I am going to create a prototype using surface mount components, any advice?

[matrixofdynamism]

This is about a personal project. So I have searched long and hard and came to the conclusion that I shall have to get TSSOP, SSOP and SOIC packages to create my prototype. I do not have experience with surface mount soldering, but lets put that aside for now.

I will need to do some "experiments" to fully understand how to use the new TSSOP (an Audio ADC) and SSOP (an Audio DAC) chips with the PIC32 in SOIC packge. I am getting a SOIC PIC32 since the other 2 data converting ICs are surface mount anyway.

Usually when we do experiments like this, we would have through hole components, we put them on a prototype board and connect it up using wires and then do what we want. The TSSOP and SSOP won't fit on a prototype board, and it seems that the PIC32 does not exist in a 0.1 inch through hole package as well.

One thing to do is to make a whole board at once. But then there will be a problem if I want to modify the connections or expand the circuit. The board shall have to be remade. The other option is to have small adapter boards for each surface mount component and then use them on the prototype board like a normal through hole component is. Do you have any advice on what I should and should not do here. The main concern I have is that each chip is new for me. So if something stops working, I need to spend a lot of time.

[DanielS]

A common work-around is SMD to breadboard adapters. If your IC has critical power decoupling or other components, you put only those on the adapter PCB and do everything else on the breadboard. Once you have a rough design you are happy with, you can start working on your first (more) integrated prototype.

[tggzzz]

Have a look at my experience, documented at https://entertaininghacks.wordpress.com/category/homebrew-pcbs/

The tl;dr is that it is easier than I expected.

[rx8pilot]

My early approach was to design and build boards that are mistake tolerant. I would fan out each pin to a hole so I could probe or add wires easily. I had every possible (practical) pad so I could add R's and C's and patch them in. When the PCB's arrive they would rarely work, but I had enough easy bodge options to work it out and learn from the mistakes without needed an all new PCB.

I also got a number of SMD breakout PCB's from SparkFun, but never really used them. As for soldering, it just takes a little time as you may guess. Solder wick, flux pen, thin leaded solder, temp controlled iron, chisel and fine tips. I got a hot-air station and LOVE it. Great for removing multi-pin devices very gently. I also pre-heat the area I need to solder if it is critical. Much easier to solder parts when the whole PCB is already very warm.

[Dago]

SOICs are very easy to solder. Because the pitch is so big I usually just solder them pin by pin (smaller pitches with drag soldering).

If I have to give a single advice then remember you can never use too much flux... Use proper gel flux and not just a flux pen.

[tggzzz]

If I have to give a single advice then remember you can never use too much flux... Use proper gel flux and not just a flux pen.

Which specific type do you use?

[matrixofdynamism]

OK, thanks guys.

The thing here is I have analogue supply, digital supply, analogue ground, digital ground and then reference voltage for the data converters as well. Where can I find advice on how to put so many power rails onto a single PCB? I think that for purpose of prototype only, in the beginning atleast, I should be able to ignore the distinction and use a single power source. The ICs will work but may be not optimally. The idea there is to have the simplest setup to get up and running since if something does not work or gets busted then it will not be clear what is the source of the problem. Then I think that maybe this is not a good idea, I should start with correct setup from the beginning. Any advice?

Should I put zener diodes on the power supply pins of the different ICs I use along with smoothing capacitors? I intend to buy linear voltage regulators that will automatically generate the correct voltage for the IC. I think I should use them on the adapter boards. I will need 3.3V and 5V.

By the way, I will need to use an oscillator crystal to supply clock, likely between 20 and 30 MHz. This is required for the data converters. I am using PCM1807 for ADC and PCM1753 for DAC. I think that it is wise to supply the same clock to the PIC as well. Can I expect havoc I mean... high speed effects when I use this crystal on a prototype board? I guess I will need to have adapter boards without any oscillator and once I put the different adapter boards onto the single prototype board, I can have the oscillator on the prototype board.

[Howardlong]

Ive done a number of mixed signal codec/PIC32 designs.

By far the best prototyping boards I have used are Schmartboards because they have a ground plane inherent throughout the boards and breakouts on every pin for 0603 decoupling caps or 0R links: this is key. That is how I start most of my mixed signal designs these days that need SMD ICs.

However, this is mixed signal, and signal integrity is everything. To do it properly, and get the benefit of your ADC's resolution you will _have_ to lay out a board, at least double sided, I'm afraid. Without it all you will be able to show a proof of concept, i.e., the PIC talks to your ADC and DAC, but beyond about 10 or 12 bits of resolution all bets are off even with Schmartboards. Trying to do it on breakout boards without ground planes will be much worse.

[tggzzz]

The thing here is I have analogue supply, digital supply, analogue ground, digital ground and then reference voltage for the data converters as well. Where can I find advice on how to put so many power rails onto a single PCB?

ADC and DAC manufacturers provide such information in their data sheets, in application notes, evaluation boards, and even books. Start with Analog Devices, and trawl the website.

Be prepared for much reading and a long learning curve, since it is a complex subject. If you can find an example layout for your device, start from there. (And if you can't, consider using a different device)

[matrixofdynamism]

@Howardlong

Are you trying to say that even though I am running at a less then several 10s of MHz, I am still in a dangerous situation and can expect high speed effects? hmmm 

OK.

hmmm

I do fully realize that 2 sided boards are required, after all, how else would a high speed signal propagate, it needs a ground plane. Can you tell me just 1 thing then. When people want to create products with high speed signals to test it out, how to do they go about doing it? Do they start from the datasheet and then move to the manufacturers example of a board layout and pretty much copy paste it for use in their prototype? What if they realize that a connection is wrong, or they want to add more functionality? Does that mean a new printed circuit board is created wasting many days in wait? In other words. Lets say you are a professional electronic engineer and have to design a product which uses components with high speed signals. However, they are complex ICs and thus some experimentation is require to fully understand the implication all their features. How would you create a prototype then?

[Howardlong]

You will have plenty of stuff generating noise well into the 10s if not 100s of MHz. Not only the PIC, but the bit clocks for the ADC and DAC. In addition, your ADC is capable of 24 bit resolution, although you'll probably only realistically get the equivalent of 18 to 20 bits if you're lucky however you roll it. Even at 16 bits with 3Vpp input, that's a resolution of 45uV. It looks like it may only be single ended too. It doesn't take much to induce a millivolt let alone 45uV.

I am sure there are many ways of approaching this. Ask three engineers the same question, you'll get three different answers. As I mentioned, I do use breakout boards for prototypes despite knowing it'll be sub-optimal in the analogue domain just to resolve any functionality problems before laying out a board... or even a schematic! Sometimes I might unit test a particular area on PCB right off the bat, but at's usually for controlled impedance stuff or high speed opamps that love to oscillate given half a chance. I am sure in other organisations that aren't one man bands there will be a guy to do the schematic, someone else to do a board layout etc etc.

If I do layout a unit test board, then I will always make the board myself too: waiting for boards always ruins my workflow. Only when I have everything unit tested and system tested as a whole perfboard/breakout board/mini PCB do I lay out the final schematic and board and submit it to a board house. It never works 100% first time for me, there's always something, but if I'm lucky I might get away with just one more spin.

[Fred27]

I totally agree with rx8pilot above. Design your boards with cutting traces and adding bodge wires in mind. It won't be as pretty or compact but will make life easier. Even if part of you board is a total screw up you may be able to salvage and test part of it - e.g. prove the PIC and DAC bits are correct even if the ADC was totally wrong.

[tggzzz]

I do fully realize that 2 sided boards are required, after all, how else would a high speed signal propagate, it needs a ground plane.

I suspect you will require more than a double sided board, given some of the performance specs that have been mentioned.

Can you tell me just 1 thing then. When people want to create products with high speed signals to test it out, how to do they go about doing it? Do they start from the datasheet and then move to the manufacturers example of a board layout and pretty much copy paste it for use in their prototype?

That's one way, and not a bad one.

For alternatives, start by searching the web for "Bob Pease", "Jim Williams", "Bob Widlar", "AN47"; the last is a useful tutorial.

What if they realize that a connection is wrong, or they want to add more functionality? Does that mean a new printed circuit board is created wasting many days in wait?

Entirely depends on the change, of course. Sometimes it is possible to bodge the changes just to check they are correct, before reworking the PCB.

In other words. Lets say you are a professional electronic engineer and have to design a product which uses components with high speed signals. However, they are complex ICs and thus some experimentation is require to fully understand the implication all their features. How would you create a prototype then?

Using your skill, knowledge, any available information, and experiments to determine where version N is inadequate and therefore what to do in version N+1.