Re: Radio S32LE samples don't go negative?
Hi Farhan,
I'm discussing about the samples I'm reading and writing from/to the wm8731 through ALSA API, not to do with high level software processing.
toggle quoted message
Show quoted text
On 2/3/24 14:09, Ashhar Farhan wrote: We zero out all bins except the central bins as a part of convolution filtering.
On Sat, Feb 3, 2024, 6:35 PM Rafael Diniz <rafael@...> wrote:
And for transmit / speaker, should such offset be used or 0 centered
sample signal can be used?
- Rafael
On 2/3/24 12:56, Ashhar Farhan wrote:
> This offset will show as dc spike (and it does) on bin zero. We
don't
> display or use that bin at all.
> - f
>
> On Sat, Feb 3, 2024, 5:01 PM Rafael Diniz
<rafael@...> wrote:
>
>? ? ?There seem to be an offset according to capture level and signal
>? ? ?intensity, but I can not (at least not easily) determine the
offset.
>
>? ? ?- Rafael
>
>? ? ?On 2/3/24 09:37, Gordon Gibby wrote:
>? ? ?> Is there a constant offset?? ?Try listening to a steady
carrier
>? ? ?which should result in a sine wave in the audio.
>? ? ?>> On Feb 3, 2024, at 04:21, Rafael Diniz
<rafael@...>
>? ? ?wrote:
>? ? ?>>
>? ? ?>> ?There is something wrong. I'm looking at the samples I get
>? ? ?from the radio rx path - I don't see negative samples, just
>? ? ?positive ones, but the sample format is signed 32 bits. When
>? ? ?reading from loopback, for example, I get the negative and
>? ? ?positive samples, as expected. This explains weird behaviors
I was
>? ? ?experiencing. Anyone can confirm this issue?
>? ? ?>>
>? ? ?>> Cheers,
>? ? ?>> Rafael
>? ? ?>>
>? ? ?>>
>? ? ?>>
>? ? ?>>
>? ? ?>>
>? ? ?>>
>? ? ?>>
>? ? ?>
>? ? ?>
>? ? ?>
>? ? ?>
>
>
>
>
>
>
|
On Feb 3, 2024, at 14:43, Dave, N1AI <n1ai@...> wrote:
The point I was trying to make is that Upton's tiny desktop concept risks getting buried by N100 or similar systems. My point was they were designed for different use cases. <snip> Overall IMO this, and the newly-announced IPO, will put pressure on Upton and team to crank up the innovation. RPi Foundation has been building selling *huge* numbers of boards for embedded applications to industrial customers. I haven't heard about an IPO, but I suspect it will allow RPi Foundation to focus on educational issues and applications and the spin-off commercial organization can focus on industrial issues and applications, as well as (I imagine) help sponsor RPi Foundation. N.B. Your linked-to alternative is priced at $279, more than 3x the cost of an 8 Gig RPi 5... Take care, Ken, N2VIP
|
Re: Radio S32LE samples don't go negative?
On Sat, Feb 3, 2024 at 03:32 PM, Gordon Gibby wrote:
I¡¯m sure you know this but a spike in the frequency domain at 0 Hz is a DC constant offset in the time domain. ?
I do, and should have pointed that out, so I thank you for your clarification. ? -- Regards, Dave, N1AI
|
On Sat, Feb 3, 2024 at 09:27 AM, Ken N2VIP wrote:
The N100 vs RPi 5 debate seems to ignore those 40 pins on the RPi that the N100 solutions lack.
Are we trying to solve a software problem or a hardware problem?
I think you'd be hard-pressed to just "drop in" an N100 system in a device like the uBitx, there would be a significant effort to interface the N100 system to a uBitx board I imagine.
True.
N100 systems are great values and solve a number of problems, but they are not suitable in some applications the RPi excels at, namely embedded applications.
If your application doesn't require the 40 pin GPIO header, why are you using an RPi?
NOTE: if $100-150 N100 systems were available at the time, Mr Upton would never have developed the RPi - his goal was a $100ish desktop that required a TV for output that could be used in education for third-world students using phone chargers, cheap keyboards/mice, etc.
Exactly. The point I was trying to make wasn't that you can drop a N100 into a sbitx/ubitx and rock and roll. The point I was trying to make is that Upton's tiny desktop concept risks getting buried by N100 or similar systems. There really is little price difference between Pi 5 and N100, and (IMO) benchmarks show the N100 is a bit faster and definitely more expandable. Overall IMO this, and the newly-announced IPO, will put pressure on Upton and team to crank up the innovation. Hopefully the sbitx community is prepared for things to keep moving rapidly. And, BTW, you can get Arduino and/or Pi-compatible GPIOs on an Intel platform if that's what you want:? ? It's a bit pricier now, but overall it seems quite possible that an Intel variant could take away a lot of Pi's lunch money sooner or later. ? -- Regards, Dave, N1AI
|
Re: Radio S32LE samples don't go negative?
I¡¯m sure you know this but a spike in the frequency domain at 0 Hz is a DC constant offset in the time domain. ?
toggle quoted message
Show quoted text
On Feb 3, 2024, at 15:21, Dave, N1AI <n1ai@...> wrote:
?On Sat, Feb 3, 2024 at 10:45 AM, Raj vu2zap wrote:
See the display difference on your scope in AC and DC mode.
You seem to be confirming Gordon's suggestion that could be a constant positive DC offset to the signal delivered to the Wolfson Codec's ADC which gets reflected in the digital values produced by the Wolfson Codec ADC. This is different from Farhan's "DC Spike" which is indeed a spike and not a constant, and is in the frequency domain, and is centered around zero Hertz. ? -- Regards, Dave, N1AI
|
Re: Radio S32LE samples don't go negative?
On Sat, Feb 3, 2024 at 10:45 AM, Raj vu2zap wrote:
See the display difference on your scope in AC and DC mode.
You seem to be confirming Gordon's suggestion that could be a constant positive DC offset to the signal delivered to the Wolfson Codec's ADC which gets reflected in the digital values produced by the Wolfson Codec ADC. This is different from Farhan's "DC Spike" which is indeed a spike and not a constant, and is in the frequency domain, and is centered around zero Hertz. ? -- Regards, Dave, N1AI
|
Update.
I fitted a Pi3 and it¡¯s booting but whilst I get noise. There is no RX of signals. Maybe I fried something else?
|
Hi, I seem to have a problem with my new sBitx.
I was transmitting a time to tune my antenna and it just stopped. It was still receiving when I stopped keying and I feared I might have popped the PAs.
I have replaced the PAs but I must have done something as the PI will now not boot up. The Pi seems to be ok and will boot ok if I remove it from the sBitx and power it through USB-C but when plugged into the SBitx, the power light comes on and the disk activity comes on briefly and then goes out. Not activity there after. I have removed the PA devices altogether to see if that is the issue but still no boot.?
Does anyone have any ideas?
Voltages to the Pi look ok. The board is labelled V2.01. The docs/schematics on the HFSigs site are a little confusing I have around three different but similar schematics but they don¡¯t seem to correspond to my board?
|
Re: Radio S32LE samples don't go negative?
See the display difference on your scope in AC and DC mode.
Raj
toggle quoted message
Show quoted text
On 03/02/2024 2:51 PM, Rafael Diniz wrote: There is something wrong. I'm looking at the samples I get from the radio rx path - I don't see negative samples, just positive ones, but the sample format is signed 32 bits. When reading from loopback, for example, I get the negative and positive samples, as expected. This explains weird behaviors I was experiencing. Anyone can confirm this issue?
Cheers,
Rafael
|
Re: SBITX v3 - Ability to use a Headset for phone modes
Thanks Ashar
I am guessing you mean sbitx_sound.c as I can¡¯t find the one you said :-)
|
Dave-
Thanks for the info - that saved me some research time!? I will look into that design and see what it looks like.Tnx agn , great group here!
Rich,
WE8T
|
The N100 vs RPi 5 debate seems to ignore those 40 pins on the RPi that the N100 solutions lack.
Are we trying to solve a software problem or a hardware problem?
I think you'd be hard-pressed to just "drop in" an N100 system in a device like the uBitx, there would be a significant effort to interface the N100 system to a uBitx board I imagine.
N100 systems are great values and solve a number of problems, but they are not suitable in some applications the RPi excels at, namely embedded applications.
If your application doesn't require the 40 pin GPIO header, why are you using an RPi?
NOTE: if $100-150 N100 systems were available at the time, Mr Upton would never have developed the RPi - his goal was a $100ish desktop that required a TV for output that could be used in education for third-world students using phone chargers, cheap keyboards/mice, etc.
Ken, N2VIP
toggle quoted message
Show quoted text
On Jan 23, 2024, at 15:00, Dave, N1AI <n1ai@...> wrote:
It also fits into some stuff I wrote earlier about how I feel Pi will be under more pressure from small form-factor (SFF) PCs using x86 processors, and that the sBITX community will have to be ready for things to change more rapidly than in the past
|
Re: Radio S32LE samples don't go negative?
We zero out all bins except the central bins as a part of convolution filtering.
On Sat, Feb 3, 2024, 6:35 PM Rafael Diniz < rafael@...> wrote: And for transmit / speaker, should such offset be used or 0 centered
sample signal can be used?
- Rafael
On 2/3/24 12:56, Ashhar Farhan wrote:
> This offset will show as dc spike (and it does) on bin zero. We don't
> display or use that bin at all.
> - f
>
> On Sat, Feb 3, 2024, 5:01 PM Rafael Diniz <rafael@...> wrote:
>
>? ? ?There seem to be an offset according to capture level and signal
>? ? ?intensity, but I can not (at least not easily) determine the offset.
>
>? ? ?- Rafael
>
>? ? ?On 2/3/24 09:37, Gordon Gibby wrote:
>? ? ?> Is there a constant offset?? ?Try listening to a steady carrier
>? ? ?which should result in a sine wave in the audio.
>? ? ?>> On Feb 3, 2024, at 04:21, Rafael Diniz <rafael@...>
>? ? ?wrote:
>? ? ?>>
>? ? ?>> ?There is something wrong. I'm looking at the samples I get
>? ? ?from the radio rx path - I don't see negative samples, just
>? ? ?positive ones, but the sample format is signed 32 bits. When
>? ? ?reading from loopback, for example, I get the negative and
>? ? ?positive samples, as expected. This explains weird behaviors I was
>? ? ?experiencing. Anyone can confirm this issue?
>? ? ?>>
>? ? ?>> Cheers,
>? ? ?>> Rafael
>? ? ?>>
>? ? ?>>
>? ? ?>>
>? ? ?>>
>? ? ?>>
>? ? ?>>
>? ? ?>>
>? ? ?>
>? ? ?>
>? ? ?>
>? ? ?>
>
>
>
>
>
>
|
Re: Radio S32LE samples don't go negative?
And for transmit / speaker, should such offset be used or 0 centered sample signal can be used?
- Rafael
toggle quoted message
Show quoted text
On 2/3/24 12:56, Ashhar Farhan wrote: This offset will show as dc spike (and it does) on bin zero. We don't display or use that bin at all.
- f
On Sat, Feb 3, 2024, 5:01 PM Rafael Diniz <rafael@...> wrote:
There seem to be an offset according to capture level and signal
intensity, but I can not (at least not easily) determine the offset.
- Rafael
On 2/3/24 09:37, Gordon Gibby wrote:
> Is there a constant offset?? ?Try listening to a steady carrier
which should result in a sine wave in the audio.
>> On Feb 3, 2024, at 04:21, Rafael Diniz <rafael@...>
wrote:
>>
>> ?There is something wrong. I'm looking at the samples I get
from the radio rx path - I don't see negative samples, just
positive ones, but the sample format is signed 32 bits. When
reading from loopback, for example, I get the negative and
positive samples, as expected. This explains weird behaviors I was
experiencing. Anyone can confirm this issue?
>>
>> Cheers,
>> Rafael
>>
>>
>>
>>
>>
>>
>>
>
>
>
>
|
Re: Radio S32LE samples don't go negative?
This offset will show as dc spike (and it does) on bin zero. We don't display or use that bin at all. - f
On Sat, Feb 3, 2024, 5:01 PM Rafael Diniz < rafael@...> wrote: There seem to be an offset according to capture level and signal
intensity, but I can not (at least not easily) determine the offset.
- Rafael
On 2/3/24 09:37, Gordon Gibby wrote:
> Is there a constant offset?? ?Try listening to a steady carrier which should result in a sine wave in the audio.
>> On Feb 3, 2024, at 04:21, Rafael Diniz <rafael@...> wrote:
>>
>> ?There is something wrong. I'm looking at the samples I get from the radio rx path - I don't see negative samples, just positive ones, but the sample format is signed 32 bits. When reading from loopback, for example, I get the negative and positive samples, as expected. This explains weird behaviors I was experiencing. Anyone can confirm this issue?
>>
>> Cheers,
>> Rafael
>>
>>
>>
>>
>>
>>
>>
>
>
>
>
|
Re: Distorted audio on DE
The resistor that supplies power to LM380 is probably charred. Replace or short it.
On Sat, Feb 3, 2024, 2:29 PM HA3HZ < gyula@...> wrote: Robert,
The audio frequency final amplifier uses LM380N, which can operate from 10V to 22V.
If you use it from an external battery, it is possible that the voltage of the battery drops below 10V.
In this case, your radio will be muted, it will only work with headphones.
I haven't tried it, I just think there might be some distortion near 10V before it turns off completely.
?
--
Gyula HA3HZ
|
Re: Radio S32LE samples don't go negative?
There seem to be an offset according to capture level and signal intensity, but I can not (at least not easily) determine the offset.
- Rafael
toggle quoted message
Show quoted text
On 2/3/24 09:37, Gordon Gibby wrote: Is there a constant offset? Try listening to a steady carrier which should result in a sine wave in the audio.
On Feb 3, 2024, at 04:21, Rafael Diniz <rafael@...> wrote:
?There is something wrong. I'm looking at the samples I get from the radio rx path - I don't see negative samples, just positive ones, but the sample format is signed 32 bits. When reading from loopback, for example, I get the negative and positive samples, as expected. This explains weird behaviors I was experiencing. Anyone can confirm this issue?
Cheers,
Rafael
|
Re: SBITX v3 - Ability to use a Headset for phone modes
Paul, See the mixer functions of sound_sbitx.c, they should help.
Essentially that is what the loopback interfaces are. So what I have done is identified the existing loopback interfaces and how they are used, I have then created a new pair ready to route the sbitx audio to a different device.
To illustrate:
Loopback 1 is used to send receive audio from the sbitx app at a fixed volume level to the digital software e.g. wsjt-x
Loopback 2 is used to capture output(mic) audio from the digital software e.g. wsjt-x and feed that back to the sbitx app for transmission
I have created a new pair of Loopback interfaces (3 & 4) to replicate this functionality with a USB connected audio device/headset but I need the source audio to have adjustable volume(gain) controls for SSB purposes.
That is the next puzzle to solve which will involve a bit of coding in the sbitx application. I will investigate this but need to better understand the coding architecture so that I can have a go at making the required updates.
At least the structural foundation is there now from a hardware perspective.
Regards
Paul G0KAO
|
Re: SBITX v3 - Ability to use a Headset for phone modes
Essentially that is what the loopback interfaces are. So what I have done is identified the existing loopback interfaces and how they are used, I have then created a new pair ready to route the sbitx audio to a different device.
To illustrate:
Loopback 1 is used to send receive audio from the sbitx app at a fixed volume level to the digital software e.g. wsjt-x
Loopback 2 is used to capture output(mic) audio from the digital software e.g. wsjt-x and feed that back to the sbitx app for transmission
I have created a new pair of Loopback interfaces (3 & 4) to replicate this functionality with a USB connected audio device/headset but I need the source audio to have adjustable volume(gain) controls for SSB purposes.
That is the next puzzle to solve which will involve a bit of coding in the sbitx application. I will investigate this but need to better understand the coding architecture so that I can have a go at making the required updates.
At least the structural foundation is there now from a hardware perspective.
Regards
Paul G0KAO
|
Re: Radio S32LE samples don't go negative?
Is there a constant offset? Try listening to a steady carrier which should result in a sine wave in the audio.
toggle quoted message
Show quoted text
On Feb 3, 2024, at 04:21, Rafael Diniz <rafael@...> wrote:
?There is something wrong. I'm looking at the samples I get from the radio rx path - I don't see negative samples, just positive ones, but the sample format is signed 32 bits. When reading from loopback, for example, I get the negative and positive samples, as expected. This explains weird behaviors I was experiencing. Anyone can confirm this issue?
Cheers,
Rafael
|
Dave, N1AI
Raj vu2zap