Sabaj A30a announced

[MCH]

2. The SW setting has been changed so that low-pass and high-pass are now combined.
SUB Cut Off Frequency: OFF, 70, 80, 90, 100, 110, 120, 140, 160 (Hz)
When SUB is set to OFF, the SW output stops and the speaker output outputs the entire band.
When the SUB frequency is set to 70 to 160Hz, the speaker output will be in high-pass mode linked to the SUB setting frequency and will be low-cut. It feels like the SUB setting frequency is the crossover point.

I've tested it and haven't found any problems so far, so I think Sabaj will issue a new FW soon.

Thank you Toku, that is good info.
Is it possible that someone measures the frequency response of those settings? It would be great!

 

[peterwen]

@Toku Thanks for the info!! Wow ... the HPF feature with sub-out is newly added to the new firmware?! I am impressed. Well done Sabaj.

Yes, it can be download and upgare it from the sabaj-audio.com web now.
Thanks

 

[Toku]

New firmware "MCU_FW_220928" for Sabaj A30a is up on HP.

Sabaj Audio Official-Sabaj音频官网

Since the establishment of our company, we have been committed to an enterprise integrating audio product design, R&D, production, market operation and service.There are mainly three categories: decoder/headphone amplifier/power amplifier. Our company is composed of a group of members who love...

www.sabaj.com.cn

 

[Tonton_ZEeD]

Yes, it can be download and upgare it from the sabaj-audio.com web now.
Thanks

Nice work with the last FW !

I was in the process of send mine back to amazon but with your last FW, it fully convinced me !

Thank you and your team, keep working, you are all doing very well !

 

[RJO]

I only have a MacBook but I might be able to borrow a windows laptop. So to apply the FW do I just connect thru USB and the downloaded file will automatically search the device once unzipped?

@Toku and the rest, appreciate all the input. I just got mine delivered and will test it this weekend.

 

[Tonton_ZEeD]

I only have a MacBook but I might be able to borrow a windows laptop. So to apply the FW do I just connect thru USB and the downloaded file will automatically search the device once unzipped?

@Toku and the rest, appreciate all the input. I just got mine delivered and will test it this weekend.

In the firmware zipfile, there is a pdf detailling the update process, take a look. It's a 5 minutes process, not automated but easy enough !

 

[Toku]

I recently started using SW and noticed that when I switch the A30a to USB input, there is noise in the SW output.
This noise appears even if the USB cable is removed, so it seems that it is mixed inside the A30a.
When I switch the input to OPT/COAX, there is no noise and it works normally.

Noise is a mixture of white noise and digital noise, and appears at a constant level regardless of the volume position or SW output level settings.

If the SW amplifier input is lowered, it becomes less noticeable, but if the input is raised, it becomes quite noticeable.
As a countermeasure, I raise the SW output setting of the A30a and lower the input of the SW amplifier as much as possible.

 

[v1adpetrov2]

I would also like to know the details of the internal processing steps, but they are not published. Even if it's a black box specification, I don't care if the actual processing is 44.1kHz because the sound quality and product quality are good.

I will try to lift the veil of secrecy a little in this matter.
Six years ago, when the mass production of amplifiers under the conditional name "Power-DAC" began, chip manufacturers were more honest with consumers and indicated specific figures in the technical documentation.
In "all-in-one" chips (digital inputs + modulator + output MOSFET) STA350BW and STA326 all input data is SRC'd to 96kHz and processed in the DAP at 96kHz. Then the PWM signal is fed from the modulator at a frequency of 384 kHz to the output MOSFET.
In more complex and powerful "Power-DAC", pairs of chips were used: a separate modulator and a separate amplifier with MOSFET-keys and a protection system.
For example: TAS5548 and TAS5614LA. In the TAS5548 modulator, all signals with frequencies from 32-kHz to 96-kHz are also is SRC'd to 96kHz and processed in the DAP at 96kHz. And the 176.4-kHz to 192-kHz data was processed in native mode. But at the end, the 384-kHz PWM signal was fed to the TAS5614LA MOSFET-keys.

AXIGEN AX5689 has a sampling rate of the input signal: 32-768 kHz, 16 to 32-bit.
But the internal DSP processing still cannot exceed 192 kHz, because the CPU power will not be enough for high-quality operation of the tone block and LFE filter at a higher signal frequency in real-time mode.
Moreover, the modulator AX5689 eventually outputs to the STA516BE all at the same PWM signal 384 kHz.

Just so as not to reduce sales, marketers should not advertise that a device receiving a 768 kHz signal will actually downsample it to 96 or 192 kHz.

 

[Toku]

I will try to lift the veil of secrecy a little in this matter.
Six years ago, when the mass production of amplifiers under the conditional name "Power-DAC" began, chip manufacturers were more honest with consumers and indicated specific figures in the technical documentation.
In "all-in-one" chips (digital inputs + modulator + output MOSFET) STA350BW and STA326 all input data is SRC'd to 96kHz and processed in the DAP at 96kHz. Then the PWM signal is fed from the modulator at a frequency of 384 kHz to the output MOSFET.
In more complex and powerful "Power-DAC", pairs of chips were used: a separate modulator and a separate amplifier with MOSFET-keys and a protection system.
For example: TAS5548 and TAS5614LA. In the TAS5548 modulator, all signals with frequencies from 32-kHz to 96-kHz are also is SRC'd to 96kHz and processed in the DAP at 96kHz. And the 176.4-kHz to 192-kHz data was processed in native mode. But at the end, the 384-kHz PWM signal was fed to the TAS5614LA MOSFET-keys.

AXIGEN AX5689 has a sampling rate of the input signal: 32-768 kHz, 16 to 32-bit.
But the internal DSP processing still cannot exceed 192 kHz, because the CPU power will not be enough for high-quality operation of the tone block and LFE filter at a higher signal frequency in real-time mode.
Moreover, the modulator AX5689 eventually outputs to the STA516BE all at the same PWM signal 384 kHz.

Just so as not to reduce sales, marketers should not advertise that a device receiving a 768 kHz signal will actually downsample it to 96 or 192 kHz.

Thank you for your detailed explanation.

The most interesting point about A30a operation is the frequency response of SMSL VMV A2 and SA400 measured by amirm.
The VMV A2 and SA400 have the same amplifier configuration as the A30a.
This characteristic is exactly the same as the 44.1 kHz frequency characteristic of the CD standard. I can't help but wonder which part produces this characteristic.

 

[ppataki]

I recently started using SW and noticed that when I switch the A30a to USB input, there is noise in the SW output.
This noise appears even if the USB cable is removed, so it seems that it is mixed inside the A30a.
When I switch the input to OPT/COAX, there is no noise and it works normally.

Noise is a mixture of white noise and digital noise, and appears at a constant level regardless of the volume position or SW output level settings.

If the SW amplifier input is lowered, it becomes less noticeable, but if the input is raised, it becomes quite noticeable.
As a countermeasure, I raise the SW output setting of the A30a and lower the input of the SW amplifier as much as possible.

Hi @Toku
Has this new firmware issue been fixed?
I am asking since my Sabaj A30A is on its way and would like to know which firmware to use
(I will be only using the USB connection)
Thank you

 

[Toku]

Hi @Toku
Has this new firmware issue been fixed?
I am asking since my Sabaj A30A is on its way and would like to know which firmware to use
(I will be only using the USB connection)
Thank you

With the latest firmware MCU_FW_220928, the problems so far have been resolved.
However, there may be updates due to functional improvements in the future, so it may be a good idea to check the HP from time to time.
Sabaj is ambitious in improving its products.

 

[ppataki]

With the latest firmware MCU_FW_220928, the problems so far have been resolved.
However, there may be updates due to functional improvements in the future, so it may be a good idea to check the HP from time to time.
Sabaj is ambitious in improving its products.

Thank you, great news!!

 

[sarumbear]

Sabaj is ambitious in improving its products.

What a brilliant manufacturer.

 

[v1adpetrov2]

Thank you for your detailed explanation.

The most interesting point about A30a operation is the frequency response of SMSL VMV A2 and SA400 measured by amirm.
The VMV A2 and SA400 have the same amplifier configuration as the A30a.
This characteristic is exactly the same as the 44.1 kHz frequency characteristic of the CD standard. I can't help but wonder which part produces this characteristic.

The whole problem with this amplifier is in the AX5689 chip.
Here are the measurements from the 2015 AXIGEN datasheet:

Due to the poor implementation of the AX5689 internal circuitry, the noise from signal oversampling is already more than -100 dBV with 25 kHz. And at a frequency of 35 kHz, it reaches a maximum value of -58 dBV.
Therefore, in the SMSL VMV A2, SA400 and Sabaj A30a amplifiers having the same type of AX5689 + STA516B chipset, a Chebyshev filter of the 20th order was used to cut off frequencies above 20 kHz.

And as a result, we get these measurement graphs:

The trouble is that such a rigid filter of the 20th order greatly distorts the phase of the audio signal, which affects the reliability of the transmission of the scene.

For comparison: a filter of the 3rd order, usually used in speaker crossovers, does not make such fatal changes to the audio signal.

 

[voodooless]

The trouble is that such a rigid filter of the 20th order greatly distorts the phase of the audio signal, which affects the reliability of the transmission of the scene.
View attachment 236128

Where are all those coils on the PCB then? It would be a massive cost factor as well.

 

[PeterOo]

Therefore, in the SMSL VMV A2, SA400 and Sabaj A30a amplifiers having the same type of AX5689 + STA516B chipset, a Chebyshev filter of the 20th order was used to cut off frequencies above 20 [[
The trouble is that such a rigid filter of the 20th order greatly distorts the phase of the audio signal, which affects the reliability of the transmission of the scene.
View attachment 236128
For comparison: a filter of the 3rd order, usually used in speaker crossovers, does not make such fatal changes to the audio signal.
View attachment 236127

thanks for attaching the Axign datasheet. I could no longer find that online.


May I ask where you got this (inside) information that a 20th order Chebyshev filter is used By SMSL and Sabaj
And where in the signal loop is this implemented? As the output filter?
Since the Axign digital feedback loop corrects the signal at the speaker terminals of the amp, would it be inside the feedback loop?

 

[v1adpetrov2]

Where are all those coils on the PCB then? It would be a massive cost factor as well.

Right! A filter with multiple inductors and capacitors is not installed here.
But in this case it is not necessary.

Measurements of the AX5689 chip were carried out by ALIGN in 2015. According to this datasheet 2015, at that time, the AX5689 supported digital inputs 44.1-, 48-, 88.2-, 96-, 176.4-, and 192 kHz sampling rates.
The modern version of the AX5689 chip (according to the 2019 specification) installed in the SMSL VMV A2, SA400 and Sabaj A30a amplifiers supports a sampling rate of 32-768 kHz.
For sure, a digital filter of the 20th order was added to the new version of the chip to eliminate the disadvantage of the old version of AX5689.
This is much cheaper and more practical than installing 20 inductors and 20 capacitors on the board to filter two audio channels.
Although, in fact, to filter PWM signals with a frequency of 384 kHz, filters are made on operational amplifiers.
This is the signal, in fact, AX5689 sends to STA516B.

 

[voodooless]

For sure, a digital filter of the 20th order was added to the new version of the chip to eliminate the disadvantage of the old version of AX5689.
This is much cheaper and more practical than installing 20 inductors and 20 capacitors on the board to filter two audio channels.

Then why talk about phase shift? A phase linear filter would be trivial to use if done digitally.

… or what we see here is simple resampling to 44.1 kHz before PWM encoding.

 

[v1adpetrov2]

May I ask where you got this (inside) information that a 20th order Chebyshev filter is used By SMSL and Sabaj? And where in the signal loop is this implemented? As the output filter?

Of course, I don't have any internal information from SMSL or Sabaj.
But this is clearly evident from the measurements made by dear amirm, screenshots of which were given in previous posts.

The STA516B chip has a standard for amplifiers class D 2-order filter at the output, which is not able to cut the signal by 20 kHz with such a steepness.

 

[v1adpetrov2]

Then why talk about phase shift? A phase linear filter would be trivial to use if done digitally.

… or what we see here is simple resampling to 44.1 kHz before PWM encoding.

Here I agree with you: it is much easier to implement a signal downsampling to 44.1 kHz in the chip than to spend CPU resources on a 20-order digital filter with a linear phase ...
The main thing is that the result will be the same - complete removal of higher harmonics and distortion after 24 kHz.

P.S. I hope that during the downsampling, at least there were no conversions from 24 bits to 16 bits.