I have had a working voice assistant satellite comprising:
Raspberry Pi Zero WH running latest updated Raspberry Pi OS on SD card
reSpeaker 2-mic HAT with HinTak driver and HermesLedControl
Rhasspy voice assistant software for recording (using arecord), playing (aplay) and wakeword detection. Other functions are performed on another machine.
It was working fine with the RasPi Zero, but noticably slower to respond than my other satellite
… so I swapped the Raspberry Pi Zero W for a Raspberry Pi 3A+. Same SD card, same PSU … but now the output has an a lot of static whether playing on speaker or headphones. The static is not present if a recording is played on my PC, so it seems to be the audio output.
Thinking maybe the Pi3 requires more power, I have checked that it is using a 5V 2.4A supply, and added power to the reSpeaker board … but no difference. And before you say it, yes I will swap power supplies around to prove it with an “official” rasPi 3 supply.
It looks as though the alsamixer settings got reset, but adjusting them has reduced the volume of static (along with volume of the recorded voice)
What else can I check / adjust to fix ?
Is the Raspberry Pi 3A known to generate electrical interference which might cause this ?
I’m not sure why changing from the Zero to the 3A+ has resulted in static on the audio output with all other things being equal.
Which audio output were you using from the HAT; the JST sockets, or the 3.5mm jack?
If you connect to the built-in 3.5mm jack on the 3A+ is there audio present there? I understand the audio from the 3A+ is implemented by filtering a PWM signal which could potentially create electrical noise, so it would be good to check if that audio is currently active.
I have swapped power supplies, and using the official RasPi 3 power supply has made no difference.
The reSpeaker 2-mic HAT has only one JST socket; and yes I tried both the speaker connected to the JST socket, and headphone on the reSpeaker board.
I plugged the headphones into the jack on the RasPi 3A+ board, and there is no sound coming through it. I had expected this, since alsamixer and the Rhasspy config have separate reSpeaker and headphone devices.
Might there be some processor- or board- specific routines which are installed or linked at first boot ?
I guess I’ll have to start from scratch and reinstall the whole thing, checking the audio output at each step As a bowl of Petunias thinks… “Oh no, not again.”
I can understand wanting to avoid a fresh install when the hardware should just be interchangeable but it may be the best way to isolate whether the problem is hardware or software based.
While each Pi board is supposed to be fully compatible with each other at the end of the day there are physical hardware differences between them so it may just be that the higher clock speeds and extra features crammed into the 3A+ board have increased the susceptibility to electrical noise when paired with that HAT.
Let us know if a fresh install yields the same results, we’ve retired that HAT from our product range so we no longer have any stock to test in the office, but we’ll help troubleshoot where we can.
Yes, that was in the /boot/config.txt file … but seems unnecessary since I am not using the RasPi’s built-in audio jack.
I have no idea what you are talking about here (I come from a software background), except that I wonder why you think a Raspberry Pi 3A+ with mic/sound HAT would need 4A power ? Incidentally the reSpeaker HAT has its own microUSB power connector (I believe for powering speakers) but connecting that as well as the official RasPi3 5.1V 2.5A power supply made no difference.
pls post cat no of the audio hat you are ussing…
amplifiers =it amplifies what you put into it if you put a dirty signal in you will get a dirty signal out…
|
that said do u have any body else like a electronic tech friend does not have to be fully qual`d some basic d.c theory…to give you some help somewhere the amp is picking up noise generally unfiltered d.c for the most part…connections matter once again …
put a 2200 uf capacitor in parallel with the input power rail to the rpi…
ok positive to pin 2 and negative to pin 6 connect the electrolytic capacitor across these pins…2200uf 16 v will do try that see if audio clears up…
My first thought is it’s just additional noise from the Pi. The Pi 3 runs faster than the Zero - maybe it’s resulting in some stronger sub-harmonics in the audible range that were below human hearing on the Zero?
Just like was common with CDMA phones near audio equipment when receiving a call.
Pis aren’t designed to be used for HiFi audio and there are a lot of possible sources of electrical noise on them.
In any case, we’ll have to do it progressively:
We know swapping the Pis resolves the issue, so we can at least narrow it down to a combination involving your hat and a Pi 3.
We know the audio file that is clean.
Next test: eliminate the hat!
Test audio playback through the onboard 3.5mm jack - with and without the mic hat.
If you’re using an official supply they’re pretty clean - 120mV pp. I’m dubious about it being the cause in this instance.
A common cause of audio interference is often sharing an audio ground with higher power equipment. The high (and variable) current through the ground path results in voltage variation being added to the audio signal.
Another common but similar issue is having a loop of conductor on the ground path (a ground loop). This becomes an inductor, and any magnetic fields will induce a current - which results in voltage differences on the ground plane.
In both cases you no longer have a common ground voltage, and it makes a right mess of things audio wise.
The Pi 3 does chew a lot more juice than the Zero so it’s theoretically possible it’s the cause - practically though I’m not sure how much of an impact it could have.
either the amp hat has a fault or reinstall software…once again a cro wold be handy here or an audio sniffer if you know what that is …basically an audio sniffer is just a small portable mini mono amp connecting to ground and a probe to the source putting the probe on the audio signal lines will find where the noise is either present or starting from…therefore finding the scarce of the noise…
Actually the reason is that I have been documenting setup of a Rhasspy Voice Assistant satellite device. Over the past couple of months I have setup various combinations RasPi 3B, Zero, and now 3A+; with reSpeaker 4-mic HAT, Adafruit Voice Bonnet and now reSpeaker 2-mic HAT - and I have written 4 documents totalling about 50 pages. I was hoping to take a break, let all that settle in my mind, before coming back to editing.
OK,
new current Raspberry Pi OS image onto a micro SD card;
open another SSH window and repeat the aplay command, but get unexpected result due to test.wav recorded at 44100, but rhasspy uses rate 16000 format S16LE
And then it dawns on me that there is a different rhasspy package for RasPi 3 !
download and copy latest RasPi 3 rhasspy .deb to RasPi 3A+
remove previous rhasspy and install correct version
run Rhasspy and test …
Blast, still too much static
Speaking the command “Porcupine, what is the time ?” I am getting:
download recorded command to PC and play it there with NO static
lots of static from speaker connected to the reSpeaker 2-Mic HAT for both the playback of my command, and the TTS of time being spoken
lots of static from headphone connected to the reSpeaker 2-Mic HAT’s jack for playback of my command, and the TTS of time being spoken
minimal static from headphone connected to RasPi 3A+ jack for playback of my command, and for the TTS of time being spoken
Have been having a read of this thread. Sight unseen I feel your problem could be related to one or both of these statements.
Everything works with the Pi Zero ?? If so there is probably nothing wrong with the audio amp.
If the Pi 3 is drawing more current than the supply can handle that may have an effect. Historically I have found that some voltage regulators create strange things and sometimes “take off” or oscillate when their input voltage gets too far down to regulate properly. In practice this tends to generate a lot of “noise”.
The second statement re earths or grounds is very important although if all the connections are the same I don’t see that happening. UNLESS the pin connections on the Pi 3 are different to the Pi Zero. I assume this audio thing just plugs onto the parent Pi. If the connections are NOT identical it is possible you are missing an earth or ground connection or getting a connection in the wrong place. You will do well to recheck the respective pin connections.
If the Pi Zero and Pi 3 were both working OK in isolation I don’t think reloading and reinstalling anything is going to do much.
Cheers Bob
Hi Guys
I think there has been a missed basic re fault finding.
I don’t see where the power rail (5V, 3.3V etc) voltages have been measured and recorded. First rule when there is a problem. Check power supply points.
Maybe it does. Has anyone checked the current requirement of all connected devices or the requirement of the set up as a whole. The Audio speaker amp will probably depend on listening level (unless it is Class A which is unlikely) or output power.
Depends on the switching and thus the ripple frequency of the power supply. At higher frequencies an electrolytic cap will look more like an inductor at the higher frequencies. You may find that a non electrolytic cap works better.
I have not looked at the circuit of any of your devices but it is possible that if a unit is powered from 2 sources one will have priority. It may be that if you are powering the speaker unit from the Pi and you connect power via the USB connector the connection from the Pi will have priority and the USB power will do nothing. This could be why you do not see any difference. Would be worth a try, power the speaker thing from USB connector but DISCONNECT the Pi power connection. This will also show if the PI and speaker thing are too much for the one supply and you will have to go for more capability.
Cheers Bob
strange you gettingn noise out of that unit maybe y its retired may have had underlying faults …hmm once again check soldering if involved or re-flash software…noise may be you actual amp equimpent your own pa system check your earthing lines…
does the pi zero use the same media as the rpi-3…me…not shure my self don’t own a zero…
sd-card…software on sd-cards are very prown to failure if not shutdown via the software …that said…
what is the of quality media material …E.G a cheap sd-card and or qualitity
…how many non software shutdows has it had over the time span sd-cards fail dramatically very easily over a short period of time if the power is pulled from the board during its read/write cycle …
Boys, boys, boys … I feel you’re all jumping on the ‘power issues’ bandwagon and bouncing ideas off each other, rather than listening to the actual OPs situation.
The official Raspberry Pi 5.1V 2.5A Power Pack works well on other RasPi’s - including a RasPi 3B with reSpeaker 4-mic HAT which I’m sure draws more power. I am not going to open it up and fiddle with it without some real evidence that this unit is faulty.
FWIW I have now swapped the reSpeaker 2-mic HAT for an Adafruit Voice Bonnet (which I had previously put aside because of an annoying whine). They are working well together, with no appreciable whine or static. Go figure.
It is impossible to draw real conclusions from a sample set of one RasPi Zero and one reSpeaker HAT. For now I’m happy to accept that some combinations just don’t work together as well as we would like.
It is not really a “bandwagon”. It is a very basic first step in trouble shooting a problem and believe me I have done a bit of that.
You don’t have to do this. Surely there are some points where you can measure some volts.
You do that. It is your 10cents.
Why not. You purchased these items as working units. You should expect them to do so to their specs otherwise it would be a very poor quality assurance system in place and the manufacturer would not stay in business long if you had to purchase 2 or 3 of everything to get something to work.
In essence you have not really fixed anything or located the reason for your static. You have just found a combination that works for you.
Cheers Bob
Seeed are still selling the reSpeaker 2-mic HATs, so I’m curious why has Core “retired” them ?
What do you recommend as a replacement - the Adafruit Voice Bonnet (poor component layout with button squeezed between left LED and mic) or Raspberry Pi IQaudio codec zero (couldn’t find any mention of it being used with a Voice Assistant) ?
I note that the reSpeaker 4-mic HATs are currently out of stock, and wonder if they are also being retired ? Is the sipeed 6+1 Microphone Array intended to be the replacement ? Sorry, but I don’t read Chinese.
And at $134.95 the ReSpeaker USB Mic Array is too expensive for multiple satellite units in a smart home. I also note your product description states " You can ask questions in our ReSpeaker forum, which is checked daily." yet I challenge you to show the last time seeed replied to any question or issue.
sorry to go back to my media/software issue is easy proved just by adding an extra sd card into the equation hag on to your old one ,just flash a fresh sd-card a good quality say sandisk number ten or ultra typoe… size is your choice and you will know with in an instant if it is a software problem…my quality`s lye more on the diagnostic side of things but the addyfruit voice bonnet looks a lot simuliar to your present card…depending…your software may need modding some it could be for the better addy gear is good…aso the strawberry stuff is said to be idio proof but does it have what you require…no punn intended…sorry for typos…
As a bowl of Petunias thinks… “Oh no, not again.”
