marce, much as I appreciate the idea of an HDI design and understand it would probably be more interesting to you; I think that doing a design using tools and knowledge that are inaccessible to others is not terribly useful as a reference. this would just add expensive PCB manufacture to the list of barriers to anyone that felt like actually making it and as much confidence as I have in your ability to design a nice PCB, I feel that unless its built and tested it cannot be considered a useful reference design, especially as there are no useful spice/device models for the ESS DAC.
perhaps I misunderstand the needs here, but I would have thought having accurate models and well predicted behaviours, impedances etc would be compulsory for something like HDI? I do think that this project needs to be followed through to producing one (btw I now have reflow ability
arduino ftw!) and verifying the results. otherwise you are kinda just on the flipside of building completely subjectively, having complete faith in an incomplete construct (but generated using objective technique). its not objective unless it can be verified in reality.
know what I mean? please dont take that the wrong way.
perhaps I misunderstand the needs here, but I would have thought having accurate models and well predicted behaviours, impedances etc would be compulsory for something like HDI? I do think that this project needs to be followed through to producing one (btw I now have reflow ability
arduino ftw!) and verifying the results. otherwise you are kinda just on the flipside of building completely subjectively, having complete faith in an incomplete construct (but generated using objective technique). its not objective unless it can be verified in reality.know what I mean? please dont take that the wrong way.
Last edited:
There will be a simple 2 layer design, a more complex multi layer design and a design that reflects the way PCB design is being forced due to the miniaturisation of components EMC requirements and how probably 30% of todays designs are done in.
It isn't really of interest to me as I do numerous types of these designs, but will be of interest to some on this site. As to the tools if you have PCB design software that can do more than two layers you have the tools to do HDI design, there are no special tools, just a basic PCB design package.
HDI is not about impedances any more than a standard multilayer design is, it is about coping with the ever decreasing size of packages.
It will be a reference for anyone who is interested, the main focus will be on designs that can be built by diyers, as mentioned earlier, as you are going to get a 2 and 4 layer design, don't moan about getting a HDI one for nothing apart from my time
others here do create some quite complex designs where HDI and the associated techniques, could be thought provoking and useful., and at the end of the day devices are going to get smaller and smaller.
One thing I will say is that most of the time I do not use any fancy tools and the most often commands I use MOVE, to move components into position, and ROUTE, to manually route the traces and that covers about 80% of my design life, so nothing high tech, just 27+years of doing it and learning and being heavily involved with fabrication and assembly.
It isn't really of interest to me as I do numerous types of these designs, but will be of interest to some on this site. As to the tools if you have PCB design software that can do more than two layers you have the tools to do HDI design, there are no special tools, just a basic PCB design package.
HDI is not about impedances any more than a standard multilayer design is, it is about coping with the ever decreasing size of packages.
It will be a reference for anyone who is interested, the main focus will be on designs that can be built by diyers, as mentioned earlier, as you are going to get a 2 and 4 layer design, don't moan about getting a HDI one for nothing apart from my time
others here do create some quite complex designs where HDI and the associated techniques, could be thought provoking and useful., and at the end of the day devices are going to get smaller and smaller.
One thing I will say is that most of the time I do not use any fancy tools and the most often commands I use MOVE, to move components into position, and ROUTE, to manually route the traces and that covers about 80% of my design life, so nothing high tech, just 27+years of doing it and learning and being heavily involved with fabrication and assembly.
I misunderstood, I thought you were saying 2 layer and HDI, no standard multilayer, ruling out the one that is probably the most interesting to those on this site, because its more accessible and familiar enough to take on.
I have no problem with HDI, or multilayer, but I think adding HDI to a long list of other barriers (ESS, price, build technique etc) is not so great. from what i've seen HDI seems to relate fairly tightly to 3D, buried micro vias and using stacked vias as 'traces' I dont see how that doesnt make things more complex and expensive; with more demand for a 3D solver. i'm quite interested in HDI myself, but I think doing HDI for a straight up dac, without FPGA, MCU etc seems overkill as there simply isnt the need for it really. we dont need to fit our dac on the back of a postage stamp, so unless you have the complexity to need it, i'm afraid I dont quite get it. its not available at ANY of the discount board-houses, nor is anything above 4 layer really.
as an exhibition though sure, whatever you want; but without application...
I agree this site really needs to get with the program, but its hard enough getting people to use SMD ... let alone SMD that requires reflow, optimised for all leadless parts.
I have no problem with HDI, or multilayer, but I think adding HDI to a long list of other barriers (ESS, price, build technique etc) is not so great. from what i've seen HDI seems to relate fairly tightly to 3D, buried micro vias and using stacked vias as 'traces' I dont see how that doesnt make things more complex and expensive; with more demand for a 3D solver. i'm quite interested in HDI myself, but I think doing HDI for a straight up dac, without FPGA, MCU etc seems overkill as there simply isnt the need for it really. we dont need to fit our dac on the back of a postage stamp, so unless you have the complexity to need it, i'm afraid I dont quite get it. its not available at ANY of the discount board-houses, nor is anything above 4 layer really.
as an exhibition though sure, whatever you want; but without application...
I agree this site really needs to get with the program, but its hard enough getting people to use SMD ... let alone SMD that requires reflow, optimised for all leadless parts.
Last edited:
its a pity there are no DACs in BGA, BGA routing and controlled impedance workflow is quite interesting to me.
Dont get me wrong Marc, i'm quite interested indeed, but in general apart from a few that seem to keep to themselves and some manufacturers, I seem to be one of the more rabid high end SMD adopters here and keen to the point of building up a reflow setup to solder the parts I have trouble seeing, or have no other real way to solder them. most others seem to want to remove them from the board ASAP...
Dont get me wrong Marc, i'm quite interested indeed, but in general apart from a few that seem to keep to themselves and some manufacturers, I seem to be one of the more rabid high end SMD adopters here and keen to the point of building up a reflow setup to solder the parts I have trouble seeing, or have no other real way to solder them. most others seem to want to remove them from the board ASAP...
Last edited:
Hi all,
@qusp:
Regarding a DAC schematic I drew up the attached schematic yesterday. I made it in KiCAD as I envision myself eventually using my PCB layout software for slightly commercial purposes. Yet, since "slightly" is most likely, I cannot justify the appr. USD 650 for the Eagle standard version - which can be used commercially. However, as we talked about I have attached it as a pdf file. I have also attached the DSD1794A datasheet.
A few comments on the schematic:
- The DAC I/V circuitry is one that I have used before and IMHO it may sound superb, however, it is single-ended. It's a "simple" design and component selection is key to the sound of this particular circuitry. I personally would be interested in seeing layout suggestions that reduce the risk of oscillation (as it in my experience may be prone to).
I realize there are many ways of doing this!
- Not being a PCB layout professional I most likely have used some symbols that are not commonplace today. I tried to find the power symbols that marce used for the schematic in post #41 in KiCAD but no luck. Polarized capacitor symbols are chosen for their small sizes ...
- The circuitry does not include the PSU as such although the DAC-near electrolytics and ferrites are included.
- It's also appropriate to say that the DAC I/V output may/will generate a rather substantial DC offset when the PSU voltages to either the DAC itself or the DAC I/V converter are switched on or off. A suitable relay circuitry - or the like - should be added to prevent this from damaging other equipment. I have made provisions e.g. for a relay that grounds the output when DC voltages get high by adding a collector resistor on the output emitter follower (to protect the transistor).
- I have not written a specific value for the ferrites in series with the various supply voltages. Personally, I would prefer a bead + wire because the low frequency impedance of a bead could be almost none but maybe there are other solutions to this?
- At the left side of the schematic I have added the output pin configuration from Amanero's Combo384. I have done so because it is a challenge for me to sort out how to "route" the connection from the Combo384 to the DSD1794. Might one of you have suggestions here?
------------------------------
In case this may be a useful starting point for a DAC layout I reckon that at some point package information on the various components is needed.
Yet in the first round might we find out if this is a place to start - and if so - is it suggested that something is added or changed in the schematic?
I'll appreciate your thoughts on this ...
Regards,
Jesper
@marce: A P.S.: I've been pondering using TI's PCM4202 as a starting point for an ADC design (it's got DSD recording capability). However, when I study the pin layout it seems to me that the clock pin arrangement is not optimum (schematic attached - pin layout is on page 6).
The SCKI pin does not have an adjacent GND pin which means that the clock signal may have to go below the IC through a GND layer plane which includes at least 2 vias. Given inductance effects at these frequencies won't this - due to impedances in the ground plane - cause small high frequency voltage shifts that are also reflected onto the AGND pin (a voltage divider between the DGND and the AGND)?
Another solution could be to place the clock below the chip with a ground plane in between and then connect it to DGND and SCKI with a track & via close to each pin. But I just wonder why TI has designed it like this? It's the same with many of their other ADCs like e.g. the TI PCM4222.
Might I have missed something?
Greetings,
Jesper
@qusp:
... when I read what you write I imagine that this package would allow for a more optimum design both for the IC designer (who may be able to better place the various sections of the IC), as well as for the person doing a PCB layout later on ..? However, I don't have any experience with this package myself ...
Regarding a DAC schematic I drew up the attached schematic yesterday. I made it in KiCAD as I envision myself eventually using my PCB layout software for slightly commercial purposes. Yet, since "slightly" is most likely, I cannot justify the appr. USD 650 for the Eagle standard version - which can be used commercially. However, as we talked about I have attached it as a pdf file. I have also attached the DSD1794A datasheet.
A few comments on the schematic:
- The DAC I/V circuitry is one that I have used before and IMHO it may sound superb, however, it is single-ended. It's a "simple" design and component selection is key to the sound of this particular circuitry. I personally would be interested in seeing layout suggestions that reduce the risk of oscillation (as it in my experience may be prone to).
I realize there are many ways of doing this!
- Not being a PCB layout professional I most likely have used some symbols that are not commonplace today. I tried to find the power symbols that marce used for the schematic in post #41 in KiCAD but no luck. Polarized capacitor symbols are chosen for their small sizes ...
- The circuitry does not include the PSU as such although the DAC-near electrolytics and ferrites are included.
- It's also appropriate to say that the DAC I/V output may/will generate a rather substantial DC offset when the PSU voltages to either the DAC itself or the DAC I/V converter are switched on or off. A suitable relay circuitry - or the like - should be added to prevent this from damaging other equipment. I have made provisions e.g. for a relay that grounds the output when DC voltages get high by adding a collector resistor on the output emitter follower (to protect the transistor).
- I have not written a specific value for the ferrites in series with the various supply voltages. Personally, I would prefer a bead + wire because the low frequency impedance of a bead could be almost none but maybe there are other solutions to this?
- At the left side of the schematic I have added the output pin configuration from Amanero's Combo384. I have done so because it is a challenge for me to sort out how to "route" the connection from the Combo384 to the DSD1794. Might one of you have suggestions here?
------------------------------
In case this may be a useful starting point for a DAC layout I reckon that at some point package information on the various components is needed.
Yet in the first round might we find out if this is a place to start - and if so - is it suggested that something is added or changed in the schematic?
I'll appreciate your thoughts on this ...
Regards,
Jesper
@marce: A P.S.: I've been pondering using TI's PCM4202 as a starting point for an ADC design (it's got DSD recording capability). However, when I study the pin layout it seems to me that the clock pin arrangement is not optimum (schematic attached - pin layout is on page 6).
The SCKI pin does not have an adjacent GND pin which means that the clock signal may have to go below the IC through a GND layer plane which includes at least 2 vias. Given inductance effects at these frequencies won't this - due to impedances in the ground plane - cause small high frequency voltage shifts that are also reflected onto the AGND pin (a voltage divider between the DGND and the AGND)?
Another solution could be to place the clock below the chip with a ground plane in between and then connect it to DGND and SCKI with a track & via close to each pin. But I just wonder why TI has designed it like this? It's the same with many of their other ADCs like e.g. the TI PCM4222.
Might I have missed something?
Greetings,
Jesper
Last edited:
yes, much better, leadless packages are revolutionising SMD layout. its got much lower inductance and since the pads can be under the part, the IC designer is free to put the inputs/outputs/power etc wherever they want, allowing the shortest path from inside the chip to the outside world and much more freedom with internal routing of the chip; rather than having to route it out to the edge. then with HDI and stacked via layout, the 'pin' can just continue down to the desired layer.
Last edited:
Hi marce - & thanks for updating me. Somehow your travelling doesn't fit in with my assumptions about PCB layout - something about being close to a computer monitor much of the time and then being part of a team working more or less in the same place to accomplish new designs ... Yet, assumptions are, well, assumptions ...
'Best for your day ;-)
Jesper
'Best for your day ;-)
Jesper
I am on a customers site in London (well Harrow), some like me to be on site sat next to the engineers some just send work to the bureau and have us do the designs remotely, it also depends on the security requirements of the design, some data at some places is not allowed off site. And when I am just consulting/training it is usually on the customers site.
Most of the time when I am laying out a board I am left to my own devices, the company's and engineers that work with me trust me to do a good job, that is why they pay for me to do the work for them, and where I am now, I am considered a part of the engineering team, hence I have spent 6 of the last 12 months down here!
Most of the time when I am laying out a board I am left to my own devices, the company's and engineers that work with me trust me to do a good job, that is why they pay for me to do the work for them
, and where I am now, I am considered a part of the engineering team, hence I have spent 6 of the last 12 months down here!Hi Marce,
Just took a look from above on Harrow via Google Earth. If Google found the right spot then it is somewhat north-west of central London, isn't it? Noticed that Wembley is close by and it also looks as if it's an area with many parks and green places between what else may be there. Without knowing I had the inclination that this could be a pleasant part of London to work in (?) - although I personally wouldn't be much attracted to 10+ hours of travel/week ;-I ... But I reckon you maybe live in the south-east part of London-)) and have to commute through the center ... (?)
Personally I live in Aarhus which is an appr. 300000 people city in the middle of Jutland. Situated between hills leading into the sea and with views across the sea to islands and other parts of Jutland. Forest, sea, people nearby & a quite hearty atmosphere - very rarely do people meet with something else than a smile & good intentions. Actually, it's called "The City of Smiles" - and will also become European Culture Capital (I think it's called so) in 2017.
Regarding the PCB layout I've also decided to move on and start doing some layout by myself. I would like to learn how to do this and for me the best way is a trial-and-re-trial process while consulting the sources available like Henry Ott, Howard Johnson & the link you posted earlier in this thread. I've also decided to start a new thread for this (DAC PCB layout for TI's DSD1794) so that should others be interested they may directly search for this thread here on diyaudio and go directly into the thread. I won't have time to set it up this evening but will do so tomorrow and post a link here in this thread.
Best wishes from a sunny Denmark,
Jesper
Just took a look from above on Harrow via Google Earth. If Google found the right spot then it is somewhat north-west of central London, isn't it? Noticed that Wembley is close by and it also looks as if it's an area with many parks and green places between what else may be there. Without knowing I had the inclination that this could be a pleasant part of London to work in (?) - although I personally wouldn't be much attracted to 10+ hours of travel/week ;-I ... But I reckon you maybe live in the south-east part of London
-)) and have to commute through the center ... (?)Personally I live in Aarhus which is an appr. 300000 people city in the middle of Jutland. Situated between hills leading into the sea and with views across the sea to islands and other parts of Jutland. Forest, sea, people nearby & a quite hearty atmosphere - very rarely do people meet with something else than a smile & good intentions. Actually, it's called "The City of Smiles" - and will also become European Culture Capital (I think it's called so) in 2017.
Regarding the PCB layout I've also decided to move on and start doing some layout by myself. I would like to learn how to do this and for me the best way is a trial-and-re-trial process while consulting the sources available like Henry Ott, Howard Johnson & the link you posted earlier in this thread. I've also decided to start a new thread for this (DAC PCB layout for TI's DSD1794) so that should others be interested they may directly search for this thread here on diyaudio and go directly into the thread. I won't have time to set it up this evening but will do so tomorrow and post a link here in this thread.
Best wishes from a sunny Denmark,
Jesper
Hi Jesper,
I live in Blackburn (near Manchester) so stay in a hotel during the week, so away from home, and have limited time to enjoy life apart from work....
As I mentioned in a reply to Qusp, even though I have lots of fancy tools, the majority of PCB layout is manual. The MOST important aspect is component placement, get that right and everything follows on, and that takes practice. The best way is if you have a system with rubber banding connections then you can see the patterns the connections create, as you do more you will be planning the routes in your head as you place the components, remembering where possible to minimise loops area of critical signals, such as clocks, feedback loops etc.
Marc
I live in Blackburn (near Manchester) so stay in a hotel during the week, so away from home, and have limited time to enjoy life apart from work....
As I mentioned in a reply to Qusp, even though I have lots of fancy tools, the majority of PCB layout is manual. The MOST important aspect is component placement, get that right and everything follows on, and that takes practice. The best way is if you have a system with rubber banding connections then you can see the patterns the connections create, as you do more you will be planning the routes in your head as you place the components, remembering where possible to minimise loops area of critical signals, such as clocks, feedback loops etc.
Marc
it just means DIYers have to become more equipped. I dont know, maybe it even makes it simpler with the right tools. I have put together my reflow oven with an old arduino, $59AUD toaster oven, the rocket scream reflow oven controller shield and a solid state relay. paste stencils are getting pretty reasonably priced to buy at the PCB fabs, but there are also ways to etch your own, they dont usually need to be so accurate and if you need to make more than one board, it becomes easier I reckon, to do it this way. you can also make up or buy stencils for the more common packages to use for the difficult parts and then add the paste for the passives with syringe, or hand solder them.
marce, its all very well to say it doesnt need specialised tools, but I would count your brain as a specialised tool in this case. years of experience dont come cheap, but I am genuinely very interested.
marce, its all very well to say it doesnt need specialised tools, but I would count your brain as a specialised tool in this case. years of experience dont come cheap, but I am genuinely very interested.
Hi marce,
Ok - so you're not crossing London but are heading north when going to your place. Actually, I just notice that below your alias name here at diyaudio (marce) it says that your location is Blackburn - missed that ....
Although I haven't been travelling much for some years now I'm what is called an exportengineer - electronic & mechanical engineering combined with international business - and remember going to distant places for some time and staying in a hotel. To me it could be a quite focussed time and pleasurable, yet it often depended on the hotel/venue I stayed in and other practical circumstances... maybe you have found a pleasant place to stay since it's so often and for a long time?
Thanks also for your suggestions regarding the PCB layout. To me interesting that you do most layout manually - I actually noticed that you wrote this in a previous post here. I will have to do so as well
About the quote above I'm not quite sure what "rubber banding connections" means ... Any hints on this?
Best wishes from DK,
Jesper
Ok - so you're not crossing London but are heading north when going to your place. Actually, I just notice that below your alias name here at diyaudio (marce) it says that your location is Blackburn - missed that ....
Although I haven't been travelling much for some years now I'm what is called an exportengineer - electronic & mechanical engineering combined with international business - and remember going to distant places for some time and staying in a hotel. To me it could be a quite focussed time and pleasurable, yet it often depended on the hotel/venue I stayed in and other practical circumstances... maybe you have found a pleasant place to stay since it's so often and for a long time?
Thanks also for your suggestions regarding the PCB layout. To me interesting that you do most layout manually - I actually noticed that you wrote this in a previous post here. I will have to do so as well
About the quote above I'm not quite sure what "rubber banding connections" means ... Any hints on this?
Best wishes from DK,
Jesper
For DIY use probably you can get away with home made stencils, for production in the commercial world, stencils (and the screen printing process) are one of if not the MOST critical factor towards getting a good reliable solder joint, and contribute over 60% to a successful run through the reflow cycle.
Yes I agree, 27 years of doing the job (the most critical learning factor) and studying the academic side has wired my brain to see patterns in PCBs especially during placement. Same as with your graphic experience, one that I am learning, and finding interesting and challenging.
Rubber banding, the connections between pins are visible while you move components, and generally jump to the nearest pin of the same net during component move.
The two pictures sort of illustrate what rubber banding is...
Yes I agree, 27 years of doing the job (the most critical learning factor) and studying the academic side has wired my brain to see patterns in PCBs especially during placement. Same as with your graphic experience, one that I am learning, and finding interesting and challenging.
Rubber banding, the connections between pins are visible while you move components, and generally jump to the nearest pin of the same net during component move.
The two pictures sort of illustrate what rubber banding is...
Attachments
Last edited:
Hi both,
Thanks again for replying.
@qusp: Interesting that reflow capability is available at such a price level & apparent level of simplicity. Don't know if I'll ever go in that direction, though ... have much equipment already & as you might have noticed I have a sustainability perspective ... but thanks again for the info!
@marce: Given your two pictures I know what you mean - have seen it - but still need to learn how to use it in practice.
BTW I've now started this thread:
http://www.diyaudio.com/forums/digital-line-level/237903-dac-pcb-layout-tis-dsd1794a.html
which means that I'm going to continue with the DAC project in that thread instead.
Hi from over here,
Jesper
Thanks again for replying.
@qusp: Interesting that reflow capability is available at such a price level & apparent level of simplicity. Don't know if I'll ever go in that direction, though ... have much equipment already & as you might have noticed I have a sustainability perspective ... but thanks again for the info!
@marce: Given your two pictures I know what you mean - have seen it - but still need to learn how to use it in practice.
BTW I've now started this thread:
http://www.diyaudio.com/forums/digital-line-level/237903-dac-pcb-layout-tis-dsd1794a.html
which means that I'm going to continue with the DAC project in that thread instead.
Hi from over here,
Jesper
Last edited:
For those hobbyists not wanting to go the DIY reflow oven route, the Atten 858D+ reflow station ($62AUD), makes SMD work very easy having not done any of this type of soldering previously. Highly recommended.
Attached pics of a test smd pcb kit... ic's were removed and resoldered with paste. R6 was the only component done by iron, easy to see the difference.
Attached pics of a test smd pcb kit... ic's were removed and resoldered with paste. R6 was the only component done by iron, easy to see the difference.
