• These commercial threads are for private transactions. diyAudio.com provides these forums for the convenience of our members, but makes no warranty nor assumes any responsibility. We do not vet any members, use of this facility is at your own risk. Customers can post any issues in those threads as long as it is done in a civil manner. All diyAudio rules about conduct apply and will be enforced.

DIY TDA1541A PCB "D3"

Would you like a through hole version of the D3?

  • Yes

    Votes: 12 52.2%
  • Maybe, depends on the performance

    Votes: 11 47.8%

  • Total voters
    23
  • Poll closed .
I would be interested in a pcb that is more hackable than the D3 board;)

For instance to bypass the LM317 combined power supply and use separate digital and analog power supplies with very low noise regulators. I could not do it on your D3 board due to the small components and 4 layer board (that is very nice by the way!)

I don’t know if it will better your dense smd design (the small paths are also beneficial and I like the closely positioned shunt regulators) but it would be nice to experiment with.

Maybe a combination of shunts, dem caps, digital data attenuators in smd very close to the dac chip and more through hole for components that not have to be very close to the dac?
Also nice to have TO-220 format ps regulators so one can change these for other types.

Regards,
 
Last edited:
Thank you for the good work Ryan. I recently heard about your boards and I'm yet to try it. But I have a bloodac with 4 TDA chips and I want to try the chips on a newer design so if you go ahead with the balanced design, I'd be up to purchasing the board(s).

Have you considered a design with dual differential balanced operation (Sorry I'm new to DIY and audio in general so I might have the wrong terminology) that can use 4 chips at once? (I'm greedy I'm sorry lol)
 
I would be interested in a pcb that is more hackable than the D3 board;)

For instance to bypass the LM317 combined power supply and use separate digital and analog power supplies with very low noise regulators. I could not do it on your D3 board due to the small components and 4 layer board (that is very nice by the way!)

I don’t know if it will better your dense smd design (the small paths are also beneficial and I like the closely positioned shunt regulators) but it would be nice to experiment with.

Maybe a combination of shunts, dem caps, digital data attenuators in smd very close to the dac chip and more through hole for components that not have to be very close to the dac?
Also nice to have TO-220 format ps regulators so one can change these for other types.

Regards,

Thanks for the suggestions Supersurfer.
 
Thank you for the good work Ryan. I recently heard about your boards and I'm yet to try it. But I have a bloodac with 4 TDA chips and I want to try the chips on a newer design so if you go ahead with the balanced design, I'd be up to purchasing the board(s).

Have you considered a design with dual differential balanced operation (Sorry I'm new to DIY and audio in general so I might have the wrong terminology) that can use 4 chips at once? (I'm greedy I'm sorry lol)

Hi santoshmanivannan,

I have no plans to make a PCB dedicated for balanced operation.
 
Hi Ryanj,

Finally last week my Sowter 1465 transformers had arrived. They have been playing a week on one board and I am impressed by the quality!
I have connected the windings in parallel for 1:5 ratio. 1:10 does not work with my volume transformer in the amps (too high impedance)

With a 2,7kohm resistor on the secondary it was very good!

But the goal for me was to drive them differentially so I soldered my second D3 board this week and connected everything in differential mode. It is now playing nicely in my work room, time to set it up on the main system for some serious listening.
The Load resistor is lowered to 1,15kohm as I do not need the gain and prefer low distortion and good driveability for the volume transformers.

Just a question regarding the current draws. How much do the 5v on i2s-pcm board and D3 board draw?
And how much current is drawn from the 26v supply?
 

Attachments

  • EA52D815-6C80-43A9-8A53-582C89ACD8BE.jpg
    EA52D815-6C80-43A9-8A53-582C89ACD8BE.jpg
    980.5 KB · Views: 321
  • 402E3151-CC99-4574-8DB0-58B35DAE3C4E.jpg
    402E3151-CC99-4574-8DB0-58B35DAE3C4E.jpg
    1,006.9 KB · Views: 308
I am an unlucky man:(

My dac had been running for a few hours nicely, and than at once the music stopped.
Checking the voltages discovered that my salas ultrabib was destroyed. I checked both of the D3 boards and the new one was dead. I think a short in the ps part and it also pulled down the salas.
When I attach an LT1893 power supply on the 26v input and turn it on it drops to 0v instantly.

Nothing to see or smell on the pcb.

Is there any usual suspect to check before I replace all active devices?
 
Hi Supersurfer,

Yes that is very unlucky. :(

The D3 26V draws about 140mA, the 5V about 20mA, and the i2s2sim 5V about 5mA - 20mA (depending on how it is connected).

Check the temperatrue of all the polymer caps, sometimes if they are over heated in the soldering process they can leak a lot more current. This excessive leakage current might be causing the noise. It could also be one of the active components - I would probably start with CC1.
 
Hi,

I used the 16v versions because the 25v versions are not available (huge backorder) and I never measured more than 10v on the DEM pins.
I did use the low leakage type you have advised though. According to the datasheet the leakage current is smaller on lower voltage ratings. See datasheet:
https://nl.mouser.com/datasheet/2/293/e_uka-1512253.pdf

Also the noise is on one board and they both are build with exactly the same components.

After I fixed the broken C10 I did make a mistake by connecting the 26v with wrong polarity, just a few seconds when I noticed the led did not light up and I powered it down again.
After that the DC voltages measured fine again.

Maybe some of the polymer caps are damaged?

If all DC voltages are ok than CC1 will not be suspect I guess?
 
Last edited:
The 16V ones may be ok, but... at their rated voltage they leak 3uA (3000nA) and the LSB is 61nA, so yes I understand they are not operating at the rated voltage but I have a feeling even at 10V you may still be getting some leakage. Then again this may be a non issue.

How much current is the 26V supply taking?

Have you checked that the DEM is running at 50hz?
 
How can I check that it is running at 50Hz.
Just put the scope on the cap legs?

If I understand the datasheet correctly the leakage is calculated like this: 0,01 x rated capacity x voltage = 0,01 x 100 x 7 =7uA
For dem voltage of 7v. This looks to be independent from the capacitor rated voltage.
 
Last edited:
I have found the mistake. When the board is connected in stereo (not differential with 2 boards) the left channel is muted. As there is not much between the input of the dac and the output, I found the defect in a loose soldering on the input ufl of the channel. It must have come loose with de-mounting the board.

Sometimes it there are separate issues that are not related but it puzzles your mind......

Now I am listening to the balanced setup again. Lets hope it keeps running :xfingers:
 
Well done finding the problem. Then I supose the leakage of the 16V rated caps may not matter but to test whether you are getting any leakage play a recoring at -60dB (set the digital volume at -60dB, or find a -60dB recording), if it sounds clean then the LSBs are not distorted.

EC mentioned leakage in this post:

Hi weissi,



TDA1541A is placed in I2S mode by connecting pin 27 to +5V, pin 4 has nothing to do with this. TDA1541A can be placed in simultaneous mode by connecting pin 27 to -5V. The pins 1 ... 4 now get a different function.

I already mentioned that this will happen in my previous post and why this happens (100uF caps need to be charged with the small bit currents). With my prototype distortion was gone after a minute or so using 100uF /25V Nichicon UKA caps or caps with a higher voltage. Do -not- use 16V capacitors, it is best to have a large voltage margin for lowest leakage current, this will also ensure faster charging with low bit currents.


TDA1541A performs -very- poorly with I2S because:

Critical sample timing signal has to be extracted from WS and BCK while the I2S stream produces continuous ground-bounce. This causes trigger uncertainty and deterministic on-chip jitter.
The permanent I2S stream causes extra on-chip switching noise.
The 0111111111111111 to 1000000000000000 change around the extremely critical zero crossing area will increase low level distortion.

I experimented for almost a decade with the TDA1541A. With I2S there comes a point where sound quality doesn't get better no matter what is tried.

TDA1541A performs a lot better when driving it in simultaneous mode with 25% data density (16 bits data followed by 48 bit silence). This also reduces average switching noise during each sample by up to 75% .
Sample timing is generated by a dedicated signal (LE) and can be generated when the data stream has -stopped- for 16 bits and ground-bounce is almost zero. This leads to lowest on-chip trigger uncertainty and lowest on-chip jitter. This is -very- important with these kind of chips that switch relatively slow (approx. 7 MHz max. switching frequency).

TDA1541A performs best by far when driving it in signed magnitude mode because:

Low level distortion and switching noise will be reduced significantly as only few bits change around the zero crossing and the monotonicity of the DAC improves.

If one is concerned about channel matching (this causes ever so slight even order harmonics distortion that is not objectionable) following can be done:

Put 100K ... 1M multi turn trimmers in parallel with each 150R passive I/V resistor and sum both outputs with say 1K resistors. Then trim for best channel / segment matching.