This is part 2 in the series of posts discussing the measurement pre-amplifier project. In part 1 I’ve covered the motivation for this project along with the circuit schematic and detailed circuit description. In this post, part 2, I will discuss the next steps related to the board design and assembly. This part won’t be as long and the first (I hope), but I would like to share some of the consideration I’ve made when laying out the board design.
The first step was deciding on a case size and topology for the front panel, as this will set some constraints on board dimensions and placement of connectors/switches/LED’s. I wanted to use a case that will be made of aluminium to use it as a shield, as at the highest gain setting the pre-amp has 60dB (X1000) of gain which makes it very sensitive to coupling to external signals. I also plan on placing the completed pre-amplifier on my work bench, so I wanted something that is relatively compact, but isn’t too cramped so that it isn’t comfortable. Something similar (or slightly smaller) than a bench DMM seemed like a good size for this as I would be able to stack it on top of my other instruments. The plan was to have all the relevant connectors and switches at the front, along with some LED’s for visual representation of the selected range, and a panel mounter voltmeter. Placing it all in a single line seemed impossible, or at least very uncomfortable to use. Therefore I’ve decide to split this into 2 different heights. This put a constraint on the height of the case, and meant I will have to split the design into 2 boards to support this. The schematics posted in part 1 of this series already represented this split board solution, with the second board used mostly for range selection.
Armed with these ball-park figures about the case size I looked for a case to match. I’ve eventually settled on a case that met my requirements but wasn’t too expensive, despite the fact it wasn’t the best looking option I’ve considered. The case is 80*160*180mm(H*W*D) and is made of aluminium. The dimensions drawing (taken from AliExpress item page) is included below in figure 1. This obviously set the dimensions of the main PCB, as I planned on having the PCB slide into the slots of the case to both hold it mechanically, and connect it to the PCB ground node.
Next we move to the actual placement of the components on the board. Since the DC-DC module and power input could be fairly noisy and inject noise into the rest of the circuit, I’ve decided I should place them at the edge of the board, as far away as possible from sensitive signals. This is especially true for the input signal path as it has a relatively high input impedance (100K), and very high gain on the most sensitive ranges. As a fail-safe, I’ve made room on-board for a shield around the DC-DC module to reduce noise further if needed. The input signal path is placed at the exact opposite side of the board to make sure it is as far away as possible from the noisy section of the board. The output path for the generator signal doesn’t have such high impedances and is therefore much less sensitive to this coupling. Figure 2 shows a top side view of the main PCB after placement and routing were completed.
The vast majority of components are placed on the top side of the board, and kept in close proximity to the parts that connect to them to minimize sensitive signals routing. The signal path with the input divider and gain stages, along with the relays that set the gain can be easily identified at the bottom of the figure. On the bottom side of the board there are mostly decoupling capacitors which are placed right under the devices they are powering.
There are multiple mounting holes in the PCB, despite the fact it is planned to slide into the case I’m using. This is to allow flexibility in case selection and mounting in the future if needed. 3 of the mounting holes are meant to be used for mounting of the second (smaller) PCB with standoffs. Out of the remaining mounting holes, 2 are connected to the exposed metal stripes at the edges. These will short the case to ground if no insulators are used when mounting. The same is true for the 2 exposed metal stripes.
The power supply section can be seen at the top of the figure, with the power input at the right most edge, following it is the DC-DC module (with optional shield), and than some filtering and the linear regulators. I didn’t use any copper flood on the top layer as I didn’t see a need for it. I have included a copper flood on the bottom layer under the low noise section of the board, as can be seen in figure 3.
The mezzanine (daughter) board for range selection is obviously much smaller and simpler, and has 3 mounting holes for mounting to the main PCB with standoffs. The larger components (rotary switch/header/LED’s) are meant to be mounted on the bottom side of the PCB.
As I’ve stated in part 1, I’m a fan of SMD parts, as the are smaller and faster to solder (no lead bending/cutting necessary), and they don’t break copper floods/take up space on the opposite side of the board. Therefore, as you can see, most used parts for these boards are SMD.
After a few weeks of wait the boards were finally here:
Since I’m assembling everything by hand, I’ve started from the SMD parts first, and then got to the larger TH parts. As always, mistakes do happen, and when I reached the DC-DC module, I’ve noticed I have made a mistake in reading the datasheet. The datasheet showed a bottom side view of the component, which I misread as the PCB footprint. This meant everything was mirrored. Since I’ve had most things assembled already, and I didn’t want to spend time waiting for new boards to be produced and shipped to me, I’ve decided to hack the available board. A few minutes later, with the help of a Dremel and a few pieces of wire I’ve modified the board to mirror the connections to the module. This might not look as nice (figure 7) as a clean PCB, but it works just as well.
After assembling both boards, everything is connected together with an 8pin cable. The panel mount meter can also be easily connected/disconnected with a 3pin header.
I’ve verified DC voltages and gain settings to make sure everything works before proceeding with performance measurements. I was happy to see that other than the power section issue which the Dremel quickly fixed, there weren’t additional layout mistakes.
This is all for part 2 of this series of posts. In the next part I will discuss some performance measurements such as TRMS accuracy, distortion, input noise and more.