Hi Terry Demol,
TDA154x chips have constant current sink outputs, constant current sources have high output impedance.
TDA1543 has an additional bias current source that feeds an additional bias current to the DAC outputs. This bias current can be varied by the current drawn from vref to GND. The TDA1543 is designed to function on a single power supply and has more relaxed DC output compliance. By varying the TDA1543 bias current, power supply noise will be modulated on the bias current, this in turn might degrade performance.
TDA1541A only has current sink outputs, and it's designed to tolerate DC voltages that are equal to DAC AC output compliance.
Both DAC chips also have specified ac output compliance, in other words the higher the AC voltage on the output, the higher non-linear distortion. In order to meet datasheet THD specs, AC output voltage MUST remain below specified value.
Specified DC / AC output compliance:
TDA1541A datasheet, page 7, AC output DC and AC) compliance +/-25mV.
TDA1543 datasheet, page 7, DC output compliance min. 1.8V, max. Vdd-1.2V, AC output compliance +/-25mV.
So passive I/V conversion that results in exceeding max. output compliance WILL result in higher distortion than specified in the datasheets, and this is clearly audible. The distortion is caused by the transistor / diode bit switches characteristics used at the current outputs. The introduced distortion is non-linear.
Using the trans-impedance converter plus correction bias current sink from DAC output to GND will allow the ac voltage on the DAC output to be trimmed down well below 25mVpp (ac voltage across the DAC output decreases with increased correction current). The trans-impedance circuit remains stable over a very large bandwidth, so it remains functioning correctly when confronted with the DAC chip large bandwidth output signal. In practice simple circuits usually provide best perceived sound quality.
In order to achieve lowest possible noise and bit clock jitter, the I2S signal amplitude at the DAC input MUST be as low as possible. In practice both chips that are based on current-steering logic, function reliably with 400mVpp I2S signal amplitude.
This greatly reduces ground-bounce, and lowers bit clock jitter inside the DAC chip at the clock input of the output latches where it really matters.
This basically applies to all external signals used in a DAC, try to minimize digital signal amplitude while ensuring reliable operation. Lowering signal amplitudes, results in much cleaner sound. It's also best to use correct impedance matching on all digital circuits. Directly connecting a TTL output to a TTL input doesn't provide optimal performance.
When connecting a DAC to external equipment (pre-amplifier, volume control, power amplifier, and digital audio source). Ground loops are created between ALL devices. Now (RF) current paths are created that super-impose (mains) interference on the GND connections and power supplies. The DAC is most sensitive to the interference caused by these ground loops.
In practice, one can run the DAC on a battery, main improvement will now be made by removing one ground loop, rather than the lower noise produced by the battery. But when the DAC source has galvanic connection to the source (USB, coax), there is still little advantage. The loop now runs from the source, through USB or Coax through the DAC electronics, back to the mains. The loops from both pre and power amplifier also pass the DAC electronics and return through the DAC power supply.
The D1M prototype now runs on a floating power supply (post #2579) and Toslink. This reduces ground loop problems and greatly improves sound quality. One can't imagine how much these ground loops can affect sound quality.
TDA154x chips have constant current sink outputs, constant current sources have high output impedance.
TDA1543 has an additional bias current source that feeds an additional bias current to the DAC outputs. This bias current can be varied by the current drawn from vref to GND. The TDA1543 is designed to function on a single power supply and has more relaxed DC output compliance. By varying the TDA1543 bias current, power supply noise will be modulated on the bias current, this in turn might degrade performance.
TDA1541A only has current sink outputs, and it's designed to tolerate DC voltages that are equal to DAC AC output compliance.
Both DAC chips also have specified ac output compliance, in other words the higher the AC voltage on the output, the higher non-linear distortion. In order to meet datasheet THD specs, AC output voltage MUST remain below specified value.
Specified DC / AC output compliance:
TDA1541A datasheet, page 7, AC output DC and AC) compliance +/-25mV.
TDA1543 datasheet, page 7, DC output compliance min. 1.8V, max. Vdd-1.2V, AC output compliance +/-25mV.
So passive I/V conversion that results in exceeding max. output compliance WILL result in higher distortion than specified in the datasheets, and this is clearly audible. The distortion is caused by the transistor / diode bit switches characteristics used at the current outputs. The introduced distortion is non-linear.
Using the trans-impedance converter plus correction bias current sink from DAC output to GND will allow the ac voltage on the DAC output to be trimmed down well below 25mVpp (ac voltage across the DAC output decreases with increased correction current). The trans-impedance circuit remains stable over a very large bandwidth, so it remains functioning correctly when confronted with the DAC chip large bandwidth output signal. In practice simple circuits usually provide best perceived sound quality.
In order to achieve lowest possible noise and bit clock jitter, the I2S signal amplitude at the DAC input MUST be as low as possible. In practice both chips that are based on current-steering logic, function reliably with 400mVpp I2S signal amplitude.
This greatly reduces ground-bounce, and lowers bit clock jitter inside the DAC chip at the clock input of the output latches where it really matters.
This basically applies to all external signals used in a DAC, try to minimize digital signal amplitude while ensuring reliable operation. Lowering signal amplitudes, results in much cleaner sound. It's also best to use correct impedance matching on all digital circuits. Directly connecting a TTL output to a TTL input doesn't provide optimal performance.
When connecting a DAC to external equipment (pre-amplifier, volume control, power amplifier, and digital audio source). Ground loops are created between ALL devices. Now (RF) current paths are created that super-impose (mains) interference on the GND connections and power supplies. The DAC is most sensitive to the interference caused by these ground loops.
In practice, one can run the DAC on a battery, main improvement will now be made by removing one ground loop, rather than the lower noise produced by the battery. But when the DAC source has galvanic connection to the source (USB, coax), there is still little advantage. The loop now runs from the source, through USB or Coax through the DAC electronics, back to the mains. The loops from both pre and power amplifier also pass the DAC electronics and return through the DAC power supply.
The D1M prototype now runs on a floating power supply (post #2579) and Toslink. This reduces ground loop problems and greatly improves sound quality. One can't imagine how much these ground loops can affect sound quality.
The D1M prototype now runs on a floating power supply (post #2579) and Toslink. This reduces ground loop problems and greatly improves sound quality. One can't imagine how much these ground loops can affect sound quality. [/B][/QUOTE]
Hi John,
Very interesting discussion, as always.
You have discarded the capacitance multiplier, why? Not needed, to be added later or what?
Thank you
Hi John,
Very interesting discussion, as always.
You have discarded the capacitance multiplier, why? Not needed, to be added later or what?
Thank you
Hi Jean-Charles,
The schematic was just a concept for testing, N1, and N2 (220uH) should be left-out as they cause ringing.
The capacitor multiplier isn't discarded, in fact, I installed it today (one BD679, 3K3 and 1000uF). Ripple voltage at the floating CT power supply is now around 1mVpp, DC output voltage around 12V with connected load.
The schematic was just a concept for testing, N1, and N2 (220uH) should be left-out as they cause ringing.
The capacitor multiplier isn't discarded, in fact, I installed it today (one BD679, 3K3 and 1000uF). Ripple voltage at the floating CT power supply is now around 1mVpp, DC output voltage around 12V with connected load.
John,
It seems that reoccurring questions are mostly about circuit drawings and how things (modules) are tied together.
I also understand you are not done and things keep changing. So it becomes kind of jig saw puzzle
Would it be an idea, you publish a circuit with everything in it as it is today, give it a revision number and just update whatever is to be updated and increase the revision number?
That would keep the thread readers in sync on what you are actually doing
It seems that reoccurring questions are mostly about circuit drawings and how things (modules) are tied together.
I also understand you are not done and things keep changing. So it becomes kind of jig saw puzzle
Would it be an idea, you publish a circuit with everything in it as it is today, give it a revision number and just update whatever is to be updated and increase the revision number?
That would keep the thread readers in sync on what you are actually doing
Hi ccschua,
I measured the output signal on the emitter of the darlingtons in the floating CT supply with an oscilloscope, after measuring slight interference at the series regulator output in the D1M. The voltage across the power-shunts was clean.
The floating CT power supply first drives a capacitance multiplier. Then it's routed to the D1M and enters a 9.6V discrete series regulator.
The series regulator feeds the connected loads through 30 ... 40 mH and 47,000 ... 100,000uF LC filters, the voltage at the loads is kept constant using power-shunt regulators, placed as close as possible to the loads.
The master clock / tracker is fed by the 9.6 volts series regulator, a precision constant current source, a 40mH choke, and finally a 5V power-shunt regulator.
D1M power distribution is as follows:
9V6 series reg > precision constant current source > LC filter > 5V shunt reg. > multi-rate master clock tracker / clock buffer.
9V6 series reg > LC filter > 4V2 shunt reg. > SPDIF cleaner (CS8416), synchronous SPDIF reclocker.
9V6 series reg > LC filter > 4V2 shunt reg. > SPDIF receiver (CS8416), WS generator (counter).
9V6 series reg > LC filter > 3V1 shunt reg. > DAC chip (1xTDA1543), I2S attenuators
9V6 series reg > LC filter > 8V9 shunt reg. > trans-impedance converters / buffers.
I measured the output signal on the emitter of the darlingtons in the floating CT supply with an oscilloscope, after measuring slight interference at the series regulator output in the D1M. The voltage across the power-shunts was clean.
The floating CT power supply first drives a capacitance multiplier. Then it's routed to the D1M and enters a 9.6V discrete series regulator.
The series regulator feeds the connected loads through 30 ... 40 mH and 47,000 ... 100,000uF LC filters, the voltage at the loads is kept constant using power-shunt regulators, placed as close as possible to the loads.
The master clock / tracker is fed by the 9.6 volts series regulator, a precision constant current source, a 40mH choke, and finally a 5V power-shunt regulator.
D1M power distribution is as follows:
9V6 series reg > precision constant current source > LC filter > 5V shunt reg. > multi-rate master clock tracker / clock buffer.
9V6 series reg > LC filter > 4V2 shunt reg. > SPDIF cleaner (CS8416), synchronous SPDIF reclocker.
9V6 series reg > LC filter > 4V2 shunt reg. > SPDIF receiver (CS8416), WS generator (counter).
9V6 series reg > LC filter > 3V1 shunt reg. > DAC chip (1xTDA1543), I2S attenuators
9V6 series reg > LC filter > 8V9 shunt reg. > trans-impedance converters / buffers.
I connected the transimpedance circuit single ended as below
I use R1 = 1K5, R2=2K7, R3 = 510R and U1 = 1N4148. T1 and T2 is C2240. C1=1200uF panasonic FC, Vin = 12.07 V DC.
The DC voltage is shown in RED. When IDAC is left open, I found 0.3V at C2 output. Connecting IDAC to my TDA1541 IoutR give the same. Wonder what is wrong here.
Hi ccschua,
T1 is not conducting, I suggest checking both T1 and T2 connections.
When viewing the 2SC2240 from the top with the rounded side facing to the right. Upper pin is emitter, center pin is collector, lower pin is basis.
Andypairo is correct. TDA1541A bias current equals 2mA when the code for bipolar zero has been latched in the TDA1541A output latches. This produces .002 * 1500 = 3 volts across R1, so the voltage at the collector of T1 should be around 12 - 3 = 9 volts DC. The voltage at the emitter of T2 should be around 9 - 0.6 = 8.4 Volts.
T1 is not conducting, I suggest checking both T1 and T2 connections.
When viewing the 2SC2240 from the top with the rounded side facing to the right. Upper pin is emitter, center pin is collector, lower pin is basis.
Andypairo is correct. TDA1541A bias current equals 2mA when the code for bipolar zero has been latched in the TDA1541A output latches. This produces .002 * 1500 = 3 volts across R1, so the voltage at the collector of T1 should be around 12 - 3 = 9 volts DC. The voltage at the emitter of T2 should be around 9 - 0.6 = 8.4 Volts.
Dear EC,
I am trying find a good use to my other DI4
I suspect that if I connect each DAC output (NIR, NIL, IR, IL) to its corresponding trans impedance/buffer circuit I could end with a balanced output NOS DAC, and, as I use a TVC as preamp, I could get rid of the output caps, since all the trans impedance circuits will be fed by the same power supply, so I hope DC output will cancel out on the transformer based preamp...am I right?
(using matched parts and transistors, that is...)
Cheers,
M
I am trying find a good use to my other DI4
I suspect that if I connect each DAC output (NIR, NIL, IR, IL) to its corresponding trans impedance/buffer circuit I could end with a balanced output NOS DAC, and, as I use a TVC as preamp, I could get rid of the output caps, since all the trans impedance circuits will be fed by the same power supply, so I hope DC output will cancel out on the transformer based preamp...am I right?
(using matched parts and transistors, that is...)
Cheers,
M