How to improve an existing SMPS to use it with class D amplifiers
- Thread starter snake69fast
- Start date
Great Thread. The Mean Wells are cheap but I'm positive that I'm hearing the switching noise. if they can be modified easily they're a great option for DIY'ers
I hope you can help out, abraxalito.
Here is 2 that I'm using with Class d amps, both 24V version http://www.mouser.com/ds/2/260/RS-100-SPEC-806359.pdf
http://www.mouser.com/ds/2/260/RS-75-SPEC-806451.pdf
cheers!
I hope you can help out, abraxalito.
Here is 2 that I'm using with Class d amps, both 24V version http://www.mouser.com/ds/2/260/RS-100-SPEC-806359.pdf
http://www.mouser.com/ds/2/260/RS-75-SPEC-806451.pdf
cheers!
Mods are certainly possible but they're not necessarily easy. They have been fun for me though. Meanwell are a Taiwanese outfit, their quality is better than the normal China mainland sourced supplies I tend to use, but even the mainland-built supplies can deliver the goods if ultimate reliability isn't a major concern.
Order up some of these cores - http://www.cwsbytemark.com/images/T80-52.pdf
You'll need 5 to build a take-no-prisoners filter for your supply.
Caps should be these - EKZH250ELL222MK30S United Chemi-Con | Mouser.
You'll need as many caps as inductors - i.e. 5.
I'll get back to you with the wire requirements for winding the toroids.
You'll need 5 to build a take-no-prisoners filter for your supply.
Caps should be these - EKZH250ELL222MK30S United Chemi-Con | Mouser.
You'll need as many caps as inductors - i.e. 5.
I'll get back to you with the wire requirements for winding the toroids.
Last edited:
Ah thanks I had totally forgotten about that 
The AL value for those cores is 42. In my design I went for a target inductance about 13uH. This translates to 18 turns on these cores. According to my sims, we want about 30mohms DCR for optimum damping.
Looking at the mechanical data the core cross-section is 3.8mm X 6.35mm so a single turn would be about 20mm in length assuming zero wire diameter. Thus 18 turns is going to be 0.4m, roughly. We therefore need wire that's 75mohm/m and that works out to be 23AWG, allowing for terminations.
The AL value for those cores is 42. In my design I went for a target inductance about 13uH. This translates to 18 turns on these cores. According to my sims, we want about 30mohms DCR for optimum damping.
Looking at the mechanical data the core cross-section is 3.8mm X 6.35mm so a single turn would be about 20mm in length assuming zero wire diameter. Thus 18 turns is going to be 0.4m, roughly. We therefore need wire that's 75mohm/m and that works out to be 23AWG, allowing for terminations.
Hi,
I have some no-name and some Meanwell SMPS and I am thinking to improve them if it is possible. (as power supply for class D amplifiers like TPA3116D2 etc)
Any smart ideas?
Thank you!
To be boring, have you actually measured the rails to see if anything funny is going on?
I don't have those particular supplies nor anything likely to be able make reliable measurements at ultrasonic frequencies. But in simulation the filter does give substantial HF reduction. If the switching noise you're hearing is common-mode induced then this filter won't have any impact on that - for CM noise reduction you'll need a CM choke.
Okay, I think I'll wait until I can take measurements.
And, sometimes its hard to audibly determine the origin, of the effect that I'm hearing.
The meanwell spec sheet states that noise+ripple measures 120mvP-P(max)
The Sima 11 linear supply states that the ripple on the DC output is 35uv RMS. I'm using 2 of these in Class D mono blocks.
And, sometimes its hard to audibly determine the origin, of the effect that I'm hearing.
The meanwell spec sheet states that noise+ripple measures 120mvP-P(max)
The Sima 11 linear supply states that the ripple on the DC output is 35uv RMS. I'm using 2 of these in Class D mono blocks.
Ah thanks I had totally forgotten about that
The AL value for those cores is 42. In my design I went for a target inductance about 13uH. This translates to 18 turns on these cores. According to my sims, we want about 30mohms DCR for optimum damping.
Looking at the mechanical data the core cross-section is 3.8mm X 6.35mm so a single turn would be about 20mm in length assuming zero wire diameter. Thus 18 turns is going to be 0.4m, roughly. We therefore need wire that's 75mohm/m and that works out to be 23AWG, allowing for terminations.
Sounds well. How do you connect them?
(But as you mentioned, CM noise is a different story, and usually this is the more critical coupling mode.)
From the speakers - treble characteristics, strident, etch, course etc.
I do listen any music that I like, regardless of the recording quality so problem could be starting from here.
The Class D amps that I have are very revealing of the recording detail and ambient type and I love this regardless.
I have no absolute conclusion yet, subjective or objective but planning on lots of experimenting, next is listening to CD vs LP through Class D with SMPS and Linear supplies.