Help with LM3875 and SMPS

[recca]

I just ordered my LM3875 Kit and am looking to build it using a SMPS for the PSU. I'm thinking of using two of these supplies . Are these suitable as PSU's for the 3875? What, if anything, do I need to put in between the PSU and amp? Caps, LPF?
Mean Well SP-150

[N-Channel]

Recca,

Are you asking if you can use one for each channel? Or are you asking if you could use two of them in series, one for the positive rail, and the other for the negative rail?

If the latter, I would advise against doing this, as most likely, the (-) terminals of the supplies are tied to earth ground, and tying the (+) terminal of one to the (-) terminal of the other could create a dead short and make bad things happen.

Even if the supplies' outputs are floating, allowing this type of arrangement, I would still advise against it for two reasons:

1) There would be separate feedback loops for the(+) and (-) rails, creating conditions for the possibility of a DC offset at the Amp's output, and;

2) There are two nearly identical switching frequencies coming from the PSUs. If their switching frequencies are not synchronized, then a beat frequency, or the difference between the two, will be generated, most likely in the audio range.

I would advise looking for an SMPS with bipolar outputs, like +/- 15V or higher. If you can not find one, there are several threads in the Power Supplies forum that help with modifying existing SMPS to give your desired voltages, or to "roll your own".

If you choose the latter, PLEASE BE ADVISED: You will be working with LETHAL AC and DC voltages, and you MUST MUST MUST take the appropriate cautions! Use an AC Isolation transformer to isolate your supply from the AC mains for safety, test the supply out first with low voltages, with light loads, etc.

One of our regular power supply contributors to this forum, EVA, has a boatload of very sage advice, as she has designed many SMPSs, and, I believe, this is her career.

I do like your desire to use an SMPS with an active PFC'ed front-end, and if you search the Power Supplies forum, you will find a few threads about Active PFC'ing. I have designed and prototyped quite a few SMPSs myself, including a couple of AC-DC switchers, but have, as of yet, to successfully tie my active PFC to my main switcher.

Hope this helps, Just my $0.02 worth.

Cheers,

Steve

[recca]

Thanks for the reply. I was asking if I could use one for each channel. Although, I don't really need that much power, I was under the assumption that the 3875's like to have at least 24v 6a from the PSU. If not please let me know what would be optimal.

Damn, I didn't realize that the 3875 and a V+ and V- tab. I guess I better keep looking for another PSU.

Anythoughts about caps or lpf in between the psu and amp?

[N-Channel]

Yep, it does have (+) and (-) pwr pins. As for the LPF between the PSU and the chips, it is probably best to roll your own.

I would go with a C-L-C (PI) filter. No idea on the compoonent values, but for the caps, use both 0.1uF, 1uF and perhaps, 1000uF to hit all the frequencies. A good brand would be Panasonic's FC-series of high-frequency, Low-ESR Electrolytics. These are available from Digi-Key.

For the inductor, you should actually use a coupled inductor to give good cross-regulation between the (+) and (-) outputs. A good source for getting such an inductor would be from a dead AT or ATX power supply. The BIG yellow toroid near the outputs is a good candidate, as its core material (#26 powdered-iron) is ideally suited for DC Chokes, and is already a coupled inductor. Obviuously, you will need to make sure that both the (+) qand (-) windings have the same turns. A good way to insure identical windings on the same core is to wind them bi-filar. Be careful using thick-gauge wire, as it is not easy to work in the tight winding process.

[recca]

Thanks for the info re: Pi Filter. I'll look into it.

As far as the PSU goes, I looked at the diagram on it and there are both V+ and V- outputs on the PSU so I think it will be okay. Right?

[jackinnj]

why not get one 48VDC supply ? -- this is a pretty common voltage -- National has a schematic for unipolar operation of the LM3875, LM3886, LM4780 -- you might find a Meanwell or Power-One devices in this range on EBay.

If you wanted you could modify an ATX supply for a higher voltage -- it would entail removing the supervisory circuitry, rewinding the transformer etc. Let me know if you would like the ATX mod article from QEX.

N-Channel has probably seen the article in EDN by Jim Williams in which he recommends a linear post-regulator AND a choke -- this consumes some energy in the linear regulator's pass element -- but the noise is reduced greatly.

[N-Channel]

Jack,

I have not seen the EDN article you refer to. I do agree with you that a linear post-regulator will significantly reduce noise on the outputs (provided proper shielding is used), however at the expense of efficiency.

why not get one 48VDC supply ? -- this is a pretty common voltage -- National has a schematic for unipolar operation of the LM3875, LM3886, LM4780 -- you might find a Meanwell or Power-One devices in this range on EBay

I have checked M-W's website, and the only dual-outputs are +5v/+12V, and +5V+24V; not +/-15V, as would be needed. I agree that a single +48V output would be worth looking at, w/AC coupling, but why not modify the +48V output to +/-24V instead? Or +/-48V? If the topology is a half-bridge, then the output rectifier is almost certainly a center-tap full-wave rectifier, using a common-cathode dual-diode. If you connect a common-anode dual diode with the same filtering circuitry, you could recover the negative voltages off the main switching xfmr, and get your symmetric +/- outputs.

Recca- the QEX article Jack refers to is a very good article on how to modify an AT or ATX supply, but it also goes into some depth as to how they work. It is not as obvious as it seems. Sure, the TL494 PWM chip controls the two NPNs or N-Channel MOSFETs, but the way the driver transformer works is perhaps more critical than the control circuitry or the main xfmr.

Jack, can you post the link to that article? I forgot it and only have the hard copy.

'73, de N8XO!

Steve

[recca]

Thanks everyone for your help. I think I definitely need to figure out how these work. In the meantime, I just need to be clear about the +/- voltage rails. The single output SP-150 produces ~6.3A at 24V. On the PSU itself there are taps for DC+ and DC-. I was under the assumption that I could connect the PSU to the amp in this manner:

PSU -----> AMP
DC+ -----> V+
DC- ------> V-

But based on what I'm getting from the posts, I'll need a seperate PSU for each polarity of V-in on the amp:

PSU -------------------> AMP
PSU#1 (DC+) --------> V+
PSU#2 (DC-) --------> V-

Is this correct? Where does the 0 come from? AC ground?

As far as the higher voltage PSU supplying both amps, what rating needs to be doubled to get enough power. I was looking at 24v 14a instead of 48v 6a. Most of these PSU's have two taps each for the DC+ and DC- would it be possible to connect an amp to each of these taps or would the PSU not like that?

This PDF has the Pin layout for the SP150

[Nuuk]

recca, you may be interested in my comments here !

I'm listening to a 3875 IGC at this moment, powered by two of the Skynet 8080 modules, one for each channel at +/-12 volts and with reasonably efficient speakers, it sings!

[recca]

Nuuk - Thanks for the reply. Actually I've gone through your site numerous times before even buying the GC. Your experience with the SMPS swayed me to go in that direction too. The problem I'm having right now is that with the Skynet's the 12v+ and the 12v- have 2 different current ratings so it seems like its a triple output PSU. Whereas the Wellbeing SP-150 is a single output. Do you think a single Wellbeing SP-150 will be able to power a single channel of a monoblock LM3785 NIGC? Did you do anything for a noise filter on the Skynets?

Thanks for input!