Hello,
From the blog of Doede ( the DDDAC priest) there is some info to be found about the use of supercaps.
I also got some info from him by private mail that he will share later after doing more test.
You can read in this thread two ideas why the supercaps perform so well.
One is the low ESR.
The other is related to the idea the French had when they first used the 5 volt supercap designed by NEC in the eighties. Back THEN they wrote the supercaps cannot replace the standard caps because their esr was just to high. THEN they were just use because they had one F in such a small package. Not sure if this is the 100% right explanation but it was not used as a bucket of energy but more like an elements that would stabilize the voltage.
Doede told me he took measurements and what surprised him was the low impedance figures all the way down to lower frequencies ( much better than usual caps) and because this will result in a much better filtering of lower frequency noise.
Andrea explained his ideas about local decoupling being the key element in some of his circuits. It seems that finding a configuration , with maybe one extra circuit costing less than 100 euro is easier than having a broad collection of circuit that all need a lot of attention to work perfectly.
It seems both Roon nucleus and etherregen need a state of the art supply too.
The 16 volt package could work well for the Roon BUT you will have to decide where you have the space to use big solution like this.
Greetings, eduard
From the blog of Doede ( the DDDAC priest) there is some info to be found about the use of supercaps.
I also got some info from him by private mail that he will share later after doing more test.
You can read in this thread two ideas why the supercaps perform so well.
One is the low ESR.
The other is related to the idea the French had when they first used the 5 volt supercap designed by NEC in the eighties. Back THEN they wrote the supercaps cannot replace the standard caps because their esr was just to high. THEN they were just use because they had one F in such a small package. Not sure if this is the 100% right explanation but it was not used as a bucket of energy but more like an elements that would stabilize the voltage.
Doede told me he took measurements and what surprised him was the low impedance figures all the way down to lower frequencies ( much better than usual caps) and because this will result in a much better filtering of lower frequency noise.
Andrea explained his ideas about local decoupling being the key element in some of his circuits. It seems that finding a configuration , with maybe one extra circuit costing less than 100 euro is easier than having a broad collection of circuit that all need a lot of attention to work perfectly.
It seems both Roon nucleus and etherregen need a state of the art supply too.
The 16 volt package could work well for the Roon BUT you will have to decide where you have the space to use big solution like this.
Greetings, eduard
Attachments
Hi Andrea,
I use the A123 LiFePo4 batteries and they have esr a bit lower than your choise of batteries at 6mohms. The ultracaps do have a clear effect with these batteries; I think this specification does not tell the whole story.
ANR26650M1-B, A123 Nanophosphate(R) 3.3V 2.5Ah Lithium Iron Phosphate (LiFePO4) Cylindrical Battery
I use the A123 LiFePo4 batteries and they have esr a bit lower than your choise of batteries at 6mohms. The ultracaps do have a clear effect with these batteries; I think this specification does not tell the whole story.
ANR26650M1-B, A123 Nanophosphate(R) 3.3V 2.5Ah Lithium Iron Phosphate (LiFePO4) Cylindrical Battery
Hi Stephan,
we provide both options for the battery daughter board so one can install LiFePo4 or LTO.
Someone claims the LiFePo4 type not very good sounding when supplying analog circuits, while the LTO sounds better.
We will make a comparison in the FIFO and in the DAC Lite.
We also will give a try to the supercapacitors but we need a board providing soft charge otherwise our battery supply detects a short circuit and goes in safety mode disconnecting the output.
we provide both options for the battery daughter board so one can install LiFePo4 or LTO.
Someone claims the LiFePo4 type not very good sounding when supplying analog circuits, while the LTO sounds better.
We will make a comparison in the FIFO and in the DAC Lite.
We also will give a try to the supercapacitors but we need a board providing soft charge otherwise our battery supply detects a short circuit and goes in safety mode disconnecting the output.
Hello,
Doede told me once that the digital circuit like their power supply to be as constant as possible. I guess this also implies as low noise as possible.
Andrea tells us his digital circuits draw pretty much a constant current so voltage changes because of load changes should be easy to tackle for him.
I presume the current load in the analogue part can be taken care of with a higher F number and low esr. But it wont be a load like the ones seen in a power amp.
I think with the right equipment one could find out what is needed.
I know the Roon takes 3 to 4 A when fed with 16 volt. Is this because it is in fact a little pc?
To a lot of us it seems a bit like black magic. Every improvement comes with an additional board which in order to work perfectly should have a well designed power supply that will usually not be a 40 euro Guide Tent board or a same seize board with an LVD regulator but transformer, rectifier, caps ans some more...
Greetings,Eduard
Doede told me once that the digital circuit like their power supply to be as constant as possible. I guess this also implies as low noise as possible.
Andrea tells us his digital circuits draw pretty much a constant current so voltage changes because of load changes should be easy to tackle for him.
I presume the current load in the analogue part can be taken care of with a higher F number and low esr. But it wont be a load like the ones seen in a power amp.
I think with the right equipment one could find out what is needed.
I know the Roon takes 3 to 4 A when fed with 16 volt. Is this because it is in fact a little pc?
To a lot of us it seems a bit like black magic. Every improvement comes with an additional board which in order to work perfectly should have a well designed power supply that will usually not be a 40 euro Guide Tent board or a same seize board with an LVD regulator but transformer, rectifier, caps ans some more...
Greetings,Eduard
You mentioned supercaps are no good as the oscillator supply, I assume you meant in the usual way - like a capacitor in front of an active supply, not like a battery.
I would have thought supercaps electrical performance would be better than batteries in some ways (noise probably being the primary one) and still like the idea of trying to use them as a battery-like power source.
Of course the application has to be able to tolerate some droop over time in the supply for super caps to be used like a battery. The more it can tolerate, the better.
One idea that was discussed by another member in the past:
You have 2 banks of supercaps, 1 charging while the other powers, which are switched when voltage falls to a certain threshold.
For a suitable application the switching frequency would probably be at least a few minutes with those 300F monster caps, too low to effect musical enjoyment.
The only obstacle is designing a soft-switching circuit that also fully isolated the charging circuit.
Alternatively, you can simply have a single super cap bank that after the initial full charge is regularly switched into the charging circuit for a top up once a voltage threshold is reached.
For supercaps an ultra fast charger could be used to keep this top up cycle very short.
Unlike batteries the charging circuit is quite simple/flexible, any good quality good linear supply could be used to keep charger influence to minimum.
The lighter the load, the lower the ratio of charge time to powering time.
The inconsistent nature of this kind of this supply will probably bother some (it bothers me) but it is a very simple to do.
I have tested this kind before with just a manual switch for the charger and smaller 30F super caps on a simple SE headphone amp and the sound quality of purely supercap power was promising.
Aside from performance I much prefer the perpetual nature of these kinds of supplies.
Having battery based supply run out and cut a listening session short or require you to wait to begin is much more frustrating to me. The need to plan ahead is added stress.
I would have thought supercaps electrical performance would be better than batteries in some ways (noise probably being the primary one) and still like the idea of trying to use them as a battery-like power source.
Of course the application has to be able to tolerate some droop over time in the supply for super caps to be used like a battery. The more it can tolerate, the better.
One idea that was discussed by another member in the past:
You have 2 banks of supercaps, 1 charging while the other powers, which are switched when voltage falls to a certain threshold.
For a suitable application the switching frequency would probably be at least a few minutes with those 300F monster caps, too low to effect musical enjoyment.
The only obstacle is designing a soft-switching circuit that also fully isolated the charging circuit.
Alternatively, you can simply have a single super cap bank that after the initial full charge is regularly switched into the charging circuit for a top up once a voltage threshold is reached.
For supercaps an ultra fast charger could be used to keep this top up cycle very short.
Unlike batteries the charging circuit is quite simple/flexible, any good quality good linear supply could be used to keep charger influence to minimum.
The lighter the load, the lower the ratio of charge time to powering time.
The inconsistent nature of this kind of this supply will probably bother some (it bothers me) but it is a very simple to do.
I have tested this kind before with just a manual switch for the charger and smaller 30F super caps on a simple SE headphone amp and the sound quality of purely supercap power was promising.
Aside from performance I much prefer the perpetual nature of these kinds of supplies.
Having battery based supply run out and cut a listening session short or require you to wait to begin is much more frustrating to me. The need to plan ahead is added stress.
Last edited:
I didn't mean supercaps weren't suitable, I meant that theoretically they are waste of money to supply oscillators and frequency doublers since they are a constant load.
As I said we have measured the phase noise of the oscillators powered by batteries and linear regulators and practically there is no difference.
This means that no noise comes from the power supply, or at least we are not able to measure any noise with our tool.
The advantage of the batteries is that they are decoupled from the main AC so the oscillators are less prone to pick up external interferences.
BTW, if someone will give it a try I'm curious to know if there are audible difference.
As I said we have measured the phase noise of the oscillators powered by batteries and linear regulators and practically there is no difference.
This means that no noise comes from the power supply, or at least we are not able to measure any noise with our tool.
The advantage of the batteries is that they are decoupled from the main AC so the oscillators are less prone to pick up external interferences.
BTW, if someone will give it a try I'm curious to know if there are audible difference.
Ok, understood.
I thought one of the reasons for providing the battery supply option was for potentially better sound quality beyond the measurements.
Circuits with constant current draw would actually make implementing this kind switching charger simpler since it can just be timed circuit, rather than the charger needing to detect the voltage on supercaps to know when to charge, making isolation more difficult.
I thought one of the reasons for providing the battery supply option was for potentially better sound quality beyond the measurements.
Circuits with constant current draw would actually make implementing this kind switching charger simpler since it can just be timed circuit, rather than the charger needing to detect the voltage on supercaps to know when to charge, making isolation more difficult.
My technical approach:
Using a switch power supply to have a good efficiency. Configure the output voltage just above the final voltage level necessary for circuit (0.5V for example). So you can use a LDO linear regulator to have a very clean power supply for audio or DAC.
Example in picture.
Using a switch power supply to have a good efficiency. Configure the output voltage just above the final voltage level necessary for circuit (0.5V for example). So you can use a LDO linear regulator to have a very clean power supply for audio or DAC.
Example in picture.
Dear Andrea,
Mouser's backorders are insane for some parts for your through hole PSU.
Could you confirm that those replacements are suitable?
647-UFG2A101MHM replaced by 647-UFW2A101MPD
512-BC33740BU replaced by 821-BC337-40-A1 or by 833-BC337-40-AP
512-BC559CTA replaced by 512-BC559BTA (different Hfe)
Thanks
Mouser's backorders are insane for some parts for your through hole PSU.
Could you confirm that those replacements are suitable?
647-UFG2A101MHM replaced by 647-UFW2A101MPD
512-BC33740BU replaced by 821-BC337-40-A1 or by 833-BC337-40-AP
512-BC559CTA replaced by 512-BC559BTA (different Hfe)
Thanks
Hi, I did not test on the same audio design with the 2 methods (linear only or SMPS + Linear). I only test an audio design with SMPS + LDO and the signal is very clear without any noise on the audio line. The system is for VOICE which is digitalized then transmit on a RF line (1GHz) to 200Nm (Nautic Miles) then transform again by the receiver in an analog signal. The communication is betwwen 2 people which have both a headset (a microphone + a headphone + a Push-To-Talk). I did some test on a VOIP line too and VOICE is identical as you talk directly with a person just in front of you
I use the following components:
If input is AC, I use a small AC/DC PS like XP POWER ECE10 or TRACO TMPW 25-112 to have a +12Vcc or +24Vcc power supply. Then I use or not a TI LMZ14203H to have the voltage I need (+12Vcc, +5Vcc, ...) --> output ripple is as low as 1mV with this IC (power supply module). I use a LDO with a PSSR of 40dB min then ripple is as low as µV -> I think it's enough for Audio power supply.
For dual supply I generally use the DC/DC TRACO TMR3 with an outpout of +/12V then a LDO TI TPS7A39 (dual LDO) to have +/-11.7V and 150mA max (an audio AOP need 4-5mA max).
Ripple at output of TMR3 is 75mV max. LDO has a PSSR of > 50dB from 10Hz to 2MHz --> ripple on the power supply audio line is 75µV. Low, isn't it ? (depend on the gain of your amp of course, if gain is very high, a lower ripple could be necessary). Of course ripple is not the only parameter so some small capacitors (some pF) or a ferrite beard to cut very high frequency could be necessary (some spikes can be at hundred of MHZ on a SMPS).
Last edited:
TWRPS-LBS-M LiFePo4 case
TWRPS-LBS-M LiFePo4 dimensions are 260mm x 195mm, but what about maximum height? The tallest component should be the big heat sink SK 68 75 SA, which is 33mm. With plastic washer and PCB thickness another 3mm should be added, so it will definitely be under 40mm?
I'm looking at this case: 270*56*235 mm (w*h*l) DAC amplifier shell aluminum chassis Instrumentation aluminum profile chassis / DIY industrial aluminum|aluminium profile|aluminium chassischassis diy - AliExpress
TWRPS-LBS-M LiFePo4 dimensions are 260mm x 195mm, but what about maximum height? The tallest component should be the big heat sink SK 68 75 SA, which is 33mm. With plastic washer and PCB thickness another 3mm should be added, so it will definitely be under 40mm?
I'm looking at this case: 270*56*235 mm (w*h*l) DAC amplifier shell aluminum chassis Instrumentation aluminum profile chassis / DIY industrial aluminum|aluminium profile|aluminium chassischassis diy - AliExpress
The maximum height of the TWRPS-LBS-M LiFePo4 is around 38 mm considering the heatsink on the top and the components on the bottom layer.
Then you have to add the height of the standoff to install the circuit in the enclosure.
The box you have chosen looks OK.