Well, it hasn't reached home yet or been plugged in for listening but I have taken it apart to have a look and it is interesting
Construction-wise it's very neat and tidy.
The Xmos USB board is easily detachable and connected to the main board on one side by four USB pins, labelled on both boards. On the other side of the Xmos board are nine pins, however the main board socket only connects six of these. We have a potentially easy direct 12s entry point for this amp using that socket!
As the mainboard socket does not have those pins labelled (although the xmos board does) I'll enumerate them here for anyone purchasing the cheaper non-xmos version of the amp. Starting from 'P4' label (closest edge of socket to RCA connectors)
1) GND
2) SPD
3) MCLK
4) SCLK
5) SDIN
6) LRCLK
then missing connections for these on the Xmos board
7) RST (I think, first letter hard to read)
8) GND (first letter hard too read)
9) STA
With the xmos board removed there is Pulse PE65612NL transformer revealed labelled USB_SPDIF on the main board. Hopefully this means that both the USB and coax/spdif inputs are isolated, I couldn't visually trace the lines to establish this but its physical positioning on the board would fit such usage.
Which amp is this?
To the mystic noise on high efficient tweeters reported earlier in this thread:
The output filter output (sic!) is higly dependent from the connected load. The speaker impedance is usually rising towards higher frequencies so there may be a mismatch between filter and load wich may produce overshots and increasing the noise.
Often Class-Ds have a Zobel after the inductors providing an "emergency load" if the speakers are not connected and also some kind of "impedance linearisation".
If there is no zobel in the output filter (as stated earlier in this thread) a impedance linearisation of the speakers towards the high frequencies may be worth a try.
The output filter output (sic!) is higly dependent from the connected load. The speaker impedance is usually rising towards higher frequencies so there may be a mismatch between filter and load wich may produce overshots and increasing the noise.
Often Class-Ds have a Zobel after the inductors providing an "emergency load" if the speakers are not connected and also some kind of "impedance linearisation".
If there is no zobel in the output filter (as stated earlier in this thread) a impedance linearisation of the speakers towards the high frequencies may be worth a try.
are there any multichannel PCI sound cards with accessible I2S lines?
another thread made a case for using network pc's instead of RPi's and Odroids etc because of the availability of soundcards.
DIYINHK's Xmos multichannel board seems expensive for what it does, no?
incidentally,
Globalegl's modded v200 had, I think, a UTP cable acting as the I2S connection - wouldn't this have been somewhat longer than the 10cm recommended max length of interconnect?
another thread made a case for using network pc's instead of RPi's and Odroids etc because of the availability of soundcards.
DIYINHK's Xmos multichannel board seems expensive for what it does, no?
incidentally,
Globalegl's modded v200 had, I think, a UTP cable acting as the I2S connection - wouldn't this have been somewhat longer than the 10cm recommended max length of interconnect?
There are a number of posts here about modifying the Julie card for i2s connectivity. Likewise sugestions that several of the creative cards with expansion connectors can be used. I haven't myself found a card providing three i2s or spdif outputs that seems more attractive than one of the xmos USB-> i2s bridges.
Considering that the DIYINHK multichannel board is AFAIK the worlds cheapest offering for what it does it seems a little harsh to call it expensive. It is admittedly a very small playing field and no doubt if such devices were manufactured in quantity and/or with cheaper components or perhaps if there was merely more competition we might see better prices.
Yes, that's why Globulegl used UTP cable to stretch that length a bit. If you want substantially longer connectors you'll need to do the same but additionally running a buffer+differential driver board at each end of the cables. I should at this point mention that just provisioning those driver boards will cost more than the DIYINHK card itself
Last edited:
Hi,
Just to share, I'm succesfully using a Nucleo Stm32F7 development board to implement:
- Asynch USB input,
- DSP processing
- output to 2x I2S (4 channels).
The very same board can manage up to 4xI2S (8 channels). The progress has been described in http://www.diyaudio.com/forums/digi...ow-jitter-achieved-stm32-microcontroller.html. All code is available ob Github.
This is from my perspective a cheap (25$) and flexible option for a FDA system.
I personnally use the device with 2 FX-Audio D802 with the SPDIF input as the board can manage both I2S, SPDIF and other options.
Last, a dedicated external clock can be connected to the output device to have an even better accuracy.
Best regards,
JM
Just to share, I'm succesfully using a Nucleo Stm32F7 development board to implement:
- Asynch USB input,
- DSP processing
- output to 2x I2S (4 channels).
The very same board can manage up to 4xI2S (8 channels). The progress has been described in http://www.diyaudio.com/forums/digi...ow-jitter-achieved-stm32-microcontroller.html. All code is available ob Github.
This is from my perspective a cheap (25$) and flexible option for a FDA system.
I personnally use the device with 2 FX-Audio D802 with the SPDIF input as the board can manage both I2S, SPDIF and other options.
Last, a dedicated external clock can be connected to the output device to have an even better accuracy.
Best regards,
JM
This is the nearest thing I've seen to just a power stage - S/PDIF in though, not I2S.
https://item.taobao.com/item.htm?spm=a230r.1.14.51.eoPN7L&id=528808249620&ns=1&abbucket=18#detail
https://item.taobao.com/item.htm?spm=a230r.1.14.51.eoPN7L&id=528808249620&ns=1&abbucket=18#detail
Alternative for a power stage with I2S digital input: http://www.st.com/content/st_com/en.../stm32-ode-translate-hw/x-nucleo-cca01m1.html
About 35€
Some code is provided to use it with Stm32 micro controller
JM
About 35€
Some code is provided to use it with Stm32 micro controller
JM
Hi Ropf,
It is maybe a bit better than you think, as they stack on top of Nucleo Stm32 cheap and powerful development board (no hardware to develop) and they propose libraries to use and configure the setup.
Cirrently I'm using 2 D802 with one of those Nucleo board and I'm astonished by the performance.
Last, there is a new partnership between ST and Audio Weaver from DSP concept that could further ease the development process.
JMF
It is maybe a bit better than you think, as they stack on top of Nucleo Stm32 cheap and powerful development board (no hardware to develop) and they propose libraries to use and configure the setup.
Cirrently I'm using 2 D802 with one of those Nucleo board and I'm astonished by the performance.
Last, there is a new partnership between ST and Audio Weaver from DSP concept that could further ease the development process.
JMF
JML11
there seems to be a profusion of kits and boards.
can you provide a link to exactly the hardware to get going with?
I used to be a developer and used to be able to think for myself too - not so much now...
warning!: newbie question about signal timing: with the use of dsp as a given is it not enough for the main computer to guarantee delivery of each 'frame' of channel signals to a demultiplexing, forwarding point such as a st326 based board?
if there is a finite and known time delay for delivery of each channel to each FDA, that a dsp engine can ultimately compensate for, where is the need for master/slave synchronisation? I recall another thread where someone attempted using two usb ports for separate channels and finding the sound stage off-centre.
is the control of usb transmission not sufficiently fine grained that synchronisation can be achieved between multiple usb ports?
I'm thinking control equals predictability equals synchronisation equals dsp compensation for time delays between channels.
there seems to be a profusion of kits and boards.
can you provide a link to exactly the hardware to get going with?
I used to be a developer and used to be able to think for myself too - not so much now...
warning!: newbie question about signal timing: with the use of dsp as a given is it not enough for the main computer to guarantee delivery of each 'frame' of channel signals to a demultiplexing, forwarding point such as a st326 based board?
if there is a finite and known time delay for delivery of each channel to each FDA, that a dsp engine can ultimately compensate for, where is the need for master/slave synchronisation? I recall another thread where someone attempted using two usb ports for separate channels and finding the sound stage off-centre.
is the control of usb transmission not sufficiently fine grained that synchronisation can be achieved between multiple usb ports?
I'm thinking control equals predictability equals synchronisation equals dsp compensation for time delays between channels.
I think the answer to this is that it very much depends on the hardware you are using.
We seemed to have reached a consensus/theory that provided you were running the USB audio in synchronous mode and providing the multiple USB sockets were originating from the same USB controller on your PC it ought to be OK.
Yet JML11 had a go and hit an exception which still had sync problems, so we are back to not being able to predict if a multiple USB setup will work OK.
Running a single multichannel output to a device that does the demux and then provides individual i2s or spdif outputs for each channel does unfortunately seem the safer route.
Alientek D8
Listening tests with this little amp are most encouraging so I've ordered up one of the DIYINHK usb->i2s interfaces to see if the i2s input is actually sensible on this amp or whether I need to go through spdif.
Comparing it to the better known (and my current favourite DDA) D802 the sound using coax spdif input is to my ears transparent and just about indistinguishable between the two amps. This perhaps isn't surprising as they both use the same (excellent) spdif->i2s receiver and essentially just channel that i2s to the input pins of the STA audio chip.
USB input (for this async xmos version) is on the D8 a very valid alternative to the spdif - so the D8 comes out ahead here.
The D8 is smaller, cuter, and comes with a decent quality remote not the usual toy town ones. It seems to stay completely cool at any reasonable volume level, switches itself off after a few minutes of inactivity and wakes up on music signal so operation can be very convenient not needing to switch it on or off each listening session.
The D8 comes with a 28v power supply instead of the 32v supply the D802 comes with. I might hear a tiny loss of gain/dynamics with the supply voltage difference but it's more likely to be my imagination/expectations rather than any real difference.
Like all these amps the display shows a context sensitive display when you operate the controls. On the 802 however the main display when playing music is the source and bitrate, on the D8 you get a dancing real time spectrum display and/or waveform instead. I don't like the graphic eq display as much - however for my intended use of 3 amps in an active setup that display would be very reassuring to know the crossovers are working and I'm not sending full frequency to the tweeters
It will be interesting to see what the v400 is like when/if it comes out but at the moment the D8 looks a good choice. I like it. A lot. But then again I also like the D802. Just not quite as much
P.S. I tried very hard to distinguish between the two amps for noise as that has been reported as a problem with the D802. But no, on my systems, I can't hear that noise on either amp.
Well, it hasn't reached home yet or been plugged in for listening but I have taken it apart to have a look and it is interesting
Construction-wise it's very neat and tidy.
The Xmos USB board is easily detachable and connected to the main board on one side by four USB pins, labelled on both boards. On the other side of the Xmos board are nine pins, however the main board socket only connects six of these. We have a potentially easy direct 12s entry point for this amp using that socket!
As the mainboard socket does not have those pins labelled (although the xmos board does) I'll enumerate them here for anyone purchasing the cheaper non-xmos version of the amp. Starting from 'P4' label (closest edge of socket to RCA connectors)
1) GND
2) SPD
3) MCLK
4) SCLK
5) SDIN
6) LRCLK
then missing connections for these on the Xmos board
7) RST (I think, first letter hard to read)
8) GND (first letter hard too read)
9) STA
With the xmos board removed there is Pulse PE65612NL transformer revealed labelled USB_SPDIF on the main board. Hopefully this means that both the USB and coax/spdif inputs are isolated, I couldn't visually trace the lines to establish this but its physical positioning on the board would fit such usage.
Listening tests with this little amp are most encouraging so I've ordered up one of the DIYINHK usb->i2s interfaces to see if the i2s input is actually sensible on this amp or whether I need to go through spdif.
Comparing it to the better known (and my current favourite DDA) D802 the sound using coax spdif input is to my ears transparent and just about indistinguishable between the two amps. This perhaps isn't surprising as they both use the same (excellent) spdif->i2s receiver and essentially just channel that i2s to the input pins of the STA audio chip.
USB input (for this async xmos version) is on the D8 a very valid alternative to the spdif - so the D8 comes out ahead here.
The D8 is smaller, cuter, and comes with a decent quality remote not the usual toy town ones. It seems to stay completely cool at any reasonable volume level, switches itself off after a few minutes of inactivity and wakes up on music signal so operation can be very convenient not needing to switch it on or off each listening session.
The D8 comes with a 28v power supply instead of the 32v supply the D802 comes with. I might hear a tiny loss of gain/dynamics with the supply voltage difference but it's more likely to be my imagination/expectations rather than any real difference.
Like all these amps the display shows a context sensitive display when you operate the controls. On the 802 however the main display when playing music is the source and bitrate, on the D8 you get a dancing real time spectrum display and/or waveform instead. I don't like the graphic eq display as much - however for my intended use of 3 amps in an active setup that display would be very reassuring to know the crossovers are working and I'm not sending full frequency to the tweeters
It will be interesting to see what the v400 is like when/if it comes out but at the moment the D8 looks a good choice. I like it. A lot. But then again I also like the D802. Just not quite as much
P.S. I tried very hard to distinguish between the two amps for noise as that has been reported as a problem with the D802. But no, on my systems, I can't hear that noise on either amp.
Last edited:
For me it would be much easier to have just some hardware pins for the chip configuration and not to have a nucleo or something similar at all.
Also all glitter things in the power stage - as filtering, volume control, drc etc. - making the setup complex and this way making the need of a µc - they are completly useless for me. All i want - take the i2s and give out the pwm with power.
Last edited:
How about the STA350BW?
This chip will accept audio via i2s and output high level pwm. You need to control it via i2c and provide suitable output filtering to the speaker(s). Given the number of setup and control permutations is not realistic to expect GPIO. At least a uP is required to drive the i2C in conjunction with i2s. If you are only trying to transfer audio streams with no format or rate changes to the STA350BW no DSP is required. In theory a PC could be used if you find suitable i2c and i2s interfaces. Software would be required on the PC to drive these interfaces (basically your own customised audio player). Generally simpler to a use uP.
Otherwise just go analog. DA --> Class D amp module.
I wish there were just powerstages with i2s input somewhere to by - not the complete amps with signalprocessing, codecs, cabinets ...
This chip will accept audio via i2s and output high level pwm. You need to control it via i2c and provide suitable output filtering to the speaker(s). Given the number of setup and control permutations is not realistic to expect GPIO. At least a uP is required to drive the i2C in conjunction with i2s. If you are only trying to transfer audio streams with no format or rate changes to the STA350BW no DSP is required. In theory a PC could be used if you find suitable i2c and i2s interfaces. Software would be required on the PC to drive these interfaces (basically your own customised audio player). Generally simpler to a use uP.
Otherwise just go analog. DA --> Class D amp module.
Man, that is cool!
I have to get one. You can drive it from just about any modern uP. It will need a heat sink on STA350BW for realistic use. Having the PCM to PWM conversion taken care of is a very big advance.
I have to get one. You can drive it from just about any modern uP. It will need a heat sink on STA350BW for realistic use. Having the PCM to PWM conversion taken care of is a very big advance.
This is what im actually doing - feeding analog input Class D modules by a signal processing board via integrated DA converters.
This is exactly what i like to do in the future. Any format and rate changes have to be handled upstream anyway.
TI has some hardware configurable chips as the TAS5760M. But they are even more limited in power as the STA350BW and i did not see any ready aviable modules.
In fact its never as easy as it looks in the eyes of the µP folks. Even major companys as SMSL have difficulties doing this right - as you can see in the first revisions of their Q5.
----------------------------
Assume something a bit aside from the stereo, 2.1 or 2way speaker mainstream - maybe just a little active 3way speaker.
The power stage of the STA350BW (or similar chips) would easy give this in 2xSE + 1xBridge configuration - but the fixed internal signal processing does not - so you have to do that by your own anyway.
Or think in something more exotic as subtractive crossovers, or Horbach/Keeles multiway constant beamwidth speakers ... at this time all the "flexibility" of the chip - causing the need of a µP control - with all the added complexity and failure potential - renders as other people needs but not yours
I dont want to blame anybody or anything. And i am happy there are modules to by with oer without µP control. But this is far away from a perfection in the sense of KISS - do just one thing - but do it right
(sorry about my rough english, hoping to be more or less understandable)
/ropf
Last edited:
Are you shure? The first time opening the SMSL Q5 i was very surprised - there is no heat sink on the STA350BW.
Then - studying the datasheet - the thermal pad seems to be on the chips downside - designed to transfer the heat via copper connectoions and groundplane. Did not dig in deeper, because i dont plan to build own modules.
I performed a test - driving 4Ohm closed cabinet speakers with a cutoff frequency of 70Hz by a sine of 12Hz (this way it was not to loud). At high level there was a point where the overcurrent protection kicks in - what is normal - but the chip stayed relativley cool.
i2c with Arduino?
i2c
About the simplest way to implement i2c is with an Arduino board (Uno, Due, Mega, ...). You would need to write software to monitor the digital inputs and depending upon the input states send commands via i2c (using WIRE library). Should be quite a bit simpler than using an ST Discovery board and low cost as well. The specification of the software requirements would actually take more effort than writing it.
Heatsink
I assumed that a heatsink is required. Given the low cost of these boards Will have to get a few and see how they work.
SMSL Q5
The cost of the SMSL Q5 is demoralising. So economical; even here in Australia. But, I would rather build something myself. See: https://voll.com.au/shop/smsl-q5/. I have been looking at getting some Voll speakers mainly for testing purposes. The SMSL amp looks to have a decent set of features.
i2c
About the simplest way to implement i2c is with an Arduino board (Uno, Due, Mega, ...). You would need to write software to monitor the digital inputs and depending upon the input states send commands via i2c (using WIRE library). Should be quite a bit simpler than using an ST Discovery board and low cost as well. The specification of the software requirements would actually take more effort than writing it.
Heatsink
I assumed that a heatsink is required. Given the low cost of these boards Will have to get a few and see how they work.
SMSL Q5
The cost of the SMSL Q5 is demoralising. So economical; even here in Australia. But, I would rather build something myself. See: https://voll.com.au/shop/smsl-q5/. I have been looking at getting some Voll speakers mainly for testing purposes. The SMSL amp looks to have a decent set of features.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.