Hi!
I am doing some diy class A headphone amplifier and would like to ask someone with more experience which power supply is suitable for my needs.
So the idea is to have two MeanWell smps modules with clclc filters feeding the regulator of some kind. The build will be small and compact - about 160x120mm
I have few ideas in my mind:
1. To have TPS7A47/33 regulators for each channel
2. To have TPS7A47/33 regulators feeding the super regulator (or kevin gilmore golden reference supply)
3. To have just super regulator (or kevin gilmore golden reference supply) with beefier filtering at the input.
4. Some better solution
Total current draw will be 300-400mA per rail and output voltage needs to be 20V.
I would like to know if regualtor output impedance has a big impact on the class A amplifier which is a farily constant load in comparison to other amplifier classes.
Any answer is appreciated.
I am doing some diy class A headphone amplifier and would like to ask someone with more experience which power supply is suitable for my needs.
So the idea is to have two MeanWell smps modules with clclc filters feeding the regulator of some kind. The build will be small and compact - about 160x120mm
I have few ideas in my mind:
1. To have TPS7A47/33 regulators for each channel
2. To have TPS7A47/33 regulators feeding the super regulator (or kevin gilmore golden reference supply)
3. To have just super regulator (or kevin gilmore golden reference supply) with beefier filtering at the input.
4. Some better solution
Total current draw will be 300-400mA per rail and output voltage needs to be 20V.
I would like to know if regualtor output impedance has a big impact on the class A amplifier which is a farily constant load in comparison to other amplifier classes.
Any answer is appreciated.
Yes I have seen it but the webpage says it can do 100mA at 18V. Is this correct?
I forgot to mention that I want to make my own pcb because I am an embedded hardware engineer and love making pcbs 😂 And of course integrate the whole thing on the main board.
I suppose that SilentSwitcher scheme is not public??
Wouldn't smps modules + passive filtering + TPS7A47/33 be very similary to the SilentSwitcher performance?
I forgot to mention that I want to make my own pcb because I am an embedded hardware engineer and love making pcbs 😂 And of course integrate the whole thing on the main board.
I suppose that SilentSwitcher scheme is not public??
Wouldn't smps modules + passive filtering + TPS7A47/33 be very similary to the SilentSwitcher performance?
The latest version of the SS can do several 100mA, depending on output voltage.
The circuit is not public unfortunately.
Member @bohrok2610 is handling the Group Buy and I understand he has still some stock.
But if you want to use the TPS7Axx chips, no need to follow that with a superreg.
That would not increase the excellent performance, especially HF-PSRR of those chips.
Jan
The circuit is not public unfortunately.
Member @bohrok2610 is handling the Group Buy and I understand he has still some stock.
But if you want to use the TPS7Axx chips, no need to follow that with a superreg.
That would not increase the excellent performance, especially HF-PSRR of those chips.
Jan
The SMPS units which I experimented with, cannot drive infinitely large filter capacitors. As I increased the output capacitance little by little, eventually it became "C_too_big" which tripped the SMPS current limit and entered Hiccup Mode (a self protection built in to the SMPS) forever. The only escape from Hiccup Mode was to power off and wait a few seconds.
So, I suggest that you do some experiments with your chosen SMPS units and find out how much capacitance is "C_too_big". Do it twice, with min-allowed-loadcurrent and again with your-application-loadcurrent. Then when designing your LCLC or CRCRC pre-filters upstream of the TPS7A47/33 voltage regulators, make sure that the total capacitance of the entire filter, is less than half of C_too_big. Now you're confident that the SMPS is comfortably away from Hiccup Mode. The nonzero impedance of the series elements (L's and R's) is beneficial too.
I'd suggest that you arrange your pre-filters in such a way that the final output node of the pre-filter (which becomes the raw DC voltage input of the voltage regulator IC) has at least (1/3)rd of the total pre-filter capacitance. You want a lot of capacitance (== a low impedance) at the regulator input, in parallel with the 10uF ceramic capacitor that the datasheet demands. So for example, if your prefilter is LCLCRC then make all three capacitors equal and presto, the final one is 1/3 of the total. And of course C+C+C < (0.5 * C_too_big).
Finally, make sure you don't accidentally create peaks in the frequency response; remember: (inductor + capacitor) = resonator after all. So ensure proper damping to keep the Q below 0.5 and thus keep peaking under control.
So, I suggest that you do some experiments with your chosen SMPS units and find out how much capacitance is "C_too_big". Do it twice, with min-allowed-loadcurrent and again with your-application-loadcurrent. Then when designing your LCLC or CRCRC pre-filters upstream of the TPS7A47/33 voltage regulators, make sure that the total capacitance of the entire filter, is less than half of C_too_big. Now you're confident that the SMPS is comfortably away from Hiccup Mode. The nonzero impedance of the series elements (L's and R's) is beneficial too.
I'd suggest that you arrange your pre-filters in such a way that the final output node of the pre-filter (which becomes the raw DC voltage input of the voltage regulator IC) has at least (1/3)rd of the total pre-filter capacitance. You want a lot of capacitance (== a low impedance) at the regulator input, in parallel with the 10uF ceramic capacitor that the datasheet demands. So for example, if your prefilter is LCLCRC then make all three capacitors equal and presto, the final one is 1/3 of the total. And of course C+C+C < (0.5 * C_too_big).
Finally, make sure you don't accidentally create peaks in the frequency response; remember: (inductor + capacitor) = resonator after all. So ensure proper damping to keep the Q below 0.5 and thus keep peaking under control.
Thanks a lot for the help.
I will integrate relay attenuator for volume control with rotary encoder for which I need dirty 24V directly from smps modules.
I might use one of your SilentSwitcher board for my next project!
Thanks a lot for a very detailed guide.
I will try to do exactly that. So smps + pre-filter + TPS7Axx it is.
Unless your Class A amp has really crap PSRR, just using the correct Meanwell SMPS with minimal extra supply capacitance is likely to sound better than the biggest 'conventional' PSU with zillion Farads.
Actually it will sound better especially if it has crap PSRR. 😊
Actually it will sound better especially if it has crap PSRR. 😊
Interesting fact. I must try that.
The amp I want to build is a dynalo amp (single ended version).
It has two current sources so I guess PSRR should be on the higher side.
The amp I want to build is a dynalo amp (single ended version).
It has two current sources so I guess PSRR should be on the higher side.