TOP242 based design: output long term drifting...
Can't believe I am evoking this same topic again after almost 2.5 years!
In some of the units out there in systems, we have been told the 15V module using TOP242 will drift up after several hrs of operation( ambient temp is high...65deg C +). As it sits just below our "upper threshold" limit, it will finally go above that. The the supervisory ckt kicks in & shuts the whole system down gradually.
This has been a nuisance in the system all these years.
Some "wise fat cat" had analyzed & expressed a deep statement that " the feedback pin is not sourced with enough current from the high resistive value in the opto circuit".
Well, we use much lower value than the model gives us due to other design considerations.
Any idea why this is happening?
We have in place the fix mentioned earlier in this post.
very much appreciate a feedback.
-b
Comments
thnx.
pl see attached. Included is the oscillo of the field return module. This has been in field at least 1 yr if not more.
I am pretty sure the issue could be solved by replacing the 2k resistor R11 to a lower value ressitor. You may try 1k or even 500 ohm resistor. But you have to make sure your loop is good after the modification since the loop gain can be 2 time higher for 1k ressitor and 4 time shigher for 500 ohm.
Another thing need to be corrected first: R9=6.8ohm, and C5//C5A=47uF, and between the Cpin and S pin using a 100nF ceramic cap.
Let me know what happen....
regards
thnx
I think we have R9/C5 fixed as you indicate. We had discussed that in 2011.
The sch I emailed is older rev.
R11 is perhaps still 2k
Are you suspecting low drive for U2 diode gets low with time?
Note that the model gives I think 7.5k under these operating conditions!
It is clear that the loop tends to go "open" as a result of long term drift phenomenon.
We know also that it tends towards a value determined by the step down ratio.
I am not sure it could be due to R11
Not all of them do so though.
What I find intriguing is that in Fig 42 of the TOP242 d/s, they show 270 Ohms with an output voltage of 19
Now that will drive the opto with almost 50mA!
It is not possible to drive the LED this hard for long term survival. Don't you think so?
U2 CTR is 100% if not >
With this drive, do you not think that we degrade the LED?
I would hesitate to lower R11...not knowing its degradation with higher current.
Is there anything in TOP242 that might drift away to make it appear as if there was lower current fed at C?
let me know...
-b
At normal condition, the current through the led is defiened by the TLV431 and control loop, not by the ressitor, so 50mA current is not make sense.
Clearly high current through the opto is must to regulate the output voltage, so you either increase the CTR or increase the current through the led. The CTR will be less and less over time, and CTR is reduced faster with higher current through the led, this should eb the root cause of the issues at field.
Honestly, use a optocoupler with higher CTR is a better choice, sinc the life of the power supply will be a lot longer.
Regards
yes, yes of course: current is set by a hostm of things ...431/LOOP ETC. Not by R11
Vce of TL431 sets where it has to as long as R11 wont limit it( by too large a value e.g.)
I like the idea of higher CTR with known long life. Sounds best way & I do not see a conflict with field ops
thnx a lot
it should be noted that R11 is very important for the life time of the power supply. Lower value resistor for R11 means longer life...
Regards
Thnx for the info.
I will have new ones built soon with CTR 300-600% range option.
Any specific test you can recommend to "predict" if the drift we have observed will no longer be a behavior with high CTR?
field dependent test takes too long & will be outright embarrassing if it occurred again!
For me, the only concern is the control loop, since the loop gain is a lot higher with the higher CTR optocoupler, so you have to double check the loop and load dynamic.
Regards
I have tried to do loop measurement on this using Venable - never got any meaningful gain/phase readings!
would you know why?
I got the new rev sch, & see that they use 1k.
That aside,. pl see attached CTR degradation: with time, tmep & If.
I find it interesting that CTR degardes at higher currents..........although, aging slope is flatter.
Do you htink I should measure the real If & Ic >> CTR in-situ & then decide on R11, type of Opto to use?..Plus obtain CTR plot with If from vendor.
I think going blind with 1k without knowing current levels will get us same results of aging!
My suspicion is that even lower value of R11 maybe warranted if If is controlled by TL431 & the loop!
-b
forgto to attach file
PLS refer to AN57 for loop design and measurement.
Normally datasheet of optocoupler shows the aging curve with different led current. And you should have some idea about the minimum CTR needed for the life cycle of your product based on the initial led current (the led current either measured or calaculated based on the initial CTR and top switch datasheet figure 7) . Once you know the minimum CTR, you can calculate the led current needed to regulate the output at the end of life cycle. Choose a R11 with lower value than your calculation should be ok for all the life cycle. As you know, the maximum led current can be calaculated by (Vo-1V-0.3V)/R11 ( 1v for the LED and 0.3V for TLV431 outputf ).The R11 should guarantee that the led current is high enough to regulate the output at the end of the life cycle.
regards
Thinking of getting rid of the opto: I have come to conclude that just by increasing If of the diode wont address CTR degradation which maybe happening independent of drive.
So: how well will the loop behave if we have the output controlled by the third winding voltage? I know it all depends upon parasitics of the flyback. LT1172 tyep controller can get away with it at low voltages...can this be done using TOP242?
thnx
Without optocoupler, power supply with TOPSwitch suffers from regulation problem, because it is not the output regulated by the control loop, but the output of the third winding regulated by loop.
But design with LNK-HP is a lot easier for the regulation issue. LNK-HP is designed for applications without optocoupler feedback, it is designed for applications with third winding feeback. Although the regulation is not as good as optocoupler feedback, it can meet +/-5% regualtion SPEC.
With LNK-HP, you donot have to worry about too much about the control loop, the loop should be stable with right compensation.
Regards
oh, that is a cool idea!
let me pursue that a bit.
thnx
Now we have a new annoying issue with 15V @1 amp module over which we have had so many discussion.If there is no load, output ripple is quite ok. uUt at 1amp non-inductive resistor load, it is a different matter. On start up, it is one thing & as the module warms up, the ripple changes completely.Pl see attached oscillogram of the module made pretty much according to design rules & "model" values.I have an external filter created by use of a "common mode choke" on the output with leakage inductance of approx 2uH. Output capacitance used is approx 470 uFMost of the time, we have excellent ripple. In several hi priced aerospace quality VME power system, 15V module will show up with this output & our system shuts down.Would you know the true origin of this? How would I stop such low freq oscillations?thnx-robin
PI_Cochrane
could you see hte new post on low frequency oscillation?
thnx
Could I have your schematic? That will give me a lot of information for your issue...
Thanks