anyone was interested. The caveat here is that there are no guarantees that it will work with anything except the CS8412/4 and SM5842 as laid out.
attached is a "life size" pic of layout as it currently stands, I'm still in the process of parts shuffling so it's only partially routed.
Attachments
Put me down for a board, let me know when you need the Paypal.
I am looking at the PMD100 datasheet and don't see how it would work with the adapter board
It states that the input data must be MSB first. Spencer has the CS8412 configured to MSB last.
Here is what the application notes say to do:
CS8412 - M3, M2, M1, M0 all to ground
PMD100 - XTIM, PROG, JUST, SMUTE, FSEL all to ground
BCPL to 5+
Looks like his will take some serious pin lifting and jumpering to get the PMD100 to work with the CS8412.
Now looking at the DIR9001 adapter kit, I think it 16 bit it outputs a MSB First Right justified, which is what we need.
I am looking at the PMD100 datasheet and don't see how it would work with the adapter board
It states that the input data must be MSB first. Spencer has the CS8412 configured to MSB last.
Here is what the application notes say to do:
CS8412 - M3, M2, M1, M0 all to ground
PMD100 - XTIM, PROG, JUST, SMUTE, FSEL all to ground
BCPL to 5+
Looks like his will take some serious pin lifting and jumpering to get the PMD100 to work with the CS8412.
Now looking at the DIR9001 adapter kit, I think it 16 bit it outputs a MSB First Right justified, which is what we need.
This is from the PMD100 application notes. There are actually 8 modes, you would have to lift the M2 pins and tie it to ground instead of +5V in order to use the PMD100.
An externally hosted image should be here but it was not working when we last tested it.
ah-ha!! i didn't see that you set JUST high on the pmd100!! So cs8412 must be set to mode 5 - 16bit to work. that would explain why it sounded like s**t.
I should have my replacement parts for the vcxo and a few replacement mlcc for the pmd100 board, but it seems like they weren't the problem. Anyway I should know by tonight if the tent vcxo has survived.
I should have my replacement parts for the vcxo and a few replacement mlcc for the pmd100 board, but it seems like they weren't the problem. Anyway I should know by tonight if the tent vcxo has survived.
If you want to connect M2 to GND (instead of +5V), you can cut the BOTTOM trace as show in picture and connect pin 18 of CS8414/12 to GND. Should be quite a easy mod in the D1V3 pcb if you want to try.
Note I have not done on this mode and thus it is take on your own risk trial.
Note I have not done on this mode and thus it is take on your own risk trial.
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.