From what i can remember from an analog devices application note:
DAC ground pins should be connected to analog gnd. ALL of them, so also dgnd. connect agnd and dgnd pins from the dac with the shortest connection possible (e.g. see pcm63 datasheet, groundplane sounds fine).
If there is one dac, the point to connect agnd and dgnd is at the dac. If there is more than one dac (or adc) on a pcb: connec at powersupplies.
But it all depends on the dac i guess. E.g. a 1541 does not have much currents go through the groundpens.
DAC ground pins should be connected to analog gnd. ALL of them, so also dgnd. connect agnd and dgnd pins from the dac with the shortest connection possible (e.g. see pcm63 datasheet, groundplane sounds fine).
If there is one dac, the point to connect agnd and dgnd is at the dac. If there is more than one dac (or adc) on a pcb: connec at powersupplies.
But it all depends on the dac i guess. E.g. a 1541 does not have much currents go through the groundpens.
CS8416 receiver
Hi I am back
So I decided that I will go with two clocks. I think we need to make the PCB that accepts both 8 pin and 14 pin oscillators. I will start with some cheep stuff and than slowly upgrade.
I divided the schematics in 3 parts:
-CS8416 and sorounding circuits
-AD1896 and sorounding circuit
-PMC1798 and sorounding circuit
In this way we can more easy clean the bugs. I am starting from your schematics and slowly checking some things. Still did not finish the first one To much to do at work...
But I have found one error (which is always good ):
CS8416 can accept on VL 3.3 to 5V and on a VD only 3.3V. Kim did right to put all on 3.3V. But, if you do that you can not use torx179 on 5V since it will demage the chip! so VL has to go on 5V. There are components like TORX179L that will work on 3.3V but I cann't find them in switzerland.
So, one should look at the datasheet and see if the difference between the VDD and VL and than add another TL or foget about the inputs and leave them on option either connected to ground or to the output device. I am thinking about this, since I will use input transformer. I do not like to connect my clean ground to some other device that I do not know and polute my signal. For my nos I made my own digital transformer (a core + 3 coper windings) but I would like to find something small for a resonable price. Second, I am not sure that I would like to use optical in, more interesting for me is USB that will be on the separate board and connected to the input again via transformer.
For the input I think it is better to put something like a buffer eg. 74VHC125M. One will use wires to connect the switch to the fron pannel and any signal that is picked up on the VDD line must be stoped to enter CS chip. I think this is the way to do it. Do we now put another TL431 for it or just decouple with 10n or 100n or 22u or with some combination?
I was browsing datasheets on capacitors and saw some tasty Panasonic SMDs that I think will be as good as OSCONS. They are EEFCD0J100R 47uF 6.3V. They have 0.05Ohm ESR on 400kHz and are around 2Eur each. So more expensive than Oscons but smaller place on PCB. This means also less posibility to roll your own capacitors after... And at farnell for these price drops for after 50pce and for oscons is after 10 (I think is to 1.7Eur). Concerning capacitors since the oscons 47u and 22u are the same footprint (biger is just taller) what would you rather use. I mean if I have 10 of them on the board the end diference in price will be smaller than 5eur so if the bigger is better why not use 47u?
I must go back to work now
Pred
Hi I am back
So I decided that I will go with two clocks. I think we need to make the PCB that accepts both 8 pin and 14 pin oscillators. I will start with some cheep stuff and than slowly upgrade.
I divided the schematics in 3 parts:
-CS8416 and sorounding circuits
-AD1896 and sorounding circuit
-PMC1798 and sorounding circuit
In this way we can more easy clean the bugs. I am starting from your schematics and slowly checking some things. Still did not finish the first one
To much to do at work...But I have found one error (which is always good
):CS8416 can accept on VL 3.3 to 5V and on a VD only 3.3V. Kim did right to put all on 3.3V. But, if you do that you can not use torx179 on 5V since it will demage the chip! so VL has to go on 5V. There are components like TORX179L that will work on 3.3V but I cann't find them in switzerland.
So, one should look at the datasheet and see if the difference between the VDD and VL and than add another TL or foget about the inputs and leave them on option either connected to ground or to the output device. I am thinking about this, since I will use input transformer. I do not like to connect my clean ground to some other device that I do not know and polute my signal. For my nos I made my own digital transformer (a core + 3 coper windings) but I would like to find something small for a resonable price. Second, I am not sure that I would like to use optical in, more interesting for me is USB that will be on the separate board and connected to the input again via transformer.
For the input I think it is better to put something like a buffer eg. 74VHC125M. One will use wires to connect the switch to the fron pannel and any signal that is picked up on the VDD line must be stoped to enter CS chip. I think this is the way to do it. Do we now put another TL431 for it or just decouple with 10n or 100n or 22u or with some combination?
I was browsing datasheets on capacitors and saw some tasty Panasonic SMDs that I think will be as good as OSCONS. They are EEFCD0J100R 47uF 6.3V. They have 0.05Ohm ESR on 400kHz and are around 2Eur each. So more expensive than Oscons but smaller place on PCB. This means also less posibility to roll your own capacitors after... And at farnell for these price drops for after 50pce and for oscons is after 10 (I think is to 1.7Eur). Concerning capacitors since the oscons 47u and 22u are the same footprint (biger is just taller) what would you rather use. I mean if I have 10 of them on the board the end diference in price will be smaller than 5eur so if the bigger is better why not use 47u?
I must go back to work now
Pred
Re: Re: CS8416 receiver
kokk
But I have found one error (which is always good ):
CS8416 can accept on VL 3.3 to 5V and on a VD only 3.3V. Kim did right to put all on 3.3V. But, if you do that you can not use torx179 on 5V since it will demage the chip! so VL has to go on 5V. There are components like TORX179L that will work on 3.3V but I cann't find them in switzerland.
So, one should look at the datasheet and see if the difference between the VDD and VL and than add another TL or foget about the inputs and leave them on option either connected to ground or to the output device. I am thinking about this, since I will use input transformer. I do not like to connect my clean ground to some other device that I do not know and polute my signal. For my nos I made my own digital transformer (a core + 3 coper windings) but I would like to find something small for a resonable price. Second, I am not sure that I would like to use optical in, more interesting for me is USB that will be on the separate board and connected to the input again via transformer.
For the input I think it is better to put something like a buffer eg. 74VHC125M. One will use wires to connect the switch to the fron pannel and any signal that is picked up on the VDD line must be stoped to enter CS chip. I think this is the way to do it. Do we now put another TL431 for it or just decouple with 10n or 100n or 22u or with some combination?
I was browsing datasheets on capacitors and saw some tasty Panasonic SMDs that I think will be as good as OSCONS. They are EEFCD0J100R 47uF 6.3V. They have 0.05Ohm ESR on 400kHz and are around 2Eur each. So more expensive than Oscons but smaller place on PCB. This means also less posibility to roll your own capacitors after... And at farnell for these price drops for after 50pce and for oscons is after 10 (I think is to 1.7Eur). Concerning capacitors since the oscons 47u and 22u are the same footprint (biger is just taller) what would you rather use. I mean if I have 10 of them on the board the end diference in price will be smaller than 5eur so if the bigger is better why not use 47u?
I must go back to work now
Pred [/B][/QUOTE]
pred said:Hi I am back
So I decided that I will go with two clocks. I think we need to make the PCB that accepts both 8 pin and 14 pin oscillators. I will start with some cheep stuff and than slowly upgrade.
I divided the schematics in 3 parts:
-CS8416 and sorounding circuits
-AD1896 and sorounding circuit
-PMC1798 and sorounding circuit
In this way we can more easy clean the bugs. I am starting from your schematics and slowly checking some things. Still did not finish the first one
Pred
kokk
But I have found one error (which is always good
):CS8416 can accept on VL 3.3 to 5V and on a VD only 3.3V. Kim did right to put all on 3.3V. But, if you do that you can not use torx179 on 5V since it will demage the chip! so VL has to go on 5V. There are components like TORX179L that will work on 3.3V but I cann't find them in switzerland.
So, one should look at the datasheet and see if the difference between the VDD and VL and than add another TL or foget about the inputs and leave them on option either connected to ground or to the output device. I am thinking about this, since I will use input transformer. I do not like to connect my clean ground to some other device that I do not know and polute my signal. For my nos I made my own digital transformer (a core + 3 coper windings) but I would like to find something small for a resonable price. Second, I am not sure that I would like to use optical in, more interesting for me is USB that will be on the separate board and connected to the input again via transformer.
For the input I think it is better to put something like a buffer eg. 74VHC125M. One will use wires to connect the switch to the fron pannel and any signal that is picked up on the VDD line must be stoped to enter CS chip. I think this is the way to do it. Do we now put another TL431 for it or just decouple with 10n or 100n or 22u or with some combination?
I was browsing datasheets on capacitors and saw some tasty Panasonic SMDs that I think will be as good as OSCONS. They are EEFCD0J100R 47uF 6.3V. They have 0.05Ohm ESR on 400kHz and are around 2Eur each. So more expensive than Oscons but smaller place on PCB. This means also less posibility to roll your own capacitors after... And at farnell for these price drops for after 50pce and for oscons is after 10 (I think is to 1.7Eur). Concerning capacitors since the oscons 47u and 22u are the same footprint (biger is just taller) what would you rather use. I mean if I have 10 of them on the board the end diference in price will be smaller than 5eur so if the bigger is better why not use 47u?
I must go back to work now
Pred [/B][/QUOTE]
pred said:
Please try to make the quotes right and only quote as much as needed. pred said:Hi I am back
So I decided that I will go with two clocks. I think we need to make the PCB that accepts both 8 pin and 14 pin oscillators. I will start with some cheep stuff and than slowly upgrade.
I divided the schematics in 3 parts:
-CS8416 and sorounding circuits
-AD1896 and sorounding circuit
-PMC1798 and sorounding circuit
In this way we can more easy clean the bugs. I am starting from your schematics and slowly checking some things. Still did not finish the first one
Pred
Yes that is a good way to go. Please make a input where you can connect an external clock (24.576Mhz-> critical for the PCM179X)
We must also remember to read the datasheets for the counters, the numbers off dividing setups must be change a bit from the qmarsch schematic (to get 12.288Mhz and 192Khz from our 24.576Mhz clock)
pred said:But I have found one error (which is always good
CS8416 can accept on VL 3.3 to 5V and on a VD only 3.3V. Kim did right to put all on 3.3V. But, if you do that you can not use torx179 on 5V since it will demage the chip! so VL has to go on 5V. There are components like TORX179L that will work on 3.3V but I cann't find them in switzerland.
So, one should look at the datasheet and see if the difference between the VDD and VL and than add another TL or foget about the inputs and leave them on option either connected to ground or to the output device. I am thinking about this, since I will use input transformer. I do not like to connect my clean ground to some other device that I do not know and polute my signal. For my nos I made my own digital transformer (a core + 3 coper windings) but I would like to find something small for a resonable price. Second, I am not sure that I would like to use optical in, more interesting for me is USB that will be on the separate board and connected to the input again via transformer.
Pred [/B]
OK great-> I wasn´t aware off that]
I can ask the guy from Dantimax about the torx179 3.3V:
http://electronics.dantimax.dk/Kits/index.html
He can buy both Torx179 and PCM1794 for a good price
I want a optical input if it not to difficult to implement.
I have not draw a input transformer to start with, because I don´t know how much influence it has on the sound quality??
But it sounds cool with your homemade transformer, can we get a manual
pred said:
You can put the TL431 between (VDD) CS8416 and the "select Buffer" and then decouple both as a normal TL431 setup in our schematic
Can you find out how to make a three inputs selector?
pred said:I was browsing datasheets on capacitors and saw some tasty Panasonic SMDs that I think will be as good as OSCONS. They are EEFCD0J100R 47uF 6.3V. They have 0.05Ohm ESR on 400kHz and are around 2Eur each. So more expensive than Oscons but smaller place on PCB. This means also less posibility to roll your own capacitors after... And at farnell for these price drops for after 50pce and for oscons is after 10 (I think is to 1.7Eur). Concerning capacitors since the oscons 47u and 22u are the same footprint (biger is just taller) what would you rather use. I mean if I have 10 of them on the board the end diference in price will be smaller than 5eur so if the bigger is better why not use 47u?
I must go back to work now
Pred [/B]
These caps are looking good
But I think I will stick to the Oscons SP "leaded" (20V 22uF), so people have better optinals, The SP serie should be ok for audio and they are also very good for decoupling Digital circuits
kimschips said:
These caps are looking good
But I think I will stick to the Oscons SP "leaded" (20V 22uF), so people have better optinals, The SP serie should be ok for audio and they are also very good for decoupling Digital circuits
I have doubts for the capacitors again
I mean do we really need anything except OSCONS? We use a ferrite bead to kill everuthing bigger than MHZ range, right? Oscons have low ESR, so they are still a good capacitor in the 100KHz range, right? If the capacitor is not good enough it can kill the perfomance of the OSCONS, right? So why parallel it with another capacitor? We can add it, it is not a problem, but then engineer in me would say, put the ceramic, it is made for the purpose. At least for the digital signals. I think that ceramics can be found in 0805 SMD package and put bellow the digital circuit, so in that way we can save space. On the other hand I saw only polyesters in SMD package in farnell catalogue. And they are mighty big, so we must put it on the top layer and put the OSCON furher away from the power pins
So again the question: Do we need this (these) extra capacitors for decoupling to improve the performance?
Pred
Digital transformer
Well it is not my ide I saw it somwhere on the net. You take a toroidal feritte (but not too high permeability if I remember well). I used something like.
Ferroxcube
Which one I do not remeber since mine is white and it should be 3C11 material, but I cannot find it here. My student did the order so I do not have a place to look for the real reference.
Then you take solid core (like 0.5mm) isolated copper wire, like they use for transformes. You wind three turns (not more than five, it is high frequency more turns more problems), on one side and the same number of turns on the other side. Feritte is making a magnetic path to create 1:1 transformer.
The number of turns will determine the frequency range so I experimented with 3 4 and 5 turnes. Used signal generator with square wave on one side, 75 ohms on primary 75ohms on secondary. On the secondary I connected the oscilloscope and looked. If I remember well, with 4 turns I had a good signal from 500KHz to 10-15 MHz.
And that is it.
Pred
PS
It is mounted on the outside of the PCB and with this project I will test the difference with or withot.
kimschips said:I want a optical input if it not to difficult to implement.
I have not draw a input transformer to start with, because I don´t know how much influence it has on the sound quality??
But it sounds cool with your homemade transformer, can we get a manual
Well it is not my ide I saw it somwhere on the net. You take a toroidal feritte (but not too high permeability if I remember well). I used something like.
Ferroxcube
Which one I do not remeber since mine is white and it should be 3C11 material, but I cannot find it here. My student did the order so I do not have a place to look for the real reference.
Then you take solid core (like 0.5mm) isolated copper wire, like they use for transformes. You wind three turns (not more than five, it is high frequency more turns more problems), on one side and the same number of turns on the other side. Feritte is making a magnetic path to create 1:1 transformer.
The number of turns will determine the frequency range so I experimented with 3 4 and 5 turnes. Used signal generator with square wave on one side, 75 ohms on primary 75ohms on secondary. On the secondary I connected the oscilloscope and looked. If I remember well, with 4 turns I had a good signal from 500KHz to 10-15 MHz.
And that is it.
Pred
PS
It is mounted on the outside of the PCB and with this project I will test the difference with or withot.
pred said:
I have doubts for the capacitors again
I mean do we really need anything except OSCONS? We use a ferrite bead to kill everuthing bigger than MHZ range, right? Oscons have low ESR, so they are still a good capacitor in the 100KHz range, right? If the capacitor is not good enough it can kill the perfomance of the OSCONS, right? So why parallel it with another capacitor? We can add it, it is not a problem, but then engineer in me would say, put the ceramic, it is made for the purpose. At least for the digital signals. I think that ceramics can be found in 0805 SMD package and put bellow the digital circuit, so in that way we can save space. On the other hand I saw only polyesters in SMD package in farnell catalogue. And they are mighty big, so we must put it on the top layer and put the OSCON furher away from the power pins
So again the question: Do we need this (these) extra capacitors for decoupling to improve the performance?
Pred
The datasheets, application notes and many NON OS dac projects uses a 100nF//Oscon capacitor.
The application note CDB42406 also using a 1nF-> Here is a 10nF better
I think also it´s good to have the ceramics because there are a lot of HF noise in a dac that needs to be lowered as much as possible
We can also use a 220uF Oscon//100nF X7R ceramic like TNT audio does in his dac?
http://www.tnt-audio.com/clinica/tnt1541_e.html
Guido or others can you help here?
Best regards
Kim
kimschips said:
Yes I know that just wondering if it is necessary. On Guido's paper on power supply decoupling it is written if you use OSCON's you do not need another capacitor.
But why we than use a ferrite bead in front of the TL431?
No way IMHO! Did you saw the prices on these 5EUR per piece and the prices drop after 50pce. Further it has a 5mm spacing between the pins as 22u and 47u have 2.5mm and are exactly the same. I am really not ready to do the PCB with several possibilities of capacitor sizes. And if I want to be close to the chip this is not an option. Plainly, it is an overkill.
If nobody answers I will put the the smaller OSCONS and on the bottom layer I will leave 1206 SMD in parallel. I think in this way one can use smaller or larger capacitors since it is not hard to solder 0805 on the 1206 footprint.
If nobody puts on the table hard eveidence about using polypropilen or polyester in paralell I am completely against it due to size. If we put all together we will have like computer motherboard for just CD receiver
Have a nice day,
Pred
Hello Pred
The ferrite Beads purpose are to separate each TL431 from the others-> a better sound quality and the risk for mixing analog digital psu is lowered.
Also look here:
http://www.tentlabs.com/Info/Articles/Supply_decoupling.pdf
Let´s stay with the 20V 22uF SP Oscons//100nF//10nF smd ceramics
If you put the ceramics at the buttom layer, how do you minimize groundloops?
Have you thought about the dac groundplane?
Best regards
Kim
The ferrite Beads purpose are to separate each TL431 from the others-> a better sound quality and the risk for mixing analog digital psu is lowered.
Also look here:
http://www.tentlabs.com/Info/Articles/Supply_decoupling.pdf
Let´s stay with the 20V 22uF SP Oscons//100nF//10nF smd ceramics
If you put the ceramics at the buttom layer, how do you minimize groundloops?
Have you thought about the dac groundplane?
Best regards
Kim
kimschips said:
But how it does separation acctualy? It is not a black magic. Bead simply kills all frequencies on PSU lines by raising the impedance. It is the same effect as the ceramic has as staying a good capacitor on high frequency and acting as a short. Two different fenomenon but with the same end result!
Yes, that is the article I am talking about. I will re-read it again when I catch the time ,to be sure about capacitors.
Why 100nf//10nf of the same type? Just put 110n. Ok ESR is smaller
but I do not see the point. I think I have a place, I fulled little bit with the positioning of the components, so maybe there will be a place. But still I am not convinced it should be done this way.
I don't know if I follow on this one. What ground loop?
Not yet, I think it is better to go step by step and pass that problem when we come to it. Still have to finish the CS part, and I am so pressed at work so I can do only a tiny portion every day.
Best regards,
Pred
I am back
Hi,
I restored, partially my portable so I am ready to work. Since this took enourmous amount of time I have even less time.
Yes I have the backup
I attached to this post my view on the CS receiver circuit. You will see that there is no torx on the board. I thought that including this would create a lot of problems, and since some people would like to use transformer for the SPDIF input I just left the capacitors and resitors.
In this way the choice is yours.
Finally, I am not sure about PSU decoupling capacitors, but since I saw that I will eventually have the place I will put the "third capacitor", most probably not SMD, with the footprint for small WIMA's (like 63V polypropilen types 5mm spacing). Please if someone has the time, look if I have not connected something wrong and I will continue, in next few days on the AD part.
I am not certain is the additional TL for the switch selector is necessary or the bead plus capacitors will do the thing?
Pred
Hi,
I restored, partially my portable so I am ready to work. Since this took enourmous amount of time I have even less time.
Yes I have the backup
I attached to this post my view on the CS receiver circuit. You will see that there is no torx on the board. I thought that including this would create a lot of problems, and since some people would like to use transformer for the SPDIF input I just left the capacitors and resitors.
In this way the choice is yours.
Finally, I am not sure about PSU decoupling capacitors, but since I saw that I will eventually have the place I will put the "third capacitor", most probably not SMD, with the footprint for small WIMA's (like 63V polypropilen types 5mm spacing). Please if someone has the time, look if I have not connected something wrong and I will continue, in next few days on the AD part.
I am not certain is the additional TL for the switch selector is necessary or the bead plus capacitors will do the thing?
Pred
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.