| Pages:
1
..
8
9
10 |
m1tanker78
National Hazard
Posts: 685
Registered: 5-1-2011
Member Is Offline
Mood: No Mood
|
|
Page 110 of the datasheet shows that the RC components of the ADC are internal to the device. Using your ICG trace as baseline for scope/probe/circuit
noise, there's still quite a bit of register imbalance showing on your OS trace. It just struck me as odd that your scope capture and ADC capture
apparently aren't in agreement.
Dave Allmon makes no mention of register imbalance in the article you linked to. OK, this is beginning to smell like a conspiracy! 
I like the idea of the cutout for a cold tip. In my early musings of CCD cooling, I started to lean toward a cold bar spanning the back of the chip
and a TEC on either side. In fact, my prototype CCD PCB has the CCD offset several mm from the board. Preventing condensation/frost is tricky
business. A sealed, purged, windowed enclosure for the CCD board with sealed connectors for conditioned and isolated signals is another level of
tricky business.
For the time being, I ordered a 'stage lighting' laser as well as a laser which may have temperature and current control. The guy who sold it to me
doesn't know much about it and it was cheap enough to give it a chance. Before I accidentally (foolishly) zapped my reds, I was working on temperature
control for the laser. I want to see what kind of results can be obtained with a 532nm ~80mW laser + temperature control. I'm still trying to avoid
expensive filters and now I'll REALLY have a reason to do so if I switch to 532nm illumination. I won't kid myself, if I ever get to the point where
the only thing missing is a good laser line filter, I'll go ahead and buy the damned thing.
Chemical CURIOSITY KILLED THE CATalyst.
|
|
|
tvaettbjoern
Harmless
Posts: 28
Registered: 25-2-2016
Member Is Offline
Mood: No Mood
|
|
The signal conditioning David Allmon is performing is not about the register imbalance, but to make sure the ADC's range is fully exploited.
Well spotted. I didn't even think of looking "inside" the MCU. I literally have no idea if the capacitors on the input have an effect. But if they do
that could certainly explain the discrepancies. I would also lend my own (lack of) consistency of method to the explanation. The captures are not
really comparable:
If we are talking about this capture:
then yes there's a lot of noise - 0.5V is not far off and ICG is cleaner than OS.
To the CCD's defense this was a measurement with the CCD connected directly to the GPIOs of the nucleo board, with free wires for all signals
including V+ and gnd.
Then there's this:
there's now a hex-inverter between the CCD and the MCU, and the IC's (both the 74HC04 and the CCD) are decoupled - poorly though. The noise is lower,
and as before we see more noise on OS than on ICG, but the difference is not as dramatic as before.
And finally we have something like this:
This was obviously not done on the scope. I can not recall if this was made with an SMD board or the very first through-the-hole boards. On the
SMD-version I paid better attention to the decoupling caps and tried to keep analog and digital signals separate.
The recordings of the Na-doublet and the CFL in the previous post was made with an SMD-version, and a firmware where I changed GPIOs to move the
digital signals physically away from the ADC-input and Vref etc on the MCU. (I also pulled a few pins down for a virtual gnd, but I'm not sure this
actually made much of a difference).
If correcting the register imbalance can remove something of the remaining noise I will be happy to apply it. Your pictures certainly look very
promising.
In the beginning I also thought about having a cold bar behind the CCD. Eventually I dismissed the idea for being too impractical. Then I read about
the Audine camera, and I think their solution even solves (or can help solve) the problem of aligning the CCD. Condensation will of course be a
problem - but I'm planning to seal the entire spectrograph off anyway leaving only the optical fiber port open (which is closed by the fiber most of
the time anyway), so a few (large) packs of silica will hopefully keep everything dry.
[Edited on 6-3-2017 by tvaettbjoern]
|
|
|
NitratedKittens
Hazard to Others
Posts: 131
Registered: 13-4-2015
Location: In the basket with all the other kittens
Member Is Offline
Mood: Carbonated
|
|
Quote: Originally posted by bfesser  | Every time I see the title of this thread, I read it as 'Cheap, Low-Resolution, Ram<em>e</em>n Spectroscopy':
<img src="http://suburbanfairytale.com/images/top-ramen.jpg" width="600" />
[Edited on 3/7/13 by bfesser] |
Now that's what I call a ramen spectrum
Basket of kittens for you ........BOOM
|
|
|
m1tanker78
National Hazard
Posts: 685
Registered: 5-1-2011
Member Is Offline
Mood: No Mood
|
|
| Quote: Originally posted by tvaettbjoern | | The signal conditioning David Allmon is performing is not about the register imbalance, but to make sure the ADC's range is fully exploited.
|
I know. In his article he also touches on subtracting dark frames, flat field, etc. but no mention of register imbalance. I'm not knocking him for it,
just wondering if I have a CCD with extraordinary R.I.
I received one of the green lasers and began setting up a mock-up on my desk (my prototype spectrometer is dismantled at the moment). I didn't get
very far when I fired up the laser and looked at the reflected light from a cuvette of powdered aspirin through a filter. It looked red and too
luminous to be Raman. With the filter still covering my eye, I put other items in the laser path and almost everything exhibits strong fluorescence.
That's a real bummer but I know some commercial Raman rigs have green lasers so it must not be a deal breaker. Still, it makes me want to jump back to
650nm and work something out so I can take scans of the Stokes side.
Chemical CURIOSITY KILLED THE CATalyst.
|
|
|
tvaettbjoern
Harmless
Posts: 28
Registered: 25-2-2016
Member Is Offline
Mood: No Mood
|
|
It's true that fluorescence will be a problem for many molecules. I hope it won't be too bad, otherwise all money spent on filters and lasers are
wasted 
A thought just struck me. Wouldn't a 750 kHz low-pass RC filter between the adc and the CCD fix at least the 2.0 MHz fM transients?
|
|
|
m1tanker78
National Hazard
Posts: 685
Registered: 5-1-2011
Member Is Offline
Mood: No Mood
|
|
| Quote: Originally posted by tvaettbjoern | | A thought just struck me. Wouldn't a 750 kHz low-pass RC filter between the adc and the CCD fix at least the 2.0 MHz fM transients?
|
A 1st-order RC filter presents a rolloff of roughly -20dB/decade. If the signals were neat sinusoids, a typical 2nd-order RC filter with
fc=750KHz would attenuate the 750-ish KHz (desired) component by about 13dB while the 2MHz (undesirable) component would be attenuated by
about 23dB. In this scenario, you'd need to amplify the filtered signal in order to boost the gain.
The problem here is that the 'video' output signal of the CCD is arbitrary in shape. The higher frequency content is an integral part of the valid
overall signal being measured. Filtering it away could distort the measurement to a certain degree. The theory behind it is a bit outside the scope of
this thread but if you're interested, search around for square wave theory or similar.
If all you want is to nip the 2MHz component, a reasonable notch filter can be constructed with an inductive component (i.e. an LC filter).
Chemical CURIOSITY KILLED THE CATalyst.
|
|
|
chat926
Harmless
Posts: 1
Registered: 9-8-2017
Member Is Offline
Mood: No Mood
|
|
Hi, can this experiment be done with a 633nm HeNe laser (laser power is 1mW). My detector is a home made spectrometer constructed with a reflective
diffraction grating, concave mirror and a CCD detector.
Also, what is the minimum laser power needed?
[Edited on 9-8-2017 by chat926]
|
|
|
tvaettbjoern
Harmless
Posts: 28
Registered: 25-2-2016
Member Is Offline
Mood: No Mood
|
|
This guy has built a very impressive Raman microscope with a HeNe-laser (and a 532nm green as well):
http://jm-derochette.be/Raman_Microscope.htm
so yes it can definitely be done (and according to one of my physics-colleagues a HeNe-laser is a cheap way to get an almost monochromatic laser)
with low laser power you'll of course need longer integration time, and so cooling of the CCD will probably be required. you should also be concerned
about focus and alignment in the raman-probe. I think (but I'm not sure) that a 180° backscattering setup is the least wasteful in terms of
Raman-signal.
|
|
|
| Pages:
1
..
8
9
10 |
![[*]](images/xpblue/default_icon.gif){kind=icon}
