Hi, no problem.
In another version i was planning on applying this DEM configuration. But no i don't know what improvements can be gained as i haven't tried it yet.
In another version i was planning on applying this DEM configuration. But no i don't know what improvements can be gained as i haven't tried it yet.
Last edited:
cool,
Just need to cut the upper trace of the voltage trace because no need about BCLK or WdCK anymore !
An idea : just keep the two pads of the voltage gnd and put its closer to the two caps pads... Why ? because the inductance : there is enough place on the same caps pads for the cap and at 90° for half on a Sussumu resistor in 805 case. So less pads, more compact and no problem of inductance if resistors are not used by some diyer like Shane e.g. So the two pads and traces for resistors beginning at the caps pads could be suppress !
Is there place at opposite for the inputs (mayb bad) I gave for the 4 pads clock at the opposite side of the TDA ?
Just need to cut the upper trace of the voltage trace because no need about BCLK or WdCK anymore !
An idea : just keep the two pads of the voltage gnd and put its closer to the two caps pads... Why ? because the inductance : there is enough place on the same caps pads for the cap and at 90° for half on a Sussumu resistor in 805 case. So less pads, more compact and no problem of inductance if resistors are not used by some diyer like Shane e.g. So the two pads and traces for resistors beginning at the caps pads could be suppress !
Is there place at opposite for the inputs (mayb bad) I gave for the 4 pads clock at the opposite side of the TDA ?
Last edited:
Yes it might be a good idea to play around with this config first to see if it makes any difference at all. The Vias in my board connected to the DEM pins could allow experimentation.
I think the best is to try to lock the signal like you do with a air cap like your photograpgh. Diyer without knowledge and scope but just ears like me can try with a test CD to find a lock around a choosen value like the input above (last design used by John) and here the icing on the cake would be the possibility or not to try the two resistors but without incidence with the layout for the inductance if no resistor (so 4 pads better than 6 six here)
Member
Joined 2006
Last call for the AYA II GB - by Friday, May 23🙂
http://www.diyaudio.com/forums/group-buys/256068-gb-aya-ii-tda1541a-dac-2014-edition-pcb-kit.html
http://www.diyaudio.com/forums/group-buys/256068-gb-aya-ii-tda1541a-dac-2014-edition-pcb-kit.html
Eldam,
I have not yet requested the quote from Laptech because the minimum quantity is 5 pcs per frequency.
The only crystal I found in NDK catalog suitable for an audio oscillator is the model NR-2B, but: no idea about the cost, relative high ESR (25 ohm for a 10MHz unit), low drive level (10uW), the package does not seems cold welded, almost sure not polished. The crystal I got from Laptech is a bit superior (generic datasheet at http://www.laptech.com/pdfs/HC-43-U_generic_data_sheet.pdf). My custom crystal is in a cold welded package, shows an ESR lower than 10 ohm, Q upper than 150K, strongly polished.
The datasheet of the 4 pads crystal oscittator from NDK does not contain any information about the phase noise, but since it costs around 2 USD I can presume it does not perform very good in phase noise.
The Crystek CCHD-957 costs around 15 times and its phase noise is not exceptional, -97dBc@10Hz from the carrier for a 22.5792MHz. I presume better than the NDK, but far from what we are looking with a very good crystal such as the Laptech or similar, at least -120dBc@10Hz from the carrier or better.
Different oscillators could need different power supply, when I said 2 pads only I meant signal and ground. For example the oscillator I'm building needs 5 and 15v. One could use 5V from the digital supply of the DAC board, but the board does not provide +15V. But obviously no problem providing 4 pads (including the power supply for the oscillator), so the diyer is free to use it or not (I suggest separate power supply for the oscillator).
Then one could fit the oscillator board close to the BCK of the DAC, soldering the 2 boards (reccomended) or build a little box for the oscillator and its power supply and link the 2 board via u.fl. cable. Anyway if one would slave the source at least 1 u.fl. cable is needed.
The power supply of the oscillator could be soldered directly to the oscillator board.
I have not yet requested the quote from Laptech because the minimum quantity is 5 pcs per frequency.
The only crystal I found in NDK catalog suitable for an audio oscillator is the model NR-2B, but: no idea about the cost, relative high ESR (25 ohm for a 10MHz unit), low drive level (10uW), the package does not seems cold welded, almost sure not polished. The crystal I got from Laptech is a bit superior (generic datasheet at http://www.laptech.com/pdfs/HC-43-U_generic_data_sheet.pdf). My custom crystal is in a cold welded package, shows an ESR lower than 10 ohm, Q upper than 150K, strongly polished.
The datasheet of the 4 pads crystal oscittator from NDK does not contain any information about the phase noise, but since it costs around 2 USD I can presume it does not perform very good in phase noise.
The Crystek CCHD-957 costs around 15 times and its phase noise is not exceptional, -97dBc@10Hz from the carrier for a 22.5792MHz. I presume better than the NDK, but far from what we are looking with a very good crystal such as the Laptech or similar, at least -120dBc@10Hz from the carrier or better.
Different oscillators could need different power supply, when I said 2 pads only I meant signal and ground. For example the oscillator I'm building needs 5 and 15v. One could use 5V from the digital supply of the DAC board, but the board does not provide +15V. But obviously no problem providing 4 pads (including the power supply for the oscillator), so the diyer is free to use it or not (I suggest separate power supply for the oscillator).
Then one could fit the oscillator board close to the BCK of the DAC, soldering the 2 boards (reccomended) or build a little box for the oscillator and its power supply and link the 2 board via u.fl. cable. Anyway if one would slave the source at least 1 u.fl. cable is needed.
The power supply of the oscillator could be soldered directly to the oscillator board.
Last edited:
thanks for the inputs Andrea,
you have to follow the way you feel as each adds to somebody else work to try to make something together. I just try to add some ideas, maybe not good as my technical background is low. Inputs of fellows, like Ryan and others are important because they have an empirical experience I haven't here !
Here are the information i found about NDK NZ2520SD Ultralow phase noise oscillators vs Laptech : there are measurements with curves i found when I was trying to find some alternative "best enough" sources to the costy Fiffo : https://hifiduino.wordpress.com/category/usb-audio/
I'm sad you have less than 5 fellows for at least 11.xxx ... maybe we are pionners to try to make such a design all togethers with asking at each step some pragmatic questions as you maid with your quotation here.... ???
What do you think of those graph on Hifiduino ? a lipo cell + a big polymer cap + a little one + smt ldo reg + tantalium + ferritebead + local smt decopling with ceramic = a ps for a crystal like a NDK or laptech, hopping tje NDK is low price in regard to the Laptech ?
you have to follow the way you feel as each adds to somebody else work to try to make something together. I just try to add some ideas, maybe not good as my technical background is low. Inputs of fellows, like Ryan and others are important because they have an empirical experience I haven't here !
Here are the information i found about NDK NZ2520SD Ultralow phase noise oscillators vs Laptech : there are measurements with curves i found when I was trying to find some alternative "best enough" sources to the costy Fiffo : https://hifiduino.wordpress.com/category/usb-audio/
I'm sad you have less than 5 fellows for at least 11.xxx ... maybe we are pionners to try to make such a design all togethers with asking at each step some pragmatic questions as you maid with your quotation here.... ???
What do you think of those graph on Hifiduino ? a lipo cell + a big polymer cap + a little one + smt ldo reg + tantalium + ferritebead + local smt decopling with ceramic = a ps for a crystal like a NDK or laptech, hopping tje NDK is low price in regard to the Laptech ?
Ryan,
The Grundig cct with 4x oversampling has WS at 176.4kHz, with no oversampling WS is at 44.1kHz
This referenced frequency is then divided internally by four, so there is a minimum to suitable values if this divided frequency is to remain outside of the pass band.
According to the tech supplement, if the interference components of the signal frequency and fDEM/4 are the same the need for DEM frequency filtering is reduced.
Based on the information, 176.4kHz would be an appropriate frequency to synch to.
Regards,
Shane
Hi Eldam,
I'm not sure the reason for the resistors.. I haven't tried, no.
If I was to move away from the capacitor across 16,17 I'd be wanting to try the cct from the philips supplement.. just like Grundig did, but in this case to be suitable for application without oversampling.
Regards,
Shane
Hi Ryan, I was thinking of trying the FIFO, but by the time I looked at all the recommended components, I could by a good complete dac for less.
Interestingly though that your configuration works with just the fifo and not the I2S to PCM board. Any idea what the I2S to PCM is for? TDA1541 takes I2S as you have implelemented so I dont get it.
Perhaps I should buy just the Fifo as you have done and get a usb to I2S source off ebay.
NDK NZ2520SD: around -113dBc@10Hz from the carrier for a 22.5792 oscillator is not bad, much better than the equivalent Crystek (-97dBc).
I don't know its price and seems it's difficult to find in small quantities:
http://www.diyaudio.com/forums/parts/171379-smd-oscillators-digital-audio-2.html
Hi Luke,
The format of the I2S data is the PCM.
I2S is the physical interface between the blocks, PCM is the data format.
Ians "PCM to I2S board" separates left and right data to allow for simultaneous data input into the TDA1541A which enables it to run at 352.1khz or 384khz.
So if you only want to run at 44.1k like myself, you can just run the 1541 in time multiplexed mode.
Hope that helps.
Hi Shane,
I remember trying it at this frequency, but can't remember hearing an improvement. Have you played around with trying to lock the DEM with WS?
The way I understand it there are advantages to simultaneous data mode at 44.1 too. I haven't looked at simultaneous data mode timing diagram in detail for a while, I did think it was shown to have advantages like allowing additional 'silence' on the input signals during conversion time.
It is Ian's half speed mode clocking that enables 352 and 384k in TDA1541 though it is still clocked above spec at those speeds and I'd love to check (measure) distortion while the chip is being 'over clocked'.
The third thing that the i2s to PCM board gives is a clean reclocked set of inverse data outputs so you can more easily set the DACs for differential output.
How audible all of those differences are is hard to know but a lot of multibit fans nagged Ian to do that board up and they all seemed like robust reasons at the time.
Cheers,
Chris
Hi Ryan,
I'm wanting to try the divided BCK.. but I'm not very good with IC's, probably need to order 5 in order to get one working without a PCB..
Not getting the WS idea, I didnt try it.. if I had 4x OS, I would try it yesterday.
Looking back through the archives, seems John did try the divided BCK at 176.4k with 17 through cap to ground. Buried in there somewhere, previously I'd seen one with 17 floating which would have 10bit operation and I sort of tuned out after that.
It was a conversation with manufacturer that suggested divide BCK after all of this, and some discussions about the grundig cct, and also the pin 17 musings. So I still would try it, but it might be easier to go to Jaycar and get 2x 6k8 resistors, take Johns word for it and not think back about it.
Shane
Hi Shane,
Sorry wrong terminology, i meant a multiple of WS.
I think it might be a good idea to try Johns simple idea.
In his thread he said "Yes, you only have to add two 6K8 resistors to the existing timing cap of 470pF. This will greatly reduce the high level of deterministic jitter on the DEM oscillator as soon as audio data is sent to the TDA1541A."
Easy enough to try!
Sorry wrong terminology, i meant a multiple of WS.
I think it might be a good idea to try Johns simple idea.
In his thread he said "Yes, you only have to add two 6K8 resistors to the existing timing cap of 470pF. This will greatly reduce the high level of deterministic jitter on the DEM oscillator as soon as audio data is sent to the TDA1541A."
Easy enough to try!
Hi Ryan,
Implemented it this afternoon, listening with headphones.. I like it quite a bit better but theres something going on at really low level details.. bit of crunch or fuzz.. not sure.
Using 6k61 and 6k67 0.25W carbon film from the pins 16, 17 to the -15V filter at pin 15, 470pF little red wima cap 16,17 as before. Lead length is less than 3mm including radial bend.. maybe metal film is better here..
Like to know what its doing but no CRO..
Regards,
Shane
- Status
- Not open for further replies.