-ecdesigns- said:
No, - 40 or -50 dB harmonics are very tiny deviations from the ideal sine curve and not distinguishable on a scope.
For FFT, any soundcard with external preamp and a freeware analyzer program would do it.
I still use a nostalgic but very versatile programmable 16 bit standalone analyzer.
Hi -EC-
Sorry to insist on this topic but, wouldn't it be great to have a volume control at the DAC or at the SD card player?
(variable I/V resistor)
There are little digipots with low power consumption, good channel matching and low tempco like these, for example:
http://www.analog.com/en/digital-to...l-potentiometers/ad5172/products/product.html
http://www.analog.com/en/digital-to...tiometers/adn2860/products/product.html#specs
I must re-view this presentation:
http://event.on24.com/event/12/97/41/rt/1/documents/slidepdf/digipots_jan09_slides__final1.pdf
Regards,
M
Sorry to insist on this topic but, wouldn't it be great to have a volume control at the DAC or at the SD card player?
(variable I/V resistor)
There are little digipots with low power consumption, good channel matching and low tempco like these, for example:
http://www.analog.com/en/digital-to...l-potentiometers/ad5172/products/product.html
http://www.analog.com/en/digital-to...tiometers/adn2860/products/product.html#specs
I must re-view this presentation:
http://event.on24.com/event/12/97/41/rt/1/documents/slidepdf/digipots_jan09_slides__final1.pdf
Regards,
M
dddac said:
You are right that things in real life can differ from paper spec, and we should be more humble to data than specs. Thanks for the FFT charts.
Nevertheless with respect to voltage compliance, I am inclined to believe that DC voltage compliance might be much more critical than AC voltage compliance. Even though I do not know the internal circuit topology of the 1543, but I can imagine that the DC compliance is largely related to the current source biasing components characteristics. Frankly 1.8V (3x0.6V) above GND and 1.2V (2x0.6V) below VDD specified for 1543 drive me to think about a number of forward biased PN junctions inside the chip. With this in mind, I believe it is not a good practice to run out of the DC voltage compliance range.
On AC compliance, any simple resistor I/V conversion will drive us out of spec, as only +/-25mV is specified. We are talking about 1V rms easily. But based on all the experience shown in these thread, it is quite clear that this violation seems to be not of a concern. Might be AC compliance spec is drawn up based on an assumption that the DAC will see a virtual ground and it will only see a small voltage swing, hence the engineer did not put much real effort in finding out the true limits??
-ecdesigns- said:
Thanks EC for the answer and the effort to plot the oscilloscope out, yes it is clear that your signal swing somewhere between 1.8V and 3.5V.
However, I would rather do the calculation the other way round:
I think you are biasing the load resistor to about 3.3-3.5V? The way you mount this R to the DAC, coupled to the fact that the power supply at 3.3-3.5V can only source current, create a ceiling for the MAX voltage that can be output by the DAC --> hence the 3.5V in your signal chart. With 23mA flowing thru 680R, you got a voltage swing of 1.56Vpp, so the MIN voltage will be just about 1.8-2V, also fall inline with your signal chart.
I think if we just put 2 alkaline batteries in series to give a 3V bias, the R load cannot exceed 520 ohm (1.2V/23mA) in order to stay within the DC compliance range. If using the LED voltage regulator you mentioned, then with 3.3-3.5V, your 680R works perfectly. This is the best you can get out with a 5V power supply. If we push up the VDD to 6V, this will allow us to use 3 alkaline batteries in series at 4.5V (1.5V below VDD conforming to DC compliance), and a possible swing of 2.7Vpp (4.5 - 1.8) with a load resistor about 1.2k Ohm.
I wil try to experiement with this. And again, thanks for this innovative idea.
hi,
i have little problem with my tda1541 dac.
everytime i insert CD to cdp and press play.. it always has "ppsstt..." for like 1 second or 2.
seems like it can't lock the signal properly when i press play..
it use CS8412CP and i using Spdif input (no optical) and 75R resistor input. PLL filter 500R+0.22uf bypassed with 3300pf.
if someone know how to fix it pls let me know.
thanks
i have little problem with my tda1541 dac.
everytime i insert CD to cdp and press play.. it always has "ppsstt..." for like 1 second or 2.
seems like it can't lock the signal properly when i press play..
it use CS8412CP and i using Spdif input (no optical) and 75R resistor input. PLL filter 500R+0.22uf bypassed with 3300pf.
if someone know how to fix it pls let me know.
thanks
Hi 2A3SET,
Yes and no, I can use both, as the SD-player runs both on a mains adapter or battery power supply.
I attached part of the SD-player schematics that shows the power supply.
The DC voltage enters a discrete low-noise LED-referenced series regulator with integrated capacitance multiplier built around L1, T2, and T3. This series regulator concept also works very well for TDA1541A power supplies (+5V, -5V, and -15V)
It then feeds LED shunt regulator (L2 and L3) through a passive filter. The voltage of this shunt regulator is then filtered again and drives a second LED shunt regulator (L8 and L9). The voltage on the anode of L8 is then used as 3.3V reference voltage. I left out decoupling caps as these color the sound.
Both master clock and high-speed synchronous reclocker are powered using a capacitance multiplier (T1), filters, constant current sources (T4 and T5), followed by some more filters, then driving a 4V LED shunt regulator. Both constant current source and LED shunt regulator form a constant power regulator.
The LM326 is unsuitable for use as external reference voltage as output noise is too high, and it's very difficult to keep integrated circuits of this complexity under control (stability), resulting in sound coloring / degradation.
Yes and no, I can use both, as the SD-player runs both on a mains adapter or battery power supply.
I attached part of the SD-player schematics that shows the power supply.
The DC voltage enters a discrete low-noise LED-referenced series regulator with integrated capacitance multiplier built around L1, T2, and T3. This series regulator concept also works very well for TDA1541A power supplies (+5V, -5V, and -15V)
It then feeds LED shunt regulator (L2 and L3) through a passive filter. The voltage of this shunt regulator is then filtered again and drives a second LED shunt regulator (L8 and L9). The voltage on the anode of L8 is then used as 3.3V reference voltage. I left out decoupling caps as these color the sound.
Both master clock and high-speed synchronous reclocker are powered using a capacitance multiplier (T1), filters, constant current sources (T4 and T5), followed by some more filters, then driving a 4V LED shunt regulator. Both constant current source and LED shunt regulator form a constant power regulator.
The LM326 is unsuitable for use as external reference voltage as output noise is too high, and it's very difficult to keep integrated circuits of this complexity under control (stability), resulting in sound coloring / degradation.
Attachments
Here are the schematics of +5V, -5V and -15V LED-referenced series regulators with integrated capacitance multiplier for the TDA1541A. Exact output voltage can be set with R3, R8, and R12 (higher value increases output voltage, lower value decreases it).
The circuit works as follows (using +5V schematic as reference):
R1 and C1 form a low-pass filter that attenuates ripple voltage. R2 feeds the cleaned-up DC voltage to Darlington T1, driving it into saturation. C2 provides frequency compensation.
The output voltage is divided by R3 and R4, and connected to the base of T2. T2 emitter is connected to a green LED (approx. 2V).
As soon as the voltage across R4 exceeds 2V + 0.6V (V-BE), T2 starts conducting, reducing the current flow through T1. This way the regulator settles around the required output voltage.
If R4 was removed from the circuit, output voltage would settle around V LED (2V) + V-BE (0.6V) = 2.6V.
The circuit responds very fast (minimum delays), and noise levels are low enough for audio applications.
The LED could be replaced by suitable low-noise zener diodes, but do carefully check effects on sound quality.
Keep in mind that the noise produced by the reference diode is amplified by the attenuation factor of the voltage divider. With the 5V regulator, the LED noise is amplified approx. 1.9 times.
With the -15V regulator LED noise is amplified by approx. 2.7 times.
So try to find an optimum between reference diode noise and voltage divider properties.
The voltage drop across LEDs with different color are slightly different:
Current = 20mA.
Infra-Red LED, approx. 1.3V
Red / Orange LED, approx. 1.9V
Yellow LED, approx. 2V
Green LED, approx. 2.1V
Use plain LEDs (no low current LEDs) for lowest noise. Don't use blue or white LEDs.
I used green LEDs for these voltage regulators as they provide highest voltage drop (2V).
The circuit works as follows (using +5V schematic as reference):
R1 and C1 form a low-pass filter that attenuates ripple voltage. R2 feeds the cleaned-up DC voltage to Darlington T1, driving it into saturation. C2 provides frequency compensation.
The output voltage is divided by R3 and R4, and connected to the base of T2. T2 emitter is connected to a green LED (approx. 2V).
As soon as the voltage across R4 exceeds 2V + 0.6V (V-BE), T2 starts conducting, reducing the current flow through T1. This way the regulator settles around the required output voltage.
If R4 was removed from the circuit, output voltage would settle around V LED (2V) + V-BE (0.6V) = 2.6V.
The circuit responds very fast (minimum delays), and noise levels are low enough for audio applications.
The LED could be replaced by suitable low-noise zener diodes, but do carefully check effects on sound quality.
Keep in mind that the noise produced by the reference diode is amplified by the attenuation factor of the voltage divider. With the 5V regulator, the LED noise is amplified approx. 1.9 times.
With the -15V regulator LED noise is amplified by approx. 2.7 times.
So try to find an optimum between reference diode noise and voltage divider properties.
The voltage drop across LEDs with different color are slightly different:
Current = 20mA.
Infra-Red LED, approx. 1.3V
Red / Orange LED, approx. 1.9V
Yellow LED, approx. 2V
Green LED, approx. 2.1V
Use plain LEDs (no low current LEDs) for lowest noise. Don't use blue or white LEDs.
I used green LEDs for these voltage regulators as they provide highest voltage drop (2V).
Attachments
Hi maxlorenz,
The problem with the volume control (poor sound quality at low volume settings) seems to be caused by limited power amplifier resolution.
Depending on power amplifier design it will have higher or lower resolution (ability of the power amplifier to accurately and smoothly track the amplified input voltage fluctuations produced by complex signals).
Most power amplifiers (especially semiconductor-based amplifiers) provide only very limited resolution.
With a given limited amplifier resolution, best use of this resolution is made when selecting highest possible output amplitude (higher volume setting). With lower volume settings, only a fraction of the given amplifier resolution is used, and sound quality degrades accordingly.
That's why the power attenuator experiment provided such good results, but it also demonstrated that my power amps had too low resolution.
Easy way to check your power amplifier, turn the volume down, if sound quality degrades (less transparency and detail), the power amplifier hasn't got sufficient resolution.
Practical example is a digital 16-bit volume control used with a 16 bit source. When set at maximum, digital audio data passes with full resolution (power amplifier outputs maximum amplitude), but as the digital volume control is turned down (power amplifier amplitude is reduced), resolution gradually drops.
Now there are some possible options:
1) Run the power amplifier with limited resolution at maximum signal level and use suitable power attenuator between amplifier and speakers.
2) Use volume controls with special properties (extreme low noise, or TVC) to patch-up the power amplifier limitations.
3) Design a power amplifier with high resolution so it offers sufficient resolution at low volume settings.
Practical example of a high resolution power amplifier is a 32 bit digital volume control used with a 16 bit source, even at lower volume settings, resolution is still sufficient.
The DC-coupled balanced bridge amplifier is my first attempt, constructing a high resolution power amplifier. I can now use a motorized 10K ALPS stereo potentiometer and still have plenty of resolution left at low (practical) volume settings.
Digipots are unsuitable for audiophile applications, the problem is caused by the (multiple) FETs in the signal path, poor quality (low wattage / high noise) of the integrated resistors, and noise / interference injected in the audio signal by the Digipot power supply.
Good quality volume controls could consist of stepped volume regulators using high wattage non-inductive (wire wound) resistors, or high wattage non-inductive (wire wound) potentiometers.
The problem with the volume control (poor sound quality at low volume settings) seems to be caused by limited power amplifier resolution.
Depending on power amplifier design it will have higher or lower resolution (ability of the power amplifier to accurately and smoothly track the amplified input voltage fluctuations produced by complex signals).
Most power amplifiers (especially semiconductor-based amplifiers) provide only very limited resolution.
With a given limited amplifier resolution, best use of this resolution is made when selecting highest possible output amplitude (higher volume setting). With lower volume settings, only a fraction of the given amplifier resolution is used, and sound quality degrades accordingly.
That's why the power attenuator experiment provided such good results, but it also demonstrated that my power amps had too low resolution.
Easy way to check your power amplifier, turn the volume down, if sound quality degrades (less transparency and detail), the power amplifier hasn't got sufficient resolution.
Practical example is a digital 16-bit volume control used with a 16 bit source. When set at maximum, digital audio data passes with full resolution (power amplifier outputs maximum amplitude), but as the digital volume control is turned down (power amplifier amplitude is reduced), resolution gradually drops.
Now there are some possible options:
1) Run the power amplifier with limited resolution at maximum signal level and use suitable power attenuator between amplifier and speakers.
2) Use volume controls with special properties (extreme low noise, or TVC) to patch-up the power amplifier limitations.
3) Design a power amplifier with high resolution so it offers sufficient resolution at low volume settings.
Practical example of a high resolution power amplifier is a 32 bit digital volume control used with a 16 bit source, even at lower volume settings, resolution is still sufficient.
The DC-coupled balanced bridge amplifier is my first attempt, constructing a high resolution power amplifier. I can now use a motorized 10K ALPS stereo potentiometer and still have plenty of resolution left at low (practical) volume settings.
Digipots are unsuitable for audiophile applications, the problem is caused by the (multiple) FETs in the signal path, poor quality (low wattage / high noise) of the integrated resistors, and noise / interference injected in the audio signal by the Digipot power supply.
Good quality volume controls could consist of stepped volume regulators using high wattage non-inductive (wire wound) resistors, or high wattage non-inductive (wire wound) potentiometers.
Hi maxlorenz, you should try the lightspeed
It's far more better than regular pot, and some people even prefer it over TVC.
http://www.diyaudio.com/forums/showthread.php?threadid=80194>
Mine is using digital POT to control the current over LDR (VCCS) and it's very easy to add a remote with balance control.
It's far more better than regular pot, and some people even prefer it over TVC.
http://www.diyaudio.com/forums/showthread.php?threadid=80194>
Mine is using digital POT to control the current over LDR (VCCS) and it's very easy to add a remote with balance control.
Attachments
Hi, I used 2 D cell batteries for the 3V ref. I kept the dac on for a long time and the voltage now drops to 2.89V.
I used 750R for I/V and I measure the I/V output voltage to be 2.06.
The tda1543 Vdd is 5.5V
I just wonder if a lower value I/V resistor would be better.
BTW, thanks EC.
The tda1543 /w EC tweak and 2sk170 output stage mate with the deq2496 sounds shockingly good. Much better my other non-os 1543 dac and a tda1541 dac I once had.
I believe the dsp in the deq2496 has already done some filtering to the digital signal so I am not sure if this dac is non-os.
Also, the output stage is extremely important, just passive I/V is not enough. Adding the 2sk170 output stage with 2+ gain really improve the overall sound.
I used 750R for I/V and I measure the I/V output voltage to be 2.06.
The tda1543 Vdd is 5.5V
I just wonder if a lower value I/V resistor would be better.
BTW, thanks EC.
The tda1543 /w EC tweak and 2sk170 output stage mate with the deq2496 sounds shockingly good. Much better my other non-os 1543 dac and a tda1541 dac I once had.
I believe the dsp in the deq2496 has already done some filtering to the digital signal so I am not sure if this dac is non-os.
Also, the output stage is extremely important, just passive I/V is not enough. Adding the 2sk170 output stage with 2+ gain really improve the overall sound.
FYC said:
Attachments
Dynamic jitter attenuator.
Thanks EC for your detailed and informative reply about volume control.
Thanks 2A3SET for you tip. I recently saw that ligthspeed attenuator. It sounds interesting and can be driven by 5V PS.
Dear EC, yesterday I made my first attempt with the dynamic jitter attenuator for my D1M. BCK is taken from PCM2707. I put a trimer resistor parallel with 220nf, in series with 150R to ground as per your diagram. The trimer set at 283R (rheostat mode) gave 1,22V DC for BCK. Ripple measured between 150R and 283R (trimer) and ground was around 10mV, even with 230nf or 320nf...is it good enough? Shall I try lower capacitances?
Today I will try to replace the trimer resistor by a better quality fixed R.
Another question> could you advice what tweak would be better for the BCK signals of the multiple DAC chips (DI4 and scrambled versions) converters-> conventional signal attenuator or dynamic jitter attenuator???
Many thanks,
M
PS> the 220R series BCK resistor is place AFTER the bat 85 (in my case) and after the dynamic jitter attenuator. Must I change it?
Thanks EC for your detailed and informative reply about volume control.
Thanks 2A3SET for you tip. I recently saw that ligthspeed attenuator. It sounds interesting and can be driven by 5V PS.
Dear EC, yesterday I made my first attempt with the dynamic jitter attenuator for my D1M. BCK is taken from PCM2707. I put a trimer resistor parallel with 220nf, in series with 150R to ground as per your diagram. The trimer set at 283R (rheostat mode) gave 1,22V DC for BCK. Ripple measured between 150R and 283R (trimer) and ground was around 10mV, even with 230nf or 320nf...is it good enough? Shall I try lower capacitances?
Today I will try to replace the trimer resistor by a better quality fixed R.
Another question> could you advice what tweak would be better for the BCK signals of the multiple DAC chips (DI4 and scrambled versions) converters-> conventional signal attenuator or dynamic jitter attenuator???
Many thanks,
M
PS> the 220R series BCK resistor is place AFTER the bat 85 (in my case) and after the dynamic jitter attenuator. Must I change it?
Hi 2A3SET,
I did built several versions of these lightspeed pots, mainly using matched LED / LDR couplers from Farnell (black cylindrical housing with 2 wires on each side), bought 50 or so for matching.
I also tested home made LED / LDR couplers, even with multiple LDR's and LEDs and experimented with different LED colors (LDR spectral sensitivity).
Although the lightspeed volume control didn't sound bad, closer examination exposed non-linear characteristic (LDR) that couldn't be corrected. I would feed in a linear saw tooth signal, and I would get out a distorted non-linear sawtooth signal.
This non-linear distortion may add some nice touch to the sound, but it's not transparent (neutral).
I achieved best (most transparent) sound quality so far, by using binary logarithmic attenuators (64 steps) with bulk metal foil resistors.
I attached a photograph of this volume control, it's still my reference.
I did built several versions of these lightspeed pots, mainly using matched LED / LDR couplers from Farnell (black cylindrical housing with 2 wires on each side), bought 50 or so for matching.
I also tested home made LED / LDR couplers, even with multiple LDR's and LEDs and experimented with different LED colors (LDR spectral sensitivity).
Although the lightspeed volume control didn't sound bad, closer examination exposed non-linear characteristic (LDR) that couldn't be corrected. I would feed in a linear saw tooth signal, and I would get out a distorted non-linear sawtooth signal.
This non-linear distortion may add some nice touch to the sound, but it's not transparent (neutral).
I achieved best (most transparent) sound quality so far, by using binary logarithmic attenuators (64 steps) with bulk metal foil resistors.
I attached a photograph of this volume control, it's still my reference.
Attachments
That looks very interesting, john.
If it isn't a restricted design, would it be possible to see the cct and the pcb layout?
I once built something along similar lines from Twisted Pear (very compact unit), but the sound didn't suit particularly well (the resistors, I understand) but it still works okay.
If it isn't a restricted design, would it be possible to see the cct and the pcb layout?
I once built something along similar lines from Twisted Pear (very compact unit), but the sound didn't suit particularly well (the resistors, I understand) but it still works okay.
Hi jameshillj,
Yes I could post schematics and PCB lay-out. The reed relays are controlled by a Microchip controller (PIC16F876) that also ensures minimum clicks / pops. It can also drive a RGB LED for volume indication.
The unit also has remote control (built around PIC16F84).
Resistors need to have low noise, so either bulk metal film, non-inductive wire wound (Mills), or high wattage metal film.
Relays also have effect on sound quality (permanent magnetic field from the coil, interference injected by the coil, and relay contact properties).
These would be nice relays, DPDT, hermetically sealed and 5,000,000 switching cycles.
http://nl.farnell.com/teledyne/172-12/relay-dpdt-12vdc/dp/9906908
If you only need volume control for the DAC (no other audio equipment connected to your set), I might have a better solution.
Put variable I/V resistors in the power amp, close to the input circuit. The DAC output current is fed to the I/V resistor in the power amp (current loop). This should eliminate impedance matching problems, and minimize interference as the current is translated into a voltage very close to the power amp inputs.
Yes I could post schematics and PCB lay-out. The reed relays are controlled by a Microchip controller (PIC16F876) that also ensures minimum clicks / pops. It can also drive a RGB LED for volume indication.
The unit also has remote control (built around PIC16F84).
Resistors need to have low noise, so either bulk metal film, non-inductive wire wound (Mills), or high wattage metal film.
Relays also have effect on sound quality (permanent magnetic field from the coil, interference injected by the coil, and relay contact properties).
These would be nice relays, DPDT, hermetically sealed and 5,000,000 switching cycles.
http://nl.farnell.com/teledyne/172-12/relay-dpdt-12vdc/dp/9906908
If you only need volume control for the DAC (no other audio equipment connected to your set), I might have a better solution.
Put variable I/V resistors in the power amp, close to the input circuit. The DAC output current is fed to the I/V resistor in the power amp (current loop). This should eliminate impedance matching problems, and minimize interference as the current is translated into a voltage very close to the power amp inputs.
Yes John, that Centigrid series is good gear alright - not cheap!
Interesting idea to run the Dac o/p current down the IC cable to an I/V at the amp - next thing would be to build the dac into the amp itself.
The next stage would be to build a current input stage and not convert it to voltage at all, then go on with the idea and build a straight out high o/p Z current gain amp and be done with it!
What is required to program those PIC chips? I have always wanted a mute button on a remote vol control (like the TV remotes) and the RGB volume indicator is novel (navel!)
Interesting idea to run the Dac o/p current down the IC cable to an I/V at the amp - next thing would be to build the dac into the amp itself.
The next stage would be to build a current input stage and not convert it to voltage at all, then go on with the idea and build a straight out high o/p Z current gain amp and be done with it!
What is required to program those PIC chips? I have always wanted a mute button on a remote vol control (like the TV remotes) and the RGB volume indicator is novel (navel!)
I have built a transimpedance circuit using 2sk170BL as common gate (with source resistance). The Vcc is +- 18 V and the output is a common source complimentary pair followed by a coupling cap to trim the DC offset.
Would it be better to replace coupling cap with DC servo say using OPA 2134 / OPA 627 in terms of clarity and resolution. This idea was done by Rudolf using OPA627. (but he uses a mirror cascode) your view pls.
Would it be better to replace coupling cap with DC servo say using OPA 2134 / OPA 627 in terms of clarity and resolution. This idea was done by Rudolf using OPA627. (but he uses a mirror cascode) your view pls.
I remember your posting up a cct like that - unfortunately, I've lost track of it, sorry - could you post it up again, or send me a copy?
I would imagine that feeding the signal thru the extra circuitry with servo would alter/change the sound a bit - a try it and see thing, tho'.
Although a rather different cct, I'm doing a "no o/p cap" version of the B1 buffer (+/- rails) and will be interesting to hear the difference.
Many people have used servo's very successfully - Beau has used a simple, rather clever servo in the Moskido amp design and obtained a quite significant improvement, for example - the Heretical III design uses one in the Cathode Follower buffer, etc.
I would imagine that feeding the signal thru the extra circuitry with servo would alter/change the sound a bit - a try it and see thing, tho'.
Although a rather different cct, I'm doing a "no o/p cap" version of the B1 buffer (+/- rails) and will be interesting to hear the difference.
Many people have used servo's very successfully - Beau has used a simple, rather clever servo in the Moskido amp design and obtained a quite significant improvement, for example - the Heretical III design uses one in the Cathode Follower buffer, etc.