[b0bb]

Yes indeed that Maxim chip charges at a frequency of 350khz, in theory I did not see it as important because in the charging process the dac will be totally inactive, once the supercap is loaded, the pic would be notified of it by an input, disconnect the charging process and the oscillator, and it would be just waiting a user start button, before all that it would not be possible to turn on the dac, you have to put an input from the pic that detects low voltage and another high voltage from the supercap, perhaps with a few simple zener diodes that control transistors that set the level high or low to each entry would be enough, some LEDs could indicate the status of the process. What I saw very interesting about this Maxim evaluation kit is that they give you everything, even the supercapacitors, it is a fully assembled and tested module for only € 27, which is very interesting to do a safe load test of the supercaps and a subsequent test of the sound with clean energy in different supply points of the 004.

Are you thinking of using this as a battery replacement?

The 175F (2x350F in series) cap will last about 2 mins with a 1300mA load before voltage dips below 3.3V if the switched mode supply is not on.
Not enough time for the 9038s to warm up.

 

[Xoverman]

Last week i had some time to measure the effect of the first primary filter before the primary core voltage regulator.
Because i didn't want to measure the tiede in honolulu divided by the distance to the moon in meters, I have decided to put SMB sockets everywhere where I want to measure. The beginning is made by the first primary filter before the primary core voltage regulator. Right behind the rectifiers.

Ripple @ the 22.000µF Filter Cap first primary filter before the primary core voltage regulator :

Ripple @ the second 22.000µF primary Filter Cap filter before the primary core voltage regulator ( same vertical resolution ):

Ripple @ the second 22.000µF primary Filter Cap filter before the primary core voltage regulator ( vertical resolution X10 ):

The high frequency ripple on the signal is actual aliasing from a 6 MHz ozilation.
It can only be measured at the second Cap before the voltage regulator.

 

[Xoverman]

Hear is a measurement using the spectrum analyzer funktion at the second capacitor with the normal measurement probe that came with the scope :

Same measurement point but this time with the smb connection.
With the SMB connection you can clearly see the 6 MHz Resonanz in the PS :

 

[Xoverman]

the first two SMB connectors in the PS :

 

[Xoverman]

Here is a low frequency FFT of the Core PS

Ripple @ the 22.000µF Filter Cap first primary filter before the primary core voltage regulator :

Ripple @ the second 22.000µF primary Filter Cap filter before the primary core voltage regulator ( same vertical resolution ):

 

[b0bb]

Ripple @ the second 22.000µF primary Filter Cap filter before the primary core voltage regulator ...

The ripple with the π filter is very low about 14mV, this is a demonstration of the filter effectiveness.

This relieves the burden from the first stage regulator and opens up the possibility of using a supercap first stage.
A supercap regulator will add much improved current handling and superior large transient handling.

Nothing comes for free, expect reduction in PSRR and degradation in noise performance due to the the periodic charging pulses from the supercap controller
The π filter and the low initial ripple makes up for some of the short comings of the first point.

Below is an example of the very nasty switching noise as the supercap controller turns on/off the charging.
This is from the MAX17701 being discussed earlier.

 

[b0bb]

Here is a low frequency FFT of the Core PS

Ripple @ the second 22.000µF primary Filter Cap filter before the primary core voltage regulator ( same vertical resolution ):

The FFT agrees with the earlier measurement if the amplitudes of the primary plus 2nd and 3rd harmonics are added up and taking 7.70V DC as the reference.
7.70V is the DC input to the first stage regulator

This gives AC noise around 18.1mV rms which is is the same ballpark of 14.1mV from the earlier measurement.
Example: -55dB down from 7.70V is 13.7mV

 

[ESL57]

Are you thinking of using this as a battery replacement?

The 175F (2x350F in series) cap will last about 2 mins with a 1300mA load before voltage dips below 3.3V if the switched mode supply is not on.
Not enough time for the 9038s to warm up.

I only seeing if there is a reasonable performance possibility with the use of supercaps in the more noise sensitive dac supplies, 1300ma is a lot of current for only 175F, so look to get to 2x 3000F in series, maybe it would only be reasonable to use it for the 3.3v XO clock, batteries is another option if more current is needed to feed more dac supplies, my idea is to make an external power supply that can switch to standard mode or clean power mode, the latter perhaps for sessions of listen for not too long, we're just catching ideas.

 

[b0bb]

I only seeing if there is a reasonable performance possibility with the use of supercaps in the more noise sensitive dac supplies, 1300ma is a lot of current for only 175F, so look to get to 2x 3000F in series, maybe it would only be reasonable to use it for the 3.3v XO clock, batteries is another option if more current is needed to feed more dac supplies, my idea is to make an external power supply that can switch to standard mode or clean power mode, the latter perhaps for sessions of listen for not too long, we're just catching ideas.

1500F (2x3000F in series) is 8x more, this gets you 17mins, not much of an improvement.
This is an incorrect use of supercaps.

Supercaps are not batteries and they perform poorly when forced into that role.
The correct use of supercaps is what Singxer did on the SU6 and LKS on the 005, the charger is always on.

XO clock is a constant load, no benefit from the quick current delivery supercaps provide.
Gen2 use LT3045 as does SU-6, Gen1 use TPS7A47, all work very well, best left alone.


Batteries have low noise but high ESR compared to caps, they were a fad for audiophiles sometime back.
Modern voltage regulators have made them obsolete.

It is a mistake to think clean power cannot be obtained onboard with the 004.

The modern regulators we have at our disposal today are so quiet that they are below the noise floor of the 9038
Examples (20Hz -22kHz):
LT3045 : 0.8uV
Belleson SPLX :1uV

9038Pro noise floor: 1.2uV

 

[Xoverman]

Yesterday i installed some more SMB Sockets into the 004.

This is the noise at the output of the primary core regulator

this is the noise at the output of the primary core regulator but triggered for the highest peak.
I guess it's when the linedrivers come into play :

 

[ESL57]

1500F (2x3000F in series) is 8x more, this gets you 17mins, not much of an improvement.
This is an incorrect use of supercaps.

Supercaps are not batteries and they perform poorly when forced into that role.
The correct use of supercaps is what Singxer did on the SU6 and LKS on the 005, the charger is always on.

XO clock is a constant load, no benefit from the quick current delivery supercaps provide.
Gen2 use LT3045 as does SU-6, Gen1 use TPS7A47, all work very well, best left alone.


Batteries have low noise but high ESR compared to caps, they were a fad for audiophiles sometime back.
Modern voltage regulators have made them obsolete.

It is a mistake to think clean power cannot be obtained onboard with the 004.

The modern regulators we have at our disposal today are so quiet that they are below the noise floor of the 9038
Examples (20Hz -22kHz):
LT3045 : 0.8uV
Belleson SPLX :1uV

9038Pro noise floor: 1.2uV

Yes, I agree, audio fads come but sometimes they don't stay for long, until now the supercaps were for use in energy maintenance for volatile memories with little consumption, maybe and it is just my thought that it is not from Engineer, the only positive thing is the ultra low internal resistance of the new high capacity supercaps that could be beneficial to reduce noise and residual AC, being a short circuit for all this, the use as a DC accumulator is true that it is little efficient as you comment B0bb.

 

[b0bb]

this is the noise at the output of the primary core regulator but triggered for the highest peak.
I guess it's when the linedrivers come into play :

What is the timescale/horizontal axis on the trace and is the primary regularor rated at 1.5A ?

 

[b0bb]

Yes, I agree, audio fads come but sometimes they don't stay for long, until now the supercaps were for use in energy maintenance for volatile memories with little consumption, maybe and it is just my thought that it is not from Engineer, the only positive thing is the ultra low internal resistance of the new high capacity supercaps that could be beneficial to reduce noise and residual AC, being a short circuit for all this, the use as a DC accumulator is true that it is little efficient as you comment B0bb.

Supercaps perform well for sort current peaks, this makes them ideal supply buffers or conditioners.

IANCanada's UCC 3.3V and 5V boards is the best use of supercaps for this type of appplication that I have come across, suggest you begin there.
Analog's LTC4425 eval board is another option at around the same price once you include the cost of the supercap.

Ultra low ESR is not unique to supercaps, the Nichicon 820uF RNL FPCaps comes very close.
You will need 3x 350F in series for the 004, this is an ESR of 6mΩ, 820uF RNL is 8mΩ

 

[Xoverman]

What is the timescale/horizontal axis on the trace and is the primary regularor rated at 1.5A ?

10ns for every square, you can see it on the top of the picture.
2x LT3081 in parallel.

 

[b0bb]

This is the trace of the same supply, regulator is LT3083
The general ripple trace shape and amplitude appear similar to @Xoverman's results.

However I think the ripple amplitudes are too high at 700-800uV rms.
Effective PSRR is only 49dB, input noise is 140mV, datasheet numbers are > 75dB

I decided to re-do the measurement on my Fluke 87V, the True RMS function has a bandwidth of 200kHz
Measument performed with DSD512 output.
Used a shielded tweezer type probe

The noise number is exactly what Analog put in their datasheet --> 40uV

This was an unexpected surprise, this shows the 004 on board regulator setup is capable of very low noise levels.
Noise at the output of the second stage Belleson SPLX is below my meter's measurement capability of 10uV