I tested two RC filters different connected either SMPS->RCfilter->F5mPCB or SMPS->StoreFilter->RCfilter->F5mPCB.
The RC filters were 0R22 and 22mF and 0R25 and 3.3mF. The filters didn't reduce the 60Hz harmonics enough to be worth the space and added PS impedance.
The store filters seem to do an excellent job of cleaning 60 Hz harmonics and other noise from the power rails. I’m not thinking of adding an extra RC filter, just bulk C right next to the channel boards. It can be easily removed if it doesn’t help.
The Mean Well supplies and the careful star ground implementation are responsible for 60Hz harmonics being low. Also the F5m amplifier topology has a good PSRR. The store filters don't do much below about 4kHz as shown in this Bode plot.
Last edited:
The inherent noise floor of my spectral measuring system is limited by characteristics of the M-Audio AP192 sound card and the external "wiring" to the deviced being measured.
This image shows the loopback spectrum of the soundcard with wiring analogous to that when measuring the F5m. The rising noise level starting at about 40kHz is due to the noise shaping of the DAC and ADC in the AP192. This needs to be taken into account when evaluating the spectral measurements of the amplifier shown in the previous posts. As can be seen, the noise levels from the amplifier are near the underlying noise floor of the measurement system, with the exception of a few low level harmonics of 60Hz (around 25uV rms) and some low amplitude SMPS noise at around 40kHz-60kHz,
This image shows the loopback spectrum of the soundcard with wiring analogous to that when measuring the F5m. The rising noise level starting at about 40kHz is due to the noise shaping of the DAC and ADC in the AP192. This needs to be taken into account when evaluating the spectral measurements of the amplifier shown in the previous posts. As can be seen, the noise levels from the amplifier are near the underlying noise floor of the measurement system, with the exception of a few low level harmonics of 60Hz (around 25uV rms) and some low amplitude SMPS noise at around 40kHz-60kHz,
Looks like the SMPS could definitely use some bulk capacitance on their outputs. I expect that the store filter boards would help startup with minimal hiccup.
Hello Ihquam,
Nice post, very interesting.
These Mean Well power supplies (I bought two LRS-150-24) have small ceramic capacitors between the negative output and ground. Mine has two ceramic capacitors in this position. My question is if you think these small ceramic capacitors can create problems, like high frequency ground loops or something? (and thus better to be removed in case +/-24V with center " tap").
Thanks
Attachments
Those are for EMI/EMC compliance and shouldn't be removed.
@TungstenAudio or anyone else who can help with my education...
Why would adding capacitors help here?
As noted previously I used a local capacitor reservoir however, upon reflection I'm not sure it helps.
It's possible that some/many/most linear supplies are undersized and that adding a large charge reservoir provides a temporary support to reduce the amount the rails droop as the result of a large transient current draw. Also, given the low frequency of the charging pulses some filtering is likely needed to remove the line frequency and harmonic noise that could creep through.
The `ripple and noise' from the SMPS is quite low and the oscillations in the rail voltage are way above the audio range. So, IF the SMPS can meet all the potential loads in a steady state condition then perhaps a charge reservoir is unnecessary. Also, one could perhaps argue that adding anything between the output of the SMPS and the amplifier board that could hinder the current delivery (like a CapMX), might not be a good thing. I don't know. I do not mean to be argumentative, but I do want to know.
Why would adding capacitors help here?
As noted previously I used a local capacitor reservoir however, upon reflection I'm not sure it helps.
It's possible that some/many/most linear supplies are undersized and that adding a large charge reservoir provides a temporary support to reduce the amount the rails droop as the result of a large transient current draw. Also, given the low frequency of the charging pulses some filtering is likely needed to remove the line frequency and harmonic noise that could creep through.
The `ripple and noise' from the SMPS is quite low and the oscillations in the rail voltage are way above the audio range. So, IF the SMPS can meet all the potential loads in a steady state condition then perhaps a charge reservoir is unnecessary. Also, one could perhaps argue that adding anything between the output of the SMPS and the amplifier board that could hinder the current delivery (like a CapMX), might not be a good thing. I don't know. I do not mean to be argumentative, but I do want to know.
No. The filter board for MeanWells such as these is a different product and further down the list of nifty store things to be introduced in the future.
That's for down the road 🙂 Right now it's my Revel F226Be, but am looking at a Madisound full range kit. Ultimately, when I get some of my woodworking tools out of storage am looking at 1 of TG's designs.
Ha! Did I say ultimately? I mean after that!
I imagine my choice of 6.375 Db was an obvious reference to the F5M, which if I'm not mistaken is 25 watts.
I imagine my choice of 6.375 Db was an obvious reference to the F5M, which if I'm not mistaken is 25 watts.
