loopyloo (14-08-17)
I'm open to suggestions about now - everything looks good but something isn't
Just doesn't make a lot of sense :O
loopyloo (14-08-17)
When testing the FETs, what was the voltage drop across the tester's 1k resistor on the gate (button switch closed of course)?
(looking for abnormal gate current)
loopyloo (14-08-17)
C118 and C119 are both 30nF measured in circuit with R124 lifted
Maybe what you said about the difference between 6599D and 6599AD is the secret. I can get one but will just have to wait from overseas.
Exactly what it should be - typical gate current for those FETs is in the microamp range
So the FETs can't be faulted but result in an abnormal loading that affects the chip Vcc (I'm assuming that because Vcc plunges when the fault is evident IE no RFmin voltage - does Vcc loading reduce significantly when the fault is not there while RFmin is good at 2V?) and where is that extra current going
C118 and C119 are only 100pF so the 30nF reading is way too high, need to lift an end of each to get a good reading on them.
Got me thinking about a shorted turn on the transformer primary but that'd be about the most remote possibility of all and doesn't really fit with all the other evidence
We're missing something here for sure?
Need Sherlock Holmes
Last edited by Skepticist; 13-08-17 at 11:20 PM.
loopyloo (14-08-17)
Just for the hell of it I'd pull out C118 & C119 to test them with a megger to see if they're leaky as well as checking with a capacitance meter. The Cap tester would read a very high capacitance if they're leaky though so the megger test would only be for confirmation in that case.
You could substitute the FET tester's LED and resistor with a more significant load like a 12V 5W lamp or similar to force more current through and compare the 2 suspect FETs with the spare one you have but I very much doubt there'll be any difference between the 3 based on the previous test. If they turn on & off they're OK.
Last edited by Skepticist; 13-08-17 at 11:48 PM.
loopyloo (14-08-17)
I pulled one of those caps out and it read 105pF.
Do you think I should get an L6599AD ?
What if I pull R126 and C114 to take the transformer out of the equation ? , might have to fake ISEN then.
PS see R119 and R 122 have 3216 written under them, what's that all about ? ah I see it now, it's a size code.
Last edited by loopyloo; 14-08-17 at 01:00 PM.
I say try anything at this stage and disconnecting the transformer primary will be a good test
ISEN won't be a problem, just connect it to GND
Still using the external supply? (fake the LINE voltage)
It could help to look at the LVG, HVG outputs with a CRO prior to (and after) putting the FETs back in circuit
Then again with and without the transformer circuit complete
Need to analyse the hell out of this to get to the truth it appears.
loopyloo (14-08-17)
I'm still not clear on what's going on with the on-board Vcc going so low
The only way I can see that happening is it's extremely overloaded and that means excessive current glowing into, and out of, IC101 based on previous test results. What's not clear is whether that excess current is flowing to ground within the chip or through one or more of the outputs to the FETs. The chip's operating current is typically about 5mA max so there's an extreme overload in there somewhere. The operating current on OUT is should be only microamps, LVG and HVG can source 300mA, sink 800mA but that's peak operating values not continuous.
What we need to know is whether those outputs are sourcing the overcurrent. They can't be sinking it with no HV connected IMO
loopyloo (14-08-17)
Yes using the external supply.
What I might do next is put the fets back in so the fault shows its ugly head and measure the Vcc current (external supply). Pretty sure the voltage didn't drop though when RFmin went down (I'd have to do it again) . Makes me wonder is the board's Vcc supplying enough current ?
Can't use my hand held oscilloscope cos it bit the dust entirely, Banggood promised to send me a new one but still waiting on their reply if they've actually sent it.
I guess if I plugged my Atten oscilloscope into my 12v-240v inverter off the car battery, it should be safe enough for it (floating ground)
Thinking, thinking a lot about this circuit. Just voicing thoughts now.
What if the circuit is fine, what outside influence can cause the issue. ZD102's section?, no it's not directly influenced by pins 11,14 or 15, the transformer?, maybe, so take it out of circuit to check, I could even remove it entirely (with difficulty) to connect a meter and check current....Then I started thinking about the power. IC501 has brown out and brown in, now I can understand the brown out bit but why have brown in?, is it for Vcc soft start?, maybe RFmin drops out if the chip's Vcc is instantly applied, but wouldn't that happen anyway when Q501 turns on?, but maybe just a small delay is enough. I keep thinking about thinking outside the square (I'm normally very good at that) but this is tricky. There is only 2 things outside IC101's circuit that could affect it, the transformer and Vcc. Hmmmm !
PS ... Just found out they're posting the new hand held tomorrow.
Last edited by loopyloo; 14-08-17 at 07:02 PM.
Your CRO will be safe if using the external 12V supply and there's no mains connection to the board. Could even connect your external 12V supply to Q101 drain to check the switching action as well as gate drives of the 2 mosfets with the CRO. With the transformer primary connected you'd see some output on the secondary but not much (if your 12V supply could provide the current without the voltage sagging too much that is)
With no secondary voltage there'll be no feedback via IC102 and the chip should be pushing its hardest to get that secondary voltage up to spec. ZD102 appears to be a 'doomsday' worst case protection for overvoltage across IC102 so you could safely lift that to see if it's breaking down and affecting operation. Even if it was shorted there's only going to be a maximum of 0.25mA flowing via R114.
The operating currents of IC101 and IC601 should be very low like 10-20mA max demand on the Vcc supply through Q501. IC501 is working fine providing the local Vcc for itself and the 3.5V standby supply, it's the only thing in there proven 100% ok so far, plus using an external supply for Vcc means it's been bypassed anyway. All the problems are downstream of Q501, even when Vcc is fed externally.
loopyloo (14-08-17)
I'll start on all of that tomorrow.
Had a busy day today including starting a repair on a wall in my home.
I'm nearly ready to start on my next job too, a 1.5 metre camera slider with 2 steppers variable speed arduino control and programmable for automatic operation. Should be a fun project.
Last edited by loopyloo; 15-08-17 at 01:04 AM.
Ok so started looking at it again and something has happened so I stopped and thought I might tell you about it before continuing to see what your thoughts are.
I put most things back into the board, R124, C108 is new cos I lost the original but we did previous tests without it anyway, Swapped C114 back to the original and left R621 unconnected and Both FETs missing.
Used the external supply still with LINE fake. Milliamp meter in series with the supply.
Turned it on and got 2.9mA and 2v on RFmin
I then soldered Q101 to the board (I chose that one because I thought if any Q102 position in the circuit would be the trouble maker) and got 2.9mA and 2v on RFmin.
I then soldered Q102 to the board and got 5.8mA and (WHAT???) 2v on RFmin ?.
Oh and the supply stayed at 12v.
I was too shocked to carry on, had to go make coffee.
Last edited by loopyloo; 15-08-17 at 05:02 PM.
Encountering more phantoms than the ghostbusters here :O
Is it possible there was a whisker of solder shorting 2 pads of one of the FETs? and fixed when desoldering them? (earlier resistance readings don't support a G-S short though but G-D or S-D are adjacent pins). S-D short would have produced a mini mushroom cloud when it vaporised which would be hard to miss. Not sure if the FET could survive a G-D short with full voltage applied either.
And was the transformer primary connected when you got those readings?
If it can't be faulted now with everything is back as it should be, it's time to power it up from the mains via the series lamp to check how Vcc and RFmin voltages hold up I'd say. Although it'd be nice to look over the gate-source voltages first with a CRO to see if all looks acceptable while still on the external supply.
Don't forget to remove the 33k resistor between Vcc & LINE before applying the mains and leave the PFC out of service for now.
C108 can be any 100nF 50V cap or even smaller and one of those tiny monolithic caps (EG Jaycar RC5490) could be fitted in there as it's only to suppress switching noise on Vcc.
loopyloo (15-08-17)
ha ha I knew you were going to ask about a whisker of solder, but no, not possible. I looked closely, checked with a couple of different meters, cleaned and scraped ... and I did that several times. It's just a mystery to me.
The cap I put in is that same Jaycar one.
Yes the transformer was in. Everything was back to normal except for the external supply with the 33k, F101 removed and R621 lifted.
Tomorrow I'll put the oscilloscope over it first then try it on its own supply. So what would you say is acceptable voltages for those FETs ?
Might be getting somewhere now, touch wood.
Last edited by loopyloo; 15-08-17 at 10:39 PM.
Skepticist (16-08-17)
The FETs are rated for 600V D-S and require a Vgs of about 10V (25V is the max) to fully turn on (minimum heat dissipation)
So you should see half cycle spikes (approx square wave) around 10V high on the gates referred to the source. I say approx square because the frequency it operates at will produce all sorts of over and under shoots as the gate capacitance is charged and discharged so there'll be a bit of a spike at every change of state, and that could be at quite a high frequency (chip rated to switch at up to 500kHz or so, minimum should be around 50kHz).
loopyloo (15-08-17)
A question: what material is the PCB? (IE fibreglass or phenolic)
There's a minimal gap between the gate and source tracks from the IC to the FETs and we know that the main fault current from a huge EHV spike burnt it's way through the FETs taking out everything in its path including the controller. I've had cases where phenolic boards have become 'carbonised' when it gets burnt leaving a partially conductive track that has to be ground back to clean material with an engraver or similar, even scratching with a hobby knife will do the trick. Fibreglass boards tend to fare much better at such extremes.
Seems you may have an intermittent fault here and the only way you know it's fixed is when it doesn't happen again so high levels of uncertainty if a definitive location of the fault is never actually found.
Bookmarks