Odd SMPS. Please, can you help me with it?

[HRMMI]

Hi to all
I was reading PSU forum in order to find something similar to what I am about to ask, but no luck. Though I did find out there is a lot of knowledgeable people here and that encouraged me to ask for help.
I am posting schematic of mine woofer that I am trying to repair. Actually I am not sure how this PWM works so I know I will be doing more bad then good. I need help.

I have many questions but will try to keep it short.
If you look at the schematics in the middle you will find Cx. It is charged by current source. Cx when charged is making first impulse for Q1. Fallowing line from Cx you will find Zener with question mark. Unfortunately that diode is not marked. Looks ordinary and it is not SMD component like many others on this schematics. Do to fact I don’t fully understand how this PSU works I don’t know how to calculate proper value for that ZD.

I have IF or IF ideas and I am stuck do to lack of experience with SMPS circuits. I assume circuit failed do to bad 200uF electrolyte on the input, BTW I don’t like that input at all.
In order to give as much useful info to anyone willing to help me I have put some notes on the schematics. I hope these will be understandable. This schematics is made in TinyCAD by me. I had to decode circuit. Power module was covered with tar.
Any help is greatly appreciated and if you need more info about the circuit or something is unclear just ask.

[kjlonergan]

The circuit appears to be self-resonant. The two power stages will alternately turn on and drive the output transformer. The mystery diode is part of a start up circuit to initially 'kick" the top power stage and start the oscillation.

Have you tried to contact the manufacturer? However, if they covered the circuit in tar it doesn't sound like they wanted the module to be repaired.

Kevin

[danb1974]

Put a scope onto Cx and see how the voltage varies on startup. Then we can guess about your zenner.

Or simulate the startup circuit if you are sure of the schematics.

[HRMMI]

Thanks for both replies

Thing is mine equipment is not at mine disposal for the moment, I do have only basic stuff around me and basic is not the scope, unfortunately.
This is mine private home woofer, Monitor audio BRW10. Not much of a woofer and as you can see they went really cheap on PSU.
Amp is D-class TDA8924(2x120W), BTL rated at 150W RMS

I do agree fully with function of Cx (10nF) as startup, though I don't understand much further then that. It seems obvious Cx is being charged by rather high RC, and does start Q1. When Q1 is ON Cx is supposed to be discharged. It should be rather high voltage on the primary winding (when Q1 is on) it is forcing Cx to empty it self through main supply. When Q1 is of Cx is being charged again. It makes sense Cx should not activate Q1 more then once during normal operation. So I was thinking to try with about 100V ZD and some other guys suggest 36V should do the same thing just fine.
There is also other CR constant from Pin2 (Pin8) on power module. When Q1 is active first impulse that came from Cx is being prolonged through C1(6nF). After some time C1 is charged and support for Q1 is lost and he is going to off state. 15V ZD and resistors are shaping impulse and protecting gate.
When Q1 is going to off state this should somehow start up Q2. Sudden stop of current through primary winding of Q1 should induce something and jumpstart Q2 right? Then Q2 should continue with supporting impulse for it self. Q2 will go to off state moment when C2 (6nF on pin8 power module) is charged enough to prevent driving of the gate.
Q2 goes off and makes impulse to start Q1 again and process is being repeated.

Question is, does this make sense with required polarity of signals, can transformer be winded in a way that helps this explanation? It seems that C1 and C2 are rather small in capacity to provide enough current, but then, what do I know…

In this case it means I only need to fiddle with higher voltage ZD and all should be fine, because this explanation of mine suggest ZD value is not critical as long as it is not to high compared to main voltage supply or to low when compared 50kCx (RC) against frequency of oscillation.

Does this makes sense and why that mystery ZD died? It is less of a problem to kill high voltage ZD then lower voltage ZD. Do to total power disipation it requres much less current to go "over the edge". It makes me think it should be rather high voltage ZD?

Comments appreciated.

[kjlonergan]

I am wondering if the mystery part might be a diac and not a zener. When the breakdown voltage af a diac is exceeded it "fires" and goes to a low impedance until the current through it drops to a low value. This would certainly make sense for a kick start circuit.

Whatever you decide to try wear safety glasses!

Kevin

[infinia]

Yea and yell "fire in the hole" before you switch it back on.
Perhaps Cx will charge up after it is running to back bias the diode string.

[megajocke]

It has to be a diac. The current through the two parallelled 100k resistors wouldn't be high enough to start it if it were a zener. (There is a 390 ohm resistor across gate-source) The diac should actually measure open with a normal multimeter.

[HRMMI]

Yea, it does make sense to have DIAC there instead of ZD. That would give really nice startup bump. Though I don't know how that would affect C1. It seems first impulse for Q1 would then be quite long.
DIAC usually does not come marked with blue stripe or stripe at all. Stripes do indicate polarity and DIAC does not have one.
Second issue is that on the power module board it is clearly ZD symbol within rectangle, though it is marked as D4. On that same board there are two more zeners for gates, but for some reason they are properly marked as ZD1 and ZD2.
So, yea, I does make me wonder now, and it does make sense to use DIAC.

[Mooly]

Safety first... the primary side is connected to the mains at all times so connecting anything like a 'scope that is grounded could have disastrous consequencies... and not just for the PSU.

You have done a great job copying the circuit, that's not easy, but I would question whether the output ground really is connected as you show... that means the whole thing is at half mains potential.

In all the thousands of SMPSU's I have worked on in commercial stuff I have never come across any with a DIAC for start up.

Sony once used a push pull arrangement with the chopper transformer primary feeding into the junction of two series connected caps.

If the FET's are intact then usual suspects are any electroylitics and any high value resistors such as the 100k's

[HRMMI]

Yea Mooly that is also really weird part with Co.
I doubled checked all this many times and did copying through rather long period of time. There is no rush here. So I am quite positive schematics are correct. I did take out power transformer and most of the components and examined all the connections closely.

I have no clue what is purpose of that Co capacitor. In a way it does remind on EMI filter designs. But here it goes to low voltage DC ground...so weird, have no clue why and how current would flow there when circuit is not closed other then through EMS of the power transformer.
BTW this system does not use earth line from AC power socket at all. Earth line is not designed in the power socket on the woofer. Only to wires.

I made error in writing value of Co it is ten times bigger value 4,7nF.
Other two 0.1uF on the input are not connected with main electrolyte capacitors they make their own attenuation of main high voltage DC.

I did find al the broken parts and made notes on schematics what died and how, but question is, why. Only way to answer that question is to try to understand how it works before I star to blow up components

Thanks for pointing out on safety, it is High voltage here, but I am ok with it. I will power this through 2 x 100W light bulbs. If it will lit it means something is seriously wrong. In standby it should draw no more then 100mA or much less then that.