On Sun, 20 May 2012 09:31:43 +1000, Franc Zabkar
put finger to keyboard and composed:

On Sat, 19 May 2012 13:52:58 +0100, "N_Cook" put
finger to keyboard and composed:

I now see someone else has been here before and ground off IC5

music-electronics-forum.com/t10881/

Did you see the following post in that thread?

================================================= ===================
Have you noticed that pins 1 and 4 (Vcc and COM) of IC5 are reversed
on the schematic (if this is IR21531)?
================================================= ===================

Sorry, I should continued reading. It appears that they are indeed
reversed.

- Franc Zabkar
--
Please remove one 'i' from my address when replying by email.

Franc Zabkar wrote in message
...
On Sat, 19 May 2012 13:52:58 +0100, "N_Cook" put
finger to keyboard and composed:

I now see someone else has been here before and ground off IC5

music-electronics-forum.com/t10881/

Did you see the following post in that thread?

================================================== ==================
Have you noticed that pins 1 and 4 (Vcc and COM) of IC5 are reversed
on the schematic (if this is IR21531)?
================================================== ==================

I noticed on the one i'm working on. I was hoping it was an error in
the schematic but it is not.

I verifyed continuity to ground at pin 1.

So the IR21531 is not a drop in replacement as far as I can tell. I
have looked at just about every self oscilating half bridge drive they
sell and nothing that has pins 1 and 4 swapped.
================================================== ==================

- Franc Zabkar
--
Please remove one 'i' from my address when replying by email.

I'll sling a 4007 under the board beefore powering up again. As my 2153
started pumping out some pulses and oscillated , at least in audible range
with low Vcc, then I can assume pins 1 and 4 are not as in that Mark Bass
schematic.
Something else I've noticed on SMPS generally before but not thought about
why. Around the elevated mains voltage area a line of slotted perforations
in the pcb material. This must be more expensive to produce than a hatched
wide line of white ink at the overlay stage. I cannot see ventilation or a
paliative to possible developement of conductive pcb material as being the
reason

Franc Zabkar wrote in message
...
On Sun, 20 May 2012 09:31:43 +1000, Franc Zabkar
put finger to keyboard and composed:

On Sat, 19 May 2012 13:52:58 +0100, "N_Cook" put
finger to keyboard and composed:

I now see someone else has been here before and ground off IC5

music-electronics-forum.com/t10881/

Did you see the following post in that thread?

================================================= ===================
Have you noticed that pins 1 and 4 (Vcc and COM) of IC5 are reversed
on the schematic (if this is IR21531)?
================================================= ===================

Sorry, I should continued reading. It appears that they are indeed
reversed.

- Franc Zabkar
--
Please remove one 'i' from my address when replying by email.


Just checked again
The polarity pip of the original was not ground off and p1 is connected to
the dropper from the + of the HV dc caps , ie Vcc and p4 is connected to the
zero/negative of the HV caps (not ground of course) . So someone getting his
pin convention wrong I think, overlay "pip" is obscured by the chip and
agrees with the original orientation of the chip

Thinking back to the original failure mode. Solder joints all looked ok ,
but this is definitely PbF although absolutely no mention anywhere. The
holes in the pcb, for the powerfets , although plated through, are 3 times
the required diameter of the width of the powerFET pins and no eyelets
padding out. If bias to a gate failed then that could easily short circuit
that mosfet, would the other one fail soon after, if the amp is in use at
the time.

With the IR2153 and 1N4007 and 10R dropper in each of HV dc supply and hot
side of hottest mosFET and 57 percent of mains there is life
Blue power LED on, SMD LED D20 on, fan on, +/- 31.6V on main rails , 7.7V on
12V 7812, -9.6V on 9812 and 9.2V on 7814.
Then hot side , 12V on new 10uF C67 and 13.5V on new 100uF C69.
But a minute in and that 4007 smokes and mains current draw rises from
nothing readable on variac AC analogue meter to about .15 amp and I switch
off. 4007 survives as does the IRF740.
Is this a result of me trying to run on too low a HV dc and removing
droppers and taking mains to 75 percent or so will remove this or something
else wrong. I cannot believe that bootstrap diode would be more than 1 amp
inside a D suffix 2153 DIP8 package.
I have the fall-back option of refitting the original driver chip but think
I'll carry on with the replaceable 2153 for the moment, monitoring the
voltage on the R78 Vcc dropper , first I think, in htat first minute

Where can that power be coming from to heat the bootstrap diode? Not the Vcc
dropper , far too ohmic, the 4148 D40, 150mA or so is fine . I will
disconnect my added 1N4007 and see what happens to those main rail voltages

A bootstrap diode is required, without it and its back to no output and that
low level descending pitch audio at switch off
Must be passing too much HV ac , ie too much leakage. Tried 2x 1N4007 in
series and no problem over 2 minutes, and no heat to finger touch soon after
switch off. Enough time to put a sniffer coil next to the SMPS Tx and
stable operating f of 430 KHz.
For 330pF and 18K the IR2153 shows expected about 120KHz, influence of T24
or low Vcc perhaps, 11.5 and climbing above 11.6V at switch off
So do you need a high frequency , and high voltage diode, here if only a few
tens of mA of DC required to pass but need to block 100s of hf V ac in
reverse. Or just low leakage and high voltage silicon diode here?

From IR application notes it looks as though the internal bootstrap diode is
actually a gated powerFET in recent manifestations of that IR215. series.
The accepted external diode for a IR2153 seems to be MUR160 so placed a
BYV96E in there and it is comfortable. Removed the droppers
Raised supply to 66 percent mains and voltages coming up to +/- 36V,
presumably between 50 and 55 V main rails with 100 percent (this amp is 250W
rated, that schematic is 500W)
Same oscillator frequency
Time has come to replace the IRF740s with pair of 20 amp /600V powerFETs

Had it running under load, osc f 380 KHz, taking temp readings of both
heatsinks until stabilised without fan cooling, top cover removed , but
monitoring the fan DC .
Can mosfets go ohmic with age as I cannot see that ps heatsink getting all
that hot. Did a single powerFET going D-S randomly short then drive too much
power into those 2 caps to make them fail and knock out the second FET?
Could there be a flaw around the design that with bad ESR caps , instead of
going into protect mode could the driver IC go into a fast cycling
restart/protect mode and somehow overdrive the powerFETs to knock them out.
Both shorted original powerFETs had a discoloured area on thge heatsink face
as though overheating,
Infuriating having so few clues, and having to conjecture failure
mechanisms. Shorted mains fuse from 2 D-S shorted powerFETs and 2 bad caps
and no other obvious overheated Rs etc

Got word back on this and its survived a number of gigs including one where
it was pushed hard.
So a pair of footprint-converted SMD STB20NM60 and that IR smps driver +
diode are an adequate replacement.
Thinking back about it the owner reckons it may have been a song sheet
sloppily laid next to the air inlet was its downfall.