We must thank @ThorstenL for providing ideas to improve our 1541 dacs.
I certainly plan on cutting the ground connection to the dem caps and connecting to -15v.
I'll try with low value film caps & 220pf dem clock capacitor first and then build another with electrolytic and low frequency dem clocking.
I can't do back to back comparisions.
I definitely favour passive IV (& transformers - the audio is very clean and does not need further ffiltering)
I'm not sure how relevant the digital signal levels are if the bit cliock is stopped when the data has been latched.
I certainly plan on cutting the ground connection to the dem caps and connecting to -15v.
I'll try with low value film caps & 220pf dem clock capacitor first and then build another with electrolytic and low frequency dem clocking.
I can't do back to back comparisions.
I definitely favour passive IV (& transformers - the audio is very clean and does not need further ffiltering)
I'm not sure how relevant the digital signal levels are if the bit cliock is stopped when the data has been latched.
There is a old ECdesign concept with a part of Riv referenced to gnd and a part refered to +5Vdc.
I corrected original formula now is OK and I check on the sims working good. But not tried in praxis.
Also it giving "dc current injection" of 2mA to Iout of TDA1541A.
True RIV vlue is Riv+Ra, but Ra value is about the 10% of Riv. I will correct on the schematic...
I corrected original formula now is OK and I check on the sims working good. But not tried in praxis.
Also it giving "dc current injection" of 2mA to Iout of TDA1541A.
True RIV vlue is Riv+Ra, but Ra value is about the 10% of Riv. I will correct on the schematic...
Last edited:
John has changed his mind on the IV resistor value. He was in favour of very low values because it wa claimed to produce least distortion.
Then he decided that a higher value (~75 I think) sounded better.
I didn't like the sound when I tried a much lower value - it compressed the audio and gave it a lifeless sound.
Then he decided that a higher value (~75 I think) sounded better.
I didn't like the sound when I tried a much lower value - it compressed the audio and gave it a lifeless sound.
Q fr All
Maybe the DEM frequency if it is aplied "extenaly" like in the PHI book style (same as Grundig style), or Hent TP way,
does not have running all of the time?
But only when data is presesent OR after data stored in the register, when the latch told DAC to go to the conversion of stored Datas?
That would be easy to implement if using an Iancanada I2s-PCM board because when he stops the bit clock, you can detect this and stop the dem clock.
With this ECdesign RIV concept You can use whatever RIV value You like. The point is that about 90% of toal RIV (that is Riv value) refered to +5V or any other +V, and 10% of RIV (Ra value) refered to gnd.
And because of incorporated Idc of 2mA there are no offet at the output...
🙂
And even simplier perhaps : you avoid the Idc and nulling circuitry, let the 0 to - 4 mA lives its life through a tube grid happy to see by nature TDA1541A negative offset (-2 mA (who learnt us that 😉 ... T. Loesch) .... then just reverse the polarity at the loudspeakers ! Choose a tube with good transconductance and Mu enough to choose up to the 50 R Riv happy limit and let the pre to add its gain ! We know the usual... well the "casual" suspects (E180EF, ECC88, 3DA with ferrites bids on the pins, 6072A, etc) ! 😊
Problem solved, no CCII, CFB op amps or OTA op amps in open loop anymore, or bad VFB !
And even simplier perhaps : you avoid the Idc and nulling circuitry, let the 0 to - 4 mA lives its life through a tube grid happy to see by nature TDA1541A negative offset (-2 mA (who learnt us that 😉 ... T. Loesch) .... then just reverse the polarity at the loudspeakers ! Choose a tube with good transconductance and Mu enough to choose up to the 50 R Riv happy limit and let the pre to add its gain ! We know the usual... well the "casual" suspects (E180EF, ECC88, 3DA with ferrites bids on the pins, 6072A, etc) ! 😊
Problem solved, no CCII, CFB op amps or OTA op amps in open loop anymore, or bad VFB !
Last edited:
If the trafos have been thought and build for an specific job there are non complications at all , sowter did it and still do it and some others for sure
and that way you really listen to the TDA analog capabilities , otherwise what you listen to are , your tubes , jfet , opamp , including their power supply , etc ....
some will say that what we listen to is the trafos , sure but this is what change/shape the signal the less
.
and that way you really listen to the TDA analog capabilities , otherwise what you listen to are , your tubes , jfet , opamp , including their power supply , etc ....
some will say that what we listen to is the trafos , sure but this is what change/shape the signal the less
.
Last edited:
The Sowter ones I use are found in some very high end commercial DACs, including, if my memory serves, an Audio Note one.
As for listening to the sound of the traffos - that comment could equally be applied to any I/V method whether solid state or tube, since their proponents all claim they sound the best.
Just a pity the Sowters are so expensive or they would have more endorsements on here.
As for listening to the sound of the traffos - that comment could equally be applied to any I/V method whether solid state or tube, since their proponents all claim they sound the best.
Just a pity the Sowters are so expensive or they would have more endorsements on here.
OT Q, when using trafo as I/V with step up ratios, how do you gents deal with the high output impedance,
surely the next stage has to be a follower/buffer of sorts right ?
surely the next stage has to be a follower/buffer of sorts right ?
In my case, I am using the preamp of a NAD 326BEE and the line inputs all have 100K input Z.
In the past I have just connected the traffs to a 50k pot and then to a power amp.
It's possible to use a resistor across the secondary to lower the primary Z and alter the sound.
In the past I have just connected the traffs to a 50k pot and then to a power amp.
It's possible to use a resistor across the secondary to lower the primary Z and alter the sound.
Thanks batteryman, I would suspect that trafo direct to a pot will not sound dynamic etc.
In anycase a high input impedance high gain stage would be needed regardless of it being a preamp etc.
In anycase a high input impedance high gain stage would be needed regardless of it being a preamp etc.
in my case ( sowters 1495 ) it is connected to my Pass F6 that has input buffer with Z at 100 k , same with the VFET2 😎
and 1.0 v rms is enough to drive them all
and yes In anycase a high input impedance "high" gain stage would be needed
.
Last edited:
Not so. Connected in series, the output is 10x the input which doesn't require further amplification any more than any CD or DAC does.
Read what Sowter say. Their FR chart is 100r input Z, into a 20k load.
https://www.sowter.co.uk/specs/1465.php
This is the honnest assertion I was waiting ! 🙂
Tubes can be very linear too; wider as well ? In the Adagio DAC, T Loesch used both traffo elevation and tubes.
@sumotan , in our quest this is not the output impedance of theI/V stage that is a problem, it is still the input impedance the DAC sees at the I/V input! One wants a low impedance to ground. The dac it self has an output impedance, capacitance as well (if you want to calcualte find the closed loop of an active I/V stage with an op amp for instance...
In that regard I am not sure tubes or traffos have a palm between each others, They still are ruled by the I/V resistor and the too much high impedance seen ... . Curent conveyor can solve that, then use any Riv value you need after.
Then for the stages that follow till the amp, impedance adaptation between two stage with a ratio of 1/10 is fine, at least my understanding. Btw, I always found passive pre with traffo to lack of something and soundstage.
Edit : I gave above to Thosrten a link about a good active I/V discrete on the paper (just lack the curent injection of 2 mA to adapt to the tda1541A) made by Sergio and measured -130 dB by Bohrok member (sligthy modified with Jfet if I remember).
I hope Thorsten says more when people have dealt with a pcb proto or even before as this story of 5V ref is a part of it
Last edited: