I'm sorry (I'm not thát into electronics), but what do you mean by cathode on the circuitry of the board?
Thanks for the suggestions
The TL431 has an anode ( the pointy side) and a cathode the opposite side, and an adjust pin (on the side)
Hi Thorsten,
Exactly my reason to purchase this kit. Quite simple layout without too much nonsense
Yes you should get the kit. It is the best TDA1541 kit available for the price. It is a very solid performer.
Thorsten, with regard to the ferrite beads on the HiFiDIY kit, are you able to advise me of the most suitable bead to chose with regard to the impedance at a certain frequency?
Am I aiming for the highest impedance at 10Mhz or should I concentrate on another frequency?
Should I choose one frequency for the CS8414 supply and another for the DAC supply?
Thank you.
Am I aiming for the highest impedance at 10Mhz or should I concentrate on another frequency?
Should I choose one frequency for the CS8414 supply and another for the DAC supply?
Thank you.
Last edited:
Hi,
Deasn't really make a huge difference, but for small stuff I use the 11DQXX, 21DQXX and 31DQXX range, they go up to 100V in 1, 2 and 3A which is okay for most applications.
For high current rectifiers I use 40CQ100 or related stuff... It is possible to stack up 3 pcs 40CQ100 to make up a bridge rectifier, the diode where both are used should be at the bottom of the stack and connected to some heatsink of so...
For really high power a small PCB with individual heatsinks for individual diodes may be required.
Ciao T
Deasn't really make a huge difference, but for small stuff I use the 11DQXX, 21DQXX and 31DQXX range, they go up to 100V in 1, 2 and 3A which is okay for most applications.
For high current rectifiers I use 40CQ100 or related stuff... It is possible to stack up 3 pcs 40CQ100 to make up a bridge rectifier, the diode where both are used should be at the bottom of the stack and connected to some heatsink of so...
For really high power a small PCB with individual heatsinks for individual diodes may be required.
Ciao T
Hi,
Not sure they are really better.
What I feared. As it is only a 2-layer board the top is groundplane and all the routing below, so it is not easy to solder decoupling cap's under the TDA1541.
You could try getting some self adheasive Kapton tape (Kapton should withstand the heat from soldering) and mask over the area below the TDA1541 (so you do not accidentaly short things) and then apply some self adheasive 3M Copper foil cut to side to this to have the decoupling ground plane below the TDA1541.
I suspect the reason is more down to the "mother of invention" (neccessity), as the CS8412 is nowadays genuine unobtanium.
Note, just a CS8414 on an adpater is actually worse than a big DIP CS8412 (not a lot though), only if you apply suitable extra decoupling cap's can you benefit from this "improvement":
Have a really close look at this above image.
The really small 0603 Size Cap's are 1uF/25V (except for the on the Filt Pin) and the bigger ones stacked at the pins are 1206 10uF/25V.
The circuit suggested by Philips is this:
(Someone uploaded this to diyaudio - I took from there and re-uploaded, I'd love to give credit to whoever posted it but forgot, appy polly logie)
While the data does show some deterioration of SNR at low frequencies note that it only reaches around 1dB @ 100Hz for 500KHz reclocking, so not much to loose sleep about.
I would suggest a frequency of 4fs or BCK divided by 8. A 74HC191 may anser the purpose when using Output C.
Just look at this page for the placement and polarity of this capacitor, Ok?:
Simple Voltage Regulators Chapter 2: Output Impedance
Ciao T
Not sure they are really better.
What I feared. As it is only a 2-layer board the top is groundplane and all the routing below, so it is not easy to solder decoupling cap's under the TDA1541.
You could try getting some self adheasive Kapton tape (Kapton should withstand the heat from soldering) and mask over the area below the TDA1541 (so you do not accidentaly short things) and then apply some self adheasive 3M Copper foil cut to side to this to have the decoupling ground plane below the TDA1541.
And again, my board already came with a small version of the CS8414 on an adapterboard. Apparently there is some positive development in chinese kits
I suspect the reason is more down to the "mother of invention" (neccessity), as the CS8412 is nowadays genuine unobtanium.
Note, just a CS8414 on an adpater is actually worse than a big DIP CS8412 (not a lot though), only if you apply suitable extra decoupling cap's can you benefit from this "improvement":
An externally hosted image should be here but it was not working when we last tested it.
Have a really close look at this above image.
The really small 0603 Size Cap's are 1uF/25V (except for the on the Filt Pin) and the bigger ones stacked at the pins are 1206 10uF/25V.
The circuit suggested by Philips is this:
An externally hosted image should be here but it was not working when we last tested it.
(Someone uploaded this to diyaudio - I took from there and re-uploaded, I'd love to give credit to whoever posted it but forgot, appy polly logie)
While the data does show some deterioration of SNR at low frequencies note that it only reaches around 1dB @ 100Hz for 500KHz reclocking, so not much to loose sleep about.
I would suggest a frequency of 4fs or BCK divided by 8. A 74HC191 may anser the purpose when using Output C.
Just look at this page for the placement and polarity of this capacitor, Ok?:
Simple Voltage Regulators Chapter 2: Output Impedance
Ciao T
Hi,
First, I would add ton's of local SMD Cap's for the CS8414 Supply (see earlier post) and pull the film cap and place the biggest os-con I can find there (in my picture you can see 470uF!)
I would use the largest value small size Low-Z capacitors you can find (my picture shows 2,200uF/6.3V ME-CA series Sanyo barely larger than the Os-Con's in the positions where Os-Con's are now...
So I'd have the TL431 with around 150 milliOhm Z-Out bypassed by 2,200uF with 66 milliOhm ESR (@ ~ 100KHz), which is bypassed with 470uF with 15 milliohm ESR (@ ~ 300KHz) which is then bypassed by 20uF Ceramic with around 5 milliohm ESR (@ ~ 1MHz) and the 1uF 0603 ceramic with around 30 milliohm (@ ~ 15MHz)...
This level of decoupling/bypassing may seem mad, but it makes sure I have a very low impedance from DC into the 50MHz range. Any better at high frequencies would not really help, as the CS8414 lead frame would negate any further improvements. I do not use ferrite beads here, but if I did they would be between 2,200uF and 470uF Os-Con.
Now to the ferrite beads.
The TDA1541 tolerates around 6MHz max, it will just not work with faster clocks and it uses current steering, so while we see some higher stuff than 6MHz, it will not be super high. So I would go for a formulation that works well at lower frequencies if I was using any at all. I probably would not and just use SMD Film Cap's and Elna Silmic Cap's.
The CS8414 can output 192 * 96KHz (at least) which is around 20MHz, but it is a CMOS IC and will have vicious current spikes at each transition of the clock. For this IC I would look at stuff that works well up to 100MHz.
Ciao T
First, I would add ton's of local SMD Cap's for the CS8414 Supply (see earlier post) and pull the film cap and place the biggest os-con I can find there (in my picture you can see 470uF!)
I would use the largest value small size Low-Z capacitors you can find (my picture shows 2,200uF/6.3V ME-CA series Sanyo barely larger than the Os-Con's in the positions where Os-Con's are now...
So I'd have the TL431 with around 150 milliOhm Z-Out bypassed by 2,200uF with 66 milliOhm ESR (@ ~ 100KHz), which is bypassed with 470uF with 15 milliohm ESR (@ ~ 300KHz) which is then bypassed by 20uF Ceramic with around 5 milliohm ESR (@ ~ 1MHz) and the 1uF 0603 ceramic with around 30 milliohm (@ ~ 15MHz)...
This level of decoupling/bypassing may seem mad, but it makes sure I have a very low impedance from DC into the 50MHz range. Any better at high frequencies would not really help, as the CS8414 lead frame would negate any further improvements. I do not use ferrite beads here, but if I did they would be between 2,200uF and 470uF Os-Con.
Now to the ferrite beads.
The TDA1541 tolerates around 6MHz max, it will just not work with faster clocks and it uses current steering, so while we see some higher stuff than 6MHz, it will not be super high. So I would go for a formulation that works well at lower frequencies if I was using any at all. I probably would not and just use SMD Film Cap's and Elna Silmic Cap's.
The CS8414 can output 192 * 96KHz (at least) which is around 20MHz, but it is a CMOS IC and will have vicious current spikes at each transition of the clock. For this IC I would look at stuff that works well up to 100MHz.
Ciao T
Thorsten,
What can you suggest on opamp decoupling on I/V conversion?
I have AD844s in WSOIC16 package, and it is problematic on decoupling - i can't throw the usual scheme of 1.5n NP0 under the chip + 4.7-10uF electrolytic under the board, as i did here
http://s3t.it/data/uploads/xo3v4ts.png
Attached is my option of bypassing - caps are on the opposite side of the board, and sit right under supply pads of opamp, connected with 4 vias/each.
The rail voltage first passes thru bypass cap's pad, and then, thru 4 vias goes directly to the opamp.
Maybe i can squeeze some TL431s in SMT package, i need to check the power consumption of AD844 first...
What can you suggest on opamp decoupling on I/V conversion?
I have AD844s in WSOIC16 package, and it is problematic on decoupling - i can't throw the usual scheme of 1.5n NP0 under the chip + 4.7-10uF electrolytic under the board, as i did here
http://s3t.it/data/uploads/xo3v4ts.png
Attached is my option of bypassing - caps are on the opposite side of the board, and sit right under supply pads of opamp, connected with 4 vias/each.
The rail voltage first passes thru bypass cap's pad, and then, thru 4 vias goes directly to the opamp.
Maybe i can squeeze some TL431s in SMT package, i need to check the power consumption of AD844 first...
Attachments
Last edited:
You have to be very careful using such an array of multiple caps for decoupling.
Everyone interested in this topic should look up the great work of MrJam here.
You can clearly see that there are combinations which will resonate due to their intrinsic inductivity or the indictivity of the traces connecting them.
JosephK also made some great measurements explaining the issues of paralleling caps.
Hi Thorsten,
How about this implementation of WM8805 I2S: http://www.diyaudio.com/forums/digital-line-level/180336-wolfson-wm8805-i2s-tda1541a-nos.html ?
If the implementation is good, I'll make my own 4 layers PCB for the WM8805 I2S and for TDA1541A after John Brown with Shunt Regulators for each section.
How about this implementation of WM8805 I2S: http://www.diyaudio.com/forums/digital-line-level/180336-wolfson-wm8805-i2s-tda1541a-nos.html ?
If the implementation is good, I'll make my own 4 layers PCB for the WM8805 I2S and for TDA1541A after John Brown with Shunt Regulators for each section.
Hi,
Hence I was quite specific on what I actually used and showed how it is arranged... The cap's with the lowest impedance and highest resonance are closest to the IC. Working our way outwards each subsequent capacitor has higher ESR, meaning it acts primarily as snubber and minimises resonances.
I agree, you cannot just throw capacitors together willy-nilly.
Ciao T
Hence I was quite specific on what I actually used and showed how it is arranged... The cap's with the lowest impedance and highest resonance are closest to the IC. Working our way outwards each subsequent capacitor has higher ESR, meaning it acts primarily as snubber and minimises resonances.
I agree, you cannot just throw capacitors together willy-nilly.
Ciao T
Hi,
Just don't use Op-Amp's. The AD844 is not great by a long stretch (neither closed loop or open loop).
There are so many better choices, not the least tubes...
Ciao T
Just don't use Op-Amp's. The AD844 is not great by a long stretch (neither closed loop or open loop).
There are so many better choices, not the least tubes...
Ciao T
Pano,
With respect, what I do goes WAY (and I mean WAY) beyond the datasheets.
Datasheets usually have one 0.1uF cap with unspecified HF resonance where I have 1uF, nothing where I have 20uF 1206, 10uF-100uF generic elcaps (which will be around 3 to 10 times higher ESR than what I use) where I have 470uF OsCon and nothing where I add 2,200uF, plus a long trace to a 3-Pin series regulator that usually accumulates way more R & L than a 431 has.
If you actually model the arrangement allowing for some trace inductance and resistance you can observe a very different supply impedance, at frequencies where no regulator whatsoever does anything useful and where it hence matters a lot.
Ciao T
With respect, what I do goes WAY (and I mean WAY) beyond the datasheets.
Datasheets usually have one 0.1uF cap with unspecified HF resonance where I have 1uF, nothing where I have 20uF 1206, 10uF-100uF generic elcaps (which will be around 3 to 10 times higher ESR than what I use) where I have 470uF OsCon and nothing where I add 2,200uF, plus a long trace to a 3-Pin series regulator that usually accumulates way more R & L than a 431 has.
If you actually model the arrangement allowing for some trace inductance and resistance you can observe a very different supply impedance, at frequencies where no regulator whatsoever does anything useful and where it hence matters a lot.
Ciao T
Hi,
It derives from raindrop-hui kits, with many of the problems retained.
I would be hesitant to recommend it's electrical design over that outlined for the vero-dac. I am also unconvinced that even 4-layer boards are a better choice than in air 3d...
I have been for a while considering design choices and techniques for an "ultimate" design. AMR currently has other priorities, but a small "fanclub" project may be on the cards, sadly without accessing AMR's technolgy or resources, though the "rejects, leftovers and tangents" that I can use readily make for a lot of material to use.
I also have access to 4 or 6 layer Teflon & Ceramic PCB fabs for small runs.
PM me.
Ciao T
It derives from raindrop-hui kits, with many of the problems retained.
I would be hesitant to recommend it's electrical design over that outlined for the vero-dac. I am also unconvinced that even 4-layer boards are a better choice than in air 3d...
I have been for a while considering design choices and techniques for an "ultimate" design. AMR currently has other priorities, but a small "fanclub" project may be on the cards, sadly without accessing AMR's technolgy or resources, though the "rejects, leftovers and tangents" that I can use readily make for a lot of material to use.
I also have access to 4 or 6 layer Teflon & Ceramic PCB fabs for small runs.
PM me.
Ciao T
Hi Thorsten,
I meant only their WM8805 I2S implementation. As for the TDA1541A, I intend to follow John Brown.
Even not with good PCB design and routing?
Thanks,
I meant only their WM8805 I2S implementation. As for the TDA1541A, I intend to follow John Brown.
Even not with good PCB design and routing?
Thanks,
Hi,
Do you mean as in "does it output I2S"or in "does it reliably provide low jitter?", the first, yes, with the right software, the second, not particulary.
Ciao T
Do you mean as in "does it output I2S"or in "does it reliably provide low jitter?", the first, yes, with the right software, the second, not particulary.
Ciao T
Part of the 'fanclub'! Looking forward to it, if designs and design considerations can be shared.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.