Bipolar SMPS with Manufacturer Recommended Circuit

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

Potato Cakes

Well-known member
Joined
Jul 1, 2014
Messages
2,258
Location
Nashville, TN
Hello, everyone.

After participating in several SMPS implementation discussions, I saw in one of the discussions where it was mentioned that the data sheets show the recommended circuit for that particular model. I have attached a modified schematic that would that a SMPS with the recommended circuit additions and make it into a bipolar PSU. I'm curious if it's better to connect C1, C2, and TVS as shown in the PDF or should just one set of those components be strapped across +VDC and -VDC.

Thanks!

Paul
 

Attachments

  • Bipolar SMPS w: Manufacturer Recommendations.pdf
    805 KB · Views: 9
Hello everyone,

This is a bust. I get the same level of noise with or without the filter circuit. I have used these same PSUs in the same bipolar configuration with another audio design and got the same noise, but was able to make it very quiet by strapping a large capacitor (2200uF/63V) across +/-VDC. The problem is that the PSU would keep power cycling as they were going into protect mode and then eventually would stay powered on. For future projects for clients, this power up behavior is not acceptable. I know that there is a simple elegant solution to this problem that has been solved many times over, but I have yet to find it. When I do I will let everyone know as these particular PSUs are very compact and are sealed (https://www.mouser.com/ProductDetail/CUI-Inc/PSK-S15C-24-T?qs=vLWxofP3U2zEsXRBFAeaNw==).

I'll see if CUI has any suggestions.

Thanks!

Paul
 

Attachments

  • PSU noise.jpeg
    PSU noise.jpeg
    407.2 KB · Views: 0
My next step is to get parts for a capacitance multiplier. I'm currently on tour right now so getting parts delivered to where I am going to be at a certain time is challenging. Plus, every workspace has to be makeshift.

I did download your rules and I have been going through them (thanks for posting those by the way). I am using SMPS's that are within the 80% range of the circuit's current draw. Right now I'm trying to find out how to figure out which values to use for the generic capacitance multiplier schematic that I found. I would like to to make a MOSFET version work to keep things simple.

Thanks,

Paul
 
a cap between 0V output and AC(N). Common mode noise will go right through an SMPS transformer

Common mode noise is by definition present on both hot and neutral conductors. If that noise gets through the transformer then connecting a cap between output and high side neutral would be connecting back to the source of the noise, and if anything make it worse.

So there needs to be a path for that to get to neutral.

Some offline switcher designs have a capacitor between output ground and earth or primary side reference if using a non-earthed power supply design.
Typically around 1nF (1000pF) safety rated capacitor is what you see in commercial designs, although that is usually based on passing EMI testing, and may not be optimal for low noise designs.

But it needs to be pretty big. You could try your 2200uF

I had to go walk around for a little bit and get two cups of coffee before I could calm down enough to use more polite language than I would use to your face for that suggestion.
For the sake of readers who may be inexperienced or uneducated enough to follow that suggestion I will walk through the basics.

Capacitive reactance is inversely proportional to frequency and capacitance, with the magnitude given by the formula:
|Xc| = 1/(2*pi*f*c)

Where |Xc| is the magnitude of the reactance (i.e. drop the phase information), pi is the standard geometry constant, f is the frequency you are concerned with, and c is the capacitance in Farads.
Plugging in North American power line frequency and 2200uF that gives:
Xc=1/(2*3.14*60*2200^e-6) = 1/0.83 = 1.2

So you have just suggested that someone connect an equivalent of 1.2 Ohms between output side ground and high voltage side.
Besides injecting all of the noise on the neutral line caused by I*R drop of power currents directly into the output ground, it creates a potentially fatal connection between power voltage and output ground in the case of a line/neutral swap (which is a relatively common mistake in wiring, and should be benign in properly built equipment).

To quote a recent post "Impedance is the key to understanding E" so I am more than a little surprised to see such an irresponsible, potentially deadly suggestion.
 
Hello everyone,

This is a bust. I get the same level of noise with or without the filter circuit. I have used these same PSUs in the same bipolar configuration with another audio design and got the same noise, but was able to make it very quiet by strapping a large capacitor (2200uF/63V) across +/-VDC. The problem is that the PSU would keep power cycling as they were going into protect mode and then eventually would stay powered on. For future projects for clients, this power up behavior is not acceptable. I know that there is a simple elegant solution to this problem that has been solved many times over, but I have yet to find it. When I do I will let everyone know as these particular PSUs are very compact and are sealed (https://www.mouser.com/ProductDetail/CUI-Inc/PSK-S15C-24-T?qs=vLWxofP3U2zEsXRBFAeaNw==).

I'll see if CUI has any suggestions.

Thanks!

Paul
Well, 2200 uF is above the maximum capacitive load for the 15W 24V model - on the datasheet it is listed at 800 uF.

The only noise you're getting is 220 Hz?

If you disconnect just one, do you get 220 Hz still?

If you add a dummy load does the noise go away?
 
AC Neutral is considered hot by some safety agencies, requiring double insulation for non-grounded chassis.

There are special (X,Y.?) rated capacitors for some mains connections (not that one).

be careful when messing around mains voltage.

JR
 
ome offline switcher designs have a capacitor between output ground and earth or primary side reference if using a non-earthed power supply design.
Typically around 1nF (1000pF) safety rated capacitor is what you see in commercial designs, although that is usually based on passing EMI testing, and may not be optimal for low noise designs.
Apologies. I have deleted the incorrect post.

I am mixing up my AC and DC SMPS. In a DC/DC converter, a large cap between 0V out and the negative input can provide a path for switching currents passing through the parasitic capacitance of the transformer back to the switch on the primary. Otherwise they have to go through the load and earth ground which has significant impedance which equates to ground noise. Of course this cap cannot work with an AC supply. I was looking at these things as a brick with four wires and that was stupid.
 
I get the same level of noise with or without the filter circuit.

You need to see the time domain behavior on the output (i.e. an o'scope). Those regulators are switching at 100kHz, so it is not obvious where a peak around 220Hz would come from.
Is that noise spectrum from the output of an audio device powered by those power supplies? Or did you AC couple the power supply output into an audio input to try to measure the noise directly?
 
Well, 2200 uF is above the maximum capacitive load for the 15W 24V model - on the datasheet it is listed at 800 uF.

The only noise you're getting is 220 Hz?

If you disconnect just one, do you get 220 Hz still?

If you add a dummy load does the noise go away?
Yes, 2200uF is above the data sheet spec. At the time I was throwing anything I could at the problem and this seemed to work as a temporary fix once the PSUs stabilized. The 220Hz is not in the audio circuit. Any problems with hum that I have had in the past with this circuit were around 60Hz. The noise problem that I posted above with the image of the issue is from the PSU itself and not the audio circuit. As I had mentioned I am using the exact PSUs in another build of the same audio circuit that had the exact same problem. The 2200uF across the two power supplies (+V on PSU 1 and -V PSU 2) made the entire circuit dead quiet, but they had to power cycle several times before stabilizing. Once they stayed powered on everything was fine, but as mentioned previously this is not acceptable for future client builds. I believe the capacitor multiplier route will be the answer, but I do not know the values of the parts needed yet.

Again, I'm on the road trying to get this issue sorted. I do not have my oscilloscope with me.

Thanks!

Paul
 
Last edited:
if it’s 220 Hz that may be a kind of hiccup mode without load. Put a dummy load on the circuit and check the PSU rails then.

Also see if it is just one supply or only the combination that makes the noise.
 
The 220Hz is not in the audio circuit.

OK, then at least the noise is in a range where the PSRR of most audio circuits is pretty good.
What is the reference level in that plot that showed the noise at -80dB?

I believe the capacitor multiplier route will be the answer

You could probably use that. You could also use the technique some large power amplifiers used, which is to put a resistance in line with the capacitors until they charged up, then short around the resistor. That could be done with a relay, or with MOSFET devices. Probably P channel so you just have to pull the gates low and don't need a gate supply above the power rail. Or I guess a P channel on the positive rail and an N channel on the negative rail would be what you want, so that for both you tie the gate to the input power rail with a large resistor, then you can pull that gate control to ground to turn on the FETs.
 
There are a lot of wasted components in the circuit and you cannot have two full wave restifiers connected in the way you have shown, to work the way you want.
You simply need the incoming AC through the noise filters, make one side the common output.
The other side, put two diodes - one in each 'direction' for the positive and negative rails. Put filter caps appropriate for the level of hum you will tolerate, including a 0.1uF disc ceramic capacitor across the main filter caps for each rail.
Thats it.
You can essentially make the circuit by eliminating all the excess components in the original circuit.

I have done this many times using a rectifier bridge and using just one of the AC inputs - it amounts to the same thing.
Essentially, it ends up as two half-wave recifiers supplying the two voltages.
That is it. Done.
 
There are a lot of wasted components in the circuit and you cannot have two full wave restifiers connected in the way you have shown, to work the way you want.
Woah! Woah! The square things in the schemo are not bridge rectifiers. They are AC/DC isolated converters.
You simply need the incoming AC through the noise filters, make one side the common output.
The other side, put two diodes - one in each 'direction' for the positive and negative rails. Put filter caps appropriate for the level of hum you will tolerate, including a 0.1uF disc ceramic capacitor across the main filter caps for each rail.
If I understand well, you are suggesting half-wave rectification for each rail. That would result in output voltages of about +/-160V (for 115V mains) and double for 230V mains. I don't think it's what the OP wants.
 
Last edited:
OK, then at least the noise is in a range where the PSRR of most audio circuits is pretty good.
What is the reference level in that plot that showed the noise at -80dB?



You could probably use that. You could also use the technique some large power amplifiers used, which is to put a resistance in line with the capacitors until they charged up, then short around the resistor. That could be done with a relay, or with MOSFET devices. Probably P channel so you just have to pull the gates low and don't need a gate supply above the power rail. Or I guess a P channel on the positive rail and an N channel on the negative rail would be what you want, so that for both you tie the gate to the input power rail with a large resistor, then you can pull that gate control to ground to turn on the FETs.
The noise level is relative. I have my interface set to unity gain input. I do not have accurate test equipment with me. The noise does not change with the gain of the audio circuit being tested. The -80dB show in actual practice is too high and clearly audible. For reference, the noise should be below the -120dB line with maybe some frequencies peaking above that. That is what I achieved with the 2200uF cap as described above.

Attached is a generic capacitance multiplier schematic that I found. I did see some others that recommended another resistor that is parallel with the capacitor which forms a voltage divider. I do not know which values to use yet.

Thanks!

Paul
 

Attachments

  • Capacitance multiplier generic.jpeg
    Capacitance multiplier generic.jpeg
    64.4 KB · Views: 0
Back
Top