speaker hiss

[jhsa]

Hey... that also has me wondering. Is the hiss ONLY there when there's no voice output? I honestly can't hear the hiss unless I'm in a very quiet place (which has me concerned my hearing should be checked)... but even when I can hear it, it seems to "disappear" when a voice is spoken. My assumption was that it was being masked or simply that the louder voice made the hiss less noticeable. But is it possible that it is only hissing when there's no output? I guess what I'm getting at... can we generate an output that is less noticeable than the hiss?

Probably not. Just a thought.

It is there also when an audio file is playing.. assign some voice to a switch, then assign a pot to control the volume.. lower the volume while the file plays. You will hear the hiss. You can also turn the volume all the way down while the file plays.. you cannot hear the file anymore, but the hiss is there..

João

[s_mack]

Gotcha. Thanks for confirming.

[ShowMaster]

The audio level seems to not be a control of the actual audio IC via a gain pin, but instead a fixed gain with the audio input from the processor. Mike and Steven are more informed, but to me, that means that we're hearing data buss ambient noise, not picked up from local traces?
Again, assuming, if the CPU is told to output something on its IO to the amp, it will become active. Audio data off, will still have the CPU IO active and sending out what ever is there, internal buss noise I'm thinking. Maybe lower the amp gain and output more audio?
Again, I don't have a programmers clue how it gets to the amp from data on a sd card?

I do know that if you go to a Big Rock Concert (AC/DC) and sit close to the big speaker clusters, you hear a lot and loud hissing before the concert starts. The inputs are off but not the megawatts amps. It's all those idle watts waiting to numb your ears and eyeballs. It works, after the first set I can't hear a car horn honking, and I can't focus. Or is it the purple smoke in the air, strangely I can't remember.

[s_mack]

After reading a bunch more... I suspect part of the problem is related to my decision to use a Class D amp. It seems that they are inherently noisy. From the wording of the datasheets ("ultra low noise", "superior noise reduction", etc) I thought it was a feature of Class D to have lower noise... but now I realize they are being relative to older Class D.

I chose Class D because they are (a lot) more efficient. They are also a lot more expensive. Since lots of people are complaining about hiss and nobody is complaining about battery life... I think I focused on the wrong parameter. A cheaper Class A/B seems like it would have been a better way to go, in hind sight.

Mike initially recommended a Class A/B. I pushed for the D because I was looking for another "hey, this is awesome" feature to boast about. So my bad.

[jhsa]

The other classes can also be noisy if the input is noisy.. Maybe someone should really break the connection between audio source and amp and check:

1 - Is the noise still there (speaker)? If yes, you got a noisy amp or is picking up external RF noise..

2 - monitor the signal coming from the DAC. Is it noisy? If yes you found the problem , or at least you would know it is not the amp.

João

[ShowMaster]

I think Mike said he did break the input path while working in this and it was quiet.

So here's where I am at.
The audio path is from pin 20 of the CPU to the 10K to the left of it. That 10K goes to a cap in line with pin 4 on the amp IC. That cap connects to a 120k going to pin 4 on the amp IC, one of its 2 inputs.
So in series from pin 20 is a 10K, a cap (value?) and then a 120K to the amp IC pin 4.
The other input pin 3, has a 130K resister to series cap to ground.
I can measure the resistors but the caps don't have a value on them?

Using chipQuick, I removed the 10K SMD resistor from the CPU pin 20.
With my ear on the speaker, I could hear data buzz, and now some CFL hum buzz at my bench. Touching the empty lands really introduced buzz/hum as would be expected. This now is an open high impedance trace running around the board. No loading to ground except the audio IC input I suspect has high fixed gain.
Putting my finger across the vacant resistor pads, a played audio file would get through at a low level with Hum.
What I did do next was to tack a wire to the close by ground pad, from the left side of the now vacant resistor pad, closest to the left side of the board, that goes to the audio IC input circuit. To my ear, all but a very very low data noise went away on the speaker!
I have several audio files and haptic assigned to switches for testing.
I'm in a very low noise room where I hear a clock ticking at the other end! I'd call that quiet for most!

My thinking is that the path and the fixed gain is what's going on. To be honest, I fired up my Taranis in this room and it's pretty noisy. I'm again wondering how critical we're being?
I'll let this info be read before any fixes are tried. I'd like Mike to think on it and weather the input impedance needs to be lowered and if possible, more driver level how ever that's done? Not my area of knowledge FW wise.

Wow, those parts and pads are small! I'm glad I sold off a few planes and bought this awhile ago!
I think I remember Mike buying one that gave me the idea.

[s_mack]

a cap (value?)

All caps that aren't marked on the schematic I provided you are 100nF. It is possible that doesn't always hold true, so I can verify when needed, but it is the case with those two you mentioned.

[s_mack]

What I did do next was to tack a wire to the close by ground pad, from the left side of the now vacant resistor pad, closest to the left side of the board, that goes to the audio IC input circuit. To my ear, all but a very very low data noise went away on the speaker!

What does that tell us? At that point, there is no audio input. You've grounded the input and it is quiet. Are you saying then that the noise isn't introduced by the amp itself? As I've admitted before, I know nothing about audio.

What happens, I wonder, if you solder a wire with a 10k resistor inline... one end to the "right" pad of the resistor you removed (the one closest to the chip) and the other end to C16? Basically physically distancing the path from the board.

Although, before you bother... have you disabled I2C for this? Because we already determined previously that the I2C line was causing some of the noise.

[jhsa]

I think Mike said he did break the input path while working in this and it was quiet.

So, if that is true it means the amplifier is not the noise source. I believe that a filter right after the DAC could probably get rid of it, once determined its frquency.

The i2C noise was different. It was a kind of tic tic tic noise.
This hiss is something else.. I think it is there when the radio is operating in bootloader mode as well I believe. Will confirm that. This could probably give Mike a clue of what causes it..
My personal opinion is that it could be some high frequency interference caused by the electronics itself, perhaps due to the proximity of some other data tracks to the audio tracks..

João

[s_mack]

perhaps due to the proximity of some other data tracks to the audio tracks.

That was the reason for my suggestion... however, the only tracks in close proximity are the i2c lines.

[jhsa]

So, I have just turned my radio on in bootloader mode.. Yes there is a noise, but more like the 50Hz hum caused by poor filtering on some power supplies.. Not saying that is the case, just a similar noise.. It could perhaps be caused by the switching regulator? maybe a cap or two could take care of this one?

The hiss most people refer to is different.. it is much higher in pitch (frequency) and is not heard in bootloader mode.

I also have another noises because I am using the i2C as a third serial data line for the bluetooth.. I can clearly hear the BT searching for other devices and trying to connect. This is a noise similar to the one caused by the i2c and of course uses the same tracks.

So, the hiss is there in normal operation, but not in bootloader mode.. What in the radio is not working in bootloader mode?
Voice? could the noise come from the DAC itself?
Mike, is it possible to somehow disable the DAC?
João

[ShowMaster]

I do believe I have the 12c disabled. I'll reread all your comments and try it again. My issue is defining the hiss vs data vs tic tic. What I hear in a quiet area, went real quiet when I grounded the 10K pad going towards the audio IC. Again the 10K removed from pin 20 of the CPU IC1.
A long trace trace with a 10K and a 120K in it, does make a pretty high impedance pickup antenna.
Especially if the audio IC gain is high.
I had to put this on hold, but since I'm taking parts off now, I'm set up to try things. I just hope I can restore this nice looking board in the end. I'm going to shop for a micro tip for my iron!
As I said, turning my B Taranis in the same location, sounds like a weed whacker in its speaker. I may have the wrong ears for this?
I will tack the audio pin 4 to ground to listen to any lower change in anything from the speaker. Probably time to find and use a transformer in place of the speaker so I can feed my scope? No ground reference for my scope.

[MikeB]

'Scoping the speaker signals may not be useful as the class D amplifier is outputting a high frequency square wave.

The audio amplifier ends with a gain of about 2.15.

Having damaged the 120K resistor when I was trying a few things, I haven't got back to trying anything else after I replaced the resistor.

Mike.

[ShowMaster]

Ok on not scoping it. I pretty much just saw a square wave and you've confirmed that.
After reading your post Mike about not hearing and offending issues after damaging the 120K on the pin 4 audio IC input, I thought I'd try it at the source, pin 20 of IC1. I as posted, had quiet audio if I shorted the line to ground with the 10K removed. This pretty much verifies your finding to me, it not the audio IC. Instead the circuitry going to it.
I'll attempt to reinstall the 10K and tack a wire to the non pin 20 side. I was wondering what some loading would do to that line to lower its impedance before the long trace run? Of course that will lower the speaker volume.
I'm open to suggestions as long as it's working and I don't kill the board off trying. Putting all the parts back will be before my glass of wine for this time for sure!

[jhsa]

So, the hiss is there in normal operation, but not in bootloader mode.. What in the radio is not working in bootloader mode?
Voice? could the noise come from the DAC itself?
Mike, is it possible to somehow disable the DAC?
João

Mike, in case you missed it..
Would this be a useful test?

Thanks

João

[MikeB]

I think I tried that, driving the output low.

Mike.

[jhsa]

And I presume the hiss is still there then..
So, what else is not working when the radio is in bootloader mode? There is some noise, but that specific hiss is not there as far as I can hear..

Question, Doesn't the DAC also output a higher frequency like the megasound module does?

João

[MikeB]

The noise from my speaker is very low.
I can hear an improvement, with a DJT module plugged in, if I turn the external module OFF. The PPM/PXX pulses are not active in the bootloader.

A general note, the volume setting will have NO effect on the noise as it only operates on the audio data before it gets sent to the DAC on the 9Xtreme.
The DAC doesn't have a higher frequency output superimposed on it as it is a true DAC, and not a PWM as on the Megasound.

Mike.

[jhsa]

No, the PXX or PPM signal is also not the cause.. I turned both modules OFF and the hiss is still there..
What else is not ON or being used when in bootloader?
I'm beginning to suspect that this high pitch hiss is caused by the processor itself..

João

[KAL]

What I experienced this weekend is the fact that this 'hiss' disappears when a switch error occurs while changing models.
But then there's another sound (through the small speaker it's quieter than the hiss) which normally gets swallowed by the hiss.
As soon as the error is cleared, the hiss starts again.

I made an audio recording via headphone plug:

hiss.mp3

(154.69 KiB) Downloaded 253 times

After about 4 seconds the switch error occurs (after model select).
After a further 4 seconds the switch error is cleared.
Then the model name is spoken while the hiss starts again.

Edit: It doesn't matter whether a TX module is powered or not.

Klaus

[ShowMaster]

Out of the CPU pin 20 , left side of the board, this is what I see with no audio playing

This pin goes to a 10K resistor that feeds a trace to the right side of the board under the audio IC.
This is what I see at the 10K on the side going to this trace.

I see this same waveform at the input pin 4 of the audio IC. I do believe it's coming from the CPU.
Very low and not a bother to me, but it does sound like the recent posted audio file.

[ShowMaster]

Ok, the pictures always get out of order. The bottom one is at pin 20 of the CPU feeding the 10K.
The top one in after the 10K that feeds the trace that goes to the amp IC input cap and 120K to pin 4.
The waveform is the same after it leaves the pin 20 CPU 10K.

Quick and dirty circuit traced schematic.

[ShowMaster]

I left my wires on in case there are suggestions of parts to hang on them. Wow, small to solder on. I'll straighten the resistor when I remove the wires. I hate crooked.
I had the resistor on and off a few times and lost it 2 times on the bench top.
I should buy a Lotto ticket because a really got lucky finding it 2 times.

[jhsa]

That's what I thought.. it comes from the CPU?

João

[MikeB]

Not necessarily, the CPU output is not that low an impedance. If noise is picked up on the tracking to the audio amp. it could feed back to the CPU, and that wouldn't be able to remove it.

Mike.

[jhsa]

As far as I can hear, on my radio there are 3 different noises, sometimes going at the same time.
1- the high pitch hiss.
2- the power supply type hum.
3- the i2c data noise. This one can be turned off I believe.
An idea someone could try is to record the noises with the radio fully ON (i2c disabled), and then record the noise from the bootloader mode.
Then in audacity capture the noise reduction profile from the bootloader noise and apply it to the track with the hiss. If this is possible and it works, you should be left with a track with just the high pitch hiss on it, that could be analized. I can't do any testing at the moment. Maybe only next week. And I will have to open the radio to record the sound directly. I don't think microphone recording is good enough for this experiment..

João

[ShowMaster]

If we agree on the audio circuit path from the CPU pin 20 to audio IC pin 4, it leave the board trace as the pick up source?
I could attempt to add a wire wrap wire jumper from the 10K to the cap going to the 120K. This would take it off of the board and any close proximity to other traces?

As posted by me, when I removed the 10K and shorted the trace to ground, it was super quiet to my ear! Now that I have a wire tacked to the 10K and one to ground, I can short those and listen. This shouldn't damage the CPU pin 20. I haven't tried that yet. I'll try a resistor to ground if it is quiet to load the line and lower the impedance, and of course, input gain.
I'm assuming the 10k, cap, and 120K in series is a filter? Not sure how that works?
Fuzzy o n that theory.

With the resistor removed the trace picked up all kinds of noises on the board as would be expected.
I'm really not thinking it's the amp IC or its power rail.

[MikeB]

The cap is simply a DC blocking cap. The 10K and 120K together (130K) match the 130K resistor on the other input to the AMP (differential input). The AMP has internal 150K feedback resistors. Reducing the 130K increases to overall gain. The DAC already outputs a 3 volt pk-pk signal, so we don't need much gain from the AMP, we have about 2.15 at present).

Mike.

[s_mack]

I could attempt to add a wire wrap wire jumper from the 10K to the cap going to the 120K.

What happens, I wonder, if you solder a wire with a 10k resistor inline... one end to the "right" pad of the resistor you removed (the one closest to the chip) and the other end to C16? Basically physically distancing the path from the board.

Yeah, try it

Any testing on my end is on hold. I had one working unit left here in my office and I mailed that yesterday to a Canadian customer in a hurry. I've got replacements on the way, and I can also attempt to repair another unit I have here.

[ShowMaster]

Thanks Mike and Steven.
I thought the resistor cap circuit was just that, DC blocking, AC coupling and resistance/impedance matching. I'll try a few things later today.
I need to buy some thinner solder today. What's the best for this tiny SMD work? I use my 60/40 but I read where it stays in the plastic molten state too long after the heat is removed. I can verify that as my resistor want to still move.
I'm read where I want maybe .020 63/37 solder for SMD as it hardens faster when the heat is removed.
I'm trying to not destroy this board or its parts as I work in it. It is however, a good practice board to work on while I'm searching for the hiss.
If I actually can work on these small parts and not destroy or lose them, I'm really good! So far so good!
I have to go out today so I'll check this thread for solder and flux recommendations and pick them up for layer.
I'll see if my local parts house has any miniature toroids I can add on a jumper wire. I don't know if they would help as I think they're more for higher freq dampening and parasitics.