Error in LM3886.pdf re: voltage drop?

[officeboy]

Maybe I’m missing part of a larger picture here, but this seems like simple math, and i can't see where I’m getting the wrong answer.

The picture below is clipped from page 20 of lm3886.pdf, basically shouldn't 17.9v+4(dov) = 21.9v, not 21.0V ?

[officeboy]

hmm, then on page 3 it says that the typical dropout voltage on + rail is 1.6 (max of 2.0), and on the - rail typical is 2.5V (max of 3)

(at -/+28V -/+100ma)

[officeboy]

Can anyone confirm or deny my… er, craziness?


*bump*

[jan.didden]

If I understand it correctly, the 4V DO is at max load (like 4 amps). The latter case apparently is with 100mA load? It seems reasonable that the DO will be much lower in that case.
Or what exactly is that 100mA spec?

Jan Didden

[officeboy]

Pretty quick reply too.

Dustin, there is nothing wrong with your math. Our wording is misleading.
Unfortunately, this is an elderly data sheet and the author is no longer around,
so I cannot verify anything with him. But I will see if I can walk through
each of the issues.

1. The most misleading item is the "Add the dropout voltage (4V for LM3886) . .
. " statement. That should have either said "approximately" or referred you to
the 4 ohm clipping curve on page 10.

2. The 4 ohm clipping curve on page 10 shows drop out vs supply voltage. That
is what you must add to the calculated 17.9V level to get the final supply
voltage. Since this is a curve and not an equation, you cannot get a very
precise number. But at a supply voltage of +/-25V the drop out is about 3.5V.
At a supply of +/-20V the drop out is about 3V. So the 21V the example ended
with is pretty accurate - it is just not too clear how he reached that 21V.

3. The page 3 drop out spec is accurate, correct and guaranteed. But it is
not very meaningful in an application because not many users would buy a hefty
part like the LM3886 and run it at 100ma load. However, the Electrical
Characteristics table is our gurantee. The conditions in the Table should be
specified so that an individual user can verify the spec. And 100ma is enough
current for us to insure we have a good junction and a good die-bond inside the
part.

4. The only thing that goes into the ABS MAX section is stuff that will cause
damage if exceeded. Ignoring the drop out voltage can give you distortion but
is not damaging to the part. So drop out is not an Absolute Maximum spec.

5. All of our curves are average values unless clearly specified as a max or a
min. A real part can vary +/-10% to 15% from the average. So power
calculation are accurate only to the extent that your particular part is an
"average" part. Supply voltages and variations due to high line and low line
will vary at least as much as the part will vary. If you design right at the
limit, you can expect some units to exhibit distortion. Your calculation can
get you close but they are only a good starting point. If you want your
production to exhibit NO distortion, you have to stay away from the limits.

One last item. Go to our Audio Web site at
http://www.national.com/appinfo/audio/. Under the center DESIGN column, scroll
down to the DESIGN GUIDE heading. The first two items are a Design Guide table
and an User Guide explaining how to use the Tables. This design guide will give
you precise (pre-calculated) values. So you don't need to go through the
exercise of doing the calculations. There are also some application notes
discussing bridging and bridge-paralleling if you need higher power levels.

I hope this is helpful.

[jan.didden]

..but the bottom line is that the 100mA DO spec is pretty useless in actual use. And the statement that the DO is depending on supply voltage is slightly misleading. It is depending on load current, and of course, with lower supply, your max load current will be less, thus the DO is less. But only because you draw less current, not because of the lower supply.
Not very convincing...

Jan Didden

[officeboy]

Yeah, they don't make it very easy for me to create my own speadsheet with a power vs. resistance graph.
I would really like a better power calculation for 22vac or about 33.5 unloaded DC supply and 5.4 ish ohms resistance.