This is part 4 in the series of posts discussing the (audio) 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 part 2, I have gone through the board layout consideration and showed the assembled boards. In part 3, I have gone through measurement results of the assembled pre-amplifier board, as well as some circuit modifications to extend its performance. In this post, part 4, I will briefly show the assembled unit, along with slight discussion of external and power supply coupling into the signal.
As with many of my recent projects, I stuck to PCB’s for the front an rear panels of the pre-amplifier. The benefits are clear, its cheap, its very easy to design in the same software tools used for all of my circuit designs, and it offers electrical shielding due to the internal copper layers that are available to us. Unlike in my previous builds, this one is significantly larger, has very large holes, and even square cut-outs. Therefore, I wasn’t sure how well it will come out. To minimize the chance of an error I’ve printed the panels on a piece of paper and measured it in place before placing the orders. You don’t want to spend a few 10’s of $’s, and wait for a few weeks before you realize you’ve made a mistake 🙂
Its a good think I did these measurements, or otherwise I would have missed the fact that one of the LED columns drills were 1mm to the side. I went with JLCPCB for the order of the panels, simply because for this size of board they were significantly cheaper. Getting 5 panels of each type (10 total) was <30$ with shipping included. JLCPCB always place their order reference code on the front side, so I’ve flipped the board so that this code will end up being on the side that is inside the case when assembled.
A couple of weeks later, the panels have arrived, and I was able to assemble the complete unit:
I was happy to see that the panels turned out quite good, I was afraid I will have to rework them manually around the square cut-outs, but it wasn’t needed. The panels obviously use a copper shield and the metal screws short it to the rest of the aluminium case. I have had to use a small file to remove a bit of the case coating from the slots the PCB slides into or else the exposed stripes at the edge of the PCB wouldn’t short to the aluminium case.
After doing this and putting the case together I’ve repeated the same noise floor measurement I did in part 3 when it was open on my bench. Just a reminder, this is done with a 50-ohm input termination, on the 2mV (60dB gain) range. 0dBFS on the scale is 4mVrms. The results can be seen below:
We see that the wide-band noise density is unchanged (~8nV/rt(Hz)), however, the 50Hz coupling is now 17dB lower than it was when the unit was open on my bench. This is still another 16dB above the noise floor, but it is already very low. Running the numbers you will see that this is just 40nVrms at 50Hz. Once you consider the fact that the pre-amplifier was placed on a shelf among many other instruments (including a couple of linear bench power supply’s) and multiple AC main (230V) cables running next to it, I think this is not bad at all.
However, I though this could be improved further. Installing the (grounded) shielding cage around the DC-DC module only aggravated the problem by a few dB. This meant that the 50Hz was now (after putting it inside an aluminium case which grounded everything including the metal shield) mostly coupling through the power input. It was either residual from the switched regulator in the USB power-brick or coupled by means of pickup by the power cable to the pre-amplifier. By installing the cage I’ve only made this coupling worse, which was very helpful as this was the clue that gave away the source of the 50Hz. The solution is now straight forward, remove the shielding cage around the DC-DC module, and disconnect the trace leading to it from the signal ground. Again, a couple of minutes worth of work, but it made a huge difference. As figure 6 shows, the AC mains pick-up was now completely gone. Again, I should reiterate, this is with 60dB of gain, so the noise floor here is at ~8nV/rt(Hz), and the 50Hz isn’t seen at all
Due to the high switching frequency of the DC-DC module used in this pre-amplifier (well over 100KHz at this load current), I see no spurs in the spectrum even when increasing the sampling rate to 192KSPS and looking at the full span of the FFT as in figure 7. The noise floor now looks higher obviously, as the FFT length is unchanged and therefore each bin is wider and contains more energy. The conclusion is very straight forward in this case, there is no need to place a shield around the DC-DC module, even with the highest gain setting of the pre-amplifier. The physical distance from it is sufficient to suppress coupling to the signal path to below the noise floor.
One small item on my to do list was measuring the input capacitance of the pre-amplifier, as I have calculated it based on device parameter, but this doesn’t take into account any parasitic capacitances from connectors/PCB/etc.
The differential impedance I have measured is 200Kohm||17pF.
For a SE mode (GND input switch active) this is 100Kohm||34pF.
Both measurements were done at 1KHz, on the 2V range where the input capacitance is maximal. This capacitance will be lower for the 20V/200V ranges due to the input divider.
This is all for part 4 of the series, it was (by far) the shortest post of them all. This sums up this series of posts describing the measurement pre-amplifier, at least for the time being. The finished pre-amplifier now allows me to measure distortion and noise of most things that come around my bench, and I will without a doubt use it for measurements in some of my future posts. As was shown in part 3, the sound-card I’m using (EMU 0404 USB) is far more limiting in measurements than this pre-amplifier. However, getting a sound-card that will be significantly better than it is far too expensive for me to be able to justify such a purchase; this is after all a just a hobby for me. If you are familiar with other sound-cards that are significantly (>6dB) better than the EMU in THD+N and could be had for a low sum on eBay, let me know in the comments or by email.