> From:         Mike Singer[SMTP:m_singer@POLUOSTROV.NET]
> Sent:         Friday, January 09, 2004 6:31 PM
> To:   PICLIST@MITVMA.MIT.EDU
> Subject:      Re: [EE]: How to measure light? Phototransistor?

> John N. Power wrote:
>> The following is from National Semiconductor application note
>> AN-72, "The LM3900 - A new current-differencing quad of
>> +/- input amplifiers. (September 1972)",

> It took me less than 1.5 min to download 300K datasheet
> "LM2900/LM3900/LM3301 Quad Amplifiers" February 1995

>> page 3:
>> " The open-loop gain changes only slightly over the complete
>> power  supply voltage range and is essentially independent of
>> temperature changes."

> I've searched the 1995 datasheet with the word "slightly" and
> "essentially": - no occurrences were found. If they used the
> 1972 vocabulary now they would get kicked off the market
> shortly.

      Read my post more carefully. The words "slightly" and
     "essentially" appear in the **application note**, not the data
     sheet.

>> Clearly, there is no expectation that the mirror gain changes
>> significantly.

> What is it "expectation" and how much is "significantly" - 0.01%
> or 10 times? Either this must  be specified in a datasheet or this
> MUST NOT be talked about at all, I think.

My copy of the data sheet, which appears in the linear data book
dated 1982, shows on the fourth page, a series of graphs titled
"Typical Performance Characteristics". In the fourth row from the
top, middle graph of three, there is a plot called  "Mirror Gain".
The horizontal axis is titled "Ta - Temperature (C)". This graph
runs from T= - 55 degrees C to +125 degrees C. The mirror gain
begins at about 1.02 at -55C and levels off at about 1.08 at 125C.
This supports the claim made in the application note.

>> There are numerous examples of triangle, sawtooth and
>> ramp generators in this apnote. A change in mirror gain with
>> input current would cause a supposedly linear output to be
>> curved.

> An app note is just a note, not a datasheet. And what is it
> "supposedly"?

If the ap note is written by the manufacturer, it should be given
some respect. The word "supposedly" is mine; it refers to an
output which one would expect to be linear if the amplifier is
perfect. That is, if the circuit is designed to produce a perfect
linear output, the output is "supposed" to be linear. Any deviation
from linearity is then an error.

>> The applications given in this note would not work if this
>> happened.

> "Triangle, sawtooth and ramp generators" will work but with
> unspecified distortions.

You keep overlooking my disclaimer that the amplifier is not
meant for precision work. The data sheet states on page 5:
"Performance as a DC amplifier using only a single supply is
not as precise as a standard IC op amp operating with split
supplies but is adequate in many less critical applications."

It then goes on to say:

"New functions are made possible with this amplifier which
are useful in single power supply systems. For example,
biasing can be designed separately from the ac gain as was
shown in the "inverting amplifier", the "difference integrator"
allows controlling the charging and the discharging of the
integrating capacitor both with positive voltages, and the
"frequency doubling tachometer" provides a simple circuit
which reduces the ripple voltage on a tachometer output dc
voltage."

*Charging and discharging* an integrating capacitor, both
with positive voltage!!

This amplifier's speciality is its "configuration", which
makes possible many unique circuit topologies with
an unusually small number of components. When used
inside a feedback loop, changes in open-loop gain make
only second order changes in regulated output. In
particular, I have just made a linear voltage regulator which
had to compare a positive output voltage with a positive
reference and produce a positive control output, with only
a single positive supply to work with. This was trivial with
this amplifier.

>> I would breadboard a circuit myself and check this,
>> but I can't do it right now.

> LM3900 production has been discontinued in former
> Millennium by National (Don't know about Motorola). Does this
> chip deserve to get breadboard now?

In addition to Motorola, Texas Instruments also produces it. Parts
from both manufacturers are listed in Newark and Allied catalogs
as of today. Even NTE, the replacement part manufacturer sells
it (NTE992). Your comment about National discontinuing it is a
cheap shot, since National has sold a large part of its linear line
to another manufacturer. This part just went along with the others.

>> I trust National to know something about these matters.

> Yeah, they are great guys. Their modern op amps are far better
> than obsolete discontinued  LM3900 (1972).


>> In any case, the amplifier does have limitations. The most pronounced
>> is its lack of speed. It is suitable only up to audio frequencies. It
>> is not intended for precision work, and that was not the original
>> application. The chip is second sourced by Motorola, has been
>> around for a respectable amount of time, as you can see from the
>> date on the apnote, and is still available. This would not happen
>> if it had serious problems with linearity.

> Regards,
> Mike.

John Power

--
http://www.piclist.com hint: To leave the PICList
mailto:piclist-unsubscribe-request@mitvma.mit.edu