Help! w/op amps

[carbonzit]

It seems that you know all this. Sorry.

No problemo. By the way, no need for that coupling cazapitor, which might only serve to cut off the low frequencies I'm after.

And yes, I did exactly that, listened (literally) to a bunch of different xistors until I found the noisiest.

 

[Roff]

Part of your problem is that the 1k resistor and 10uF cap in your preamp is a high pass filter, rolling off at 20dB/decade starting at 16Hz. Then you follow that with a low pass filter with a corner frequency of 3.2Hz.
The net result is that your noise is filtered as shown in the plot below.You can change the cap to 100uF, and scale both resistors up by a factor of 10, which will lower the corner frequency to 0.16Hz. Then you will have a filter which is flat over a range of 0.16Hz to 3.2Hz, which is a decade and an octave.

 

[carbonzit]

Part of your problem is that the 1k resistor and 10uF cap in your preamp is a high pass filter, rolling off at 20dB/decade starting at 16Hz. Then you follow that with a low pass filter with a corner frequency of 3.2Hz.
The net result is that your noise is filtered as shown in the plot below.You can change the cap to 100uF, and scale both resistors up by a factor of 10, which will lower the corner frequency to 0.16Hz. Then you will have a filter which is flat over a range of 0.16Hz to 3.2Hz, which is a decade and an octave.

OK, so use 100 uF, 10M and 10K, right? Leave the low-pass filter stuff as is? I'll try it.

By the way, nobody has commented on why this doesn't work with the inverting input feedback connected (dotted line in schematic). If connected, output goes dead. So as it is, the inverting input is just floating. Not a good thing.

 

[Roff]

OK, so use 100 uF, 10M and 10K, right? Leave the low-pass filter stuff as is? I'll try it.

By the way, nobody has commented on why this doesn't work with the inverting input feedback connected (dotted line in schematic). If connected, output goes dead. So as it is, the inverting input is just floating. Not a good thing.

The inverting input has to be connected to the output. otherwise, you have what amounts to sophisticated junk. Maybe it wasn't working because of the inadvertent highpass filter I just pointed out.
You might even want to make that cap in the preamp 1000uF, which will push your corner down to 16mHz.
You can also add a 4.7nF cap across the 10Meg feedback resistor, which increase the slope of your lowpass filter to 60dB per decade, out to about 400Hz.
Change the filter resistors up by a factor of 100 (330k each), and the caps down by a factor of 100 (330nF and 100nF). This is because the big cap should really be nonpolarized, and 330nF is easier to get. This also improves the filter performance at higher frequencies, where it starts to be nonideal.
Keep in mind that none of this may matter, if the noise does not exist at these low frequencies.

 

[MrAl]

Hello again,


Oh yeah i forgot to mention that the reverse voltage rating for base emitter junctions of transistors is usually fairly low so i wonder how a source made using a transistor like this is going to hold up over time. The zener circuit does not have this problem because the zener is actually made for this breakdown action.
Check out the rating of your transistor and then wonder how it is going to work over time