Reply by Jan Panteltje April 6, 20122012-04-06
On a sunny day (Thu, 5 Apr 2012 19:47:29 -0500) it happened "Tim Williams"
<tmoranwms@gmail.com> wrote in <jlleet$v2p$1@dont-email.me>:

>Yuck, an LDO. Don't you read Joerg at all? ;-) > >It could very well be audio. Hook up a speaker and see :)
No kidding, some time ago I had a shortwave station coming in, it faded in and out...
>The op-amp >probably won't be making much, but the BJT junctions will rectify RF if it's >around. I wouldn't think a little circuit like this would really be prone >to anything, even with a 20' power cable hanging off it. > >More likely, it's amplified residual noise from the reference. References >tend to be quite noisy, even with a bypass cap. MCP1525 says 500uVpp from >10Hz-10kHz, which will easily make >> 50mV at the op-amp output. It should >be heavily weighted towards low frequencies though -- it's 1/f noise to >begin with (including the 1uF filter cap), and the (essentially) open loop >op-amp integrates that further, making more-or-less brown noise (1/f^2).
That is a very good point, and I admit I overlooked that 500uVpp, so I will investigate it a bit more. Maybe a simple RC filter after the MCP1525 before the opamp input will decrease this, and will not affect response there. Lots of things to investigate this weekend. The EEPROMS and some later silicon versions of the PIC 18F14K22 arrived too, along with some other stuff that I will address some other time I hope. I finished the design of the switching mechanism that switches in the batteries if the input power from the wall wart fails, but still need to build it and test it. That is next.
Reply by Tim Williams April 5, 20122012-04-05
Yuck, an LDO.  Don't you read Joerg at all? ;-)

It could very well be audio.  Hook up a speaker and see :)  The op-amp 
probably won't be making much, but the BJT junctions will rectify RF if it's 
around.  I wouldn't think a little circuit like this would really be prone 
to anything, even with a 20' power cable hanging off it.

More likely, it's amplified residual noise from the reference.  References 
tend to be quite noisy, even with a bypass cap.  MCP1525 says 500uVpp from 
10Hz-10kHz, which will easily make >> 50mV at the op-amp output.  It should 
be heavily weighted towards low frequencies though -- it's 1/f noise to 
begin with (including the 1uF filter cap), and the (essentially) open loop 
op-amp integrates that further, making more-or-less brown noise (1/f^2).

Tim

-- 
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms

"Jan Panteltje" <pNaonStpealmtje@yahoo.com> wrote in message 
news:jlk1ho$ttf$1@news.albasani.net...
> How about it? Experiments of the third kind , take 999999. > Update > opamp and PIC power supply design: part 1 > > > > PNP > --------------------------- e > c--------------------------------------------------------------- + 3.5V > +/- 1mV @ 20mA noise free to PIC and EEPROMS. > | | b Q1 | | | > | > +4 to + 8 V | | | | | | > | > | | R1 | | | [ ] 40k > | > | [ ] 100k | | | | > | > | | | | | + | +2.5V > | > === 470u | +2.1V c c / > --------- | > --- |------ b NPN NPN b-----< | | +2.5V | > MCP1525 | | > | | e Q2 e Q3 \ |------)-------------| > |-----| > | --- | | 1/4 | | | | > | | > | \ /-> ------ TCL274| | > | --------- | > | --- LED | | | === > | === > | | green [ ] 1k | [ ] | 1u > | --- 470u > | | | R2 | | 100k | > | | > ------------------------------------------------------------------------------------------------- > > Transistors: PNP BC557B, NPN BC548B. > The 'Tim' capacitor from the opamp output to its - input is missing, > even a small one ruins step load recovery. > There is LF noise from the the CMOS opamp on its output, looks like audio, > but is really just noise in the audio range, and it does not bother me a > bit. > The rather large output caps maybe help too. > The MCP1525 is a Microchip 2.5V precision reference. > > I did some impulse response test, > and, as the opamp is basically open loop here, > needed to know the stability of the control loop, > and how the system handles impulses. > The PIC will cause supply current variations, > but especially the EEPROMs will cause 5mA or bigger current impulses > during a write. > Here a scope screenshot: > http://panteltje.com/pub/opamp_output_20mA_impulse_load_IMG_3426.JPG > This is a 100 x per seconds repeated 20mA (resistor switched on output > with a BJT). > The response is very nice, opamp output goes 100mV negative during load, > (negative disables Q3, and then Q2 delivers more base current to Q1 to > compensate for > any drop in output voltage. > I have been unable to detect any drop in output voltage with my > scope at 10mV / division AC, no surprise due to the huge loop gain. > The 'noise' is CMOS opamp low frequency noise as far as I know[1]. > With this this part of the supply has passed design review... > > Looks so much like 'audio' maybe it is trying to say something?
Reply by Jan Panteltje April 5, 20122012-04-05
How about it? Experiments of the third kind , take 999999.
Update
  opamp and PIC power supply design: part 1



                                 PNP
    --------------------------- e   c--------------------------------------------------------------- + 3.5V +/- 1mV @ 20mA noise free to PIC and EEPROMS.
              |         |         b Q1   |          |       |                             |
+4 to + 8 V   |         |         |      |          |       |                             |
              |         |  R1     |      |          |      [ ] 40k                        |
              |        [ ] 100k   |      |          |       |                             |
              |         |         |      |          |  +    | +2.5V                       |
             === 470u   | +2.1V   c      c         / |------|              ---------      |
             ---        |------ b NPN  NPN b-----<   |      |      +2.5V  | MCP1525 |     |
              |         |         e Q2   e Q3      \ |------)-------------|         |-----|
              |        ---        |      |      1/4 |       |       |     |         |     |
              |        \ /->       ------     TCL274|       |       |      ---------      |
              |        --- LED        |             |       |      ===         |         ===
              |         | green      [ ] 1k         |      [ ]      | 1u       |         --- 470u
              |         |             |  R2         |       | 100k  |          |          |   
    ------------------------------------------------------------------------------------------------- 

Transistors: PNP BC557B, NPN BC548B.
The 'Tim' capacitor from the opamp output to its - input is missing,
even a small one ruins step load recovery.
There is LF noise from the the CMOS opamp on its output, looks like audio,
but is really just noise in the audio range, and it does not bother me a bit.
The rather large output caps maybe help too.
The MCP1525 is a Microchip 2.5V precision reference.

I did some impulse response test,
and, as the opamp is basically open loop here,
needed to know the stability of the control loop,
and how the system handles impulses.
The PIC will cause supply current variations,
but especially the EEPROMs will cause 5mA or bigger current impulses during a write.
Here a scope screenshot:
 http://panteltje.com/pub/opamp_output_20mA_impulse_load_IMG_3426.JPG
This is a 100 x per seconds repeated 20mA (resistor switched on output with a BJT).
The response is very nice, opamp output goes 100mV negative during load,
(negative disables Q3, and then Q2 delivers more base current to Q1 to compensate for
any drop in output voltage.
I have been unable to detect any drop in output voltage with my
scope at 10mV / division AC, no surprise due to the huge loop gain.
The 'noise' is CMOS opamp low frequency noise as far as I know[1].
With this this part of the supply has passed design review...

Looks so much like 'audio' maybe it is trying to say something?