Yes I just want ed to point the same thing...
How LE pulse should be after stopped sys clk burst?
and how long it should be last 1 sys clk?
is it enough to be 4 pulses (2 word bits) to 16bit length, or larger silence?
And does the LR data bus should be splitted into LL and RR for literarry simultaneous or coud be one after another?
sorry for poluting the topic with Qs about different dac chip model...
Zoran, all this info is in the data sheet of TDA1540P and TDA1541A. PM me if I you still have any questions.
Its ok.
What I can say is that the tube is essentially a 2C51. I've used WE396A the same.
Not so high Gm, you'll be fine..
My Take - others will disagree.. DC heat the tube, schottky bridge CM 10R then 33kuF to LT1085A as CCS regulator.
Load the tube with a CCS try 5mA to 8mA, take output from 'mu' or upper resistor of lower ring of two transistors for Zout 1/Gm or there abouts - 2.2uF coupling cap to keep phase shift negligible at low frequencies and also keep output impedance (XC) low at the same.
Mu is like a 6922, 33 or so.. bias with tda1541A current through 39ohm, grid leak.. add 22ohm to each cathode.. use one triode per section D/A output left/right. No bypasses on cathode.
Add 220 carbon composite between I/V resistor and series grid. For good measure add 50R carbon composite between CCS plate load and plate.
You could bypass cathode resistors 22ohm with 0.001uF poly but with Gm being low enough, you'd be likely ok.
Thats a nice tube.. I'd swap for WE396A any day.
Use a tube rectifier with a heater for slow ramp up.. let A supply heaters come up first.
If you can, choke input (meeting critical inductance) to 100uF then LC split 10H to 55uF or such to each branch and you're laughing. Bleed 10mA across each final filter to help lower noise… and on one branch form a voltage divider to raise the heater (faux) centre tap to +35VDC or so, bypass with 2 to 10uF.
You wont look back..
.. Yeah ok you will. but it'll be fun.
Shane
What I can say is that the tube is essentially a 2C51. I've used WE396A the same.
Not so high Gm, you'll be fine..
My Take - others will disagree.. DC heat the tube, schottky bridge CM 10R then 33kuF to LT1085A as CCS regulator.
Load the tube with a CCS try 5mA to 8mA, take output from 'mu' or upper resistor of lower ring of two transistors for Zout 1/Gm or there abouts - 2.2uF coupling cap to keep phase shift negligible at low frequencies and also keep output impedance (XC) low at the same.
Mu is like a 6922, 33 or so.. bias with tda1541A current through 39ohm, grid leak.. add 22ohm to each cathode.. use one triode per section D/A output left/right. No bypasses on cathode.
Add 220 carbon composite between I/V resistor and series grid. For good measure add 50R carbon composite between CCS plate load and plate.
You could bypass cathode resistors 22ohm with 0.001uF poly but with Gm being low enough, you'd be likely ok.
Thats a nice tube.. I'd swap for WE396A any day.
Use a tube rectifier with a heater for slow ramp up.. let A supply heaters come up first.
If you can, choke input (meeting critical inductance) to 100uF then LC split 10H to 55uF or such to each branch and you're laughing. Bleed 10mA across each final filter to help lower noise… and on one branch form a voltage divider to raise the heater (faux) centre tap to +35VDC or so, bypass with 2 to 10uF.
You wont look back..
.. Yeah ok you will. but it'll be fun.
Shane
SD2-Player
Hi John,
could you please explain in a few words how to create the 3 supply voltages for the TDA1541A (+5V, -5V, -15V) from a single 24V supply.
Hi John,
could you please explain in a few words how to create the 3 supply voltages for the TDA1541A (+5V, -5V, -15V) from a single 24V supply.
Hi Goh,
As requested.
This one is for d3A but 6688 E180F drops right in (e280F also).
You could probably reduce cathode resistor to 22ohms, perhaps even 10 ohms.. easy enough to play around with. The main idea is using the -2mA current offset from the 1541A to help minimally bias the tube, credit to Thorsten for the idea. The removal of the CCS injection scheme (to null for 0V at the 1541A output) probably also had some positive impact.
No doubt there are better circuits out there.. but this one is quiet, has good drive and low output impedance and was fun to build ( I cant solder solid state bits very well)
Shane
As requested.
This one is for d3A but 6688 E180F drops right in (e280F also).
You could probably reduce cathode resistor to 22ohms, perhaps even 10 ohms.. easy enough to play around with. The main idea is using the -2mA current offset from the 1541A to help minimally bias the tube, credit to Thorsten for the idea. The removal of the CCS injection scheme (to null for 0V at the 1541A output) probably also had some positive impact.
No doubt there are better circuits out there.. but this one is quiet, has good drive and low output impedance and was fun to build ( I cant solder solid state bits very well)
Shane
Attachments
Last edited:
Hi JOSI1,
You need 3 shunt regulators, two 5V shunt regulators and one 10V shunt regulator.
Connect these as follows:
+24V -> current limiting resistor or CCS -> 5V shunt reg -> 5V shunt reg -> 10V shunt reg -> 0V.
+5V present between CCS / resistor and the plus of the 5V shunt reg #1.
GND present between the minus of 5V shunt reg #1 and the plus of 5V shunt reg #2.
-5V present between the minus of 5V shunt reg #2 and the plus of the 10V shunt reg.
-15V present between the minus of the 10V shunt reg and 0V of the 24V supply.
TDA1541A draws following current from the supply voltages:
+5V, 27 … 40mA
-5V, 37 … 50mA
-15V, 25 … 35mA.
So a bias current of 100mA should be sufficient and holds some reserve for powering other circuits as well.
You need 3 shunt regulators, two 5V shunt regulators and one 10V shunt regulator.
Connect these as follows:
+24V -> current limiting resistor or CCS -> 5V shunt reg -> 5V shunt reg -> 10V shunt reg -> 0V.
+5V present between CCS / resistor and the plus of the 5V shunt reg #1.
GND present between the minus of 5V shunt reg #1 and the plus of 5V shunt reg #2.
-5V present between the minus of 5V shunt reg #2 and the plus of the 10V shunt reg.
-15V present between the minus of the 10V shunt reg and 0V of the 24V supply.
TDA1541A draws following current from the supply voltages:
+5V, 27 … 40mA
-5V, 37 … 50mA
-15V, 25 … 35mA.
So a bias current of 100mA should be sufficient and holds some reserve for powering other circuits as well.
-ecdesigns-; The analogue circuit was next. I wanted to use passive I/V conversion with the highest possible output signal (while keeping distortion low). I ended up using a 140 Ohm I/V resistor (between TDA1541A output and GND). Without further measures this will result in massive distortion as the output signal clips on the lower side. This can be solved by injecting a bias current (on the DAC output pin). I used a 1 K Ohm resistor in series with a hybrid choke connected to +5V. This way one can squeeze around 550mVpp directly out of the TDA1541A while maintaining low distortion. Then it's quite easy to amplify this signal to a suitable level to drive a power amp. [/QUOTE said:
Where did you find that quote?
Which post #?
John is talking about bias injection to the analog output of the TDA1541a , not the DEM circuit you are referring to. (6K8 resistor is from DEM to +5V)
Koldby
Hi niamex,
TDA1541A output compliance (absolute maximum rating!) equals +25mV and -25mV.
Output current varies between (almost) 0 and -4mA (only goes negative).
Without signal output current equals -2mA.
When using a passive I/V resistor between output and GND without a +2mA bias current (to compensate for the -2mA bias current of the TDA1541A) the voltage can only go negative. Voltage swings above 25mVpp will now already exceed output compliance of -25mV. Absolute maximum value for I/V resistor would be 0.025 / 0.004 = 6.25 Ohms.
When using +2mA bias current, voltage swing is extended to +25mV and -25mV (instead of 0mV and -50mV) as signal swings around, not below 0V. Provided suitable CCS is used, this will always lead to lower distortion with given output signal swing. Absolute maximum value for I/V resistor now equals 0.05 / 0.004 = 12.5 Ohms.
Exceeding output compliance not only leads to increased distortion (non-linear response), it also leads to increased TDA1541A induced deterministic jitter.
TDA1541A performs best by far with (almost) zero ac on its outputs.
Leaving out the 14 filter caps on the TDA1541A will lead to clearly measurable switching spikes on the TDA1541A output. This may or may not be a problem, depending on desired performance.
When leaving out the filter caps, make sure f DEM is at least 4 times higher than fs (sample rate). So for 44.1 KHz NOS this translates to 176.4 KHz f DEM and for 44.1 KHz with 4 times oversampling (SAA7220) this translates to 705.6 KHz f DEM.
TDA1541A output compliance (absolute maximum rating!) equals +25mV and -25mV.
Output current varies between (almost) 0 and -4mA (only goes negative).
Without signal output current equals -2mA.
When using a passive I/V resistor between output and GND without a +2mA bias current (to compensate for the -2mA bias current of the TDA1541A) the voltage can only go negative. Voltage swings above 25mVpp will now already exceed output compliance of -25mV. Absolute maximum value for I/V resistor would be 0.025 / 0.004 = 6.25 Ohms.
When using +2mA bias current, voltage swing is extended to +25mV and -25mV (instead of 0mV and -50mV) as signal swings around, not below 0V. Provided suitable CCS is used, this will always lead to lower distortion with given output signal swing. Absolute maximum value for I/V resistor now equals 0.05 / 0.004 = 12.5 Ohms.
Exceeding output compliance not only leads to increased distortion (non-linear response), it also leads to increased TDA1541A induced deterministic jitter.
TDA1541A performs best by far with (almost) zero ac on its outputs.
Leaving out the 14 filter caps on the TDA1541A will lead to clearly measurable switching spikes on the TDA1541A output. This may or may not be a problem, depending on desired performance.
When leaving out the filter caps, make sure f DEM is at least 4 times higher than fs (sample rate). So for 44.1 KHz NOS this translates to 176.4 KHz f DEM and for 44.1 KHz with 4 times oversampling (SAA7220) this translates to 705.6 KHz f DEM.