SYMSI70 gas heater, TOP247Y exploded, power supply, snubber network

[hikomalek]

Hello, I have a SYMSI70 gas heater board, which has issues with its primary side power supply, controlled by TOP247Y chip.
When I replace a TOP247Y it explode in 1 second. I have desoldered transformer and measured a winding ratio, so transformer looks like OK.

Transformer:

|Info| Primary | Aux | AB | CD|
|-|----------|----------|-|-|
|Ratio|1|25|40|12|
|Voltage AC|230V|9.2V|5.8V|19V|
|Voltage DC (*1.41)|230V|13V|8.1V|27V|

I have disabled a undervoltage protection of TOP247Y, disconnected secondaries and tried to power it up with voltage from 50 to 150V DC (I have connected in series multiple 32V DC printer power supplies).
When I connected all secondaries, all voltages were correct, display was showing an error "F10" as nothing was connected and "some relays" on board were switching. So board works correctly on 150V DC.
A Drain-Source current(on primary side) was about 50mA, power consumption about 8W. On the board, nothing is hot, so there are no hot components.
But when I connected it to 230V AC, which is about 325V DC, TOP247Y exploaded. This happen everytime. I have another older board which has some "logic issues" and chip TOP247Y works correctly in this old board, so the chip TOP247Y is not broken.
I have transformer out of the board on about 30cm wires.

It means that error must be in a snubber network, there are two, one around a transformer and the second around a Drain-Source of TOP247Y chip. I do not have ESR meter or oscilloscope, I have only simple "Non-True RMS multimeter".

I have even tried to put a P6KE300A tvs diode to primary snubber network,
Diode BA159 was OK, but I have replaced that with UF4007.
Snubber capacitor C803 (10n) I have replaced with two 6.3nF 400v capacitors
Second snubber capacitor C1, about 70pF was "white SMD ceramic MLCC" I have replaced with "simple ceramic" which is probably for low voltage only.

When I'm measuring voltages even with 50V DC power supply, snubber voltages are high, about 200-500 V on multiple places.

I have some P6KE130A (tvs diode), should I use it?

How can I check it safety? I wanted to use a tyristor dimmer to slowly rise a voltage (I do not have a variac), but it is just cutting some pieces of waves and does not limit max voltage, so TOP247Y exploded too. So I'm thinking of using some "capacitive divider" to make a lower voltage.

So there is no overcurrent, just an overvoltage caused by a snubber network malfunction. Please what can I do with that? Everything looks like correct

[CapLeaker]

Means Jack sh*t to me. These TOP, LNK, TNY etc switchers are quite delicate. If the board isn’t hospital grade clean in between the pins, the switcher will also just fail. I fixed a device where someone else didn’t properly clean around the pins. device ran for a few hours and bang, the switcher went. I did the second repair and it is still running today 24/7/365.
make sure everything is absolutely clean. The problem with these TO220-6 is that the pins are too close together, HV doesn’t have to work hard in order to get into a different pin. To me these TO220-6 is the worst design ever.
I wouldn’t disable UVLO etc. Snubber caps I always replace them with the same value, but 2KV rated. Symptoms of TOP etc, lid blown off to looking brand new but still dead. Had found them do all that on various devices. Good luck!

[stj]

topswitches have a long history of exploding in 2 situations.
1: poor startup capacitor causing it to hickup till the boom.
2: low supply voltage causing it to not start oscillating properly.

[stj]

just a thought,
make sure it's getting smooth 340v-ish dc,
a bad rectifier or input cap could blow one.

[hikomalek]

Thank you all for your responses.
Regarding disabling under voltage protection (UVLO), I had to do that, because I was testing that on "low voltage", to prevent explosion. At 50V Drain-Source current was 200mA, at 150V just about 50mA. TOP247YN was not hot (it has even nice heatsink).

About clean board and between pins/board track interaction. Is it real? Yes, when measuring by multimeter, Drain-Source voltage is over 1000V (so my multimeter, cannot measure it), at least multimeter didn't break yet. My board looks like very bad, as I broke a board tracks, so I had to "replace them" by thick wires. But I consider, the distance between them is "large" and nothing bad can happen.

About smooth 340V DC at the input. Last time I wanted to test it on higher voltage (more than 150V, as I do not have more printer power supplies). So I had an idea to use a diac-thyristor dimmer, to regulate an input voltage. I was aware that it is just cutting the output, but not limiting an amplitude (which is still 325V). But I set it to minimum, which can be about 40V (diac threshold). And at this configuration TOP247YN exploded in a second. So when I was powering it by 150V DC it was working correctly, but about 40V AC (with 325V peaks) broke that.

But what about snubber network and TVS diode? I would expect that as I have a tvs diode P6KE300A around a primary winding, voltage should not be so high. I have a P6KE130A, should I use it? Then I guess, max primary winding voltage will be 130V + 325V from "mains" so about 450V.

Is it possible to use an external MOSFET and drive it using a TOP247YN (maybe drain)?

This power source has output voltage of 8V (for a logic) and 27V (for relays), then, it is driving a fan which blowing an air, I guess this can be PWM controlled. So maybe TOP247YN is powering this fan too.
A have tried connect 8V and 24V external power supplies to transformer output and it was working (but of course without connecting a board to gas heater).


Is there any way how to test it and do not break it again?
How to get a DC voltage of 200V or 300V? I do not have a variac.

I will try to put insulated tube to each TOP247YN pin and clean a board. And I replace an input smooting capacitor (after a rectifier).


Thank you :-)

[CapLeaker]

First: no… you can’t replace a 300V transient diode with a 130V one.
Second: while I do have a variac, I hardly ever use it. Mostly we are using the incandescent dim bulb current limiter in series trick. If there is a lot of current drawn, the incandescent bulb will limit the current to the PSU and light bright. Since this is a small power supply you need a low wattage incandescent bulb.

[stj]

a variac can kill a topswitch, keep it for valve based stuff..

[Agent24]

About clean board and between pins/board track interaction. Is it real? Yes, when measuring by multimeter, Drain-Source voltage is over 1000V (so my multimeter, cannot measure it), at least multimeter didn't break yet. My board looks like very bad, as I broke a board tracks, so I had to "replace them" by thick wires. But I consider, the distance between them is "large" and nothing bad can happen.

Show us a photo of this area... It doesn't matter if the distance is large, the pins are still close together in the end, and if they are not clean (flux residue etc) there can be leakage current that turns into electrical breakdown and failure.

[hikomalek]

Hello, so update.
As you have mentioned, it likes smooth input DC voltage.
- Input smoothing capacitor (after a rectifier) was wrong, it has about 50nF capacitance and it should be 47uF. I have replaced it with some other.
- removed a TVS diode (there was not any TVS diode before)
- replaced faster diode UF4007 by a slower BA159 (original), maybe too fast is not good, even somewhere there are using just a 1N4047
- enabled UVLO protection

And it didn't exploded immediately! But later I was disconnected and connected Drain, to "Primary of transformer" when it was turned on and now TOP247Y is shorted... Probably, I have connected that at the time when it was switching with high duty after a restart, so there was a high current suddenly.


But I have another board (older), that one have working TOP247Y chip (last one). I thought that board has some logic error, but it looks like that secondary smoothing capacitor was bad, so I have replaced capacitors and relays are not switching as crazy as before (maybe relays didn't have enough power).

Now on secondaryB there is a voltage on smoothing capacitor about 42V, which is I guess too high. Capacitors are rated to 35V, there is LM317, which is only to 40V.
I have a feeling, that when i was powering that with a lower DC voltage on this secondaryB was about 28V.

There is no feedback from secondaryA and secondaryB to TOP247Y. Only AUX has a feedback, by about 6V zener diode, so after AUX rectifier diode there is 12V.
So voltage on secondaryA and secondaryB is only based on transformer winding ratio and "power transfer" among AUX and secondaryA and secondaryB.

What can cause a higher voltage on secondary? Can it be a higher current on AUX, so more "power" is transferring to both secondaries?

I think the issue is again in capacitors. I do not have an ESR meter (I have a plan to make it soon). My components come from a scrap from electric trash.
I'm expecting new TOP247YN from China next week (10 pieces by 0.86 EUR). In our local shop one TOP247YN is for about 5 EUR and 8 EUR postal fee.

Thank you for your valuable info, it was useful...

Good document about power supply:
https://www.onsemi.com/pub/collateral/and8461-d.pdf

[CapLeaker]

Switching out UF4007 to a BA159 isn’t a good idea as the recovery time is much higher on the BA159. If there was a slower diode in there, you can certainly switch it out for a faster one.

[Agent24]

Switching out UF4007 to a BA159 isn’t a good idea as the recovery time is much higher on the BA159. If there was a slower diode in there, you can certainly switch it out for a faster one.

I think he said the board came with a BA159 originally, so it would be fine, but I agree... photos!!!

[CapLeaker]

I think he said the board came with a BA159 originally, so it would be fine, but I agree... photos!!!

I do not think so. Usually you see ultra fast diodes with a Topswitch. Anything lower you’d have a switch loss and stresses the Topswitch 147yn. So the BA159 isn’t a correct part for it, unless it is a very low supply output, but at the same time why using a TOP147 then? Doesn’t make sense. The UF4007 is the correct part to use, same as MBR or SR diodes. Depending on what diodes are used and how much power this PSU has to produce, it can affect the output voltage if a wrong diode is used. Also creates more heat too.

[Agent24]

I have transformer out of the board on about 30cm wires.

I don't think that's going to be good for it... why are you doing this?

I'm expecting new TOP247YN from China next week (10 pieces by 0.86 EUR)

At that price they sound suspicious... could be fakes

Maybe the original chip blew thanks to the dead input filter capacitor...
Then, the new chips you had (if from China as well) maybe just can't handle 230v or have other issues...

Incorrect output voltage could also be due to fake chips.

[stj]

check the input diodes incase they damaged the cap too.

[hikomalek]

Hello, thanks for your feedback as usually.
Short recapitulation:
- power supply is using a TOP247Y chip which has a feedback from AUX winding, so windings secondaryA, secondaryB are without feedback
- when I replaced a smoothing capacitor (47uF/400V) after a (4-diodes bridge rectifier), TOP247Y is not exploding anymore ;-)
- I have disabled under voltage protection (again) and tested it on 60V, 90V, 120V DC
- based on https://www.onsemi.com/pub/collateral/and8461-d.pdf, it is better to use a slower snubber network diodes
- the issue is too high voltage on secondaryB, about 42V, but capacitors there are rated to 35V and LM317 here is just to 40V
- I have just a simple non-true RMS multimeter with capacitance meter; no ESR meter, no oscilloscope, no inductance meter
- I have connected that through a 25W light bulb (but of course it did not prevent explosions of former TOP247Y chips)
- I have removed a transformer from the board (so it is on 30cm long wires) and removed both secondaries from it, just primary and AUX is connected as it should be
- I made a "test load" on secondaryB which consist of: diode, smoothing capacitor (22uF) and parallel resistive load

| Resistor on secondaryB | Voltage on this resistor |
|----------|----------|
| 1K | 51.5V |
|3K3 |64.3V|
|5K6 |67.8V|

- when I doubled a capacitor value (2x 22uF) on "test load" values were the same


So I think it works as expected, with a higher current/power, there is a lower voltage on secondary side. So maybe if I use a "very high load", then voltage will lower to 30V as expected.

There was a zener diode Z800, 7.5V; voltage on AUX, after a diode D1-BAS21 was about 11.4 - 11.7 V
When I replaced a zener diode with 3.9V diode, then that voltage was about 7.43V, but output voltage on secondaryB was the same!!! So I think this is the issue.
From my point of view as AUX voltage is lower then secondaryB voltage should be lower too.

Transformer
|Info| Primary | Aux | secondaryA | secondaryB|
|-|----------|----------|-|-|
|Ratio|1|25|40|12|
|Voltage AC|230V|9.2V|5.8V|19V|
|Voltage DC (*1.41)|230V|13V|8.1V|27V|


Based on transformer ratio (measured some time ago) AUX vs secondaryB is 2:1. So secondaryB voltage should be about 2 times higher than AUX voltage.

So as with 7.5V Zener diode on AUX was 11.5V, then secondaryB should be 23V (two times higher).
With 3.9V Zener diode on AUX was 7.43V, then secondaryB should be about 15V.


There are some russian forums where people have similar issues with high voltage on secondaries:
- they have solved high voltage by replacing a Zener diode (Z800) and C805 (Control pin capacitor)
https://forum--monitor-net-ru.transl...r-service-info
https://forum--monitor-net-ru.transl..._x_tr_pto=wapp

I have tried about 10 different capacitors on position C805 (47uF, 50V), which is a capacitor connected to Control pin of TOP247Y.
Voltage on TOP247Y Control pin is 5.8V as expected, I think it fluctuates little bit (5.7-5.8V), but never under 4.8V which should be interal under voltage protection, which cause a restart cycle.

What is measuring a non-true rms multimeter on 66 kHz rectangular signal?

As AUX voltage is always a constant (even under lower voltages and different secondary load) does it mean that TOP247Y is working correctly even it is from China for a few cents?

How TOP247Y works?
- at start Control pin capacitor is charged to 5.8V
- switching start from 0 to max duty cycle (it takes max 10ms)
- current flowing into a Control pin is changing a duty cycle, more current -> less duty

So it must be changing its internal resistivity on Control pin based on flowing current (this is probably that Shunt regulator/error amplifier).

It means what is controlling a duty cycle is just a current into Control pin.
What can cause a broken capacitor into a Control pin? Less current, it means higher duty cycle and more voltage on secondaries?
What I'm measuring on AUX (11.5V) is just a voltage drop on zener diode (opened in reverse), 2R2 resistor and capacitor in 66kHz by simple multimeter.

I'm trying to simulate a LTSpice model of that...

Why other winding voltage is not based on transformer ratio?

But I'm just a simple person with minimal electronics skills, so I'm asking for a help.



OK, end, sorry for long brainstorming.

[Agent24]

I don't know why you have the transformer attached with 30cm long wires? What is the purpose of this?
Perhaps it is upsetting the measurements?

[hikomalek]

Hello,

I don't know why you have the transformer attached with 30cm long wires? What is the purpose of this?
Perhaps it is upsetting the measurements?

I had to a remove a transformer from the board as it has high secondary voltages to prevent breaking a secondary circuit. So transformer is on "wires" which connect Primary and AUX windings. And my wires are 30cm long. They are "straight".

Anyway I found that they have a great designer site:
https://piexpertonline.power.com/piexpert/design/index#

You can design your power supply here, changing parameters, changing types of circuits. You can even specify you want a "zener based AUX feedback". Based on this you can see that snubber network in "my power supply" is not well designed.

Based on that I found that output voltages depends only on transformer winding ratio (as I always thought). So it does not depends on zener diode. So it transformer windings are not shorted, voltages should be correct.

Transformer design
https://www.thierry-lequeu.fr/data/TOPSWITCH/AN32.pdf

When I lowered a capacitance of C2 from 2.5nF to 0.5nF, then secondaryB voltage was lower. I think R8 and C2 should be just low pass filter or something like that.

Attached "example power supply.pdf" has AUX voltage 27V, so for that reason there is a 22V zener diode. So difference is 27-22=> 5Volts

So as Control pin current is limiting a duty cycle and voltage, "something is consuming a current" and there is not enough current to flow into a Control pin. So maybe C805 is leaking current.
Diode D1 and Z800 (Zener) are OK.
I have tried about 10 different 47uF capacitors on position C805 (of course all of them was from trash and they are old, but have the right capacitance). I hope at least one was good.

How to measure an output secondaryB voltage if with increasing resistive load, that voltage is lower? Should I completely disconnect an output resistor? Voltage on output capacitor should be "the right" as it is already rectified and smoothed.

Do you have idea what can be wrong or what I can try?

[hikomalek]

I was thinking about AUX circuit here and I did some simulations:
What is controlling a duty cycle is just a current into a Control pin. So more current, less duty cycle.

At "Rising AUX DC voltage.png" is it showing that current into Control pin starter to rise just after a Zener voltage of Z800 diode.

Another pictures are just simulation which describe, that with higher input amplitude (7, 9, 12, 15 V) you have higher current in Control pin.
Then you can clearly see range of fluctuation of voltages based on a measuring position (giant fluctuation).


Nice articles about measuring a PWM voltages with a multimeter.
https://lygte-info.dk/info/DMMPWM%20UK.html
https://www.allaboutcircuits.com/tec...nalog-voltage/

So even if multimeter is somehow averaging a voltage, its result depends on everything, duty cycle, curve format.

I understand it this way:
- on primary TOP247Y is generating a PWM signal which is opening/closing its internal transistor
- primary and secondaries are in opposite phase (transformer windings dots)
- all secondaries have a diodes which flows only in one direction
- when transistor is open, a power is putting into a magnetic field of the core
- when transistor close, then giant current change happen in primary and giant current peak is made
- that peak goes through a "fast" snubber network diode and consume in its RC filter
- this operation can possibly influence a power in a transformer core
- when transistor is closed, then secondary diodes are open and they "consume a power" stored in the transformer core
- that power is regulated by PWM signal duty cycle
- TOP247Y starting from 0 to its max duty cycle and negative influence a duty cycle based on its Control pin current

So primary windings count is responsible only for a magnetic power stored in a transformer core, it does not control a secondary voltages.
Snubber network should not influence secondary voltages too.

Secondary voltages depends on its winding ratio. So as AUX and secondaryB has ratio of 1:2, so when an AUX has 15 volts, then secondaryB must have 30 volts.
Rising a voltage on AUX side will stop after enough current in Control pin (it will lower a duty cycle, or stop rising it).

What is telling a TOP247Y how much Control current is enough? I guess Control voltage must not fall to 4.8V, so should be in range of 4.8-5.8V.

Please check "Control current, duty cycle.png". It is showing that 2.2 - 6.6mA are enough to control a duty cycle.

And finally please check "Control pin voltage.png", this show that 5.8V of Control voltage is a threshold value to rise a Control current.
It means at external 5.8V it will start to rise current in Control pin to lower a duty cycle.
It means as there is a diode (0.7V) + zener diode (7.5V)+ threshold voltage 5.8V => 0.7+7.5+5.8=14V, this should be AUX voltage, and secondaryB should be 28V.

Any differences can be caused by a leaking a current on C805 (47uF/50V) or somewhere else.

Please correct me if I'm wrong.