I just made some updates to the landing page of my YouTube channel. Will be uploading more videos soon and probably kick it off with a short intro video about XRK Audio.
https://m.youtube.com/c/xrkaudio
https://m.youtube.com/c/xrkaudio
Hi Folks,
This thread has been a little inactive lately. I thought it would be good to kickstart it with a Halloween Sale for all things in my shop. Here a 10% off coupon good for all items in the shop through Oct 31, 2023.
https://xrkaudio.etsy.com/?coupon=10OFF
Have a good Halloween!🎃
This thread has been a little inactive lately. I thought it would be good to kickstart it with a Halloween Sale for all things in my shop. Here a 10% off coupon good for all items in the shop through Oct 31, 2023.
https://xrkaudio.etsy.com/?coupon=10OFF
Have a good Halloween!🎃
Happy Holidays everyone! Hope you are all able to relax and spend time with friends and family. Maybe have some time to do a little amplifier/speaker building or listening too!
With a week to spend at home with family, sometimes it is good to find time for a little DIY after everyone has gone off to bed.
I got to play with my new Desktop Diamond Buffer headphone amp last night. I made a high bias current Class A variant by changing the emitter resistors from 4.75ohm to 1.0ohm. The bias current went from about 18mA to 78mA. Heatsinks are definitely warmer but not hot so operationally this is going to be fine. I was surprised what a difference that made to harmonic profile. I was also surprised to see how clean a cheap little DCDC converter from 12vdc to +/-12vdc followed by some CRCLCRC filtering can be. I compared it to my best PSU from the Yarra Mk1 (linear toroidal trafo / active bridges / CRCLC / ultra low noise LDO regulators - and the noise floor was actually as good and a tad less mains noise. I might start a thread on just discussing various types of PSU’s for different uses. Having a Focusrite and REW for doing FFTs really shows quantitatively what an amplifier is doing. I always try to listen to an amp for a long time before doing measurements to make sure I don’t get mental biases. What’s interesting is that my ears can hear the subtle differences in harmonic profile.
Stay safe and healthy everyone, and happy holidays and wishing you a great new year to come!
With a week to spend at home with family, sometimes it is good to find time for a little DIY after everyone has gone off to bed.
I got to play with my new Desktop Diamond Buffer headphone amp last night. I made a high bias current Class A variant by changing the emitter resistors from 4.75ohm to 1.0ohm. The bias current went from about 18mA to 78mA. Heatsinks are definitely warmer but not hot so operationally this is going to be fine. I was surprised what a difference that made to harmonic profile. I was also surprised to see how clean a cheap little DCDC converter from 12vdc to +/-12vdc followed by some CRCLCRC filtering can be. I compared it to my best PSU from the Yarra Mk1 (linear toroidal trafo / active bridges / CRCLC / ultra low noise LDO regulators - and the noise floor was actually as good and a tad less mains noise. I might start a thread on just discussing various types of PSU’s for different uses. Having a Focusrite and REW for doing FFTs really shows quantitatively what an amplifier is doing. I always try to listen to an amp for a long time before doing measurements to make sure I don’t get mental biases. What’s interesting is that my ears can hear the subtle differences in harmonic profile.
Stay safe and healthy everyone, and happy holidays and wishing you a great new year to come!
I find this statement very interesting, X. I remember using a DCDC converter without any filtering in the PSU of the AKSA Lender preamp (after you recommended it's use) and was surprised how quiet it is. Where can I get more info on the CRCLCRC filter you mention? A search brought up page 129 of the Super Regulator thread, but I'm not keen on reading 172 pages.
Hi Skylar,
I made this filter years ago by cobbling loose cheap caps and inductors lying around on my bench. Here is a closeup photo of it, and basically I made it using 470uF caps and 4.7R resistors plus a 1mH ferrite core inductor and some 100nF film caps at the exit. There is also a 6k8 bleed resistor to make sure it doesn't hold a charge for long. I sketched it up on LTspice and ran an AC analysis and surprisingly, it has a lot of attenuation at the DCDC switch frequency of 100kHz. Minus -210dB in fact - an amount that exceeds the noise floor of any measurement tool.
Here is the schematic:
And here is the AC response:
The DC simulation shows that for a 15vdc input, at 68mA of current flow through a 200ohm dummy load, there is a 1.3v dropout. So you probably don't need all of these stages for an effective filter. Anything better than -120dB is more than enough.
If we simplify to this it still works very well with -140dB at 100khz and only a 0.35v dropout for 73mA current:
Here is the AC response:
If you leave the 4R7 in series with the inductor, the small peak at 400Hz will probably be flatter.
Here is a closeup photo:
For those of you who have built the Aksa Lender preamp using my PCB's, you will notice that it comes with a CLRC filter board with similar values (single rail). Hence, using a SMPS with the Aksa Lender with this type of filter is very effective.
So finally, if you make it really simple CLRC filter, you get -90dB at 100kHz and still about 0.35V dropout:
It seems worthwhile to add the extra stage CR in front to get basically complete blockage of SMPS noise.
I made this filter years ago by cobbling loose cheap caps and inductors lying around on my bench. Here is a closeup photo of it, and basically I made it using 470uF caps and 4.7R resistors plus a 1mH ferrite core inductor and some 100nF film caps at the exit. There is also a 6k8 bleed resistor to make sure it doesn't hold a charge for long. I sketched it up on LTspice and ran an AC analysis and surprisingly, it has a lot of attenuation at the DCDC switch frequency of 100kHz. Minus -210dB in fact - an amount that exceeds the noise floor of any measurement tool.
Here is the schematic:
And here is the AC response:
The DC simulation shows that for a 15vdc input, at 68mA of current flow through a 200ohm dummy load, there is a 1.3v dropout. So you probably don't need all of these stages for an effective filter. Anything better than -120dB is more than enough.
If we simplify to this it still works very well with -140dB at 100khz and only a 0.35v dropout for 73mA current:
Here is the AC response:
If you leave the 4R7 in series with the inductor, the small peak at 400Hz will probably be flatter.
Here is a closeup photo:
For those of you who have built the Aksa Lender preamp using my PCB's, you will notice that it comes with a CLRC filter board with similar values (single rail). Hence, using a SMPS with the Aksa Lender with this type of filter is very effective.
So finally, if you make it really simple CLRC filter, you get -90dB at 100kHz and still about 0.35V dropout:
It seems worthwhile to add the extra stage CR in front to get basically complete blockage of SMPS noise.
Last edited:
How important is inductor value? You show 1mH here, but in the AKSA Lender CRLC filter, the BOM calls for 22mH.
Second question: Is the winding conductor size important at low current levels? The previously mentioned 22mH toroidal uses 1.68mm dia wire. I can get one of the same value for much cheaper elsewhere, but it’s wound with 1mm magnet wire. Just trying to save a few pennies if wire thickness doesn’t matter.
It's for powering a ESP32 MCU and small LCD screen running on 5V and drawing about 20mA max, but I want to use a filter after the DCDC.
Second question: Is the winding conductor size important at low current levels? The previously mentioned 22mH toroidal uses 1.68mm dia wire. I can get one of the same value for much cheaper elsewhere, but it’s wound with 1mm magnet wire. Just trying to save a few pennies if wire thickness doesn’t matter.
It's for powering a ESP32 MCU and small LCD screen running on 5V and drawing about 20mA max, but I want to use a filter after the DCDC.
It depends on how you size the system and what kind of current you need. For your 20mA all you need is a $0.25 inductor. A 20mH inductor will cost a pretty penny and let you use smaller caps. Just scale inversely. You definitely don’t need 1.6mm wire (that’s 14ga and good for 10A)!
If you want a filter to power your big Class A amplifier (which is what Hugh designed it for) and I think he wants to filter 50Hz whereas you only need to go down to 20kHz to filter out SMPS at 100kHz.
For your Class ESP32 MCU just use a small 1mH or even 0.5mH inductor and some 470uF or 1000uF caps. Make it similar to the schematic I showed above. That will crush all signs of SMPS noise.
I am designing a mains filter for SMPS noise and it’s important to keep the caps small since they need to be Y2 rated safety caps (fail open) which are not made larger than 2.2uF ($10ea).
I measured a 400Hz signal injected into it and nothing came out the other side (like a black hole). The sims predicted -130dB attenuation but only -36dB for 100kHz. Still quite good and can pass 10A at 250v.
Those are 1.5mH common mode chokes and 2.2nF to 3.9nF Y2 ceramic caps (blue ones).
If you want a filter to power your big Class A amplifier (which is what Hugh designed it for) and I think he wants to filter 50Hz whereas you only need to go down to 20kHz to filter out SMPS at 100kHz.
For your Class ESP32 MCU just use a small 1mH or even 0.5mH inductor and some 470uF or 1000uF caps. Make it similar to the schematic I showed above. That will crush all signs of SMPS noise.
I am designing a mains filter for SMPS noise and it’s important to keep the caps small since they need to be Y2 rated safety caps (fail open) which are not made larger than 2.2uF ($10ea).
I measured a 400Hz signal injected into it and nothing came out the other side (like a black hole). The sims predicted -130dB attenuation but only -36dB for 100kHz. Still quite good and can pass 10A at 250v.
Those are 1.5mH common mode chokes and 2.2nF to 3.9nF Y2 ceramic caps (blue ones).
Last edited:
Hi X, I was hallucinating when I said 20mA max draw for the ESP32. I’ve just measured again and it’s 100mA. So, speccing it as 200mA with a healthy safgety margin.
I’m assuming (never a good thing) that 1mH for the inductor is still fine, with 470uF caps. Would you agree?
I’m leaning towards a Kemet 1mH ferrite inductor with 450mA rated current: Mouser Part No 80-SBC9-102-451
I’m assuming (never a good thing) that 1mH for the inductor is still fine, with 470uF caps. Would you agree?
I’m leaning towards a Kemet 1mH ferrite inductor with 450mA rated current: Mouser Part No 80-SBC9-102-451
I redid the calc with your particular inductor which has 15ohm DCR. I reduced the series resistor values to 0.47 ohm to decrease your voltage loss. I get -120dB at 100kHz for this:
This has a 2v drop for 204mA of current with an input of 7.2Vdc and output of 5.16Vdc. Here is the schematic:
Use 1W resistors if you can, but 0.5W will also work.
Edit: you can probably leave off R2 and it won't matter since L1 already has 15ohms DCR.
This has a 2v drop for 204mA of current with an input of 7.2Vdc and output of 5.16Vdc. Here is the schematic:
Use 1W resistors if you can, but 0.5W will also work.
Edit: you can probably leave off R2 and it won't matter since L1 already has 15ohms DCR.
Last edited:
Thank you X, you are very generous with your time. Much appreciated.
I was hoping to use a 5V smps (Hi-Link) directly off the grid instead of a DCDC, but if the circuit requires 7.2V, I'll have to use the DCDC running off another type of PSU. Hi-Link doesn't have a 7.2V, but there is a 12V unit that I could use + DCDC adjusted to 7.2v. It also allows for voltage fine-tuning.
I was hoping to use a 5V smps (Hi-Link) directly off the grid instead of a DCDC, but if the circuit requires 7.2V, I'll have to use the DCDC running off another type of PSU. Hi-Link doesn't have a 7.2V, but there is a 12V unit that I could use + DCDC adjusted to 7.2v. It also allows for voltage fine-tuning.
You are welcome. You might want to check out the AmyAlice filter designed by Mark Johnson. It’s a bit more high tech and uses SMT parts. There’s a special “feed through” capacitor that alone filters out a lot of stuff. It has the advantage of very low dc drop as the resistors are very low.
https://www.diyaudio.com/community/...apacitors-smd-assembly-max-3a-max-48v.405099/
https://www.diyaudio.com/community/...apacitors-smd-assembly-max-3a-max-48v.405099/
The voltage drop is indeed very low with the AmyAlice filter: 56mA at 200mA load. It means I'll be able to use the 5V Hi-Link smps directly without needing a higher voltage smps and DCDC converter. It's a no-brainer, except it requires SMD soldering which isn't a problem for me.
So, you did help me, although I won't be using your filter design for this particular project. It's nice to be able to choose an optimal solution, though.
No problem. Glad you have a solution. That AA filter is using these “feed through capacitors”. They provide some nice filtering behind their capacitance rating by design. Sort of like those screw in bulkhead feed through connectors for power input to a RF box. But no bulkhead and still same filter ability.