What happens when:
0:00 Introduction
0:10 Size comparison
0:25 What’s inside?
0:46 Building our own linear power supply
1:07 JLCPCB
2:00 The mains
2:18 Input fuse
3:29 Input switch
3:44 Transformer – Introduction
3:59 Transformer – Structure
4:23 Transformer – Magnetising current
5:35 Transformer – Reactive power
7:40 Transformer – Magnetic coupling
8:06 Transformer – Secondary winding
9:08 Transformer – Why? (isolation & voltage change)
10:30 Transformer – Secondary (load) current
11:16 Transformer – Real-world voltage and current waveforms
14:04 Sometimes it’s best to keep things simple
15:03 AC to DC – Diode
17:02 AC to DC – Full bridge rectifier
19:09 AC to DC – Split secondary
20:10 AC to DC – Output ripple
21:02 DC capacitor
23:21 Pulsed input current (bad)
24:39 Output regulation
26:19 Zener diode
27:40 Open loop linear regulator
30:44 Closed loop linear regulator
32:48 Complete circuit summary
33:29 Outro

If you do the analysis of the circuitry there is a capacitive coupling that pumps the isolated output at the oscillation frequency that leakage is coupled to the supply line and ground by the class y capacitor, if it’s not there the leakage would be superimposed on the output, a grounded supply could have a screen to stop this , but it’s 2 wire

A couple of people have asked whether it’s worth substituting the PSUs currently supplied with the Spartan 10 as they certainly feel rather ordinary. While they look like normal modern switching wall-warts, they are in fact linear transfomers with much quieter operation and far better capacitive isolation from the mains than switchers. A low leakage capacitance is important as it stops the mains coupling into the ground path which will be connected in some form to the low voltage output to derive relative power supply rails.

The first photo shows the linear type used by Classic Audio with only 30pF measured between the mains and low voltage side (in reality it’s more like 20pF due to leakage between the test leads). Very little AC current from the mains can be coupled from the high voltage to the low voltage side.

The isolation is excellent and there are no components connected between the high and low voltage sides. The low voltage low frequency AC output is then rectified and regulated to very quiet DC rails inside the Spartan 10, although it’s not really necessary as the amplifiers inside the S10 have superb rail rejection. The regulation is there mainly to protect from power surges and ensure the headroom doesn’t vary with mains voltage.

The second photo shows a switching supply commonly found with a lot of similar audio gear. Here the isolation is not great due to the need to connect a ‘suppression’ capacitor between the high and low voltage sides of the switching supply to prevent the high frequency switching transients from coupling across to the low voltage side and causing radio frequency emission from the output lead. It measures at 660pF of leakage capacitance, some 30 times higher than the linear transformer. In reality it is more than double this as it sits behind the high-voltage rectifier which skews the reading on the meter. This isn’t good news as 30+ times more HF noise from the mains can get into our ground path and our equipment will now ‘float’ on the end of this leakage capacitance if not grounded. Switching transients from the high voltage rectifier will also make their way into the ground path at a considerably greater level.

The switching supply is cheaper and doesn’t require internal regulation inside the piece of equipment it’s connected to (although such regulation is likely to be ineffective if required as the noise on the output will be high frequency and harder for a linear regulator to reject).

The external PSU for the S10 is a plain simple low frequency transformer. The power supply rails are derived within the unit itself with the rectifiers, smoothing capacitors and regulators. The internal components determine the performance of the power supply which as already mentioned is much better than it already needs to be, the transformer simply supplies the working voltage and ensures excellent isolation from the mains.

Contrary to marketing lies told by others regarding highly dubious claims of ‘transient headroom’, or ‘high instantaneous current for punch and slam’ (whatever that means!) a bigger transformer will do nothing except reduce the isolation due to the larger area between the high and low voltage sides. Anyone making these claims should be treated with suspicion as they either don’t know what they’re talking about or are just straight up making fraudulent claims.

Instantaneous current demand is met by the internal reservoir capacitors, not the external transformer which only charges the reservoirs for a few milliseconds 50 times per second. The rectifier essentially disconnects it from the reservoir and DC rails for 80% of the time! As long as the reservoirs hold up the voltage ahead of the regulators then the power supply delivers the voltage and the current. This is not a linear process and the regulator doesn’t care what the exact voltage is in the reservoir so long as it’s above 18 volts or so. In the Spartan 10 it remains around 22V at full load, ensuring an excellent margin of operation. The transformer is rated to supply the average current and then some, so the small size is quite adequate.

So there we are! Fortunately they really are the best devices for the job! You can try this one at home also if you have a capacitance meter handy. Rant over !

Low leakage current is most common. Sometimes it is possible higher leakage on device but you were well grounded.

“Out of curiosity when i change polarity of ac power cord everything becomes normal. What’s the reason?”

Many modern equipment that use swich mode power supplies have filter capacitors between mains side and chassis. Depending on equipment design there can be two capacitors (designed to be grounded equipment) or one capacitor from one of of the ungrounded power connector pins to case. Depending on mains plug polarity you have it from live to chassis (leakage) or neutral to chassis (almost no leakage).

Traditional transformer can have some leakage that can vary depending on which side of primary is on live or neutral.

“is it dangerous for human being?”

When equipment is properly designed, im good condition and built using proper parts, it is not dangerius.

Today, let’s rewind the faulty transformer of my 12V 50A 600W switching power supply. I wound it partially the original way (with some original shortcomings), just using copper instead of aluminium, and splitting the secondary into more parallel wires, leaving the original design of the primary. Of course it could be redesigned much better, fitting higher cross section of copper into it to reduce the current density. It could also use many more of thinner wires to further reduce the skin effect. The returns that waste space could be eliminated by winding each section in two layers. The secondary could be a copper strip which is more space-efficient for high current low voltage windings.

Today, the autopsy of the important components of the 24V 41.7A 1000W switching power supply from eBay. I opened the primary smoothing capacitors, which are apparently fake Nippon Chemi-Con, actually, used relabelled capacitors desoldered from e-waste (scrapped electronics). I also opened and reverse engineered the main power transformer and the GDT (gate drive transformer). I measured the winding inductances, checked the insulation safety, counted the numbers of turns and measured the wire diameters. I checked the capacitances and ESR (impedance) of all electrolytic capacitors. I also checked whether the transformers, the output inductor and the EMI inductor use copper or just copper coated aluminium (CCA). I also reverse engineered the gate drive circuitry of the power MOSFETs, quite a questionable design.

A very common PSU fault on a fairly nice power supply from a media player.

Don’t be fooled by the cheap SRBP (Synthetic Resin Bonded Paper) style PCB. It’s been designed with common sense and safety in mind to comply with UK standards.

The sizing of the diode array is probably mainly for the increased passive thermal dissipation.