I'm currently looking at using switch mode power supply 'bricks' for my active loudspeaker project, and have a concern about permitted load capacitance. The modules I have looked at specify no more than a few 10s of uF, otherwise they may be unstable or fail to start-up.
Does anyone have any experience of this ? I am concerned that the low reservoir capacitance will be a problem on large load current transients, and that the rapid decay of the supply on turn-off will result in loud 'pops'. (With a normal linear supply you usually have a few 1000 uF to take care of both issues).
Does anyone have any experience of this ? I am concerned that the low reservoir capacitance will be a problem on large load current transients, and that the rapid decay of the supply on turn-off will result in loud 'pops'. (With a normal linear supply you usually have a few 1000 uF to take care of both issues).
You should be OK if you use CRCRC approach. The R's will reduce the turn on transient current that might trip the PSU. You won't need 10000s of uF as the switching frequency is very high. You might get away with a few 100uF between the resistors.
Thanks for that.
I wasn't thinking so much about the switching ripple, but rather the ripple due to load current. For example, suppose I have a 48V supply and my amp is designed for +/-20V into 4 ohms (50W). The supply current waveform will be a half-wave rectified sinewave with a peak value of 5A and an average of about 1.8A. Using a small reservoir cap, I think I will need a SMPS rated at 5A continuous (240W), whereas a large cap would smooth out the current pulses, reducing the power requirements to about 90W. But the supply might oscillate !
I wasn't thinking so much about the switching ripple, but rather the ripple due to load current. For example, suppose I have a 48V supply and my amp is designed for +/-20V into 4 ohms (50W). The supply current waveform will be a half-wave rectified sinewave with a peak value of 5A and an average of about 1.8A. Using a small reservoir cap, I think I will need a SMPS rated at 5A continuous (240W), whereas a large cap would smooth out the current pulses, reducing the power requirements to about 90W. But the supply might oscillate !
There are several supplies for audio that has no issues with extra capacitance, for startup surge reduction you can connect them (the extra capacitor bank) through an inductor, that way you also reduce the high frequency ripple.
You also change (increase) the output impedance of the power supply when you put an inductor in series.
I have no issue putting 20,000uF on my 350W 24V SMPS. I do have a few uH inductor from output to the cap.
No you don't, as the extra banks will define the impedance with the ESR,
In my power amp 56v I use 32x4700uF caps (I use 2x56V medical Resonant supplies pr channel) the caps are orginised in sets of 2 right next to the Power pin on each of the 16 output devices. On the button of my PCB I have 4,7uf stack foil and 0,47 ohm metal strip Remitter.
depends on the topology of the SMPS
is it regulated? if yes then why are you adding such large C , Re this isn't your dads 120 Hz filter the outputs are being recharged at 40KHz plus AND most importantly under control of feedback loops. ( the outputs caps are designed to a specific size and esr ) also adding large LCs at the output will with a high probability, actually make transient behavior much worse. use smaller decoupling caps at the ends of the loads , if any extra dynamic control is needed. Arbitrary design decisions are dumb.
is it regulated? if yes then why are you adding such large C , Re this isn't your dads 120 Hz filter the outputs are being recharged at 40KHz plus AND most importantly under control of feedback loops. ( the outputs caps are designed to a specific size and esr ) also adding large LCs at the output will with a high probability, actually make transient behavior much worse. use smaller decoupling caps at the ends of the loads , if any extra dynamic control is needed. Arbitrary design decisions are dumb.
Try build it and listen, SMPS are good as you don't have rail sagging, extra capacitance makes them even better.
agreed.
Many people do not understand the difference a smps makes to conventional power supplies in terms of reservoir caps.
In a conventional power supply you find them behind the rectifier on the secondary side.
In an smps you find them behind the rectifier on the primary side.
And yes, in most cases stuffing smps with big additional secondary caps is counter productive.
Last edited:
Nope, electrolytics have ESL as well as ESR. Here's the standard capacitor model, and impedance vs frequency from a paper by Cornell Dublier:
With an SMPS you have to deal with radiation with a lot of power from the low tens of kHz -- this can cause all the bipolar junctions in your circuit to conduct when you don't want them too. Analog Devices has a paper on this, but i am too lazy to find it right now.
The control loop of an SMPS isn't necessarily designed to be audio friendly either.
Attachments
Just build and listen,
With the extra supply you increase the transient current capability far beyond the current rating of the Smps, this is a clear fact and a clear benefit. Very nice with regulation but that's totally in vain if you rail collapse af 5A.. You really need at lot more peak current than that.
You need the inductor to filter the 200 mV or so 200 KHz smps ripple.
Actually I use resonant technology supplies, they switch softer and you have less noise from squares to deal with.
With the extra supply you increase the transient current capability far beyond the current rating of the Smps, this is a clear fact and a clear benefit. Very nice with regulation but that's totally in vain if you rail collapse af 5A.. You really need at lot more peak current than that.
You need the inductor to filter the 200 mV or so 200 KHz smps ripple.
Actually I use resonant technology supplies, they switch softer and you have less noise from squares to deal with.
Last edited:
I am glad to see this thread going. I never understand why people keep saying SMPS is no good. My main issue is they get expensive when you want above 24V. I got my 2 24V 15A supplies for $47 total including shipping!!! I asked this question, no body give me a reasonable answers.
I believe you want inductor in series and drive a bank of capacitor to filter out the glitches from the SMPS, the caps after further filter the glitches and serve as reservoir for peak current.
I am using these two SMPS in my test platform for my heavily biased Class AB with a big Class A region. I use 160mF total for both rails. I have no problem power up at all. I put so much cap so I can easily switch to the toroidal transformer and rectifiers to go 40V. The SMPS have no problem power up.
I believe you want inductor in series and drive a bank of capacitor to filter out the glitches from the SMPS, the caps after further filter the glitches and serve as reservoir for peak current.
I am using these two SMPS in my test platform for my heavily biased Class AB with a big Class A region. I use 160mF total for both rails. I have no problem power up at all. I put so much cap so I can easily switch to the toroidal transformer and rectifiers to go 40V. The SMPS have no problem power up.
There is the problem with SMPSs and audio amp loads. A class AB amplifier draws a peak current of over three times the average. With a sine wave. For typical use the ratio is even higher. They draw power in little spurts, first off one rail, then the other. Switchers are more comfortable (and rated) for constant DC loads. If you have a 600 watt amp at 75% class AB efficency you buy an 800 watt supply, right? WRONG. Unless it's good for 2400 watts (at least in the short term) you won't get 600 watts at low frequency. And when switchers limit they do not do so gracefully. The stiffening caps will help (like in a car stereo install) but the real solution is the expense of a LOT bigger supply.
its true most SMPS don't have a smart current limit.
for sine waves Ipeak is 1.4 times I(ave). certainty you might expect some mythical 3 times figure for figuring the SOA of the OPT for some added engineering safety margins? even if the OP is overbuilt it would be of some advantage for the power supply to protect it? CL reset and repeat is easy compared to replacing rows of transistors.
it might be useful for the power amp to sense low speaker impedance and command the SMPS to lower the Vrail and raise Ipeak accordingly. Then SMPS primary CL senses ave power correctly.
for sine waves Ipeak is 1.4 times I(ave). certainty you might expect some mythical 3 times figure for figuring the SOA of the OPT for some added engineering safety margins? even if the OP is overbuilt it would be of some advantage for the power supply to protect it? CL reset and repeat is easy compared to replacing rows of transistors.
it might be useful for the power amp to sense low speaker impedance and command the SMPS to lower the Vrail and raise Ipeak accordingly. Then SMPS primary CL senses ave power correctly.
Last edited:
Alan, you can stack your supplies and get 48V and a lot of amperes. I have made 2x72V by using 4 x 36V medical supplies. Solution do the same. They add close to 1F extra caps loaded thru a coil. Try google and see a picture of their gigantic power supply for the 700 series power amp
That's why it's a good practice to have reservoir caps. You don't draw heavy current for extended period of time. If that happens, even the transformer can collapse. It is quite cheap to get a 20A+ SMPS now a days.
Good idea. It is still going to be cheaper than transformer/rectifier/large caps.
You can even use low voltage cap across each supply as it will serve as good divider.
My application is for 2-way active speakers, and ~50W per driver (4 ohms). In principle there could be a full scale transient on both woofer and tweeter amps at the same time, giving a peak current of 10A (at 48V). So I either need a PSU rated at 480W continuous, or I need to find a supply which is tolerant of a large reservoir cap. As some have mentioned, a series inductor before the cap would help, although it would need to be quite big to be effective at low audio frequencies.
I'd be interested in knowing what brands of PSU people use which would be suitable (and available in the UK).
I'd be interested in knowing what brands of PSU people use which would be suitable (and available in the UK).
I find the discussion very interesting. My question is how come I never see high end power amplifier using SMPS? You can buy 36V or 48V supply really cheap, just stack it up. Most important thing is they are so much lighter. Amps like Krell are over 100lbs. You can reduce half the weight of the amp using SMPS.
You don't need to use as many filter caps that cost money also.
Worry about collapsing, use SMPS with higher current rating. I see 48V 10A SMPS for like $55 on ebay.
You don't need to use as many filter caps that cost money also.
Worry about collapsing, use SMPS with higher current rating. I see 48V 10A SMPS for like $55 on ebay.
- Status
- Not open for further replies.