OK here's an update. First, thanks for the generous feedback.
It is the regulator that's getting hot, not the bridge.
I chucked the 25V cap and found a 470uf 63V cap in my box and put that
in there (it does have the negative pin connected to ground - I will
update schematic). I also found a .001k ceramic disc cap and placed it
between pins 1 and 2 of the regulator.
I'm not sure what the value of a ".001k" cap is. I also have these:
104Z 50v, 102K 1kv, 103M KCK, 104k 1KV, and one that just says 331. I
understand the 1kv = 1000 volts but I'm not sure what the values are.
If elected president of the narcoleptic neophyte society, my first
legislation would require manufacturers to print all values in
microfarads so a .001uf cap would say ".001uf" on it.
I don't really have anything to measure oscillations. If I hit the
lotto I will run straight to the Fluke dealer to pick up a $3500
Scopemeter. I'd love to get one of those and sign up to be a cadet at
the Scopemeter academy. Then I could look at oscillations and zoom in
and decode words in an RS232 data stream and be delighted beyond
imagination.
Back on earth, I found a cement block power resistor in my box, a 10W
2.7 ohm. I put it in series between the bridge's positive DC output
and the regulator. Surprisingly the cement resistor does not even get
warm (but it obviously is not blown because the circuit powers up). I
think I may have this wrong as I have seen a load resistor on some
schematics that connects to ground. Should it go one end to the bridge
and one end to ground or in series as I have it now?
The transformer is actually a 26v AC 300ma output and it feels OK to
the touch with no appreciable increase in temperature.
My 7805 is in the TO-220 package and the heatsink is one of those small
clip-on types. I cannot seem to locate a picture of this type.
Something similar would be this:
http://www.ladyada.net/make/x0xb0x/fab/images/to220-heatsink_t.jpg
I do have some white thermal contact paste on it.
There is a reason I'm using a 24v AC transformer. Mainly it's because
I don't want to run two power supplies and the item ultimately being
powered via relay requires 24v AC. I have to be cautious as to how
much I reveal of the entire circuit because -sooner or later it seems
like someone always winds up saying "just go out and buy one off the
shelf, Pokey, you idiot" and I'd like to make this work.
It would be nice if I had a way to measure the current being consumed
but my cheap DMM does not have this function. I found these specs from
the relay's data sheet:
Minimum Contact Load: 10mA @ 5VDC.
Initial Contact Resistance: 100 milliohms max. @ 100mA, 6VDC.
Nominal Coil Power: Contact rating 3 = 200mW.
Contact rating 10 = 450mW.
Throw in a friendly 2n2222, a midrange pic processor, and a max233 and
you pretty much have it. I will try to work on a complete schematic
with the crappy but free software that I have.
Speaking of DMM, why do I read 10V AC at the 5V DC regulator output? I
thought the bridge and filter cap knocked off any AC.
The term "red hot" is misleading as the part does not actually glow. I
will now reference it as "bitch hot" which should clear things up.
Also, this isn't "let's go spend $50 at digikey" to fix the problem. In
my dreams, I order my $3500 Scopemeter from digikey and become a cadet.
This is more like "let's see if Pokey can dig enough parts out of his
junk box to make this relatively simple thing work reasonably".
BTW, the new reading after doing all of above is 170F at the heatsink
tab so there has been an improvement, but still seems high. This is
measured with a non contact thermal thermometer so accuracy is what it
is.
I think I may have a 12V or 13V zener diode around. What about
replacing two of the 1N4004s used in the bridge with two zeners? I
think this would then reduce the voltage to the regulator.
Pokey.