DAC, i2c, PCM5122, surface mounted (SMT)

3 Steps to Planning Parts Selection for Your Next Electronic Hardware Product

As an engineer or hardware designer, one of the hardest but most important things when starting a new design for a product is the part selection. It is common to select a couple of IC’s first that perform the functions your product needs to do, and then design the rest. After this, the passives and other “less” important parts are picked. However, this can lock you into having specified parts that do not exist or are over budget. With some planning during the design phase, you can tune your design and bill of materials for optimal costs and manufacturability.

Perform a Rough Pass of Your Electronic Design

This is an Audio DAC (Digital to Analog Converter) board that I’ve been working on for a pinball related project. The requirements for the DAC were that it had to accept I2S Digital Audio and be controlled by I2C. I chose the PCM5122 IC from Texas Instruments as it was widely documented to work with the Raspberry Pi and has drivers built into the Raspberry Pi environment. I designed the schematic first as you would normally do. We will be building this PCB in quantities from 100 to 250 unit batches at a time, so that’s the pricing I will look at when designing the PCB. I am hoping to hit a build price of around $20.

First pass of the Audio DAC design.

To reduce the labor for building the DAC boards I really wanted to build the board with as much surface mounted (SMT) parts as possible. I wanted to use film caps becasue they are regarded as the best performing type of capacitor for audio devices. The interesting thing is there are not a lot of SMT film caps out there (compared to through-hole film caps) and even less that can handle lead-free reflow temps. I figured this is why most Audio devices still have through-hole capacitors. The polymers used in film caps are very susceptible to high temperatures. I eventually discovered that the Panasonic series ECHU film capacitors can handle up to 260C peak reflow and AVX had the CB series film capacitors with a peak reflow of 255C so I was set here.

Next, I went with some thin film audio grade SMT resistors by Susumu (RS Series). I chose these resistors because they were the least expensive I could find that had “audio” in the datasheet. Scientific, I know!

For connectors, I went with what I’ve used before. For the 3.5mm audio jack output, I went with the SJ1-3533NS. This isn’t the least expensive 3.5mm audio jack but it has proven itself to be very robust and virtually indestructible due to the part’s full metal shell and beefy through hole lugs. I have tested lots of different 3.5mm audio jacks in field service and this one seems to work the best. Then, to connect the DAC board to the main Pinball Controller I selected the 70246-1201 a 2×6 2.54mm header. This was selected simply because I had the part already designed in my Eagle library.

At this point I had the entire schematic mocked up and only had two (2) through-hole parts to hopefully limit the labor pricing. Next steps were to create a preliminary PCB in Eagle, where I arranged the parts in a somewhat compact fashion, then I drew a border around it. From there, I uploaded the PCB to the MacroFab PCB (demo here) platform and filled out the bill of materials.

Preliminary PCB Layout. No traces routed.

Preliminary DAC design uploaded into the MacroFab platform.

Analyzing the Bill of Materials

Once all of my information was uploaded and filled out, I punched 100 into the quantity section of the interface and looked at the price. At 100, my rough preliminary design hit a tad above $26. Not too far off from where I needed to be. There are three ways to reduce costs here:

  1. Reduce PCB specs and size
  2. Cut the Bill of Materials
  3. Change parts to reduce labor.

At this point, reducing the PCB size wasn’t realistic, since I needed to add mounting holes to the design and it was 2-layer board, without anything special for the specifications. I checked the bill of materials, for the big-ticket” items.

The populated bill of materials for the preliminary DAC design.

The lion’s share of the BOM went to the PCM5122. Well, that part cannot change and it’s already an SMT part, so the labor for placing it is already the lowest it could be.

The next largest cost of the PCBA was the 70246-1201 connector. This is not an inexpensive part but the through-hole labor was eating up my budget. I started searching around and found M20-8760642, which is an SMT mounted 2×6 connector with 2.54mm pitch which matches the 70246-1201! The M20-8760642 was less expensive than the 70246-1201 and being SMT, it was less expensive on the labor end. I typed in the M20-8760642 part number into the J1 search selection on BOM screen to see the difference. Over a run of 100 PCBs, I can save $267.33 ($2.67/unit) with this change to the design. Seems like a win to me!

J1 with the original 70246-1201 connector part.

J2 with the M20-8760642 SMT header replacement.

Surprisingly the 0.1uF film capacitors (CB027D0104JBA) were the next most expensive thing on the bill of materials. Since these are digital bypass capacitors and are not in the path of the audio signal, I decided to change them to normal ceramic capacitors. I did decide to choose the more expensive C0G (NP0) dielectric as the microphonics is the lowest out of all other dielectrics for ceramic capacitors. This netted me a $317.49 ($3.17/unit) change, more savings than changing the connector!

C5-C8 Bypass cap with the original film capacitor selected.

C5-C8 Bypass cap with a C0G ceramic capacitor selected.

I decided to do this with the other capacitors and audio resistors just to see what a heavily cost-optimized board would come to.

C1-C2 are film capacitors that are in the Audio pathway.

C1-C2 swapped with some C0G dielectric capacitors for further cost reduction.

R4-R5 are thin film audio grade capacitors.

R4-R5 replaced with inexpensive house part thick film resistors.


Overall, I was able to bring the price from $26.49/unit to $20.50/unit. Success!

Building The First Prototype

Using the BOM screen on MacroFab, I was able to get my price down to my target but I don’t know if the lower price parts would affect the audio quality of the DAC. I took the parts I was swapping and decided to build three different configurations of the DAC PCB. The idea will be to conduct blind listening tests of the Audio DACs and see if a $6 difference in the bill of materials is worth it. Here’s the breakdown:

  1. DAC board fully loaded with “audio” parts
    1. C1-C2: ECH-U1C222JX5
    2. C5-C8: CB027D0104JBA
    3. R4-R5: RS2012P-471-D-T5-3
  2. DAC board with audio parts in the audio pathway and with inexpensive bypass capacitors
    1. C1-C2: ECH-U1C222JX5
    2. C5-C8: 885012207072
    3. R4-R5: RS2012P-471-D-T5-3
  3. DAC board with the least expensive parts everywhere
    1. C1-C2: VJ0805A222KXXPW1BC
    2. C5-C8: 885012207072
    3. R4-R5: MF-RES-0603-470

The final prototype PCB ended up about the same size as the preliminary layout. I chose to go with the SMT 2×6 header, it should work fine and not be subject to lots of side loading.

Audio DAC board prototype design finished.


With three different revisions of the DAC PCB designed and ordered, it is now time to talk about testing. For testing, I will hook up the DAC board to a Raspberry Pi. The testers will not know which DAC they are listening to. I will also use a frequency tone generator on the Raspberry Pi to capture the output on the DAC to see how clean (or not) the signal is. With this information I can decide which version of the design will go into production.

Was this post helpful? Have other questions? Let us know in the comments below.