TDA1541A DUAL DIFFERENTIAL MK1

[batteryman]

More on my dual 1541a Rainbow_hui dac saga..

This has now been configured for differential working using a pair of Sowter 9545 transformers in MK1 test version.

Using one dac for normal and the other for inverted data operation meant just interposing a spare inverter in IC11 - a 74HC14 before pin3 of dac 2.

To prove that I had connected the transformers correctly, with both dacs receiving the same data, there was no ouput ( as good as!) as expected.

Next step is to use the dacs in simultaneous mode with one dac per channel. However this requires 2s complement to offset binary conversion and a simple method of doing so has been devised which I hope to test next week.

In the meantime more listening to be done for the consequences of the lsb error due to inverting the data. (Nothing audible with a quick listen)

I found a mistake on the pcb, the unused gates of IC11 are shown correctly as having the inputs tied to ground except that its actually pin 6 (output) that was grounded instead of pin 5.
Ianj

[guido]

One way is to use a DF (like PMD100) and put the tda's in simulatious mode. There are designs on the net with a cpld to do this also (but with i2s feeds iirc).

[batteryman]

Originally Posted by guido

One way is to use a DF (like PMD100) and put the tda's in simulatious mode. There are designs on the net with a cpld to do this also (but with i2s feeds iirc).

It only needs the msb inverting so doesn't seem a difficult job with a some 74F logic..
The data stream is msb first, and commences after ws goes high so we know where it is and 'lasts' for one bck cycle....
(I am running the dacs in nos mode and want to keep everything as simple as possible so that would rule out a PMD100)
Ianj

[oshifis]

Interesting project, I have something similar in my mind, but the many gates needed for a perfect I2S inverter (like the one designed by Pedja Rogic) deterred me. If you succeeded, please share it with us. I would like to combine a dual differential TDA1541A DAC with my tansformer/tube I/V converter...

Laszlo's Valve Output Stage with Lundahl transformer

[batteryman]

Originally Posted by oshifis

Interesting project, I have something similar in my mind, but the many gates needed for a perfect I2S inverter (like the one designed by Pedja Rogic) deterred me. If you succeeded, please share it with us. I would like to combine a dual differential TDA1541A DAC with my tansformer/tube I/V converter...

Laszlo's Valve Output Stage with Lundahl transformer

Why not try the simple method I am currently evaluating of just inverting the data to one dac? (assuming they are currently paralled)
I have been listening to my favourite cds for several hours and cannot hear any unwanted side effects.
I am in fact surprised with how clean the results are, with no digital noise, aliasing, nor other artifacts considering that there is no filtering at all (apart from that afforded by the transformers) and nos mode is being used. Its not perfect and not a major advance over parallel dacs but more work is planned.

The lsb errors resulting from imperfect inversion surely do not matter because when a track is playing there is always enough noise in the recording* to swamp the errors?
I'll let my ears be the judge everytime rather than pursue perfection & theory.

*from the mics, mixers, audience, performers and general ambient noise, air conditioning etc etc.

[batteryman]

This is a summary of the mods and I/v transformer connection to the raindrop_hui dual dac pcb. I have taken a holistc approach and modded the preamp input as well.

a) One DAC receives inverted data using a spare inverter in U11 (74hc14) i.e. its one dac for true and one for inverted data. There are no other mods to the I2S signals.

b) All four dac outputs are loaded with .1% Beyslag 100R resistors.

c)The 9545 transmformer primaries are connected between the true and inverted dac outputs. The centre tap is not connected to ground, this prevents any dc flowing through the transformer due to the dac output offset. (I cannot hear any difference when I do ground it)

d) The CD input of the Quad 34 has been modified by removal of the input coupling cap, and series 39k resistor, just leaving a 1M resistor from non inverting input to ground.

The results are excellent but the next step made a surprising difference...

e) The secondary of the transformers (series connected) were loaded with 2.7K at the preamp input socket. This is reflected to the primary as about 90r and had a very surprising effect on improving the 3 dimensionality and depth and also boosting the low end a little at the expense of a drop in signal. This isn't a problem as the Quad 34 Cd input is very sensitive compared to modern standards (100mv).

Until I get round to trying the one dac per channel configuation its hard to know whether this will produce an even better result. Maybe the extra complexity of the logic needed to process the I2S signals will cancel out any gains....

Next I will increase some of the decoupling caps - all are 100n currently.
The result so far is superior to my MF X-dac. (an X-ray player supplies the spdif input)

[batteryman]

I dug out an unused TC Electronics M350 audio effects processor which I remembered has an spdif output. The inputs were fed from a sig gen and the bypass butotn pressed.
The resulting 1Khz square wave suffers from plenty of ringing (taken from the M350 analogue output).

The final traces were with the transformer connected between the dac L & -L outputs, in parallel with 100R (no resistors to ground on the dac output. The centre tap was grounded and the secondary loaded with 47k.

The previous traces were with 100R to ground on all dac outputs and the secondary loaded with 2K, centre tap not grounded.

image 1 shows the dac outputs - dac 2 is the inverting dac.
image 2 shows input and output of the transformer
image 3 shows analogue output of the M350 v output of the transformer.

[batteryman]

Its often said that the TDA1541 does not like dc on its outputs and so needs a very low load value load resistor ege 15 to 33r. However I can find no mention of this on the 1991 datasheet.

Well using Sowters 9545 transformers, and dropping down from 100r to 47r produces a much less pleasant sound. It is thinner, harsher and flatter.

I have put up with this all week to see if I can find any merit at all in using a lower resistor and then today, switched back to 100r resistors.

The result is more bass, greater dimensionality and depth, and lack of harshness, yet no lack of treble. If anything it is clearer and more delicate. A vast improvement on the lower value.

My next step is to interpose EC Designs common gate fet (2sk170) I/V converters between dacs and transformer. The common gate configuration offers very low input impedance, unity current gain and with a 500r drain resistor, an output signal of 2vp-p (from the 4ma dac output current).

However the output impedance is highish so I'm not sure what the audible result will be. The transformers are connected between the true and inverted dac outputs with the centre tap not grounded so no dc flows through the windings. I will no doubt need to experiment with the secondary load resistors which are currently 2k7. The preamp's (Quad34) input opamp was also changed from a OPA2604 to a NE5532.

[analog_sa]

Sorry for the off-topic post but why would anyone with diy abilities use something as profoundly awful as a Quad 34?

[batteryman]

Originally Posted by analog_sa

Sorry for the off-topic post but why would anyone with diy abilities use something as profoundly awful as a Quad 34?

Please keep your off-topic opinions to yourself.