It is VERY (unbelievably) strange.
Here are the measurements of your trafos 0.5+0.5:6 in my setup for reference (for differential mode of TDA1541A) and 0.64Vrms output:
I understand that for 2Vrms THD will be higher, but the same at 1kHz as for 20Hz
Hi Ivan,
Yes I was a bit confused as well!
I think the high noise floor of my test setup is adding to the level of the harmonics. So like i said before, take them with a grain of salt, maybe a few grains...
To compare apples with apples - is it possible you could measure 20hz at .707V rms? - maybe not really much different.
At the end of the day I let my ears have the final say, if the bass sounds a bit thick and the vocals have an edge to them - I simple cant enjoy the music in the same manner. There is a good chance the dominant H3 could be the culprit.
Maybe when running in differential mode with the cancelation of H2 the H3 is more noticeable?
If to answer short, no I can't, because your trafos was in single vers
But I can say there will be almost no difference with the plot I showned, as 0.707 vs 0.64 is less than 1dB difference. Will be pretty the same
Well, I can say that the differential mode is less natural for my ears, I will better prefer a bit more harmonics, but with its organicaly natural sequence (it is a bit vice versa for LF, but H2 should be there).
Another story is a parallelling, which is not works for me as well, as the sound precision will be determined by the best piece from batch in best case scenario. Paralleling is good for SigmaDelta, not for our hero...
So that is why I do not see any benefits for my ears to use more than one single TDA1541A.
Horses for courses. Points taken.
I found that going from single ended to differential gave a cleaner more hifi sound, particularly in the vocals - they sounded less "bloomy" for lack of a better word - not that I had an issue with how the vocals sounded. I would surmise that this would be from the attenuated H2. There are probably other factors at play here like common mode noise attenuation and bit averaging - but that was my experience anyway.
While on the topic, I took a differential measurement on my scope using the "poor mans differential measurement technique" to see what noise was left over. Two probes each 50 ohm terminated and in this case added together instead of subtracting to cancel out the music signal. The probes were connected to each side on the primary winding with the grounding leads left off - this way we can see the noise left over without the signal that will be attenuated by the transformer. Music was playing at full swing while taking this shot.
The amplitudes in reality are larger because of the passive probe impedance, but still, its a good indication of the noise being attenuated by the transformer in differential mode.
I found that going from single ended to differential gave a cleaner more hifi sound, particularly in the vocals - they sounded less "bloomy" for lack of a better word - not that I had an issue with how the vocals sounded. I would surmise that this would be from the attenuated H2. There are probably other factors at play here like common mode noise attenuation and bit averaging - but that was my experience anyway.
While on the topic, I took a differential measurement on my scope using the "poor mans differential measurement technique" to see what noise was left over. Two probes each 50 ohm terminated and in this case added together instead of subtracting to cancel out the music signal. The probes were connected to each side on the primary winding with the grounding leads left off - this way we can see the noise left over without the signal that will be attenuated by the transformer. Music was playing at full swing while taking this shot.
The amplitudes in reality are larger because of the passive probe impedance, but still, its a good indication of the noise being attenuated by the transformer in differential mode.
Attachments
Audial USB to Simultaneous PCBA
Just to say Pedja Rogic is looking at doing a run of USB XMOS input to simultaneous output PCBA's, it would likely have external clock inputs and balanced output option. Ryans I2S to simultaneous converter is super handy but this could be a decent alternative, if pricier, one board front end.
USB interface and AYA 2021 editions [poll] - Audial
Just to say Pedja Rogic is looking at doing a run of USB XMOS input to simultaneous output PCBA's, it would likely have external clock inputs and balanced output option. Ryans I2S to simultaneous converter is super handy but this could be a decent alternative, if pricier, one board front end.
USB interface and AYA 2021 editions [poll] - Audial
Hi analog_sa,
Your right, I will look into this and correct my mistake, thanks.
R16 and R17 need a combined value of about 290R to give a 24V supply, 4V across the CCS and 20V for -15, -5, +5.
Last edited:
Just to say JLSounds USB/i2s board has been doing this for a while
I have one and it sounds very good with separated power supplies and the D3 in simultaneous!
Pedja's board would have the option of external clocks and also balanced out as a nice option, but it looks like there's not enough demand for him to progress it. Luckily Ryan's simultaneous PCB is in my in-tray
I want to try balanced.
I don't think it is a big deal as the actual regulation is determined by the separate shunts. Would be very surprised if it's audible at all.
Yeah I couldn't detect any difference, maybe slightly 'darker background' but my mind could be influencing this.
For all that have the D3 up and running its not an urgent mod, but what I changed was the value of R21 from 5k1 to 4k7 which gives about 24.3V at the output. Ill update the BOM and Schematics. Sorry for any inconvenience!
4k7 resistor
Hi All,
After further listening with the correct resistor in place there is clearly an improvement in SQ.
If you have already ordered all your parts and cant easily/cheaply get hold of this resistor value and dont want to place an order just for one 50 cent resistor then please click the link below and ill get one to you.
If you have a 4k7 in a 0805 package these will also fit the 0204 footprint.
click here
Hi All,
After further listening with the correct resistor in place there is clearly an improvement in SQ.
If you have already ordered all your parts and cant easily/cheaply get hold of this resistor value and dont want to place an order just for one 50 cent resistor then please click the link below and ill get one to you.
If you have a 4k7 in a 0805 package these will also fit the 0204 footprint.
click here
Hifi Collective has various voltages of 100uf KA, here is the 25V one
Hifi Collective 25V
Farnell has the 16V (for other voltages additional fee have to be applied)
Farnell 16V
Hifi Collective 25V
Farnell has the 16V (for other voltages additional fee have to be applied)
Farnell 16V
There is also a lower leakage one, the KL series, as another member pointed out
https://www.diyaudio.com/forums/dig...e-nos-dac-using-tda1541a-656.html#post5562405
These ones can be found in EU from TME
TME 100uf UKL series
https://www.diyaudio.com/forums/dig...e-nos-dac-using-tda1541a-656.html#post5562405
These ones can be found in EU from TME
TME 100uf UKL series
Digikey has em @ .59 US.
Blocked
Don't know why the link says it's Blocked. It's safe.
Hope this helps!
Aguaazul
Last edited:
Ryan,
That is really nice of you to offer to send us some! I have a bunch of D3's so I don't mind buying my own.
At Mouser US, I changed [MELF Resistors 0.4W 5.11Kohms 1% 02 04 MELF 200V 50ppm] to [MELF Resistors 0.4W 4.7Kohms 1% 02 04 MELF 200V 50ppm] & searched Mouser.
They have 2 exactly the same specs. Different part numbers,
MMA02040C4701FB300 @ .26$ US.
MMA02040C4701FB000 @ .12$ US.
I'll be ordering 10, & that drops the .26 down to .19.
It's not the money... Looking to get the correct part!
What is the part number.
Cheers,
Aguaazul
Hi Aguaazul,
Either of those parts should be fine, the only difference is the factory pack quantity for buying full reels.