Hi diyAudio!
I've been reading this forum since quite some time but now that I have a large project going on, I finally registered.
First of all, I'd like to introduce myself, so you know a bit of my background. I am a student of electrical engineering, 4th semester, in Germany. I have been building digital circuits with micro controllers since 2007. Digital is nice, because you can design a lot with the rule of thumb - especially when you work with "slow" processors of 20MHz.
Now I started something different:
I am currently building a nice preamp/amp combo which uses a new kind of controll concept to simultaniously use multiple amps. I will be sharing my details when its runing.
For the preamp, I could need your help. I'd like to use the CS3318 because it is said to deliver a solid performance and it offers everything I need. Please find attached my current schematics and board layout.
The whole thing is planned as two PCBs, one for the relay input switching, one for the actual preamp. The two PCBs can be stacked and interconnect with a fairly common pin header.
The digital part of the circuit (digital supply of the CS3318 and the input switching incl. relays) is powered from a 7805 and 7812 power supply. The 3,3V for the CS3318 is generated from the +12V supply via a LM317 circuit.
The analog supply consisty of a LM317/LM337 supply with +-15V (mounted off-board). This supplies other things like the phono stage (which is not shown here). The whole preamp however needs +-8V so those +-15V rails are converted down again with a LM317/LM337 circuit.
The CS3318 can not drive low impedance loads, and it should be driven from a low impedance source. That's why I added unity gain opamp buffers for both the inputs and the outputs.
I have only little experience when it comes to analog circuitry - so I'd like to hear your advice on my setup.
What do you think about the schematics?
I read that the CS3318s reset pin must be low whenever the digital supply is not present - otherwise the analog circuitry might be destroyed. So I added pulldown resistor for the reset pin - the level shifter should have high-impedance outputs when not powered. Is that alright or do I need a watchdog IC?
And probably more important: What about the layout?
Are the bypass caps allright?
What about the signal flow?
I read a lot, that its not a good idea to separate digital and analog ground. However, those two grounds come from different supplies with different transformers and are connected to the chassis in a star point. If I connect them again on the PCB i will have a ground loop. So I separated them on the pcb - is that alright?
Is it okay to have vias connecting the top and bottom analog ground plane all over the place?
Thanks in advance for your suggestions,
TheSlowGrowth
PS: I know there is another CS3318 Layout thread, but it uses a fairly different setup, so I thought I'd better open a new topic for this.
I've been reading this forum since quite some time but now that I have a large project going on, I finally registered.
First of all, I'd like to introduce myself, so you know a bit of my background. I am a student of electrical engineering, 4th semester, in Germany. I have been building digital circuits with micro controllers since 2007. Digital is nice, because you can design a lot with the rule of thumb - especially when you work with "slow" processors of 20MHz.
Now I started something different:
I am currently building a nice preamp/amp combo which uses a new kind of controll concept to simultaniously use multiple amps. I will be sharing my details when its runing.
For the preamp, I could need your help. I'd like to use the CS3318 because it is said to deliver a solid performance and it offers everything I need. Please find attached my current schematics and board layout.
The whole thing is planned as two PCBs, one for the relay input switching, one for the actual preamp. The two PCBs can be stacked and interconnect with a fairly common pin header.
The digital part of the circuit (digital supply of the CS3318 and the input switching incl. relays) is powered from a 7805 and 7812 power supply. The 3,3V for the CS3318 is generated from the +12V supply via a LM317 circuit.
The analog supply consisty of a LM317/LM337 supply with +-15V (mounted off-board). This supplies other things like the phono stage (which is not shown here). The whole preamp however needs +-8V so those +-15V rails are converted down again with a LM317/LM337 circuit.
The CS3318 can not drive low impedance loads, and it should be driven from a low impedance source. That's why I added unity gain opamp buffers for both the inputs and the outputs.
I have only little experience when it comes to analog circuitry - so I'd like to hear your advice on my setup.
What do you think about the schematics?
I read that the CS3318s reset pin must be low whenever the digital supply is not present - otherwise the analog circuitry might be destroyed. So I added pulldown resistor for the reset pin - the level shifter should have high-impedance outputs when not powered. Is that alright or do I need a watchdog IC?
And probably more important: What about the layout?
Are the bypass caps allright?
What about the signal flow?
I read a lot, that its not a good idea to separate digital and analog ground. However, those two grounds come from different supplies with different transformers and are connected to the chassis in a star point. If I connect them again on the PCB i will have a ground loop. So I separated them on the pcb - is that alright?
Is it okay to have vias connecting the top and bottom analog ground plane all over the place?
Thanks in advance for your suggestions,
TheSlowGrowth
PS: I know there is another CS3318 Layout thread, but it uses a fairly different setup, so I thought I'd better open a new topic for this.
Attachments
Hello,
Nice to see others using the CS3318 for a multichannel pre-amp.
1) Bypass cap? Suggest to use SMT 100n/X7R/25V/0805(0603) and place on same side as the CS3318, as close to input power pins as possible. Route the supply through the cap then directly to the pins. Try not to make a tee connection
2) MCU with 5V supply? Not such a good idea/choice. Reasons, you need a level translator. 5V MCU's , are old technology. I use Atmel xmega, it works really well. Also uses your 5V only for the relays. Since the relays switch some I, it is best to manage the return paths back to the source.
3) O/P buffers? So you anticipate a load less than 2kohm? Not a very usual situation.
4) So you select 1:8 inputs, using relay selector, then pass the sig through 8 buffers to each CS3318 input which are spec'd at 8Kohm. You probably, do not need one buffer, per input, you can probably get away with using only 2 or 4 of them (share).
5) I always reference the evaluation pcb to try to mimic the way that they did it. If you copy almost exactly then you have a very good chance of success.
From the DS, The use of a unified ground plane is recommended for optimal performance and minimal radiated
noise. The CS3318 evaluation board demonstrates the optimum layout and power supply arrangements.
Quick review of pcb looks okay. Suugest to put everything on one pcb. I like the RCA conns from Mouser, 161-4221-E, cheap, simple. I try to minimize on wiring as it is a PITA, can pick up hum/noise. Always best to shield or reference to a ground plane, as much as possible.
6) What MCU are you planning on using? software?
7) Where are going to to get the pcb made?
8) More vias connecting top/bottom grounds are better than not enough of them. Full contact vias are better than ones with thermal reliefs
9) Ground plane? The eval pcb, uses one common ground plane. Place an array of small vias under the device to connect the top/bottom planes
10) Do not rely on the CS3318 o/ps to be stable at power up even if the part is in reset. Suggest to have a way to short/open the connections to the following stages.
Good luck
Rick
Nice to see others using the CS3318 for a multichannel pre-amp.
1) Bypass cap? Suggest to use SMT 100n/X7R/25V/0805(0603) and place on same side as the CS3318, as close to input power pins as possible. Route the supply through the cap then directly to the pins. Try not to make a tee connection
2) MCU with 5V supply? Not such a good idea/choice. Reasons, you need a level translator. 5V MCU's , are old technology. I use Atmel xmega, it works really well. Also uses your 5V only for the relays. Since the relays switch some I, it is best to manage the return paths back to the source.
3) O/P buffers? So you anticipate a load less than 2kohm? Not a very usual situation.
4) So you select 1:8 inputs, using relay selector, then pass the sig through 8 buffers to each CS3318 input which are spec'd at 8Kohm. You probably, do not need one buffer, per input, you can probably get away with using only 2 or 4 of them (share).
5) I always reference the evaluation pcb to try to mimic the way that they did it. If you copy almost exactly then you have a very good chance of success.
From the DS, The use of a unified ground plane is recommended for optimal performance and minimal radiated
noise. The CS3318 evaluation board demonstrates the optimum layout and power supply arrangements.
Quick review of pcb looks okay. Suugest to put everything on one pcb. I like the RCA conns from Mouser, 161-4221-E, cheap, simple. I try to minimize on wiring as it is a PITA, can pick up hum/noise. Always best to shield or reference to a ground plane, as much as possible.
6) What MCU are you planning on using? software?
7) Where are going to to get the pcb made?
8) More vias connecting top/bottom grounds are better than not enough of them. Full contact vias are better than ones with thermal reliefs
9) Ground plane? The eval pcb, uses one common ground plane. Place an array of small vias under the device to connect the top/bottom planes
10) Do not rely on the CS3318 o/ps to be stable at power up even if the part is in reset. Suggest to have a way to short/open the connections to the following stages.
Good luck
Rick
Hello rsavas!
Thanks for your suggestions!
First of all: I want to make the pcb myself with toner transfer. This is why I can't use a solid ground plane - the printer does not print a solid area thick enough - there will be little holes through which the acid will etch the copper. So I always use the "hatched" ground planes.
Selfmade pcbs won't have vias with a connection through them. So I always drill a hole through the pcb, stick a resistor lead through the hole and solder it on the top and the bottom side. This is the reason why I can't make vias under a chip. Another limitation is that I can't start a track on the top layer, when the pin can only be soldered from the bottom side. As an example: The opamps on the top left of the pcb have a track starting on the bottom layer (pin 3) and then change to the top layer through a via.
Now to your points:
1) I think I already did this.
2) I know that lots of modern uCs run at 3.3V - I have a user interface board and the controller board already finished, so I'd like to avoid switching them to 3.3V if not absolutely necessary. I have the level shifter already included before the CS3318, so it should work.
3) I know, that most amps (or other devices connected to the preamp) wont have such a low impedance at all. But many here on the forums and elsewhere have stated that it is better to have an output buffer nevertheless to keep distortion at a minimum. I included the opamps with the thought: "I can easily use a wirelink instead". Are the buffers fine - in terms of their schematics?
4) Basically its the same as 3). I thought: Why not - I can easily connect the inputs of the CS3318 so they share the same input buffer if it prooves to result in the same performance.
5) I did not find the reference pcb anywhere - could you post a link for me?
About the ground plane: As stated in my last post, the analog and digital ground come from different supplies using independent transformers and are connected in the chassis star point. Wouldn't I create a ground loop when I connect the two ground planes to a unified ground?! I always fear that there will be hum...
6) Atmega32. Good old thing, does what it is supposed to do. The controller board is already up and running on a vero board. When everything is working, I will create a solid etched version because thats more likely to withstand the force of time (the enamelled wire used on the current pcb is not fail proof at all), but so far it works fine.
7) See above. I will try to make it myself. We'll see if it works. The fine tracks around the CS3318 will be the hardest part.
8) I guess my "wire link" vias will have good contact 🙂 Nice, so that means I can stuff some more vias on there
9) what do you mean with that?
10) So you mean it would be best to have an additional 10 relays to switch the conenction on after the CS3318 has powered up? Hmm, that sounds resonable, but it will be a pain to fit them in there without creating additional sources of noise.
Thanks again!
Thanks for your suggestions!
First of all: I want to make the pcb myself with toner transfer. This is why I can't use a solid ground plane - the printer does not print a solid area thick enough - there will be little holes through which the acid will etch the copper. So I always use the "hatched" ground planes.
Selfmade pcbs won't have vias with a connection through them. So I always drill a hole through the pcb, stick a resistor lead through the hole and solder it on the top and the bottom side. This is the reason why I can't make vias under a chip. Another limitation is that I can't start a track on the top layer, when the pin can only be soldered from the bottom side. As an example: The opamps on the top left of the pcb have a track starting on the bottom layer (pin 3) and then change to the top layer through a via.
Now to your points:
1) I think I already did this.
2) I know that lots of modern uCs run at 3.3V - I have a user interface board and the controller board already finished, so I'd like to avoid switching them to 3.3V if not absolutely necessary. I have the level shifter already included before the CS3318, so it should work.
3) I know, that most amps (or other devices connected to the preamp) wont have such a low impedance at all. But many here on the forums and elsewhere have stated that it is better to have an output buffer nevertheless to keep distortion at a minimum. I included the opamps with the thought: "I can easily use a wirelink instead". Are the buffers fine - in terms of their schematics?
4) Basically its the same as 3). I thought: Why not - I can easily connect the inputs of the CS3318 so they share the same input buffer if it prooves to result in the same performance.
5) I did not find the reference pcb anywhere - could you post a link for me?
About the ground plane: As stated in my last post, the analog and digital ground come from different supplies using independent transformers and are connected in the chassis star point. Wouldn't I create a ground loop when I connect the two ground planes to a unified ground?! I always fear that there will be hum...
6) Atmega32. Good old thing, does what it is supposed to do. The controller board is already up and running on a vero board. When everything is working, I will create a solid etched version because thats more likely to withstand the force of time (the enamelled wire used on the current pcb is not fail proof at all), but so far it works fine.
7) See above. I will try to make it myself. We'll see if it works. The fine tracks around the CS3318 will be the hardest part.
8) I guess my "wire link" vias will have good contact 🙂 Nice, so that means I can stuff some more vias on there
9) what do you mean with that?
10) So you mean it would be best to have an additional 10 relays to switch the conenction on after the CS3318 has powered up? Hmm, that sounds resonable, but it will be a pain to fit them in there without creating additional sources of noise.
Thanks again!
11) I'd like to review your phono design, as well, if that is okay with you. Here is a paper, selling parts but gives good evidence to use a jfet as a i/p device for a MM cartridge.
http://www.linearsystems.com/assets...LSK489_Application_Note_Preliminary_Draft.pdf
Depending upon size, you could very well put all on one pcb. pcb's are expensive, so minimize size and qty.
12) I see that you are doing the layout in Eagle. I need to learn Eagle so if you want to post the design, I can take a look at it further.
13) Look at the eval pcb, right underneath the CS3318, they have connected all i/p ref's, dgnd to a small fill that through multiple paths eventually meet to a common point. It could have been done better. Many times, layout is done by people who do not understand return paths. You have to look at traces as RLC. If you minimize them, you have less parasitics, which is the aim of a good layout. Think of every layout as if it is running at RF frequencies. Even if you are running at DC, external RF can kill you.
Okay layout 101 over.
http://www.linearsystems.com/assets...LSK489_Application_Note_Preliminary_Draft.pdf
Depending upon size, you could very well put all on one pcb. pcb's are expensive, so minimize size and qty.
12) I see that you are doing the layout in Eagle. I need to learn Eagle so if you want to post the design, I can take a look at it further.
13) Look at the eval pcb, right underneath the CS3318, they have connected all i/p ref's, dgnd to a small fill that through multiple paths eventually meet to a common point. It could have been done better. Many times, layout is done by people who do not understand return paths. You have to look at traces as RLC. If you minimize them, you have less parasitics, which is the aim of a good layout. Think of every layout as if it is running at RF frequencies. Even if you are running at DC, external RF can kill you.
Okay layout 101 over.
http://www.cirrus.com/jp/pubs/rdDatasheet/CDB3318_DB2.pdf
Okay no 3.3V MCU, use SPI with level translator.
Muting outputs. I can not be certain what the CS3318 will do, as I have not read/heard or made one myself. I am just warning you that this is a very hi probability, that it will be the case. I have seen them use transistors/logic mosfets as muting devices as they are a lot cheaper than relays.
If you have hi gain PA directly connected & power up the pre-amp, while the PA is powered up, you will find out. If your electronics manage the PA output mute, then this will probably do.
Toner method for fine pitch (0.5mm) pqfp. Have you done this before.
I used a xmega128 BOB from Sparkfun, to get my feet wet.
Assembler/C/BASCOM-AVR?
Okay no 3.3V MCU, use SPI with level translator.
Muting outputs. I can not be certain what the CS3318 will do, as I have not read/heard or made one myself. I am just warning you that this is a very hi probability, that it will be the case. I have seen them use transistors/logic mosfets as muting devices as they are a lot cheaper than relays.
If you have hi gain PA directly connected & power up the pre-amp, while the PA is powered up, you will find out. If your electronics manage the PA output mute, then this will probably do.
Toner method for fine pitch (0.5mm) pqfp. Have you done this before.
I used a xmega128 BOB from Sparkfun, to get my feet wet.
Assembler/C/BASCOM-AVR?
Last edited:
I'm using the p06 preamp (Elliot Sound Products) - it sounds great with the components I used, though I have the feeling that there is a tiny bit of distortion in the high frequencies. It sounded so agressivly crisp (but in a positive way) compared to my old amp (Yamaha AX-500). But the yamaha is crap anyway, so the p06 might be just fine and most likely it is my "acclimatization" to the poor yamaha sound that makes me think there is something wrong about the high frequencies. I will try to take good measurements when I have the chance. But even if there is a slight distortion - I might (at least for now) stick to the sound, because so far I really like it. Especially electronic music (Jon Hopkins - Immunity) sounds brilliant with this preamp. But hey, thats OT 🙂
13) If I look at the eval pcb (finally found it on the cirrus website) and then at my version - It looks quite comparable. The only thing they did differently: They have small traces of GND between the different inputs and outputs - I should do the same, but that would mean to make the traces even smaller, which is harder to etch at home...
What would you have done differently compared to the eval pcb?
13) If I look at the eval pcb (finally found it on the cirrus website) and then at my version - It looks quite comparable. The only thing they did differently: They have small traces of GND between the different inputs and outputs - I should do the same, but that would mean to make the traces even smaller, which is harder to etch at home...
What would you have done differently compared to the eval pcb?
How do you mean that?! Switching the signal path with mosfets?! I have never seen that before - am I misunderstanding something?
It needs patience, but it works if you have some experience with the method.
Programming will be C (most of it is already written)
I read about muting with Mosfets, but it seeems to introduce distortion - so I went the good old relay way.
So I just updated my approach to support muting relays. The relays mount under the preamp board on the board that has the input selection relays on it as well. From the outputs of the preamp, a simple wirelink or pin header connects to the bottom board where the relays can short the signal to AGND.
I still don't feel 100% comfortable with that, as I'm afraid of the power consumption of these muting relays. They will have to be switched on all the time when the preamp is running. The off position of the relay contacts has to be the one connected to AGND - this is the only way I can be sure the signals will be shorted on power-up. I know, using relays with multiple switches in them could reduce the number of coils that have to be powered, but I have plenty of those simple one-switch relays flying around here and I just can't afford placing another parts order (would be the third within a week...).
I attached the new schematics and layout. I also attached two images highlighting the power traces. Are they fine this way?
Any comments on the reset pin of the CS3318? Do I need a separate watchdog IC to be sure the reset pin is low when the digitral supply is not present? This is very important - it seems like the CS3318 could be destroyed when the reset pin is high und the analog supply is on while the digital supply is missing.
So I just updated my approach to support muting relays. The relays mount under the preamp board on the board that has the input selection relays on it as well. From the outputs of the preamp, a simple wirelink or pin header connects to the bottom board where the relays can short the signal to AGND.
I still don't feel 100% comfortable with that, as I'm afraid of the power consumption of these muting relays. They will have to be switched on all the time when the preamp is running. The off position of the relay contacts has to be the one connected to AGND - this is the only way I can be sure the signals will be shorted on power-up. I know, using relays with multiple switches in them could reduce the number of coils that have to be powered, but I have plenty of those simple one-switch relays flying around here and I just can't afford placing another parts order (would be the third within a week...).
I attached the new schematics and layout. I also attached two images highlighting the power traces. Are they fine this way?
Any comments on the reset pin of the CS3318? Do I need a separate watchdog IC to be sure the reset pin is low when the digitral supply is not present? This is very important - it seems like the CS3318 could be destroyed when the reset pin is high und the analog supply is on while the digital supply is missing.
Attachments
No I mean, shorting the outputs to ground temporarily while the device settles during a power up/down cycle.
If you are good at making fine pitch pcb's, do you want to make a few for me, for some extra $ ?
Distortion of a semi switch, yes if it is in the signal path, but I am saying to use them to ground the outputs during a powerup/down cycle.
Put some vias underneath the CS3318 and change the way the ground fill is done on the top side. Would it make a difference, it is hard to say, for sure, just the way I do it.
"p06 preamp", a simple phono amp, but I am sure it is fine.
Relays, are a pain but I think necessary. You can use small reed relays since the current drawn is very small. A latching type of relay can save on power.
CS3318, reset, you have the reset pin at gnd potential using a R and it can only go high if you set the bit high in sw. explicitly.
Each opamp on the left side, do they have the same local decoupling, one per pin, as is done with the CS3318? The CS3318 supply bypassing looks okay.
Last edited:
I know - but it seems to be a bit more complicated. See here:
Mosfet as shunt/mute device
I'm from Germany - so if your board settings are correct (Ontario) then shipping would probably make things expensive. And be aware that there are many drawbacks of diy pcbs, especially when it comes to dual layer boards. As I said earlier, soldering all those resistor leads to create vias can be a PITA.
The system is modular - Maybe I replace the circuit later on?! Right now it is fine and I probably should use my money to get a better record player instead. I have a Dual CS 626 direct drive turntable, but the needles are quite old and only cheap chinese ones are still produced.
Yea, they have 100n directly under them.
Tomorrow the parts will arrive, so I'll try to get the thing etched now. I'll check back to tell you how its going.
Thanks for your help so far!
TheSlowGrowth
- Status
- Not open for further replies.