Hi everyone,
I'm putting my enclosure together and I am wondering how to handle the grounding of the aluminum enclosure I have with respect to the WIFI antennas and shielding on my super noisy wifi board.
The wifi board is behind an isolated power supply, therefore the antenna shielding is also connected to the isolated ground. The antenna jacks will be attached to the aluminum back plate for the enclosure, thereby potentially forming a ground loop and introducing noise to the "clean" ground for the rest of the system.
The DC power inlet is also connected to the back plate of the enclosure, providing the single point of contact between the large single-sided copper PCB i'm using as a ground plane and the enclosure.
I've been researching this for quite a while and I feel like all I see is conflicting information (obviously case dependent). Some sources say connect the grounds with a ferrite, others say a capacitor, still others mention a pair of shottkey diodes back to back.
In this application, I think that using the enclosure to shield the analog signal line from the WIFI is the most important consideration.
Attached is a simplified schematic showing the problem.
I'm putting my enclosure together and I am wondering how to handle the grounding of the aluminum enclosure I have with respect to the WIFI antennas and shielding on my super noisy wifi board.
The wifi board is behind an isolated power supply, therefore the antenna shielding is also connected to the isolated ground. The antenna jacks will be attached to the aluminum back plate for the enclosure, thereby potentially forming a ground loop and introducing noise to the "clean" ground for the rest of the system.
The DC power inlet is also connected to the back plate of the enclosure, providing the single point of contact between the large single-sided copper PCB i'm using as a ground plane and the enclosure.
I've been researching this for quite a while and I feel like all I see is conflicting information (obviously case dependent). Some sources say connect the grounds with a ferrite, others say a capacitor, still others mention a pair of shottkey diodes back to back.
In this application, I think that using the enclosure to shield the analog signal line from the WIFI is the most important consideration.
Attached is a simplified schematic showing the problem.
Doubtful there is a one-size-fits-all answer to this. Various problems and solutions probably exist. Maybe the best way to go would be to prototype, measure, and experiment. Sometimes that's the only way, unless perhaps you have sophisticated EM modeling resources available. Seems unlikely for most of us here.
One experiment might be to disconnect the analog output ground and rely on the roundabout ground the analog ground return path. Does an FFT look different? If so, in what exact way? Does it sound different, better or worse, something else? Etc. In other words, first try to tease out if you have a problem and if so how big is it? Then start trying to tease out what improves it.
One experiment might be to disconnect the analog output ground and rely on the roundabout ground the analog ground return path. Does an FFT look different? If so, in what exact way? Does it sound different, better or worse, something else? Etc. In other words, first try to tease out if you have a problem and if so how big is it? Then start trying to tease out what improves it.
Last edited:
The big issue is that it is coming from the 2.5ghz wifi and my scope only goes up to 500mhz, so I can’t measure it directly. The interference manifests as digital sounding “pt, pt, pt”.
If i forget the interference and wifi antennas for a moment, how should i tie the enclosure ground to the ground coming from the SMPS?
If i forget the interference and wifi antennas for a moment, how should i tie the enclosure ground to the ground coming from the SMPS?
Hard to say. Depends. If the SMPS output is floating from the AC line, that might make it easier. Probably what is best will depend on how much switching noise remains on the PS output and what mix of common-mode verses differential-mode. If common mode then its output may act like a radiating antenna. If differential mode then maybe you could ground it at the load only.
Regarding the 2.5GHz issue, at least you have a symptom you can check by listening. If you had an alternate input device such as a USB to I2S board, then you could maybe compare any FFT spurs from it verses from the wireless module (with or without antennas in place when testing with USB, etc.) Sometimes there will be some small audio FFT spurs from demodulation of RF in semiconductor junctions that will cause a nonlinearity to be stationary enough to catch some evidence of it.
In any case, once the RF interference is low enough probably other imperfections in the wi-fi audio board and switching amps will mask it. After all, wi-fi or BT board analog outputs often come from a very ultra-cheap dac inside the RF module. One can only expect so much from them.
Regarding the 2.5GHz issue, at least you have a symptom you can check by listening. If you had an alternate input device such as a USB to I2S board, then you could maybe compare any FFT spurs from it verses from the wireless module (with or without antennas in place when testing with USB, etc.) Sometimes there will be some small audio FFT spurs from demodulation of RF in semiconductor junctions that will cause a nonlinearity to be stationary enough to catch some evidence of it.
In any case, once the RF interference is low enough probably other imperfections in the wi-fi audio board and switching amps will mask it. After all, wi-fi or BT board analog outputs often come from a very ultra-cheap dac inside the RF module. One can only expect so much from them.
Last edited:
indeed. I am not excited about the analog output. It is more of a backup so that I can order parts. I actually had been using the digital outputs from that wifi board, but I'm waiting on some better isolation parts from Digikey (accidentally killed my first isolator chip somehow). Whatever reduces the interference to prevent analog issues will at least not hurt when/if I get the digital working again.
I think that something in the wifi board is actually resonating with the SMPS in a nasty way. I had to put a lot of filtering on the power to prevent reflections upstream, and ultimately went to a better isolated supply for it and the PO89ZB filters to keep the power nice for the amp modules.
Naturally, using a huge linear laboratory supply fixed the problem, but not in a practical way.
Good suggestion on the USB-i2s bridge. I actually have one sitting here for exactly that purpose but didn't know how far down the rabbit hole I wanted to go.
I'll dig up my i2s signals that I captured that have the switching in them. Perhaps they will be more diagnostic to you than they were to me. Pretty ugly stuff. Unfortunately the workaround to get the i2s output from the wifi board is to make it the i2s master, which brings its ugly clock signal over to the DSP.
I think that something in the wifi board is actually resonating with the SMPS in a nasty way. I had to put a lot of filtering on the power to prevent reflections upstream, and ultimately went to a better isolated supply for it and the PO89ZB filters to keep the power nice for the amp modules.
Naturally, using a huge linear laboratory supply fixed the problem, but not in a practical way.
Good suggestion on the USB-i2s bridge. I actually have one sitting here for exactly that purpose but didn't know how far down the rabbit hole I wanted to go.
I'll dig up my i2s signals that I captured that have the switching in them. Perhaps they will be more diagnostic to you than they were to me. Pretty ugly stuff. Unfortunately the workaround to get the i2s output from the wifi board is to make it the i2s master, which brings its ugly clock signal over to the DSP.
Maybe power supply needs more filtering very near to the power supply, so power is already pretty clean by the time it gets to the amplifiers, etc. I have used a portable shortwave radio with an external antenna input as a probe to evaluate noise radiated from a switcher and down its output wire. Did it for more than one project. Might see a lot of hash, say, between 1MHz - 2MHz.
Another less sensitive way to do that is to use a scope probe with the ground lead attached to the tip so as to make a little loop antenna. Wave it around near the power supply and cable to see if you can pick up anything. If so, and if you can follow it down to your project box... Well, you know.
Another less sensitive way to do that is to use a scope probe with the ground lead attached to the tip so as to make a little loop antenna. Wave it around near the power supply and cable to see if you can pick up anything. If so, and if you can follow it down to your project box... Well, you know.