Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

Filters - do you think you know what you bought?

Filters Report
  • Please log in to reply
15 replies to this topic

#1 dobrychemik

dobrychemik

    Lift Off

  • -----
  • topic starter
  • Posts: 7
  • Joined: 18 Oct 2021

Posted 19 October 2021 - 07:03 AM

Hi everyone!
I’m new here, but I’m not a newbie to astro. I’m a chemist with some experience in spectroscopy and have access to two spectrophotometers. Therefore, some time ago I started to verify the parameters of astronomical filters. During almost two years of measurements, I collected a lot of spectra. Moreover, I have found manufacturers’ declarations often differ enormously from reality. In this thread I will present you the results of my measurements from time to time.
Oskar

 

EQUIPMENT

SPEC1 - Shimadzu UV-2101PC (the better one)
SPEC2 - Hitachi U-2900

 

PROCEDURE

1. Warming up of the spectrophotometer, especially of both lamps - 15 minutes.

 

2. Automatic calibration - baseline measurement over the entire spectrum measurement range.

 

3. Measurement parameters:
- temperature 20 ⁰C
- SPEC1, standard measurements: 300-850 nm, resolution 0.2 nm, slit 0.2 nm
- SPEC1, high resolution measurements: resolution 0.05 nm, slit 0.05 nm
- SPEC2, standard measurements: 200-1100 nm, resolution 1 nm, slit 1.5 nm
- SPEC2, high resolution measurement: …………………………….

 

4. Four-fold measurement of the empty spectrum (T = 100%) and two-fold measurement of the spectrum for the masked detector (T = 0%). These spectra are used to calibrate the filter transmittance measurements more precisely.

 

5. Control measurement of the spectra of my two filters: Optolong SII 6.5nm and Optolong L-eX. These measurements are used to check the correctness of the wavelength scale. If the difference in the position of the transmission bands is greater than 0.1 nm compared to the measurements taken as the reference*, then all measurements from a given session are appropriately rescaled.
Maximum wavelength error for standard measurements using SPEC1: 0.1 nm.
The above measurement calibration procedure allows me to obtain repeatable measurement results for a given filter and a reliably compare the spectra obtained during different measurement sessions.

 

* Reference measurements: the spectra of the selected two filters were measured on two different spectrophotometers with the maximum possible resolution, minimum slit width and minimum scanning speed. Both devices were controlled by separate computers with different control software. The obtained results were fully consistent, i.e. the position of the transmission bands was the same with accuracy of 0.05 nm. This procedure was done twice. In this situation, I assumed that these measurements were correct and the parameters of the selected filters determined in this way would constitute a reference for all subsequent measurements.

 

6. Angular measurements and filter efficiency simulations
In order to simulate the efficiency of the filter depending on the lens speed and the size of the central obstruction of the telescope, it is necessary to determine the effective refractive index. For this purpose, I measure the spectra in high resolution (0.05 nm) for the filter tilted by the given angle: 8, 10 13 and 15.2 degrees and again in the same order to minimize the measurement error. I count the refractive index to two decimal places with an estimated error of +/- 0.02.
The relationship between the shift of the transmission bandwidth and the refractive index of the filter is given by the formula:

 

(I can't paste the graphic, so I attach as a separate file)

 

In the efficiency simulation model I developed, I divide the entire surface of the lens into many concentric circles and I count the shift separately for each of them. The widths of these circles are calculated automatically so that each gives an shift greater than the previous one by a value equal to the spectral resolution of the measured spectrum (0.2 nm). Thanks to this approach, it is very easy for me to count the resultant, shifted spectrum in Excel, because I have no problem with the changing (shrinking) wavelength scale. The steps present in the simulations are an artifact resulting from the spectral resolution of the spectra (0.2 nm). Higher resolution spectra and calculations would result in a smoother curve, but its averaged course would not change significantly. You can also smooth the resulting curve, but any differences in the results are insignificant.

 

Attached Thumbnails

  • formula.jpg

  • Dagobert, DavidWasch, Norup and 8 others like this

#2 dobrychemik

dobrychemik

    Lift Off

  • -----
  • topic starter
  • Posts: 7
  • Joined: 18 Oct 2021

Posted 19 October 2021 - 07:04 AM

Principles of “conscious filtering”

 

1. Every interference filter is different, there are no two identical ones, even if they are from the same manufacturer, from the same batch.

 

2. If the manufacturer presents the spectrum of a given filter model on his website, you can be sure that it is the spectrum of the best selected filter. If you buy this model, your copy will almost certainly be worse.

 

3. A user who does not have access to a spectrophotometer cannot verify the quality of the filter, thanks to which most manufacturers can sell defective filters as full-fledged with impunity, because even they usually allow to obtain satisfactory results, but at the cost of much more time

 

4. The quality of the filter should be considered in the context of the set with the telescope - the effectiveness of the interference filter depends on the parameters of the telescope (its lens speed and the size of the central obstruction)

 

5. Every filter is different…


  • littleredstar, ChrisMoses and mikemarotta like this

#3 dobrychemik

dobrychemik

    Lift Off

  • -----
  • topic starter
  • Posts: 7
  • Joined: 18 Oct 2021

Posted 19 October 2021 - 07:16 AM

For a good start, I present a test of a set of Baader narrowband filters dedicated to work with high lens speed optics:

Baader Ha 3.5 nm F/2 CMOS
Baader OIII 4nm F/2 CMOS
Baader SII 4nm F/2 CMOS

 

The reports are available here:

 

https://drive.google...Yfk?usp=sharing


  • psandelle, littleredstar, ChrisMoses and 2 others like this

#4 TOMDEY

TOMDEY

    Voyager 1

  • *****
  • Posts: 10,921
  • Joined: 10 Feb 2014
  • Loc: Springwater, NY

Posted 19 October 2021 - 08:00 AM

Very thorough indeed! Do you also take into account whether the modeled optical system is telecentric on the image side (very few are)... which introduces additional angular deviations from normal incidence across the field and should be factored into your resultant summation/integration. Sometimes this effect overwhelms the pupil mapping contributions.

 

Nearly all of the arcane modeling nuances have very little effect, except for very narrowband applications, (like those you are characterizing!). The more traditional visual and broaderbands will perform close to factory specs. I measured a few tubs of filters a few years ago and found that many had degraded... so trashed them. This is especially true of the high-performance ones. Some others hang right in there as good as the day they emerged from the coater. This makes buying used filters a craps shoot.

 

Final potential performance effect is the Doppler Shift of the targeted nebula... there's the local component and for stuff in our own galaxy and then the universal expansion component for farther galaxies. Also rotations of clouds and even individual stars. Add to that polarization effects and it gets just about as immersive as we care to account. These interesting effects are most easily witnessed on our own sun viewed in Hα and the proms off the E and W limbs. Thankfully, those etalon filters are either tilt or pressure-tunable in telecentric-space.

 

Nice work; thanks for sharing!   Tom


  • Jon Isaacs likes this

#5 cytan299

cytan299

    Apollo

  • *****
  • Posts: 1,144
  • Joined: 24 Nov 2014

Posted 19 October 2021 - 08:07 AM

For a good start, I present a test of a set of Baader narrowband filters dedicated to work with high lens speed optics:

Baader Ha 3.5 nm F/2 CMOS
Baader OIII 4nm F/2 CMOS
Baader SII 4nm F/2 CMOS

 

The reports are available here:

 

https://drive.google...Yfk?usp=sharing

Looking at the OIII and SII locations which are not even close to the filter peaks ( or am I reading your plots wrong) what does it mean? Is it supposed to be this way?

 

cytan

 

Suggestion: Is it possible for you to put grid lines on your plots so that we can see how high the peaks are easily by reading off the y axis. Thanks!


Edited by cytan299, 19 October 2021 - 08:46 AM.


#6 PiotrM

PiotrM

    Skylab

  • *****
  • Posts: 4,139
  • Joined: 03 Jan 2010
  • Loc: Poland

Posted 19 October 2021 - 08:56 AM



Looking at the OIII and SII locations which are not even close to the filter peaks ( or am I reading your plots wrong) what does it mean? Is it supposed to be this way?

 

cytan

 

Suggestion: Is it possible for you to put grid lines on your plots so that we can see how high the peaks are easily by reading off the y axis. Thanks!

f/2 filters will have shifted bandpass when measuring head on. When you direct the light beam at an angle (like in a f/2 scope) the bandpass will shift, theoretically to the desired wavelength. If you look through your interference filters at an angle you will see they pass different "color".

 

Check the "optics without central obstruction" chart. At f/2 it has 70% total [S II] transmission from both peaks but at f/4 it's already 20% as the blueshift isn't strong enough to move the band enough.

 

And another part that for RASA it has above 80%. In a telescope without central obstruction (lens, refractor) even set to like f/2.2 not every light ray will come at such angle - those closer to the centre will come at lower angle which can cause that you get star light but not the emission line. in RASA, Hyperstar and alike those centre rays are blocked by the obstruction.


Edited by PiotrM, 19 October 2021 - 09:00 AM.

  • astrokeith, OldManSky and MaciejW like this

#7 astrokeith

astrokeith

    Apollo

  • -----
  • Posts: 1,432
  • Joined: 14 Mar 2010
  • Loc: Surrey, UK

Posted 19 October 2021 - 09:16 AM

As PiotrM points out, its not that simple.

 

An F/2 filter should be optimised to get maximum energy through the passband, integrated over the F/2 cone of light. This will almost inevitably mean that transmission at 90 degrees to the surface will be bad. It is necessary to measure at an angle of incidence approaching the F/2 cone angle.

 

This exposes a 'can of worms'. An F/2 filter quoted as say 4nm bandwidth, can only be 4nm at a particular angle. Integrated over the whole f/2 cone angle it will be very considerably wider. But this is perhaps a necessary evil if fast telescope are to be  used. The only solution is to use a telecentric optical arrangement at the filter, which I dont believe is available for amateur systems.


  • ChrisMoses and OldManSky like this

#8 dobrychemik

dobrychemik

    Lift Off

  • -----
  • topic starter
  • Posts: 7
  • Joined: 18 Oct 2021

Posted 19 October 2021 - 02:51 PM

Very thorough indeed! Do you also take into account whether the modeled optical system is telecentric on the image side (very few are)... which introduces additional angular deviations from normal incidence across the field and should be factored into your resultant summation/integration. Sometimes this effect overwhelms the pupil mapping contributions.

I don't model telecentric lenses because it's virtually impossible. Perfect telecentric lenses do not exist. Really existing systems are only approximately telecentric and only the manufacturer knows exactly what the characteristics of the light cone behind the last lens are. If the lens were fully telecentric, the distance between it and the sensor could be freely changed without any impact on the image obtained. It is obvious that this is not the case. I'd rather not model it at all than just anyhow smile.gif


Edited by dobrychemik, 19 October 2021 - 02:53 PM.


#9 cuivienor

cuivienor

    Vanguard

  • *****
  • Posts: 2,104
  • Joined: 06 Oct 2010
  • Loc: Tokyo, Japan

Posted 28 November 2021 - 10:58 PM

Why isn't this thread pinned or something?? This is very important information! I wish I had seen this before buying my own Baader Highspeed Ultra NB filters.

 

I've been looking at the discussion on Astropolis, and I'm glad to see that a further OIII filter seems to be much better (flat top, not shifted too much to the red).

 

I just tested my Baader filters on my C6 Hyperstar (with massive CO). While Ha and SII seem fine, I'm getting much, much worse results on OIII, almost no structure comes through. Attached are two images taken under similar conditions:

- Baader 4nm High speed on the left, 3 hours integration at F2

- Astrodon 3nm on the right, 3 hours integration at F4

 

I'm shocked at how bad the Baader is...

 

Attached Thumbnails

  • Baader vs Astrodon.jpg


#10 Norup

Norup

    Lift Off

  • -----
  • Posts: 7
  • Joined: 16 Jan 2009

Posted 03 December 2021 - 05:31 AM

Very nice analysis!

I came here to post my own scans of the different and older model Baader f/2 Highspeed H-alpha and O-III filters, and found this post.

My scan is not as carefully performed, but should still be reasonably ok. FWHM of scan bandpass is 0.5nm, and wavelength calibration error below 0.5nm. It has been made in parallel light, perpendicular to the filter. A wider scan 320-1080nm was done, with no significant leaks found.

It did not occur to me to measure the actual effective refractive index. At first I just guesstimated a value, but in the figure below, dobrychemiks value of n=1.93 has been used for calculating the shift for the telescope intended for use, the f/2.8 Takahashi Epsilon 180 ED. The filters appear to be an ok match for this telescope and even faster ones.

In the figure, the filled curve is for parallel beam and the dashed for the e-180, taking the changing f-ratio between central obscuration and mirror edge into account.

NB_scan_shift.png


Edited by Norup, 03 December 2021 - 05:34 AM.


#11 mikemarotta

mikemarotta

    Messenger

  • -----
  • Posts: 420
  • Joined: 09 Dec 2019
  • Loc: Austin, Texas

Posted 03 December 2021 - 06:22 AM

 I’m a chemist with some experience in spectroscopy and have access to two spectrophotometers. Therefore,

 

Thank you for your work and your analysis.


Edited by mikemarotta, 03 December 2021 - 06:27 AM.


#12 cuivienor

cuivienor

    Vanguard

  • *****
  • Posts: 2,104
  • Joined: 06 Oct 2010
  • Loc: Tokyo, Japan

Posted 04 December 2021 - 02:09 AM

Very nice analysis!

I came here to post my own scans of the different and older model Baader f/2 Highspeed H-alpha and O-III filters, and found this post.

My scan is not as carefully performed, but should still be reasonably ok. FWHM of scan bandpass is 0.5nm, and wavelength calibration error below 0.5nm. It has been made in parallel light, perpendicular to the filter. A wider scan 320-1080nm was done, with no significant leaks found.

It did not occur to me to measure the actual effective refractive index. At first I just guesstimated a value, but in the figure below, dobrychemiks value of n=1.93 has been used for calculating the shift for the telescope intended for use, the f/2.8 Takahashi Epsilon 180 ED. The filters appear to be an ok match for this telescope and even faster ones.

In the figure, the filled curve is for parallel beam and the dashed for the e-180, taking the changing f-ratio between central obscuration and mirror edge into account.

attachicon.gifNB_scan_shift.png

Those spectra look excellent and pre-shifted just well, including OIII. Thanks for sharing!



#13 aukropov

aukropov

    Lift Off

  • *****
  • Posts: 13
  • Joined: 12 Dec 2020
  • Loc: Oshawa, ON, Canada

Posted 28 December 2021 - 11:00 AM

Hi everyone!
I’m new here, but I’m not a newbie to astro. I’m a chemist with some experience in spectroscopy and have access to two spectrophotometers. Therefore, some time ago I started to verify the parameters of astronomical filters. During almost two years of measurements, I collected a lot of spectra. Moreover, I have found manufacturers’ declarations often differ enormously from reality. In this thread I will present you the results of my measurements from time to time.
Oskar

 

EQUIPMENT

SPEC1 - Shimadzu UV-2101PC (the better one)
SPEC2 - Hitachi U-2900

 

PROCEDURE

1. Warming up of the spectrophotometer, especially of both lamps - 15 minutes.

 

2. Automatic calibration - baseline measurement over the entire spectrum measurement range.

 

3. Measurement parameters:
- temperature 20 ⁰C
- SPEC1, standard measurements: 300-850 nm, resolution 0.2 nm, slit 0.2 nm
- SPEC1, high resolution measurements: resolution 0.05 nm, slit 0.05 nm
- SPEC2, standard measurements: 200-1100 nm, resolution 1 nm, slit 1.5 nm
- SPEC2, high resolution measurement: …………………………….

 

4. Four-fold measurement of the empty spectrum (T = 100%) and two-fold measurement of the spectrum for the masked detector (T = 0%). These spectra are used to calibrate the filter transmittance measurements more precisely.

 

5. Control measurement of the spectra of my two filters: Optolong SII 6.5nm and Optolong L-eX. These measurements are used to check the correctness of the wavelength scale. If the difference in the position of the transmission bands is greater than 0.1 nm compared to the measurements taken as the reference*, then all measurements from a given session are appropriately rescaled.
Maximum wavelength error for standard measurements using SPEC1: 0.1 nm.
The above measurement calibration procedure allows me to obtain repeatable measurement results for a given filter and a reliably compare the spectra obtained during different measurement sessions.

 

* Reference measurements: the spectra of the selected two filters were measured on two different spectrophotometers with the maximum possible resolution, minimum slit width and minimum scanning speed. Both devices were controlled by separate computers with different control software. The obtained results were fully consistent, i.e. the position of the transmission bands was the same with accuracy of 0.05 nm. This procedure was done twice. In this situation, I assumed that these measurements were correct and the parameters of the selected filters determined in this way would constitute a reference for all subsequent measurements.

 

6. Angular measurements and filter efficiency simulations
In order to simulate the efficiency of the filter depending on the lens speed and the size of the central obstruction of the telescope, it is necessary to determine the effective refractive index. For this purpose, I measure the spectra in high resolution (0.05 nm) for the filter tilted by the given angle: 8, 10 13 and 15.2 degrees and again in the same order to minimize the measurement error. I count the refractive index to two decimal places with an estimated error of +/- 0.02.
The relationship between the shift of the transmission bandwidth and the refractive index of the filter is given by the formula:

 

(I can't paste the graphic, so I attach as a separate file)

 

In the efficiency simulation model I developed, I divide the entire surface of the lens into many concentric circles and I count the shift separately for each of them. The widths of these circles are calculated automatically so that each gives an shift greater than the previous one by a value equal to the spectral resolution of the measured spectrum (0.2 nm). Thanks to this approach, it is very easy for me to count the resultant, shifted spectrum in Excel, because I have no problem with the changing (shrinking) wavelength scale. The steps present in the simulations are an artifact resulting from the spectral resolution of the spectra (0.2 nm). Higher resolution spectra and calculations would result in a smoother curve, but its averaged course would not change significantly. You can also smooth the resulting curve, but any differences in the results are insignificant.

Hi Oskar,

I was wondering if you would be able to share your finding on Antlia filters? I saw your comments on youtube about testing those and you mentioned that the maker falsified spectrum graphs to the filters. Could you please provide more details? 
Thanks,

Artem


  • Poynting likes this

#14 dobrychemik

dobrychemik

    Lift Off

  • -----
  • topic starter
  • Posts: 7
  • Joined: 18 Oct 2021

Posted 11 January 2022 - 04:13 PM

Hi Oskar,

I was wondering if you would be able to share your finding on Antlia filters? I saw your comments on youtube about testing those and you mentioned that the maker falsified spectrum graphs to the filters. Could you please provide more details? 
Thanks,

Artem

Here you can find the results of my Antlia ALP-T measurements:

 

https://astropolis.p...&comment=942529

https://astropolis.p...&comment=942739

 

and conclusions:

 

https://astropolis.p...&comment=942941


  • LYAstro likes this

#15 dobrychemik

dobrychemik

    Lift Off

  • -----
  • topic starter
  • Posts: 7
  • Joined: 18 Oct 2021

Posted 14 January 2022 - 04:34 AM

The leading Polish store with astronomical equipment already offers Antlia ALP-T 2.0" filters with individually measured spectra and with simulation of effectiveness depending on the lens speed of the telescope. And the best thing is the price: lower than for unproven filters in other stores. Treat it as an campaign to make people aware that it is possible to enter a higher level of reliability and honesty towards the customer. I would also like to remind you that Antlia adds its own spectrum charts to its filters, but they are false - all filters get the same, hardly readable printout.

 

Below are links to the individual filters. There you can see the measurement reports as a PDF file. The reports are in Polish, but you can easily guess what's going on or use the Google translator. I sincerely recommend.

 

OM-1482:
https://teleskopy.pl...m widma OM 1482

 

OM-1486
https://teleskopy.pl...m widma OM 1486

 

OM-1488:
https://teleskopy.pl...m widma OM 1488



#16 KLWalsh

KLWalsh

    Apollo

  • -----
  • Posts: 1,227
  • Joined: 19 Feb 2014
  • Loc: North Georgia, USA

Posted 14 January 2022 - 11:26 PM

I use a pair of spectrometers at work that can cover the spectrum from 350 to 1100nm. I typically set the bandwidth to 2 nm.
I’ve measured and posted some filter info here, such as the graph of the SkyGlow filter, below.
I use a NIST-traceable, broadband Lambertian Luminance Standard as the source and an F/2.8 optics head with a dual/split fiber optic feeding the two spectrometers simultaneously. The spectrometers have sensors cooled to -10C.

 

 

Attached Thumbnails

  • A530227A-1FBB-453C-AB20-962728D9B693.jpeg



CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics





Also tagged with one or more of these keywords: Filters, Report



Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics