Battery power supply
Hi MGH,
Don't forget that I live near an industrial area with a highly polluted mains voltage, so the DI DACs (and audio-set) are already designed to operate correctly under worst-case mains voltage conditions.
Most DACs / pre-amplifiers are single-ended designs that are very sensitive to (power supply) noise and hum. So that explains why they may well perform better on a battery. That's is the reason why I used a balanced DAC configuration.
I solved the problem by using a good mains power supply:
- two-stage mains filters (Shaffner)
- Schottky and fast, slow recovery diodes
- Separate filters (Murata) in all major power supply lines (DI 8M)
- Local voltage regulators with over-dimensioned decoupling caps (design is less critical now)
- Voltage pre-regulators
- Gyrators on each tube anode
- Regulated DC voltage for the tube filaments
Using a balanced design that makes the DAC quite immune to noise and interference.
The new ultra low jitter clock oscillator module (< 1ps), is quite insensitive to small power supply fluctuations.
The high amplitude signal levels used internally are quite insensitive to noise and hum.
Getting rid of the pre-amplifier all together (I use passive resistor volume control now).
I used lead-gel batteries for many years in other equipment, they are quite troublesome. They sulfate quickly (lose their capacity) when voltage drops below a certain level for longer time periods.
The DI 8M needs multiple / high voltages (tubes) requiring some sort of switch-mode DC to DC converter, these produce so much switching noise that the advantage of using batteries is basically gone.
Hi MGH,
Don't forget that I live near an industrial area with a highly polluted mains voltage, so the DI DACs (and audio-set) are already designed to operate correctly under worst-case mains voltage conditions.
Most DACs / pre-amplifiers are single-ended designs that are very sensitive to (power supply) noise and hum. So that explains why they may well perform better on a battery. That's is the reason why I used a balanced DAC configuration.
I solved the problem by using a good mains power supply:
- two-stage mains filters (Shaffner)
- Schottky and fast, slow recovery diodes
- Separate filters (Murata) in all major power supply lines (DI 8M)
- Local voltage regulators with over-dimensioned decoupling caps (design is less critical now)
- Voltage pre-regulators
- Gyrators on each tube anode
- Regulated DC voltage for the tube filaments
Using a balanced design that makes the DAC quite immune to noise and interference.
The new ultra low jitter clock oscillator module (< 1ps), is quite insensitive to small power supply fluctuations.
The high amplitude signal levels used internally are quite insensitive to noise and hum.
Getting rid of the pre-amplifier all together (I use passive resistor volume control now).
I used lead-gel batteries for many years in other equipment, they are quite troublesome. They sulfate quickly (lose their capacity) when voltage drops below a certain level for longer time periods.
The DI 8M needs multiple / high voltages (tubes) requiring some sort of switch-mode DC to DC converter, these produce so much switching noise that the advantage of using batteries is basically gone.
Hi maxlorenz,
Yes, it's time to demonstrate the set this way, I am very curious about the feedback.
Still no 100nF caps? what types did you order? can't you use other types?
I have to admit that an audio set will never be perfect, so very slight sound colouration (compensations) that enriches the (non perfect) sound quality, might just add the finishing touch.
There is much more to interlinks than one could possibly imagine, the entire audio spectrum passes them, the entire audio spectrum is distorted by them. They are just as important as the audio equipment used. The higher the audio set quality gets, the more critical interlinks become. Each type of device needs a fully matched interlink, an interlink can perform very well on one set, and perform bad on another, it's a matter of tuning. In my opinion there isn't a "perfect" interlink, as each application requires specific interlink properties.
Yes, it's time to demonstrate the set this way, I am very curious about the feedback.
Still no 100nF caps? what types did you order? can't you use other types?
I have to admit that an audio set will never be perfect, so very slight sound colouration (compensations) that enriches the (non perfect) sound quality, might just add the finishing touch.
There is much more to interlinks than one could possibly imagine, the entire audio spectrum passes them, the entire audio spectrum is distorted by them. They are just as important as the audio equipment used. The higher the audio set quality gets, the more critical interlinks become. Each type of device needs a fully matched interlink, an interlink can perform very well on one set, and perform bad on another, it's a matter of tuning. In my opinion there isn't a "perfect" interlink, as each application requires specific interlink properties.
Hi tubee,
Basically you need a high quality capacitor to start off with, call it the main capacitor, this will usually be a polypropylene cap. However this cap will never be perfect, air pockets between the foils allow mechanical movement of the foil, introducing unwanted harmonics. The termination of the foil and wires can introduce series resistance, causing problems with peak currents / power. The winding itself is inductive, this could cause problems with frequency range and phase accuracy. Then there is the loss factor.
The capacitor can have copper, TIN, aluminum foil, or a metallized foil. The more mass, and the better conductance the foil has, the better. The higher foil mass will lower the likeliness of mechanical movement. I used a TIN foil cap for the tweeter crossover for this reason.
The Audyn polypropylene plus has good termination between foil and wires (high peak current capability), two counter-wound foils in series (low inductance), and low loss factor. It seems vacuum techniques are used to minimize air pockets between the foils, minimizing unwanted harmonics. Audyns feel like a solid block of plastic / resin.
But like many other capacitors they are still not optimal, by using a matching bypass cap, both capacitors "fuse" into a single capacitor with better properties. Finding the correct bypass cap is very tricky. The Audyn polypropylene plus may very well boost / distort trebles when using the wrong type, or no bypass cap.
Yes I think his review is very useful, I can only confirm most of the remarks he made. Capacitors are a critical part of an audio set, and need to be selected very carefully.
I don't agree with the remark of just putting a bypass cap (MKP1837) across every capacitor. These capacitor types and values need to be selected very carefully, in order to optimally compensate the imperfections of the main capacitor, otherwise they can make matters even worse.
Basically you need a high quality capacitor to start off with, call it the main capacitor, this will usually be a polypropylene cap. However this cap will never be perfect, air pockets between the foils allow mechanical movement of the foil, introducing unwanted harmonics. The termination of the foil and wires can introduce series resistance, causing problems with peak currents / power. The winding itself is inductive, this could cause problems with frequency range and phase accuracy. Then there is the loss factor.
The capacitor can have copper, TIN, aluminum foil, or a metallized foil. The more mass, and the better conductance the foil has, the better. The higher foil mass will lower the likeliness of mechanical movement. I used a TIN foil cap for the tweeter crossover for this reason.
The Audyn polypropylene plus has good termination between foil and wires (high peak current capability), two counter-wound foils in series (low inductance), and low loss factor. It seems vacuum techniques are used to minimize air pockets between the foils, minimizing unwanted harmonics. Audyns feel like a solid block of plastic / resin.
But like many other capacitors they are still not optimal, by using a matching bypass cap, both capacitors "fuse" into a single capacitor with better properties. Finding the correct bypass cap is very tricky. The Audyn polypropylene plus may very well boost / distort trebles when using the wrong type, or no bypass cap.
Yes I think his review is very useful, I can only confirm most of the remarks he made. Capacitors are a critical part of an audio set, and need to be selected very carefully.
I don't agree with the remark of just putting a bypass cap (MKP1837) across every capacitor. These capacitor types and values need to be selected very carefully, in order to optimally compensate the imperfections of the main capacitor, otherwise they can make matters even worse.
-ecdesigns- said:
Hi John
Yes i have had bad results with bypassing of coupling-caps too, but silver-mica cap as bypass?? Never tried.
You mentioned about a air pockets in MKP cap, my CDP has 6.8 uF MKP motor starter caps, nice solid wound cap, dampened with some sort of stuff, maybe oil or wax. They sound sometime a little coloured in bass-mid, but rather good for the price.
I still have the audyn cap plus caps 6.8 uF lying around they will be used some time. But all projects of me are having a rest, i want to slow down all my project idea's for my own sake.
I admire your addiction to the subject.
Off topic: There are photographs of my surround speakers displayed in german magazine Klang & Ton aug/sept 2007, (with the round radiating tweeters, remember?)
Thanks, EC, for your reply (post #1662)
I ordered different kinds of polyester caps, all from Epcos brand. No Wima could be found. I trust Epcos industrial caps. I chose different VDC with an uneducated guess about higher VDC caps performing better...maybe they have higher inductance(?) but I seem to like better a bigger cap. Maybe you addressed this topic in your previous answer to Tubee...
They should be arriving soon.
Amen!
The fact is that we (I) know little about electromagnetic waves (or particles?
).
I wish you the best of lucks
I think your description about the sound of DI16 (and surely of DI8) is very accurate...so accurate that I can't improve it or add anything to it, so I remain silent, enjoying my little audio setup on a daily basis ...the warrior's reward!
The spacial clues the DI16 retrieves and that turns the music so enjoyable, are particularly difficult to describe and imagine, before listening to them...CD's are not that bad after all!!
Bye,
M
I ordered different kinds of polyester caps, all from Epcos brand. No Wima could be found. I trust Epcos industrial caps. I chose different VDC with an uneducated guess about higher VDC caps performing better...maybe they have higher inductance(?) but I seem to like better a bigger cap. Maybe you addressed this topic in your previous answer to Tubee...
They should be arriving soon.
Amen!
The fact is that we (I) know little about electromagnetic waves (or particles?
).I wish you the best of lucks
I think your description about the sound of DI16 (and surely of DI8) is very accurate...so accurate that I can't improve it or add anything to it, so I remain silent, enjoying my little audio setup on a daily basis
...the warrior's reward!The spacial clues the DI16 retrieves and that turns the music so enjoyable, are particularly difficult to describe and imagine, before listening to them...CD's are not that bad after all!!
Bye,
M
Hi,
I began the built of the towers for the DI8*4 DAC:
First pick shows a 4*TDA1543 tower and cooling system...
Kids, you don't want do that at home...believe me.
Better look at Doede's offer (unless EC has some alternative ).
One ready, seven more to go!
Another pick, showing general view of the main PCB. Note that some components are placed on the bottom side, to free some space.
The other stages, PSs and USB/DI2S receiver are ready to connect and use, after voltage check out. I hope 7806 regs are up to the task...
Bye,
M
PS: I believe it is better to connect those heatsinks to ground, right?
I began the built of the towers for the DI8*4 DAC:
First pick shows a 4*TDA1543 tower and cooling system...
Kids, you don't want do that at home...believe me.
Better look at Doede's offer (unless EC has some alternative
).One ready, seven more to go!
Another pick, showing general view of the main PCB. Note that some components are placed on the bottom side, to free some space.
The other stages, PSs and USB/DI2S receiver are ready to connect and use, after voltage check out. I hope 7806 regs are up to the task...
Bye,
M
PS: I believe it is better to connect those heatsinks to ground, right?
Hi maxlorenz,
The heatsink looks very good, I know how much work is involved in constructing these.
When using 4 x 8 TDA1543 chips, both 7806 regulators are up to the task.
Each TDA1543 dissipates 250mW (typical) with 32 chips, power dissipation equals 8 watts typical. At 6 volts this results in approx. 1.4A
each group of 16 x TDA1543 will draw approx. 700mA typical, the voltage regulators (LM7806) limit at 1.5A.
When the regulators are supplied with a typical voltage of approx. 12 volts, the voltage drop equals 6V as well, dissipating approx. 4 watts in each LM7806 regulator. Make sure you use sufficient cooling for these regulators.
You can connect the heatsinks to chassis (screening), I did the same with both the DI 16 (8 * 3) and the DI 8M.
The active I/V resistor value should be approx. 330 Ohms with this configuration, in order to avoid clipping.
The heatsink looks very good, I know how much work is involved in constructing these.
When using 4 x 8 TDA1543 chips, both 7806 regulators are up to the task.
Each TDA1543 dissipates 250mW (typical) with 32 chips, power dissipation equals 8 watts typical. At 6 volts this results in approx. 1.4A
each group of 16 x TDA1543 will draw approx. 700mA typical, the voltage regulators (LM7806) limit at 1.5A.
When the regulators are supplied with a typical voltage of approx. 12 volts, the voltage drop equals 6V as well, dissipating approx. 4 watts in each LM7806 regulator. Make sure you use sufficient cooling for these regulators.
You can connect the heatsinks to chassis (screening), I did the same with both the DI 16 (8 * 3) and the DI 8M.
The active I/V resistor value should be approx. 330 Ohms with this configuration, in order to avoid clipping.
apologies
While searching for info on ideal IV resistors, I stumbled on this thread
TDA1541 NOS DAC
Contains some beautiful pictures, though I cannot understand the Chinese text, for example this :
mother of all IV resistors ?
thought I'd share, sorry if this is old news
While searching for info on ideal IV resistors, I stumbled on this thread
TDA1541 NOS DAC
Contains some beautiful pictures, though I cannot understand the Chinese text, for example this :
mother of all IV resistors ?
thought I'd share, sorry if this is old news
Ideal (IV) resistors
Hi Jives11,
I assume it's used as active I/V resistor (1K5), Bernhard may be right about it being inductive, but we are talking about a few hundred nano henries typical (depending on resistance value), I think the inductance isn't such a big problem here, and I doubt if it's even audible. Much more important are low noise, and low temperature coefficient (stability).
In my opinion the best (I/V) resistor is a pure resistor, meaning very low noise, TC, reactance, capacitance, and very high stability. Resistors that meet these requirements are bulk metal foil types. They are often used in high performance (aerospace) applications.
Each of the four DI 8M I/V stages generate 4 * 0.004 * 500 = 8V full scale voltage across the I/V resistor. the resolution is halved to 18 bits (balanced design with 2 I/V converters for each channel), so a LSB change will result in 30.5uV.
I now use 500 Ohm bulk metal foil I/V and 2K5 NFB resistors. They are basically inaudible / neutral, they provide very low distortion / high resolution.
This might be interesting, to mask the LSB (Least Significant Bit) in the 19 bit DI 8M, it only takes an impedance fluctuation of 500 / 262144 = 0.0019 Ohm ! That's approx. 0.00038% ! Just think about this when choosing an I/V or NFB resistor. It also indicates how critical these circuits actually are.
Even better, what would happen when passing this signal through a pre / power amplifier with plain metal film or carbon resistors?
When used in negative feedback loops, small impedance fluctuations will change gain / add distortion, depending on the processed audio signal, this in turn covers details, and lowers usable resolution.
For these reasons, I started using Arcol 0.1% precision resistors in DI 8M, passive volume control and power amplifier. I don't need the close resistance tolerance, although it's welcome, but I do need the low temperature coefficient.
I only use bulk metal foil resistors for feedback and I/V circuits, because the feedback resistors are outside the control loop and aren't corrected.
Plain (MRS25) metal film resistors appear to cause clearly audible loss of resolution (masking), so I try to avoid them where possible, especially in the signal path.
I read some discussions about audible differences between resistors, from my own tests, I can only confirm that they can introduce clearly audible distortion / loss of resolution, provided the surrounding audio equipment / interlinks are capable of resolving this high resolution.
With the correct test setup, these dynamic resistance fluctuations are measurable, and optimal resistor selection could be made based on these measurements.
Hi Jives11,
I assume it's used as active I/V resistor (1K5), Bernhard may be right about it being inductive, but we are talking about a few hundred nano henries typical (depending on resistance value), I think the inductance isn't such a big problem here, and I doubt if it's even audible. Much more important are low noise, and low temperature coefficient (stability).
In my opinion the best (I/V) resistor is a pure resistor, meaning very low noise, TC, reactance, capacitance, and very high stability. Resistors that meet these requirements are bulk metal foil types. They are often used in high performance (aerospace) applications.
Each of the four DI 8M I/V stages generate 4 * 0.004 * 500 = 8V full scale voltage across the I/V resistor. the resolution is halved to 18 bits (balanced design with 2 I/V converters for each channel), so a LSB change will result in 30.5uV.
I now use 500 Ohm bulk metal foil I/V and 2K5 NFB resistors. They are basically inaudible / neutral, they provide very low distortion / high resolution.
This might be interesting, to mask the LSB (Least Significant Bit) in the 19 bit DI 8M, it only takes an impedance fluctuation of 500 / 262144 = 0.0019 Ohm ! That's approx. 0.00038% ! Just think about this when choosing an I/V or NFB resistor. It also indicates how critical these circuits actually are.
Even better, what would happen when passing this signal through a pre / power amplifier with plain metal film or carbon resistors?
When used in negative feedback loops, small impedance fluctuations will change gain / add distortion, depending on the processed audio signal, this in turn covers details, and lowers usable resolution.
For these reasons, I started using Arcol 0.1% precision resistors in DI 8M, passive volume control and power amplifier. I don't need the close resistance tolerance, although it's welcome, but I do need the low temperature coefficient.
I only use bulk metal foil resistors for feedback and I/V circuits, because the feedback resistors are outside the control loop and aren't corrected.
Plain (MRS25) metal film resistors appear to cause clearly audible loss of resolution (masking), so I try to avoid them where possible, especially in the signal path.
I read some discussions about audible differences between resistors, from my own tests, I can only confirm that they can introduce clearly audible distortion / loss of resolution, provided the surrounding audio equipment / interlinks are capable of resolving this high resolution.
With the correct test setup, these dynamic resistance fluctuations are measurable, and optimal resistor selection could be made based on these measurements.
Re: Ideal (IV) resistors
Hi EC,
Interesting thoughts about resistors. Audiodesign is like software, never "ready"
On the point above a thought. These fluctuations need to happen very rapidly to mask the signal. If the fluctuation are slow compared to the sampling rate, I see no problem, other than volume differences in that range, but who cares. As long as the resistance stays "constant" among a number of samples this should be no problem? May be there other things we cannot measure what makes the audible difference ?
Also, assuming 20kHz bandwith and a R of 500 Ohm the thermal noise is 0,4 uVolt, conveniently lower (-38dB) than 30uVolt of the LSB. So no real masking by this I guess?
your thoughts ?
best regards
doede
Hi EC,
Interesting thoughts about resistors. Audiodesign is like software, never "ready"
On the point above a thought. These fluctuations need to happen very rapidly to mask the signal. If the fluctuation are slow compared to the sampling rate, I see no problem, other than volume differences in that range, but who cares. As long as the resistance stays "constant" among a number of samples this should be no problem? May be there other things we cannot measure what makes the audible difference ?
Also, assuming 20kHz bandwith and a R of 500 Ohm the thermal noise is 0,4 uVolt, conveniently lower (-38dB) than 30uVolt of the LSB. So no real masking by this I guess?
your thoughts ?
best regards
doede
Thermal noise, loss of resolution
Hi dddac,
I ran listening tests with the DI 8M, the eight TDA1541A chips with 10 times lower THD than the TDA1543, and the use of bulk metal foil resistors for I/V and diff amp reveal much more detail. The 19 bit resolution and 352.8 KHz virtual sample rate pick-out everything present on a CD recording. Test tracks from Chesky Audio (xylophone sample at -60dB) is reproduced crystal clear. Weak sound reflections in recording studios that are usually inaudible are very clearly audible with the DI 8M. I can even hear pre-echos (probably magnetic imprint from one tape winding on the next) on CDs created from analog master tapes.
So even low distortion levels that are usually inaudible on existing audio sets, become very clearly audible now, and need to be corrected.
I think there are two issues that are related, resistance, and thermal noise fluctuations. When a current passes the thin metal film layer, it will produce heat, this heat in turn will result in changes in resistance and thermal noise levels.
I am not sure about the frequency of the resistance fluctuations, perhaps the masking of details is mainly caused by modulated thermal noise.
The effect is clearly audible with my audio set, MRS25 resistors (50ppm) do cover-up details, even in my passive volume control. Replacing all MRS25 resistors with Arcol 0.1% types of exactly the same resistance value resulted in a much clearer sound with more detail.
One way to reduce resistance / thermal noise (fluctuations) is to use low TC / low noise resistors. Bulk metal foil resistors can achieve TC levels of less than 0.5ppm, and are virtually noise free components. So when using these on critical positions (I/V conversion, Feedback loops, input circuits), the thermal noise is significantly reduced, lowering distortion and revealing more detail.
If bulk metal foil resistors are too expensive, or the required resistance value isn't standard, the Arcol / Tyco 0.1% precision resistors are a good alternative (15ppm / lower noise).
Another method is the use of higher wattage resistors, they have a thicker metal film layer resulting in lower noise levels and reduced resistance fluctuations. I have good results with Beyschlag 1W metal film resistors from Conrad Electronics, I am using them in the DI 8M output attenuator.
About the thermal memory issue, I think something similar happens with semiconductors, when a current passes them they also heat-up and produce more (thermal) noise.
The constant power circuit (to tackle thermal memory distortion) wasn't performing optimal, because of the HiFi show 2007, I postponed developments regarding this issue. Now it has become my major objective.
I solved some thermal memory and negative feedback problems in the MPA80 monoblock power amplifiers. I use modified cascode circuits for both current sources and differential stages. Next I replaced all MRS25 resistors at critical positions with Arcol 0.1% resistors, used Vishay S102 (250R and 10K) for the feedback loop, and significantly reduced open loop gain (less negative feedback). Because the new cascode circuits result in ultra-linear operation with vanishing low distortion, the reduction of NFB didn't result in increased THD.
I also replaced all existing current sources (tubediff amp, LM4562 class A bias circuits) with the improved cascode current sources.
One stunning result is that bad (early) CD recordings that contain jitter (flat uninvolving sound with little spatial information and detail) now come to live, and sound surprisingly well. It seems I hit one of the mechanisms that is responsible for jitter related sound degradation.
The results are so promising that I plan to design a discrete OP-amp with similar cascode techniques, for use in the DI 8M active I/V and diff amp stages.
Does this thermal noise level remain 0.4uV regardless of resistive material temperature? noise levels are likely to increase with increased temperature.
The effects of (modulated) thermal noise levels can be amplified.
When listening at low volume settings, maximum signal amplitude is quite low (100mV or less), then (thermal) noise is likely to become problematic, and could very well reduce resolution.
Hi dddac,
I ran listening tests with the DI 8M, the eight TDA1541A chips with 10 times lower THD than the TDA1543, and the use of bulk metal foil resistors for I/V and diff amp reveal much more detail. The 19 bit resolution and 352.8 KHz virtual sample rate pick-out everything present on a CD recording. Test tracks from Chesky Audio (xylophone sample at -60dB) is reproduced crystal clear. Weak sound reflections in recording studios that are usually inaudible are very clearly audible with the DI 8M. I can even hear pre-echos (probably magnetic imprint from one tape winding on the next) on CDs created from analog master tapes.
So even low distortion levels that are usually inaudible on existing audio sets, become very clearly audible now, and need to be corrected.
I think there are two issues that are related, resistance, and thermal noise fluctuations. When a current passes the thin metal film layer, it will produce heat, this heat in turn will result in changes in resistance and thermal noise levels.
I am not sure about the frequency of the resistance fluctuations, perhaps the masking of details is mainly caused by modulated thermal noise.
The effect is clearly audible with my audio set, MRS25 resistors (50ppm) do cover-up details, even in my passive volume control. Replacing all MRS25 resistors with Arcol 0.1% types of exactly the same resistance value resulted in a much clearer sound with more detail.
One way to reduce resistance / thermal noise (fluctuations) is to use low TC / low noise resistors. Bulk metal foil resistors can achieve TC levels of less than 0.5ppm, and are virtually noise free components. So when using these on critical positions (I/V conversion, Feedback loops, input circuits), the thermal noise is significantly reduced, lowering distortion and revealing more detail.
If bulk metal foil resistors are too expensive, or the required resistance value isn't standard, the Arcol / Tyco 0.1% precision resistors are a good alternative (15ppm / lower noise).
Another method is the use of higher wattage resistors, they have a thicker metal film layer resulting in lower noise levels and reduced resistance fluctuations. I have good results with Beyschlag 1W metal film resistors from Conrad Electronics, I am using them in the DI 8M output attenuator.
About the thermal memory issue, I think something similar happens with semiconductors, when a current passes them they also heat-up and produce more (thermal) noise.
The constant power circuit (to tackle thermal memory distortion) wasn't performing optimal, because of the HiFi show 2007, I postponed developments regarding this issue. Now it has become my major objective.
I solved some thermal memory and negative feedback problems in the MPA80 monoblock power amplifiers. I use modified cascode circuits for both current sources and differential stages. Next I replaced all MRS25 resistors at critical positions with Arcol 0.1% resistors, used Vishay S102 (250R and 10K) for the feedback loop, and significantly reduced open loop gain (less negative feedback). Because the new cascode circuits result in ultra-linear operation with vanishing low distortion, the reduction of NFB didn't result in increased THD.
I also replaced all existing current sources (tubediff amp, LM4562 class A bias circuits) with the improved cascode current sources.
One stunning result is that bad (early) CD recordings that contain jitter (flat uninvolving sound with little spatial information and detail) now come to live, and sound surprisingly well. It seems I hit one of the mechanisms that is responsible for jitter related sound degradation.
The results are so promising that I plan to design a discrete OP-amp with similar cascode techniques, for use in the DI 8M active I/V and diff amp stages.
Does this thermal noise level remain 0.4uV regardless of resistive material temperature? noise levels are likely to increase with increased temperature.
The effects of (modulated) thermal noise levels can be amplified.
When listening at low volume settings, maximum signal amplitude is quite low (100mV or less), then (thermal) noise is likely to become problematic, and could very well reduce resolution.
Hi all,
i finally heared what John talking about.
I build DI-16 (my plan was DI8M) because, i was little sceptical, because sounds to good when he is talking about DACs and also other components.
But DI16 works amazing good, sound is very clean, natural, musical.
Everything looks correct. Now everything else became weak point (speakers Magneplan, amplifier SKA, LDR attenuator).
I got now big trust in John designs, that i wanted to build all setup of EC designs: amplifier, attenuator, and sonic resonators, if i get chance for that.
I use USB module with DI-16, miniPC (800Mhz) with installed Ubuntu 7.04.
John said Ubuntu 6.06 LTS works better-faster.
My plan is also going to OSX, still better than Linux Ubuntu.
I think also i will go for DI8M.
regards, Bostjan
PS: just wanted to share
i finally heared what John talking about.
I build DI-16 (my plan was DI8M) because, i was little sceptical, because sounds to good when he is talking about DACs and also other components.
But DI16 works amazing good, sound is very clean, natural, musical.
Everything looks correct. Now everything else became weak point (speakers Magneplan, amplifier SKA, LDR attenuator).
I got now big trust in John designs, that i wanted to build all setup of EC designs: amplifier, attenuator, and sonic resonators, if i get chance for that.
I use USB module with DI-16, miniPC (800Mhz) with installed Ubuntu 7.04.
John said Ubuntu 6.06 LTS works better-faster.
My plan is also going to OSX, still better than Linux Ubuntu.
I think also i will go for DI8M.
regards, Bostjan
PS: just wanted to share
Hi
J have a question about the new dem clock published by eccdesign.
on the schématic the 74HC161 is directly connect to the Res which is just under the module. it means no more Res, transformer and murrata on the line ?
We can also notice that now the Dgnd is connect to the Agnd, do we have to do this directly on the module.
Thanks for your help.
J have a question about the new dem clock published by eccdesign.
on the schématic the 74HC161 is directly connect to the Res which is just under the module. it means no more Res, transformer and murrata on the line ?
We can also notice that now the Dgnd is connect to the Agnd, do we have to do this directly on the module.
Thanks for your help.
DI 8M (DEM clock) mods
Hi gil-garcia,
I think it's time to provide some more detailed information of the mods I made to both DI 16 and DI 8M. I will take some pictures to illustrate the mods. I will post this information as soon as possible.
I already posted photographs / description of this mod, but perhaps they weren't clear enough.
About the DEM clock,
DI8 Timing module:
- Cut the DEM clock trace (via close to the text 2007 on the component side), the via is located close to the number 7 near the edge of the PCB.
- Run a patch wire from U11 pin 12 to the via (solder side).
This will increase the DEM clock rate to 352.8 KHz, by selecting HC161 Q2 instead of Q3.
DI8 mainboard:
R1, C1 > replace with wire link
R2, C2, N33 > remove / open circuit
Place wire link between N33 terminal closest to C2 and N33 terminal closest to N30 / R2.
N31, N32 place wire link on outer connections (centre connection open / not connected)
N30, > leave it installed
C3...C10 (470pF Wima), > replace with small 680 Ohm metal film resistors.
This basically connects the 74HC161 DEM clock output directly to all eight 680 Ohm resistors. GND connection is provided by N30.
DA1541A modules:
- Connect a 100nF foil cap (BC components) between TDA1541A pin 17 and GND.
- Solder a 330 Ohm SMD MELF 1206 size low noise resistor between DEM clock input terminal and GND.
- cut the trace between DEM clock input terminal and TDA1541A pin 16
- Solder the previously removed Wima 470pF cap between the DEM clock input terminal and TDA1541A pin 16, make sure it doesn't touch the DI8 mainboard surface when the module is installed.
Keep connections as short as possible!
This basically connects the DEM clock input pin to GND through a 330 Ohm resistor, and feeds it into pin 16 by 470pF. Pin 17 is decoupled to GND with a 100nF film cap.
Both Analog and Digital GND circuits remain connected by ferrite bead N30 (capacitor of filter is not used / centre pin disconnected).
I hope this information will help for now.
Hi gil-garcia,
I think it's time to provide some more detailed information of the mods I made to both DI 16 and DI 8M. I will take some pictures to illustrate the mods. I will post this information as soon as possible.
I already posted photographs / description of this mod, but perhaps they weren't clear enough.
About the DEM clock,
DI8 Timing module:
- Cut the DEM clock trace (via close to the text 2007 on the component side), the via is located close to the number 7 near the edge of the PCB.
- Run a patch wire from U11 pin 12 to the via (solder side).
This will increase the DEM clock rate to 352.8 KHz, by selecting HC161 Q2 instead of Q3.
DI8 mainboard:
R1, C1 > replace with wire link
R2, C2, N33 > remove / open circuit
Place wire link between N33 terminal closest to C2 and N33 terminal closest to N30 / R2.
N31, N32 place wire link on outer connections (centre connection open / not connected)
N30, > leave it installed
C3...C10 (470pF Wima), > replace with small 680 Ohm metal film resistors.
This basically connects the 74HC161 DEM clock output directly to all eight 680 Ohm resistors. GND connection is provided by N30.
DA1541A modules:
- Connect a 100nF foil cap (BC components) between TDA1541A pin 17 and GND.
- Solder a 330 Ohm SMD MELF 1206 size low noise resistor between DEM clock input terminal and GND.
- cut the trace between DEM clock input terminal and TDA1541A pin 16
- Solder the previously removed Wima 470pF cap between the DEM clock input terminal and TDA1541A pin 16, make sure it doesn't touch the DI8 mainboard surface when the module is installed.
Keep connections as short as possible!
This basically connects the DEM clock input pin to GND through a 330 Ohm resistor, and feeds it into pin 16 by 470pF. Pin 17 is decoupled to GND with a 100nF film cap.
Both Analog and Digital GND circuits remain connected by ferrite bead N30 (capacitor of filter is not used / centre pin disconnected).
I hope this information will help for now.
Thanks for your response
your last post was very clear, j just didn'yt understand where you have connected the Agnd with DGND.
In fact not directly on the module but on the main board. In fact there is a small line (kind of digital ground ) wich connect each Dgnd module . This dgnd line is conect to to the Agnd with N30 on the main board.
J thought according to the schema that you did that directly on the TDA Module.
Is it possible to post a part list for your project in order to estimate the price of the DI8. After reading the post, j think it is possible to start with just 4 TDA.
Just an other point, as you have decided to use an USB source (j hope not saying something wrong) it is not very convenience in my case. Do you have a pcb or a schématic for using a spdif receiver with good reclock or something in order to convert the I2S from the cd pro. No computer in my listenning room.
Philippe
your last post was very clear, j just didn'yt understand where you have connected the Agnd with DGND.
In fact not directly on the module but on the main board. In fact there is a small line (kind of digital ground ) wich connect each Dgnd module . This dgnd line is conect to to the Agnd with N30 on the main board.
J thought according to the schema that you did that directly on the TDA Module.
Is it possible to post a part list for your project in order to estimate the price of the DI8. After reading the post, j think it is possible to start with just 4 TDA.
Just an other point, as you have decided to use an USB source (j hope not saying something wrong) it is not very convenience in my case. Do you have a pcb or a schématic for using a spdif receiver with good reclock or something in order to convert the I2S from the cd pro. No computer in my listenning room.
Philippe