Kripton wrote in news:4e7c30a0$0$16475$ snipped-for-privacy@news.free.fr:
I've had to replace a lot of 3842's.
it depends on the PS design,some designs leave the IC vulnerable,particularly if the switcher FET/XSTR blows.
but the 3842 makes a nice PS circuit.
-- Jim Yanik jyanik at localnet dot com
No ? There are a number of chips in the 35 and 38 series pwm controllers, which are all pretty similar in function, and whilst I'm not certain that I have specifically had a 3842 faulty, I have certainly replaced many of those series over the years, usually for a dead symptom, due to the internal voltage reference having failed. I agree that the caps decoupling the supply and reference voltages are by far the commonest problem though, and that was what I first suggested to the OP as a possibility for his problem. The symptoms that he has, are pretty weird for a switcher. I, like others on here, would initially have suspected the filter cap for the rail that the regulation feedback is derived from, but the fact that he replaced all of the secondary caps right off, seems to knock that one on the head.
Arfa
"Approx worst case ESR values for new capacitors at 20deg C"
is what it actually says. Bob once told me that when he put that chart on the front, he took it directly from a manufacturer's data sheet - Pan IIRC
I guess that there are two ways you can take that. I have always taken it on the positive side, rather than the negative view that you take. As I'm sure you well know from your own experience, using an ESR meter is a black art anyway, as much magic as it is science, but given that, in the many years that I have owned and used daily, a Bob Parker (DSE) meter, I have always found the chart to be an excellent guide. From many years of experience with SMPS's, and having found and replaced many pwm chip supply decoupling caps, I would not feel the immediate need to replace one of that value and voltage rating, that went 0.5 ohms. I would not consider it 'marginal' and potentially the source of the problem, until it had got a lot closer to 1 ohm. Typically, caps that have failed in that position, will be up around the 10 ohm mark, or open circuit.
Yes, agreed, but you've got to admit that the supply to the chip fluctuating between 7 and 15 volts is a pretty bizarre state of affairs, and there's not actually too much bar the startup R and it's decoupling cap, connected to that pin. Of course, the self-powering supply, also connected to this point, might be interfering with the reading, which is why I suggested disconnecting it. As to whether the supply will run in this condition, depends on the value of the startup resistor, but running or not, it might at least give an indication of where next to look, depending on whether the voltage on the chip side of the resistor is now steady, indicating that the fluctuating voltage is coming back via the self-powering supply, and in which case is a red herring, or still varying as a result of the chip drawing current erratically. I might have felt inclined to think that the fluctuation was the startup bleed via the startup resistor at 7 volts, followed by the self-power coming on line at 15 volts as the supply gets going, followed by the supply shutting back down and so on, due to some secondary side problem. However, that then doesn't stack up with the secondary side voltages being steady, but wrong, and worse than wrong, unbalanced.
It is an odd state of affairs to be sure, but of course, as we all know, some problems on switchers can be very obscure.
Arfa
Talking about switching supplies. Today I decided to look at a supply removed from an allen Bradley Panel computer, this is the second time the supply has quick, we were able to buy exact replacement supplies so it was cheaper to do that at the time.
This time I decided to investigate the issue on one of the broken supplies. I found the main cap on the HV side to be shorted, it was something like 230Uf 450VDC. This cap did not cause any components to open that I could see other than a varistor in series to it for the inrush currents. Any way, I could not get that damn cap out of the board, I don't know what type of solder they used on that cap but the only thing I had that would melt the solder was a 140 watt soldering pistol, and that didn't do me a lot of good since I needed to get to the other side of it.
My regular station set to 850F would not melt this solder but items around it had no issues breaking down with that temp.. I guess If I really want to fix this thing, I will have to use my air iron and some added heat with another iron on the area to blow the solder out the back side.
Oh well.
Jamie
I find adding fresh solder to difficult melts and sucking it out repetitively eventually will fill the joint with regular solder. Of course if theres a heavy plane of copper, its a problem. JC
On Fri, 23 Sep 2011 16:52:09 +0100, "Arfa Daily" put finger to keyboard and composed:
As has already been mentioned, there could be something wrong with the OP's measurements. He states that "currently it is outputting +17 on one output and somewhere between +22 and +30 on the other output."
One would expect that the +/-15V supplies would have identical windings. Moreover, since they would be wound over the same core, then one would expect that the two outputs should track each other reasonably well, allowing for slight load differences.
Maybe there is a clue in the fact that the OP's measurements are both positive. Could it be that he is uing the wrong 0V reference?
For example, if he is mistakenly using the -15V rail as the 0V reference, then he would be measuring +15V for ground, and +30V for the +15V rail, if the supply were working correctly.
- Franc Zabkar
-- Please remove one 'i' from my address when replying by email.
Hmmm. That's a very valid point, Franc. Although possibly not specifically in switchmode power supplies, I got the impression that the OP was reasonably savvy in electronic repairs in general, so did not even consider that the measurement technique could be in error. To be honest, I never even noticed that he was declaring both to be positive ... :-\
Arfa
I don't see that a fluctuating supply would cause a polarity reversal on the -15v rail.
On Fri, 23 Sep 2011 01:56:37 +0100, "Arfa Daily" put finger to keyboard and composed:
If the feed is coming directly from the +170V rail, then I can't see that there will be much difference in the power dissipation of the startup resistor after the power supply has settled into run mode.
For example, if we assume that the chip's supply voltages are 5V at startup and 15V in run mode, the the voltage across the resistor will vary between 165V and 155V.
- Franc Zabkar
-- Please remove one 'i' from my address when replying by email.
Typically, the startup supply doesn't come *directly* from the high voltage rail, It is normally fed via a diode, and the self-powering rail is arranged to be a higher voltage than the startup supply. This results in the diode feeding the startup supply, to become reverse biased once the supply fully starts, which then results in the current draw from that supply reducing to as good as zero, and thus likewise dropping the power dissipation in the startup resistor to virtually zero.
Arfa
No, you're right, it wouldn't, which is why Franc and I elsewhere in the thread, have now entered into the discussion of how exactly the OP is taking his measurements i.e. what point he is using as his ground, for a start.
Arfa
On Sun, 25 Sep 2011 01:29:05 +0100, "Arfa Daily" put finger to keyboard and composed:
Your comment took me by surprise. Just to make sure I wasn't having a brain fart, I consulted an application note:
I confess that I haven't really thought about this subject before, but according to page 4 of the PDF, there is no back-biased diode. AISI, a diode that is fed from a 170VDC supply cannot become reverse biased by a much lower bootstrap voltage.
The app note suggests that the UC3842 will remain in the off state until the capacitor on its Vcc pin charges up to the UVLO (under voltage lockout) turn-on voltage of 16V. During this time the IC draws only 1mA.
After the UVLO turn-on threshold is reached, the IC turns on and pulses the chopper. The bootstrap winding then generates the Vcc supply for the IC and prevents the capacitor from discharging below the IC's UVLO turn-off threshold of 10V.
Fig 31 on page 13 of the same document has an application circuit for an isolated +/-12V supply. The 56K resistor (R2) can pass about 3mA. This means that 2mA goes toward charging the capacitor. The UC3842 datasheet specifies a typical operating current of 14mA.
If you were to disconnect the bootstrap winding, then the output voltage will only remain alive for as long as it takes the Vcc capacitor to discharge from 16V to 10V.
I = C . dV/dt
so ...
dt = C x dV / I
= 100uF x (16V - 10V) / (14 - 3)mA = 55 msec
- Franc Zabkar
-- Please remove one 'i' from my address when replying by email.
Hmmm. OK. I'll buy that. I must admit that I haven't taken too much notice of how that bit of the circuitry works in recent years. Perhaps I'm just going back to the early days when I was first taught about these things on manufacturers' service courses. Maybe there used to be a zener on the end of the startup resistor, and then a diode. The diode could then be reverse biased by the self-powering supply, and the current in the startup resistor would decrease to that of the zener's draw only. I'll have to have a look back on some of the older schematics for VCRs and DVDs etc
Arfa
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.