diodes have become dirt cheap nowadays if 1n4003's are chap, so is a 10A ampere bridge...
i built a 16kt88 tube amp years ago wherein i used a 16A 600v bridge diode for B+ psu...
Attachments
Not to mention big inrush current when turned on and caps charging.
You want ratings many times of classA bias.
You want ratings many times of classA bias.
There definitely is significant inrush current. Takes ~ 15 seconds for DCV to settle to near nominal value.
I ran the amp with the hood off for an hour. Transformer is warm but not hot. It was getting warmed up along with everything else inside. The real heat generators are the outputs and regulator. Definitely need to increase the venting.
I replaced the 1n4003's with 3A 200V diodes. DCV and ripple are unchanged. I then added 0.47Ω between diodes and filter cap. Lowered RMS ripple from 120mV-ish to 85mV. It definitely sounds smoother, less pushed forward than before.
It's a great-sounding little amp. I think it could be even better if the regulator didn't have to deal with so much ripple. I found a 15V 3.5A switcher in the warehouse today and if it checks out, will replace the xfmr and caps with it this weekend and test my theory..
I ran the amp with the hood off for an hour. Transformer is warm but not hot. It was getting warmed up along with everything else inside. The real heat generators are the outputs and regulator. Definitely need to increase the venting.
I replaced the 1n4003's with 3A 200V diodes. DCV and ripple are unchanged. I then added 0.47Ω between diodes and filter cap. Lowered RMS ripple from 120mV-ish to 85mV. It definitely sounds smoother, less pushed forward than before.
It's a great-sounding little amp. I think it could be even better if the regulator didn't have to deal with so much ripple. I found a 15V 3.5A switcher in the warehouse today and if it checks out, will replace the xfmr and caps with it this weekend and test my theory..
Inrush current is one reason you don’t want to undersize or skimp on diodes. Even if running 2 amps thru a 1N4003 doesn’t kill it, repeated 50 amp turn on surges eventually will.
The 15V 3.5A switcher I have here (Phihong) looks like new but has no output. Unfortunate, because if would fit nicely inside the amp. Given how cheap these things are, it's probably not worth putting time into fixing it. So I'll need to order one in.
I do have a couple 15V indiustrial switchers (10A Lambda, 5A Boschert) but they're physically way too large. However, testing them brought up the issue of output noise voltage. They both are relatively quiet into a 100kHz-BW meter, 500uV and 700uV respectively. And most of that is at their 100kHz switching freq. Looking through data sheeets of Mouser's volumnous stock of 15V 2-4A switchers (nearly a hundred), most of them switch at 65-70kHz, many give no output noise spec, and the ones that do are in the 120mV range (20MHz BW). If most of that noise is in the audio BW then it's no better than what the filtered rectified xfmr is giving.
Does anyone have experience with these small enclosed and open-frame switchers? What does their noise look like in the audio range?
I do have a couple 15V indiustrial switchers (10A Lambda, 5A Boschert) but they're physically way too large. However, testing them brought up the issue of output noise voltage. They both are relatively quiet into a 100kHz-BW meter, 500uV and 700uV respectively. And most of that is at their 100kHz switching freq. Looking through data sheeets of Mouser's volumnous stock of 15V 2-4A switchers (nearly a hundred), most of them switch at 65-70kHz, many give no output noise spec, and the ones that do are in the 120mV range (20MHz BW). If most of that noise is in the audio BW then it's no better than what the filtered rectified xfmr is giving.
Does anyone have experience with these small enclosed and open-frame switchers? What does their noise look like in the audio range?
15 seconds is way too long...
consider the 50 hz line, how many cycles of current will it take for inrush current to dampen down?
with 20ms, even if it takes 100 cycles,(which is highly unlikely) that is just 2000ms or just 2 seconds...
5 cycles or 100ms is all it takes....more than enough for me...
@TonyTecson The long settling time is partly due to a large capacitance on the LT1085 adjust pin.
I hooked up the outboard Boschert switcher. Yes, it works fine as a raw DC supply. Quality in the bass-midbass is improved, noticeably but not drastically. All things considered, it's worth getting a switcher that fits inside the case.
I'm still a bit uncomfortable with the 120mV noise spec on the small switchers. Surely they can't be that bad...
I hooked up the outboard Boschert switcher. Yes, it works fine as a raw DC supply. Quality in the bass-midbass is improved, noticeably but not drastically. All things considered, it's worth getting a switcher that fits inside the case.
I'm still a bit uncomfortable with the 120mV noise spec on the small switchers. Surely they can't be that bad...
smps ripples are very much high up there in the audio band, 70khz, and you will never hear it on your speakers, so why not stop measuring and start listening to music?
So I got an XP Power LCS50US15 switcher 15V @ 3.4A . Looking at its noise at 1A output, the switching freq (65kHz) is nowhere to be seen, buried in the noise floor. The dominant noise components are at 500kHz and 1.5MHz easy enuf to filter. Not bad for such a teeny, inexpensive unit.
A typo in my previous post , 1.5MHz should be 1.5kHz. It is the dominant noise component.
After installing it into the amp, I can not recommend this supply. The reason the switching freq was "buried in the noise floor" is because the noise is so high. Broad low-freq noise below 100Hz. A 5mV rms spike at 1.5kHz. The 60kHz switching freq is 2.5mV, and contaminates the ground, feeding straight through the regulator into the amps. With 1500uF maximum capacitor on the output allowed, none of this can be filtered satisfactorily. Compared to the Boschert industrial switcher, the sound quality is worse.
Can anyone recommend a good, compact, less-noisy switcher?
After installing it into the amp, I can not recommend this supply. The reason the switching freq was "buried in the noise floor" is because the noise is so high. Broad low-freq noise below 100Hz. A 5mV rms spike at 1.5kHz. The 60kHz switching freq is 2.5mV, and contaminates the ground, feeding straight through the regulator into the amps. With 1500uF maximum capacitor on the output allowed, none of this can be filtered satisfactorily. Compared to the Boschert industrial switcher, the sound quality is worse.
Can anyone recommend a good, compact, less-noisy switcher?
OK, will do. The RS-50-15 is equivalent output, a tad larger but will fit. How is one supposed to know if in fact the noise is better? The specs are identical to the XP Power one. The noise specs on all of these switchers across all brands vary so little, they're useless.
I'd use the Boschert (I have 5 of 'em) but it's too large.
I'd use the Boschert (I have 5 of 'em) but it's too large.
Call them up and ask. It's also possible to add an output filter to a SMPS. Maybe this one could be scaled up a bit.
https://diyaudiostore.com/collections/power-supplies-accessories/products/smps-dc-filter-p089zb-kit
Zachary Potter
Electronic Engineer｜VAC
T: 510-683-8886 ext.216 or 816-601-4934
MEAN WELL USA, INC. ｜w: www.meanwellusa.com
E: zachary.potter@meanwellusa.com
Last edited:
Thanks Ray, I'll definitely call them.
I looked at that filter, it won't do anything for the 1.7k spike and its harmonics, which is the most audible anomaly.
I looked at that filter, it won't do anything for the 1.7k spike and its harmonics, which is the most audible anomaly.
Yeah, I dunno, it's the worst noise spectra I've ever seen from a switcher. Might be ok for non-audio digital gear.
I'm thinking we need a sticky thread where folks can post noise measurements of a switcher when they get them. That would be the best way to take some uncertainty out of the process, because the specs are useless. The noise rating is like a bridge built so high that anything can pass under it.
I'm thinking we need a sticky thread where folks can post noise measurements of a switcher when they get them. That would be the best way to take some uncertainty out of the process, because the specs are useless. The noise rating is like a bridge built so high that anything can pass under it.