Xilence XP600 600w ATX PSU - No PWM Controller on primary side - how does it work?

[dicky96]

Hi guys
I have a faulty Xilence ATX PSU here

The fault basically is I have two short circuit power transistors on the primary side of the large SMPS transformer - they are 2SC3320.

There is another smaller transistor on the primary side C3866 which tests ok

The PSU has three transformers - it looks like the 2SC3320's drive the big one and the C3866 drives one of the small ones. There is a separate PFC PCB with it's own controller and MoSFETS

There is also a short on the 3.3V output and a near short on 5V - I suspect this probably blew the two 2SC3320

However, What has me interested (not for the repair but for increasing my own knowledge) is how this PSU actually works

The only 'connection' between HV and LV sides are the three SMPS transformers and one opto isolator

There is no PWM controller on the primary side. Normally I would expect to find two - one for the 'big' transformer and another for 5VSTB

The only PWM controller I can see is on the low voltage side KA7500C. There is another 8pin IC nearby, but the datasheet says this is a Diode array. SG6510D

It is concievable (to me) that either the opto-isolator or one of the small transformers provides drive from the KA7500C to the 2SC3320 pair. In any case no feedback via an opto-isolator would be needed as the controller appears to be on the low voltage side

And the other small transformer could provide 5VSB and also power the KA7500C - but this begs the question, where is the controller for that SMPS transformer? I did also check just to see if it is a small linear transformer connected to 220V input but that also appears not to be the case - it does appear to go to the C3866 but there is no IC driver here, just another small NPN transistor part 8055S and a few resistors/capacitors. The opto-isolator transistor also seems to connect into this area.

So possibly I am quite wrong in my guess how it works - the main thing I would like to know is how does the PWM controller on the low voltage side get it's power and why is there nothing like a 'TNY' chip or similar to drive the Standby supply?


See pics.

[SMDFlea]

Re: Xilence XP600 600w ATX PSU - No PWM Controller on primary side - how does it work

A diode array wouldn`t have a notch on it ,might be this https://cdn.badcaps-static.com/pdfs/...2c03b8ed95.pdf

[dicky96]

Re: Xilence XP600 600w ATX PSU - No PWM Controller on primary side - how does it work

Hi SMDFlea
Yeah that could be the other IC, but what really puzzles me is how this PSU starts up.

The only PWM controller (apart from PFC) I can see is on the LV side. There are three transformers and on opto-isolater connecting between LV and HV side

As I said it seems reasonable that I have one (large) transformer driving the 3.3V, 5V and 12V rails, one small transformer driving the 5VSB and power to the PWM controller, and the third transformer driving the Base signal from the PWM IC to the Bipolar transistors to power the large SMPS transformer

There is also an opto-sisolator though the output voltage control feedback would not be required to pass from LV to HV side if the main PWM controller was on the LV side so it is presumably doing something else

All the above would make sense to me if the STBY/PWM PSU had a controller IC on the HV side - but is doesn't.

And that is why I can't work out how it works - unless there is a discrete oscillator circuit somehow generating the drive to the 5VSB/PWM controller Vcc transformer (there are a few resistors/capacitors and transistors around there - including the opto-isolator transistor.

But really I am puzzled as to how it works.

[PeteS in CA]

Re: Xilence XP600 600w ATX PSU - No PWM Controller on primary side - how does it work

The KA7500C is the PWM, basically a TL494 clone. You have three transformers. One is the main output transformer, one is the Standby transformer, and the third is probably the base drive transformer for the half bridge inverter transistors, the 2SC3320s. The "C3866" is probably actually a 2SC3866, probably for the Standby supply.

The inverter base drive circuit may be self-oscillating, generating Vcc for the KA7500C, which when it comes up takes over control of base drive to the inverter transistors. Or the Standby circuit may provide Vcc for the KA7500C.

[momaka]

Re: Xilence XP600 600w ATX PSU - No PWM Controller on primary side - how does it work

And that is why I can't work out how it works - unless there is a discrete oscillator circuit somehow generating the drive to the 5VSB/PWM controller Vcc transformer (there are a few resistors/capacitors and transistors around there - including the opto-isolator transistor.

That's exactly what you have!

These are called self-oscillating designs. Often they are implemented either with just one transistor on the primary side (and therefore no feedback, which will be evident by a lack of an optocoupler) or with 2 transistors on the primary side (which usually does have feedback from an optocoupler). The latter is usually a lot more prevalent and we/I/me usually call it 2-transistor self-oscillating design, or just 2-transistor 5VSB for short.

Both 1 and 2-transistor 5VSB designs are very crude in terms of protections - that is, they really don't have any! (Therefore, make sure to recap the 5VSB output caps and "startup cap" - if any - with good reliable brands, or the failure of the 5VSB can be catastrophic to both the PSU and attached PC hardware.) However, in the case of a 1-transistor design, it always uses some kind of a linear regulator (typically 7805 for 5VSB) on the secondary side, because it doesn't have feedback and cannot regulate the voltage by itself. In that regard, 1-transistor 5VSB circuit really does have a slightly better protection, as it will have whatever protection the linear regulator might offer. But in the case of 2-transistor 5VSB, there is none, aside from power self-limiting... which is often say way too high or not implemented at all. Thus, any kind of malfunction on the output of a 2-transistor 5VSB will eventually result in a meltdown.

On the other hand, the one redeeming quality of self-oscillating designs is that they are built entirely from off-the-shelf parts. So even 20 years down the road, you should still be able to find parts for them. In contrast, 5VSB circuits done with PWM ICs or more often now PWM-FET combo ("offline switch") may become hard to find replacements after many years. Obviously, ones like the TNY- and TOP- family have been around for a while and probably still will be. But anything more "exotic" might not stick around for too long.

Anyways... regardless of which self-oscillating design you have... for bridge topology main PS with a voltage control IC (such as KA7500, TL494, SG6105, SDC2921, ATX2005, AT2005b, and etc.), the 5VSB circuit's transformer will also have a secondary side auxiliary rail (or sec. aux., for short) that provides unregulated voltage to the PWM IC and any other logic ICs used for the protections and PS-ON signal.

Both 1 and 2 -transistor 5VSB circuits will also have a primary side auxiliary rail (pri. aux., for short) that is used for driving OFF the main (big) switching 5VSB transistor.

I've posted the schematics of a few of these 2-transistor 5VSB designs from the various PSUs I have worked on.

Here is one of a HEC Orion HP585d that I posted in the PSU pictorial thread:
https://www.badcaps.net/forum/attach...1&d=1578459617
... and it's worth noting that it's almost an exact copy of a Delta DPS-300ab-15b (that I haven't posted yet, but should be someday soon.)

And here is another 2-transistor design like the above one, but this one with 2x BJTs, rather than a MOSFET and BJT, like the HEC Orion above:
https://www.badcaps.net/forum/attach...9&d=1443215464

There is also a short on the 3.3V output and a near short on 5V

Double-check that.
Some of the cheaper tier PSUs (especially anything based on the now ancient half-bridge design) may have low-resistance "minimum load" / "dummy" load resistors on the outputs to keep the PSU working when there isn't much of a load on the PSU. It depends on the output filter design, though, as not all PSUs will have them, and the values for these load resistors can vary quite a bit between designs. I've seen them as low as 4.7 Ohms on the 3.3 and 10 Ohms on the 5V rail - typically in big 2-5 Watt resistors too, so that they can dissipate the heat (and they will run very hot!)

So if your PSU has anything like that, just beware, as these could trick you into thinking that's a "short-circuit" (especially when testing with continuity or diode check), when in reality it isn't.

There is no PWM controller on the primary side. Normally I would expect to find two - one for the 'big' transformer and another for 5VSTB

For bridge -type topologies (half-bridge, full bridge, push-pull) with BJTs, the controller will almost always be on the secondary side.

If you're further interested on that topic, you can see a partial PSU schematic just of the driver circuit (for the main PS) that I did for the PSU in this thread (it's in post #8) while trying to fix it.

It is concievable (to me) that either the opto-isolator or one of the small transformers provides drive from the KA7500C to the 2SC3320 pair.

That single optocoupler is for the 5VSB. If you trace the 5VSB output, you will see that it feedbacks through that optocoupler through a 431 shunt. On the other side of the opto (output / primary) will be driving a small transistor that drives the bigger transistor for the 5VSB (hence, 2-transistor 5VSB.)

[dicky96]

Re: Xilence XP600 600w ATX PSU - No PWM Controller on primary side - how does it work

Hi PeteS CA & Momaka thanks for the extensive info - like I said I didn't need it to diagnose the fault but I did want to understand how it works. It seems my best guess supposition was correct anyway

I also came to the conclusion there probably is not a short on the 3.3V and 5V outputs - eventually by comparing with another working PSU (different type but it gave me enough to go on)

This Xilence PSU is now the topic of one of my videos where I describe how the circuit works based on my best guess before this your reply (especially PFC section) - and diagnose the problem hopefully in enough detail that beginners can follow it.

The only slightly weird thing I have now is an unusual reading in diode mode across the bridge rectifier when reverse biased - but if you watch the video you will see I make a note of this, but decide to not worry about it at the moment

Anyway if you liked my YT stuff, or you would like to critique it and help me improve, here is the video

https://www.youtube.com/watch?v=f7NmQHH-xUg

[PeteS in CA]

Re: Xilence XP600 600w ATX PSU - No PWM Controller on primary side - how does it work

That's exactly what you have!

These are called self-oscillating designs. Often they are implemented either with just one transistor on the primary side (and therefore no feedback, which will be evident by a lack of an optocoupler) or with 2 transistors on the primary side (which usually does have feedback from an optocoupler). The latter is usually a lot more prevalent and we/I/me usually call it 2-transistor self-oscillating design, or just 2-transistor 5VSB for short.

Both 1 and 2-transistor 5VSB designs are very crude in terms of protections - that is, they really don't have any! (Therefore, make sure to recap the 5VSB output caps and "startup cap" - if any - with good reliable brands, or the failure of the 5VSB can be catastrophic to both the PSU and attached PC hardware.) ... But in the case of 2-transistor 5VSB, there is none, aside from power self-limiting... which is often say way too high or not implemented at all. Thus, any kind of malfunction on the output of a 2-transistor 5VSB will eventually result in a meltdown.

On the other hand, the one redeeming quality of self-oscillating designs is that they are built entirely from off-the-shelf parts. So even 20 years down the road, you should still be able to find parts for them. In contrast, 5VSB circuits done with PWM ICs or more often now PWM-FET combo ("offline switch") may become hard to find replacements after many years. Obviously, ones like the TNY- and TOP- family have been around for a while and probably still will be. But anything more "exotic" might not stick around for too long.
...

Self-oscillating flyback designs have been around for a long time. They were already in use when I got my first power supply job back in 1980. Examples from that era would be the Apple II+ power supply, made by Astec, or Boschert's OL25 models. In the early-mid 80s Boschert (Computer Products) brought out a whole series of newer models ranging in power from 30W through 165W (180W if air-cooled). These all had resistors in the switch transistor emitter circuit that provided power limiting (not very precise, since the P/Ss were rated to work down to 85 VAC (and usually worked down to 75 VAC), and had an OVP crowbar, usually for the 5V output (able to sink enough current to trip the power limit). In the later SL## and XL## models the 12V or 15V windings were connected to the 5V winding to improve regulation some.