Sony introduced HAD(TM) CCDs in 1984. The HAD (Hole Accumulation Diode) means there is an additional accumulation layer added to a n+P diode junction. It drains-off thermally generated electrons and thus reduces thermal noise - most notable at low light levels
Boosting the quantum efficiency through
several development stages starting with on-chip microlenses
(OCL) over each cell focusing light to each detector's effective
area (Hyper HAD) and not wasting any photons on the
ever-narrowing boundaries between the detector cells. This led
to Super HAD CCD (TM), which
has a second convex microlens close to each cell well, which is
useful with fast, low F-number lenses, where the light rays
might have a very low incident angle. The other more infra-red
sensitive branch, also with double OCLs, is called EXview HAD
The EXview HAD CCD (TM), introduced in 1998, is able to convert deep-penetrating IR (infra red) photons to electrons as result, of a new design of the detector, which simply uses a thicker silicon layer. It makes the sensor's IR-end responce to extend further and be more quantum-efficient than earlier CCDs. The adopted charge-drain structure reduce vertical smear. Absolute quantum-efficiency of these EXview HAD chips peaks at about 70% at 630 nm wavelength, and remain at least doubly more efficient than Super HAD CCDs at any wavelength longer than 900 nm. The sensitivity at shorter-end of light spectra (<500 nm) is slightly better than for most Super HAD CCDs. EXview chip may be cooled, but only needs cooling down to -30° C by a single-stage Peltier cooler for long exposure astronomical imaging application. There is one negative aspect on the EXview chip: it is initially more prone to develop hot-spots and therefore would prefer digital signal processing (DSP) circuitry around it to remove the hot-spots from the image (such as WAT-902DM2s and LCL-902Q).
Note: Typical sensitivity figures derived from datasheets refered to ICX039BLA. Minimum sensitivity can be -2 dB from typical.
ICX429ALL's value for VIS+IR would be + 18dB which does not "show" on the QE graph.
ICX255AL is a 1/3" sensor.
There are two different EXview CCDs: the discontinued (June 2001) Sony ICX249AL used in cameras manufactured before 2003 and the ones currently in production with Sony ICX429ALL (these are just the CCIR types! EIA type = last number is smaller by 1).
For more detailed information, see Sony's
data sheets for ICX249AL and ICX429ALL
Simply, since the pixel size is smaller
(the area where photons hit), the sensitivity can't be as good
on a 1/3" CCD compared to a 1/2" format CDD of same chip
construction. The relative QE-graphs above of two otherwise
identical cameras (actually a ICX429ALL 1/2" EXview HAD CCD and
ICX259AL 1/3" EXview HAD CCD) also show, that the spectral
responce is more flat on the 1/2" sensor on visible wavelengths
(looks partially as a coating issue).
Watec 902H uses EXview CCD, but the Watec WAT-120N camera however, does not use EXview chip, but a Super HAD CCD chip from Sony. Super HAD CCDs are said to be less prone to hot-pixel problems and have smoother dark-frame, but they fall behind in sensitivity against EXview CCDs.
I noticed, like many others on WAT-902H (and on LCL-902K), the image with as-supplied HIGH gain setting, is very noisy, but evidently, MetRec can pick out more meteors from a noiser picture, but I still would lower the gain some. The trimer on the older 902H model is inside and an you need to open the enclosure to access the trimmer. MetRec's "Grab" utility's video histogram (no integration on) is one tool to use: turn CCW the trimmer next to HI/LO switch with a plastic tool and leave it somewhere halfway for reasonable gain (random noise spikes extend to about 50% on the x-scale of the histogram). Do not switch to LO gain - the adjustment trimmer will not work with it. Another weird thing on 902H was the lack of DC- iris lens support (unless you bought the HC-01 adapter cable), and the unreliable RCA (not BNC) video output connector.
These issues were fixed by the new WAT-902H2 Superior and WAT-902H2 Ultimate camera models, which support both video- and DC-iris lenses and come with BNC video output connector. They also have switch and potentiometer for gain;: HI, LO and MCG (manual) at the rear of the camera. Please note the H2 stands for 1/2" CCD, not the less sensitive H3 model with 1/3" CCD!
With the aberration correcting screw-on lens installed (supplied with some of the Computar lenses) and at the same time increasing the camera gain, the limiting magnitude improved by about 1 magnitude, from mag. +2.5 to mag +3.5, which shows positively on the meteor counts. In September 2005 the average sporadic meteor rate for the best 6 nights was 7.3 (+-1.1) SPO meteors / hour, peaking at 8.4/h on a single night. The mean meteor magnitude has varied between mag. +1.3 and mag. +1.8 depending on sky conditions and was +1.37 as whole month's average. About 7% of detected meteors belong in the mag. +3.0 (+-0.5 mag) bin. The system misses 19 out of 20 mag. +3 meteors due to low S/N-ratio with these faint meteors compared to mag.+1.7 meteors. Brightest recorded meteor was mag. -1.4 and dimmest identified with PostProcessing was mag. +3.8.
Since the optical window change, the best sporadic hourly rate in Sept. 2007 was 9.2 SPO/h and limiting magnitude improvement of 0.1 mag.
WAT-120N should be less sensitive than WAT-902H when used for meteor astronomy with MetRec and with no integration. If you like deep-sky imaging with long exposure times (integration) with WAT-120N, in spite of the less sensitive CCD chip, you could see very faint objects invisible to the naked eye depending on the optics you have. I tried WAT-120N with a 4.5" refractor scope and observed stars down to mag. +15 using the 10-second exposure setting.
For meteor work with MetRec, only singe frame integration (one step on the RC-switch CW from 12 a-clock) can be used, since the software looks for moving transient events and needs to determine the velocity from object's positions on successive odd video fields. The IR emissions from a meteor that EXview chip could pick up more efficiently, have to pass through the atmosphere, possible optical window and optics. Looking at CCD chip data sheets, the difference between EXview HAD CCD and HAD CCD cameras should be clear, the EXview (for instance the 902H) being more sensitive. The MetRec RefStar-procedure related "Grab" utility with 32x integration seems to be a bad way to compare cameras - the differences are small, Lm mag. +5.1 against mag. +5.3 with these two cameras (with Dedal Gen. 2 intensifier and F2 lens, Lm of 5.6 was achieved by that method). With no integration, the difference is much greater than just 0.3 magnitudes or so.
|Watec WAT-902H 6 mm F0.8 lens
Software: MetRec V3.6
Date : 2003/12/12
|Watec WAT-120N 6mm F1.3 lens, integration OFF, Gamma:
Software: MetRec V3.6
Date : 2003/12/12
|19:44:11 Meteor #37 at
(0.516,0.137)->(0.521,0.104) frames=3 dur=0.10s pixel=2
dir=279ø vel=17.7ø/s å=1.4
shower=MON bright=2.6mag raddist=4.6ø exp vel=18.3ø/s (22.043h,48.90ø) -> (21.932h,48.02ø) acc=1.5'
19:44:11 Saving sum image of meteor #37 made of 9 frames ... ok!
|19:48:03 Meteor #38 at
(0.559,0.623)->(0.450,0.629) frames=13 dur=0.54s
pixel=3 dir=177ø vel=12.6ø/s å=1.9
shower=SPO bright=0.5mag (23.942h,64.74ø) -> (0.365h,58.93ø) acc=4.1'
19:48:04 Saving sum image of meteor #38 made of 20 frames ... ok!
|19:48:03 Meteor #5 at
(0.540,0.583)->(0.445,0.575) frames=9 dur=0.50s pixel=3
dir=185ø vel=12.4ø/s å=2.1
shower=SPO bright=1.2mag (23.936h,64.53ø) -> (0.336h,59.29ø) acc=4.8'
19:48:03 Saving sum image of meteor #5 made of 19 frames ... ok!
|19:48:22 Meteor #39 at
(0.723,0.341)->(0.721,0.324) frames=2 dur=0.06s pixel=3
dir=265ø vel=19.3ø/s å=1.9
shower=GEM bright=1.4mag raddist=2.9ø exp vel=17.0ø/s (21.582h,63.10ø) -> (21.535h,62.40ø) acc=0.0'
19:48:22 Saving sum image of meteor #39 made of 8 frames ... ok!
|19:59:22 Meteor #40 at
(0.883,0.674)->(0.890,0.458) frames=13 dur=0.54s
pixel=11 dir=272ø vel=17.5ø/s å=3.2
shower=GEM bright=0.2mag raddist=0.6ø exp vel=16.6ø/s (21.549h,79.87ø) -> (20.607h,71.42ø) acc=1.6'
19:59:23 Saving sum image of meteor #40 made of 20 frames ... ok!
|19:59:22 Meteor #19 at
(0.873,0.595)->(0.885,0.513) frames=6 dur=0.22s pixel=7
dir=278ø vel=17.7ø/s å=1.8
shower=GEM bright=1.0mag raddist=0.4ø exp vel=16.6ø/s (20.974h,76.06ø) -> (20.686h,72.71ø) acc=1.0'
19:59:21 Saving sum image of meteor #19 made of 12 frames ... ok!
|20:01:15 Meteor #41 at
(0.049,0.107)->(0.039,0.108) frames=2 dur=0.06s pixel=1
dir=171ø vel=13.8ø/s å=1.1
shower=SPO bright=2.8mag (23.718h,28.69ø) -> (23.746h,28.27ø) acc=0.0'
20:01:15 Saving sum image of meteor #41 made of 8 frames ... ok!
|20:02:01 Meteor #42
at (0.801,0.995)->(0.703,0.270) frames=29 dur=1.14s
pixel=14 dir=262ø vel=28.4ø/s å=4.8
shower=HYD bright=-0.4mag raddist=0.8ø exp vel=27.1ø/s (3.523h,80.29ø) -> (21.691h,59.15ø) acc=19.3'
20:02:02 Saving sum image of meteor #42 made of 35 frames ... ok!
|20:02:01 Meteor #28
at (0.739,0.850)->(0.721,0.329) frames=21 dur=0.82s
pixel=12 dir=268ø vel=28.4ø/s å=2.5
shower=HYD bright=-0.2mag raddist=0.9ø exp vel=27.9ø/s (1.256h,79.95ø) -> (21.843h,62.27ø) acc=10.9'
20:02:02 Saving sum image of meteor #28 made of 27 frames ... ok!
|20:07:32 Meteor #43 at
(0.753,0.233)->(0.753,0.092) frames=10 dur=0.38s
pixel=5 dir=270ø vel=16.4ø/s å=3.4
shower=GEM bright=0.6mag raddist=0.5ø exp vel=15.5ø/s (21.393h,59.87ø) -> (21.109h,54.43ø) acc=1.9'
20:07:32 Saving sum image of meteor #43 made of 16 frames ... ok!
|20:13:20 Meteor #44 at
(0.410,0.265)->(0.406,0.200) frames=4 dur=0.18s pixel=2
dir=266ø vel=17.4ø/s å=1.3
shower=GEM bright=1.2mag raddist=0.9ø exp vel=16.9ø/s (23.363h,48.42ø) -> (23.171h,46.44ø) acc=2.9'
20:13:20 Saving sum image of meteor #44 made of 11 frames ... ok!
|20:21:15 Meteor #45 at
(0.469,0.550)->(0.464,0.416) frames=9 dur=0.34s pixel=5
dir=268ø vel=18.7ø/s å=2.5
shower=GEM bright=0.5mag raddist=1.2ø exp vel=18.0ø/s (0.454h,58.41ø) -> (23.855h,55.00ø) acc=0.9'
20:21:16 Saving sum image of meteor #45 made of 15 frames ... ok!
|20:21:15 Meteor #73 at
(0.467,0.462)->(0.469,0.431) frames=2 dur=0.10s pixel=1
dir=274ø vel=17.8ø/s å=1.1
shower=GEM bright=1.2mag raddist=0.8ø exp vel=18.1ø/s (0.258h,57.52ø) -> (0.113h,56.73ø) acc=0.0'
20:21:15 Saving sum image of meteor #73 made of 8 frames ... ok!
|20:23:47 Meteor #46 at
(0.063,0.128)->(0.069,0.090) frames=3 dur=0.10s pixel=4
dir=278ø vel=19.0ø/s å=1.6
shower=MON bright=1.7mag raddist=1.1ø exp vel=17.4ø/s (0.120h,29.69ø) -> (0.011h,29.16ø) acc=0.4'
20:23:47 Saving sum image of meteor #46 made of 9 frames ... ok!
|20:24:27 Meteor #47 at
(0.643,0.053)->(0.643,0.009) frames=3 dur=0.14s pixel=1
dir=270ø vel=14.9ø/s å=1.2
shower=GEM bright=2.2mag raddist=1.4ø exp vel=14.7ø/s (21.905h,50.28ø) -> (21.816h,48.71ø) acc=1.8'
20:24:28 Saving sum image of meteor #47 made of 10 frames ... ok!
|20:39:04 Meteor #97 at
(0.190,0.802)->(0.201,0.799) frames=2 dur=0.06s pixel=1
dir=342ø vel=14.9ø/s å=1.1
shower=SPO bright=2.8mag (2.811h,48.62ø) -> (2.788h,49.17ø) acc=0.0'
20:39:05 Saving sum image of meteor #97 made of 8 frames ... ok!
|20:39:46 Meteor #48 at
(0.475,0.918)->(0.475,0.870) frames=4 dur=0.14s pixel=6
dir=271ø vel=17.0ø/s å=1.6
shower=GEM bright=0.9mag raddist=1.0ø exp vel=16.3ø/s (2.891h,62.74ø) -> (2.594h,62.71ø) acc=0.5'
20:39:46 Saving sum image of meteor #48 made of 10 frames ... ok!
|20:39:46 Meteor #99 at
(0.434,0.847)->(0.435,0.830) frames=2 dur=0.06s pixel=3
dir=275ø vel=17.8ø/s å=1.4
shower=GEM bright=1.2mag raddist=2.5ø exp vel=16.2ø/s (2.827h,62.73ø) -> (2.723h,62.70ø) acc=0.0'
20:39:46 Saving sum image of meteor #99 made of 8 frames ... ok!
|21:07:26 Meteor #49 at
(0.242,0.617)->(0.241,0.581) frames=3 dur=0.10s pixel=2
dir=269ø vel=19.4ø/s å=1.5
shower=GEM bright=1.5mag raddist=1.1ø exp vel=16.7ø/s (2.088h,47.22ø) -> (1.946h,46.69ø) acc=0.1'
21:07:27 Saving sum image of meteor #49 made of 9 frames ... ok!
|21:07:47 Meteor #50 at
(0.171,0.881)->(0.169,0.832) frames=4 dur=0.14s pixel=4
dir=268ø vel=16.7ø/s å=3.3
shower=GEM bright=1.0mag raddist=0.5ø exp vel=15.1ø/s (3.230h,45.51ø) -> (3.042h,45.19ø) acc=1.5'
21:07:48 Saving sum image of meteor #50 made of 10 frames ... ok!
|21:07:47 Meteor #144 at
(0.138,0.743)->(0.138,0.733) frames=2 dur=0.06s pixel=1
dir=269ø vel=10.5ø/s å=1.2
shower=SPO (is GEM, wrong velocity!) bright=2.2mag (3.098h,45.35ø) -> (3.060h,45.23ø) acc=0.0'
21:07:48 Saving sum image of meteor #144 made of 8 frames ... ok!
|21:10:00 Meteor #149 at
(0.120,0.686)->(0.126,0.688) frames=2 dur=0.10s pixel=2
dir= 19ø vel=3.7ø/s å=1.2
shower=SPO bright=2.7mag (2.937h,43.76ø) -> (2.938h,44.06ø) acc=0.0'
21:10:00 Saving sum image of meteor #149 made of 9 frames ... ok!
|21:19:17 Meteor #51 at
(0.271,0.982)->(0.273,0.911) frames=6 dur=0.22s pixel=6
dir=272ø vel=14.6ø/s å=3.3
shower=GEM bright=0.8mag raddist=0.0ø exp vel=14.2ø/s (3.840h,51.17ø) -> (3.528h,51.26ø) acc=0.6'
21:19:18 Saving sum image of meteor #51 made of 12 frames ... ok!
|21:19:17 Meteor #166 at
(0.229,0.870)->(0.234,0.830) frames=4 dur=0.14s pixel=5
dir=276ø vel=14.2ø/s å=1.9
shower=GEM bright=1.1mag raddist=1.7ø exp vel=14.2ø/s (3.753h,51.23ø) -> (3.572h,51.19ø) acc=2.0'
21:19:18 Saving sum image of meteor #166 made of 10 frames ... ok!
|21:20:00 Meteor #52 at
(0.860,0.593)->(0.844,0.509) frames=4 dur=0.14s pixel=2
dir=259ø vel=30.7ø/s å=1.7
shower=SPO bright=0.4mag (22.806h,76.34ø) -> (22.641h,72.72ø) acc=3.2'
21:20:00 Saving sum image of meteor #52 made of 10 frames ... ok!
|21:24:47 Meteor #53 at
(0.312,0.220)->(0.318,0.096) frames=9 dur=0.34s pixel=7
dir=273ø vel=16.5ø/s å=3.9
shower=GEM bright=0.5mag raddist=0.8ø exp vel=15.6ø/s (0.754h,42.77ø) -> (0.367h,39.81ø) acc=1.5'
21:24:48 Saving sum image of meteor #53 made of 15 frames ... ok!
|21:24:59 Meteor #54 at
(0.535,0.868)->(0.606,0.775) frames=9 dur=0.34s pixel=5
dir=307ø vel=18.1ø/s å=2.5
shower=SPO bright=1.0mag (3.306h,66.14ø) -> (2.478h,69.58ø) acc=2.0'
21:24:59 Saving sum image of meteor #54 made of 15 frames ... ok!
|21:26:18 Meteor #55 at
(0.836,0.633)->(0.876,0.508) frames=6 dur=0.26s pixel=5
dir=288ø vel=24.0ø/s å=2.3
shower=SPO bright=1.1mag (23.461h,77.10ø) -> (22.340h,73.21ø) acc=2.8'
21:26:18 Saving sum image of meteor #55 made of 13 frames ... ok!
|21:26:55 Meteor #56 at
(0.546,0.680)->(0.535,0.660) frames=3 dur=0.10s pixel=6
dir=243ø vel=13.4ø/s å=3.2
shower=SPO bright=0.5mag (2.023h,64.94ø) -> (1.933h,64.04ø) acc=0.3'
21:26:55 Saving sum image of meteor #56 made of 9 frames ... ok!
|21:26:55 Meteor #182 at
(0.524,0.637)->(0.521,0.628) frames=2 dur=0.06s pixel=7
dir=249ø vel=10.5ø/s å=2.1
shower=SPO bright=0.9mag (2.012h,64.90ø) -> (1.976h,64.55ø) acc=0.0'
21:26:55 Saving sum image of meteor #182 made of 8 frames ... ok!
|21:30:48 Meteor #57 at
(0.174,0.290)->(0.176,0.181) frames=8 dur=0.30s pixel=9
dir=271ø vel=16.5ø/s å=3.8
shower=GEM bright=0.5mag raddist=0.9ø exp vel=15.5ø/s (1.451h,38.06ø) -> (1.117h,35.77ø) acc=1.8'
21:30:48 Saving sum image of meteor #57 made of 14 frames ... ok!
|21:30:48 Meteor #186 at
(0.200,0.127)->(0.204,0.098) frames=3 dur=0.10s pixel=4
dir=277ø vel=15.0ø/s å=1.6
shower=GEM bright=1.1mag raddist=1.9ø exp vel=15.4ø/s (1.258h,36.78ø) -> (1.169h,36.22ø) acc=2.4'
21:30:48 Saving sum image of meteor #186 made of 9 frames ... ok!
|21:31:19 Meteor #58 at
(0.611,0.750)->(0.616,0.733) frames=2 dur=0.06s pixel=2
dir=286ø vel=19.6ø/s å=1.3
shower=GEM bright=1.6mag raddist=2.0ø exp vel=16.4ø/s (2.366h,69.57ø) -> (2.216h,69.60ø) acc=0.0'
21:31:19 Saving sum image of meteor #58 made of 8 frames ... ok!
|21:32:18 Meteor #59 at
(0.575,0.598)->(0.611,0.412) frames=12 dur=0.46s
pixel=15 dir=281ø vel=19.4ø/s å=5.3
shower=GEM bright=-0.7mag raddist=0.7ø exp vel=16.9ø/s (1.481h,64.89ø) -> (0.317h,61.52ø) acc=1.8'
21:32:19 Saving sum image of meteor #59 made of 18 frames ... ok!
|21:41:06 Meteor #60 at
(0.628,0.450)->(0.637,0.408) frames=3 dur=0.14s pixel=3
dir=282ø vel=15.8ø/s å=1.4
shower=GEM bright=1.5mag raddist=2.0ø exp vel=16.8ø/s (0.533h,63.39ø) -> (0.294h,62.43ø) acc=0.1'
21:41:06 Saving sum image of meteor #60 made of 10 frames ... ok!
|21:42:59 Meteor #62
at (0.732,0.941)->(0.853,0.614) frames=22 dur=0.86s
pixel=56 dir=290ø vel=18.5ø/s å=10.0
shower=GEM bright=-2.0mag raddist=0.8ø exp vel=15.8ø/s (4.388h,76.95ø) -> (23.459h,77.03ø) acc=10.5'
21:43:00 Saving sum image of meteor #62 made of 28 frames ... ok!
|21:42:59 Meteor #207
at (0.702,0.882)->(0.827,0.638) frames=18 dur=0.70s
pixel=33 dir=297ø vel=18.7ø/s å=5.9
shower=GEM bright=-1.2mag raddist=0.8ø exp vel=16.0ø/s (3.763h,78.17ø) -> (23.580h,77.27ø) acc=7.3'
21:42:59 Saving sum image of meteor #207 made of 24 frames ... ok!
|total 186 frames||total 71 frames|
|Observing Statistics for Watec WAT-902H with 6 mm F0.8 lens||Observing Statistics for Watec WAT-120N with 6 mm F1.3 lens|
|start of observation : 2003/12/12
end of observation : 2003/12/13 21:45:27
effective observing time : 2 h 2 m 0 s (12.3 meteors/h +-3.5)
# active meteor showers : 5
: N=6, mm= +0.8 mag
|start of observation : 2003/12/12
end of observation : 2003/12/12 21:45:27
effective observing time : 2 h 2 m 0 s (5.9 meteors/h +-2.4)
# active meteor showers : 5
: N=4, mm= +2.0 mag
The difference in the numbers of obseved meteors, suggest a
difference of 0.8 magnitudes between 902H and 120N cameras.
Download (save file on disk) and watch this video clip
WAT-902H: mag +0.3 sigma-Hydrid: 29 frames+ WAT-120N: the same mag +0.3 sigma-Hydrid: 21 frames
WAT-902H: -1.6 Geminid: 22 frames WAT-120N: the same mag -1.6 Geminid: 18 frames
Left: WAT-902H, the EXview sensor
camera. Right: WAT-120N, the HAD sensor
With this test setup 902H did beat 120N in the above observation contest only by 6 against 4 with sporadic meteors, but for some reason, the Geminid meteors by stunning 16/7 !
But... there were two technical bias factors favoring 902H: the
120N video was taped with VHS video recorder for later playback to
MetRec and it also had the less powerful F1.3 lens, which in
theory falls 1 magnitude short of the F0.8 lens. So one more test
was set up with identical lenses on both cameras:
These test images were taken with same lens, the 6 mm F0.8
Eneo and gamma set to 0.45. These windows are zooms of Cas- area
and the limiting magnitudes are estimates only. Since there are
several minutes between grabbing the image pairs, the conditions
may have changed!
Frustrated on earlier night-sky test's scatter of results, an
indoor test with LEDs in series at 20 uA current (Red, Green,
Blue, IR) as the light source, gave these results: (the upper part of image is WAT-902H and lower
WAT-120N (gamma 0.45, no integration, gain set to max.) with
some HP digicam shots below). Both cameras were at room
temperature and turned on for 10 minutes before image was grabbed.
Luminescence of LEDs used is unknown.
The Blue LED with WAT-120N was just above the noise (pixel value
21 (+-14)) and the difference on infra-red is very clear, as
it should be. WAT-902H beats 120N on all wavelengths used in this
test. The best color for 120N was Red, but for 902H, the
differences between colors were not that big, except for the Blue
LED. HP-850 digicam rejects IR totally and reveals the Blue
LED obviously has lower luminescence that those other LEDs.
Calibration was done with neutral density filters.
Watec WAT-902H (left - EXview) and Watec
WAT-120N (right - Super HAD). Both cameras at identical
settings; no integration, gamma 0.45.
Test images produced with incandescent light source, test card slide and a stack of two neutral density filters (1.0 and 2.0). Video grabbed at full VGA resolution, 8-bit, no integration.
Comparison images taken with WAT-120N using different camera
integration settings form 1 to 4 frames and single 8-bit frame
of WAT-902H and LCL-902K and WAT-902H2 ULTIMATE video. Light source: incandescent lamp with 1.0 and 2.0 ND filters stacked.
In brief, the old 902H lacked some HW features that became available with 902H2 SUPERIOR and ULTIMATE as well as the Watec America's LCL-902K (more dip switches to set gain, select shutter, gamma, backlight compensation and Video or DC auto-iris support). Comparing sensitivity the 902H and 902H ULTIMATE no diffence was observed. The LCL-902K was just a little less sensitive, by 2...3 dB. On the LCL-902K a slight brightening occurs on the left side of the image in low-light conditions, which most likely is CCD preamplifier glow - faint light scattering inside the chip is produced by the on-CCD-chip amplifier transistors on the edge of the CCD.
The latest model on WAT-902- series is the 902H2 ULTIMATE, that
is advertised to beat the old 902H in sensitivity by 3- fold, but
as you see from the diagonally cut image with upper left part
grabbed from the new ULTIMATE model and right lower
part from old 902H (after 5000 h usage) , there is no difference!
The gain setting of 902H2 ULTIMATE was MCG (manual) and set as
identical as possible with the 902H. Neither this or the night sky
test indicated any noticeable difference, so the 100 ulux
advertised sensitivity over the previous 300 ulux must once again
be some marketing driven testing-method trick. Of course the 902H2
ULTIMATE does have external settings for gain, gamma, back
light, as well as shutter, plus the support for both autoiris
types, but for meteor observing the only benefit in replacing the
old 902H are fewer hot-spots on CCD - for a while - for a price of
It seems the gap between EXview'ed WAT-902H and the "plain" HAD
WAT-120N is real and since you get two 902Hs for the price of one
120N, and if you are planning to do just meteor observing, the
choice is clear: the 902H. If you have the money and want to play
around with longer exposure times with non-meteoric objects,
possibly by attaching the camera to your telescope with tracking
drive, the 120N is the thing to have, and as a bonus, you still
can make fairly good meteor observations, if you set integration
to 1 frame. Increasing frame integration (exposure time) on
WAT-120N will not do much good and MetRec will not work
properly with slowly updating static frames. Integration was not
used in the tests and as mentioned - all camera settings were
Some Watec (Japan) (WAT-###) and Watec America Corp.'s (2) (LCL-###) B&W camera
models using EXview HAD CCD:
|Type||Manufacturer announced sensitivity F1.4||Measured sensitivty (5)
(F1.2 at 20 dB S/N)
|CCD sensor (4)*||misc.|
|WAT-902DM2s||600 ulx (1)||2200 ulx||ICX429ALL (1/2")||DSP|
|300...150 ulx||2700 ulx||ICX429ALL (1/2")||902H; video AI only,
902H2; DC and AI
|WAT-902H||300 ulx?||3000 ulx||ICX249AL (1/2")||older CCD version of 902H|
|WAT-902H2 Supreme||300 ulx||-||EXview (1/2")||Newer version|
|WAT-902H2 Ultimate||100 ulx||-||EXview (1/2")||Latest version, price about 310 €|
|WAT-902DM3s||1200 ulx||4100 ulx||ICX259AL (1/3")||DSP|
|WAT-902HB3s||600 ulx||5200 ulx||ICX429ALL (1/2")|
|LCL-902Q||150 ulx F1.2 (3)||-||ICX429ALL (1/2")||DSP offsets up to 10 hot spots, DC-AI|
|LCL-903K||300 ulx F1.2||-||ICX259AL (1/3")||DC&Video AI|
|WAT-100N||1000 ulx||-||ICX429ALL (1/2")|
|LCL-902K||150 ulx (3)||-||ICX429ALL (1/2")||DC&video AI|
|Toshiba IK-1000A||1000 ulx (F1.2)||-||E2V's EMCCD|
There are several 902-series cameras with
EXview chips, older versions, PCB version, DSP, or just
different AI and other setting
options: 902H2, 902H3, 902H, 902HB2S, 902HS, 902K, 902DM2s, 902Q.
* Cameras with older type ICX249AL
(ICX248AL) EXview chip are no longer marketed, the chip is
not produced or sold by Sony
Typical sensitivity figures for intensified cameras (ICCDs):
|Intensifier Generation||Typ. sensitivty at 6 dB S/N||Typ. sensitivity (20 dB S/N)||price|
|Gen. 2||10 ulx||200 ulx||$3300 & up|
|Gen. 2.5||3 ulx||70 ulx|
|Gen 3||0.5 ulx||15 ulx|
Watec's B&W camera models using Super HAD CCD:
|Type||MFGR's announced sensitivity F1.4||Measured sensitivty (5)
(F1.2 at 20 dB S/N)
|CCD sensor (4)||misc.|
|WAT-127LH||1500 ulx||3200 ulx||ICX419ALL (1/2")|
|WAT-902DM2||1800 ulx||2900 ulx||ICX419ALL (1/2")||DSP|
|WAT-902DM3||2000 ulx||4400 ulx||ICX409AL (1/3")||DSP|
|WAT-902B||3000 ulx||-||ICX419ALL (1/2")|
|WAT-525EX||3000 ulx||-||ICX419ALL (1/2")|
Note 1 : 1 ulx = 0.000 001 lux
Note 2 : Watec (Japan) and Watec America Corp. have NO business relationship since April 2000.
Note 3 : Watec America Corp. claims better sensitvity for a seemingly similar camera model: this is very likely due to a different measurement method than any real sensitivity difference! I would at this time read the most sensitive cameras to be all in the same ballpark.
Note 4 : The CCD chip types on the above table are based on information received directly from the manufacturer from Japan and to a small part; matching sensor spectras per camera type as they are on Watec web site against those on Sony's datasheets.
Note 5: The "Measured sensitivity" (per Lebedev&Snipe of EVS) using F1.2 lens and 20 dB S/N ratio and presumably low IR content image source, reduces the tested sensitivity figure of EXview CCD cameras dramatically compared to manufacturer's own data most likely based on IR rich source and perhaps a 6 dB S/N ratio or 50 or 80% video? Another question is has the illumination been measured from the CCD (as reflected light) or is it measured from the scenery? Even so, manufacturer given sensitivity figures are doubtful and LCL- vs. WAT- marketing-driven.
There is false information on the web claiming some Watec models use EXview CCD, while they really do not!
Many OEM cameras which became available in
late 1990's, are using Sony's CCD and related signal processing
technology, but partly due to CCD production process
limitations, the production has been limited at times. Though
Sony's CCD chips are sensitive, Kodak introduced frontside CCDs
in 2001 with peak QE of 85% - there is not much room for
improvement in QE, since the theoretical 100% limit is getting
so close. So, how to reduce the nasty Read-Noise?
With 902H MetRec can detect meteors down to mag. +3.5, but for
the colour SK-2146XAIP that limit should be at about mag. -3,
between the brightness of Jupiter and Venus. All of the meteors
detected with MetRec with SK-2146XAI, would be fireballs. Using
normal lenses, for ordinary meteors, none-DSP CCD colour video
cameras - even the ones with EXview chips, are blind as bats
compared to B/W video cameras.
When reading a CCD at video rates, Read-Noise becomes the major problem at low light levels. Even cooling won't help since it mostly reduces Dark-Noise. There is a new innovation (patented 1990, 1994, in production since 2002) which uses an additional on-CCD-chip electron multiplier amplifier, which E2V Technologies (ex. Marconi Applied Technology) calls "EMCCD" (CCD types CCD60, CCD65, CCD87, CCD97) and Texas Instuments (TI) as "ImpactronTM" (CCD types TC247, TC235SPD, TC285SPD). It is based on Impact Ionisation electron multiplication phenomena in silicon. The E2V's extra output register, which delivers the electron multiplication (gain), is designed to handle about 40 V clock signal instead of the normal 10 V clock, while TI uses just 15 V.
The E2V chips are about a decade more expensive than Texas TC253 wich price is only about 500 €.
EM-CCD surveillance (non-scientific) cameras are also in production:
lx (F1.2 CMG x 1000, AGC 12 dB, White 50%)
Hitachi KP-E500 0.00003 lx (Monochrome in full motion, maximum sensitivity setup, F1.2, 50 IRE) (B/W camera)
Hitachi KP-DE500 0.00005 lx (Monochrome in full motion, maximum sensitivity KP-DE500 setup, F1.2, 50 IRE), 0.0009 lx (Color in full motion, maximum sensitivity setup, F1.2, 50 IRE)
Hanwha Techwin SHC-750 0.0005 lx @t F1.2 (Monochrome), 0.008 Lux @ F1.2 (Normal color mode)
These are advertised with price tags up from around 2300 € (SHC-750, others quoted at 6000€ to 9000€). They are Peltier-cooled and draw 4 A at 12 V DC.
NEC and Goto Corporation produce the NC-R550a, a three- EM CCD chip
cooled video camera that costs around 44 000€. See the video of a
Please note Watec has introduced newer versions of 902; the 902H2Superior and 902H2Ultimate with 1/2" CCD chip and support DC and video auto iris.
Don't buy a lens with too short focal length that gives you trees
and TV-antennas or a lens intended for 1/3" CCD which gives you
more or less vignetting .
CCD-camera's image distortion graph with Eneo G0608NDDC-560 lens w. WAT-902H
Coverage map plot of southern Finland at 90 km meteoric
Automated camera power-up and -down is handy, though a CCD camera (WAT 902H or the Mintron) with autoiris lense could well be powerd on permanently, but you may save some energy using the twilight switch. A 20€ twilight switch module (JO-EL CDS-20K) installed on a wall, powers up the camera after sunset for the dark hours of the night. Typically twilight switches can be set to 5 LUX and it powers up the camera when sun is about 3...6 degrees below horizon - way before meteor recognition can be initiated and likewise recognition ends way before twilight switch kills the camera power. The twilight switch has to be shielded from bright artifical lights such as car headlights not to cause unwanted interruptions for camera's power feed.
If the camera would use an intensifier, which is too sensitive for dusk and dawn light levels, Theben has an astronomical clock module programmed for any country with many cities in the built-in the menu and with up to +- 2 h of offset for the switching time from local sunset and sunrise. The approximate price of the clock module is about 90€ at wholesale dealers and about 150€ at retail shops. Theben module will not take in to account the extending duration of twilight evident at higher latitudes. Powering on and off the camera at exactly right time each day could be implemented with a 365-day programmable clock module by preprogramming the on-off times for every day. Modules like the Grässlin timer, or the GE Digi 4T model with TAXISET programing tools should be able to do this.
The lens/optical window's heater power depends on climate and season. In Finland, I have the heater powered at 50% only during nighttime in April, May, August and September and 24/7 from October to end of March. During winter the power is 50% when outdoor temperature is 0 °C or higher, and full power is applied when it is below freezing to keep the window free of snow, defrost and dry at both sides and to keep the lens & camera dry and a little warm. Excessive heating will not do good for the camera since it may shorten it's life and increase the noise. At 0 °C outdoor temperature and the camera powered, producing about 4 W, and the resistor at 50% power (11 W), the lens temperature was + 15° C and with 100% heating power (22 W) +30° C. At - 30 °C outdoors, I barely get the lens to 0 °C.
Watec WAT-120, with the HAD sensor favored by Deep Sky enthusiasts due to integrated mode and the more simple WAT-902H with the EXview chip, were both tested for CCD imperfections. The WAT-120N's CCD chip has 2 hot- pixels as new, while the WAT-902H and LCL-902K with EXview has none, though the dark field is not as smooth and if the recommended DSP would be there, it would mask the spots. Bye the way, Watec gives 3 year warranty and it's no wonder - there are no Al-electrolytic capacitors in these two cameras, all bigger caps are tantals.
WAT-902H hot-pixels after 16 months (contrast enhanced, 32 frame integration)
After 16 months, WAT-902H has 10 bright
hot-pixels and WAT-120N when used with longest exposure setting,
has tens of hot-pixels.
total: 556 h 22 min 27 s
2004/2005 Teff total: 779 h 45 min 23 s
2005/2006 Teff total: 837h 45 min 20 s
2006/2007 Teff total: 812 h 43 min 52 s
2007/2008 Teff total: 746 h 6 min 14 s
2008/2009 Teff total: 762 h 27 min 14 s
2009/2010 Teff total: 717 h 7 min 4 s
2010/2011 Teff total: 632 h 12 min 7 s
2011/2012 Teff total: 655 h 59 min 57 s
2012/2013 Teff total: 709 h 37 min 30 s
2019/2020 Teff total: 840 h 13 min 8 s
2020/2021 Teff total: 744 h 43 min 11 s
2004/2005 Nobs: 132
2005/2006 Nobs: 152
2006/2007 Nobs: 153
2007/2008 Nobs: 121
2008/2009 Nobs: 136
2010/2011 Nobs: 149
2011/2012 Nobs: 159
2012/2013 Nobs: 142
2019/2020 Nobs: 156
2020/2021 Nobs: 133
|QTY of all
|First obs: Nov. 2003 - latest obs: 15. May. 2021|
2005/06 2006/07 2007/08
2008/09 2009/10 2010/11
2011/12 2012/13 2019/20 2020/21
N SPO: 999 1493 2943 2809 2004 1770 3778 2334 1965 1666 1621 2997
9 5 1
2005/06 2006/07 2007/08
2008/09 2009/10 2010/11
2011/12 2012/13 2019/20 2020/21
N MON: 10 1 11 6 6 3 6 15 20 1 9 10
N HYD: 14 9 9 0 2 7 8 7 4 5 8
N GEM: 75 165 109 12 269 384 164 20 0 5 55
N COM: 16 12 20 36 13 21 44 7 45 18 17 4
N URS: 15 15 34 44 34 13 0 6 20 0 0
N QUA: 25 11 2 0 94 127 10 0 0 0 131 8
N DLE: 3 1 10 3 0 6 1 12 6 4 3 6
N LYR: 17 16 36 25 21 24 11 21 29 24 17 10
N ETA: 1 0 1 15 0 0 0 0 0
N ELY: 3 5 2 3 5 4
N PAU: 2 1 0 0 0 0 2
N SDA: 5 4 18 4 4 4 9 7 11 12 12 13
112 107 271
709 143 129
| NOTE: table covers only observations with
CCD cameras from Nov. 2003 onwards, Nov.: WAT-120N,
Dec. 5. onwards: WAT-902H, Oct. 2009 onwards WAT-902H2S,
Jul. 2020 onwards WAT-902H2U
||"not observable": radiant never rises above horizon, or summerlight night skies from 15. May ... 1. August. Note: meteor shower list was revised in 2007.|
"Radiant" plot analysis below prepared by ykChia, thanks!
Floated glass flats used as optical windows loose about 10% of total light by reflection from both surfaces. Though a small disk of ordinary window glass costs almost nothing, if you want to make the window's transmissivity better, replace it with anti-reflective (AR)- coated flat. Of course the real optical windows sold by Edmund Optics are optically perfect (flat), but the float glass is nowadays good enough for CCTV and it's also now available with AR-coatings and these glasses you can order to your preferred size and shape for about for 50...60 € (150 mm diam.). Antoher option is you find
a scrapped LCD-display, which happens to have a shield-faceplate of AR-coated glass, remove it and have it cut by a craftsman to your preferred size and shape.
The search for a supplier ended up with just one firm willing to
deliver a cut-size disc from stock on 2 days notice; the importer
of Luxar glass
(made in Switzerland)(importing a single box or even a sheet is
very expensive). Luxar(TM)is
available globally in different thicknesses, I opted for 4 mm both
sides broad-band 5-layer AR-coated (BBAR). The coating is created
by Magnetron sputtering of successive oxide layers (TiO and SiO)
and produces very low reflectivity. Reflections are by far the
most significant cause for loss-of-light in the optical window.
The pass-band of Luxar (400...770 nm) could be slightly wider, but
it will do. The whole instrument, including optics and CCD, works
perfectly from 450 to 700 nm, but not on UV because of
non-quarz-grade glass absorbtion. Performance is impaired on NIR
by all coated elements and the Silicon-based CCDs gradual
reduction of sensitivity towards 1100 nm upper limit. Some
SyncMaster 710MB displays have AR-glass but not all. The
comparison plot of the transmission is found below.
Optical windows: two multi-coated and a non-coated
Note: transmissivity scale is imaginery and adjusted for each
lens - not comparable.
There seems to be one basic "made in Japan" coating on CCTV lenses and a few exceptions.
Note: transmissivity scales for camera lenses and oculars are imaginery and adjusted for each item - not comparable. Graphs taken on flats and spectacle lenses should be close to actual real transmittance.
|lambda [nm]||net transmittance with Luxar||net transmittance with soda-lime glass||proportions of meteoric line emissions on 50 nm bands between 375...675 nm|
Total system transmittance including all optical surfaces and
CCD's efficiency vs. wavelength
(presuming Eneo lens' peak transmittance being 85%). 82% of light at 600 nm is passed to
CCD, which converts 70% of the photons to electric charge.
Note: Lenses of same series have similar transmittance passband shape, but there are differences between different manufacturers.
Since the AR- coatings are based on 1/4- wavelength film layer(s), the optimal (and specified) performance is valid for light passing through the flat at right-angle. When the angle-of-incidence becomes very shallow, in such cases as wide-field camera's FOV corners (very short focal length optics), rays passed at the corners of the image suffer from minor loss of light by reflection and transition of the AR- coating's spectral pass-band (towards shorter wavelengths). The transmission curve has a cut-off point at about 50° angle (AoI). Reflectivity is identical for Luxar and soda- lime glass at angle of 60° and for Luxar it gets worse beyond that angle.
At 70° angle from center to picture's corner equals to using a lens with focal lenght ~ 2 mm (1/2" CCD). Wide angle optics needs a dome-type window. The mechanical durability of coated Luxar is good, and not an issue, unless you polish your windows with sandpaper. The right way to clean the glass is use a cotton swab and 1/1 alcohol-water mixture.
The idea was to reduce loss of light before it hits the lens and
then the CCD. Window's losses were reduced by 7...8% (from
400...650 nm) by BBAR- coated optical window. Considering only
meteoric line-emissions and their average relative strengths, the
downside of having limited bandwidth window is not significant
since the CCD and specially the camera's lens alrerady has several
surfaces with optical coatings limiting the bandwidth. The strong
line emissions below 400 nm are severly attenuated, but this would
mean the lens and the CCD would have to be made to work also on
for UV-A - such equipment are not produced for the commercial
market. All non-quartz glass has a fairly steep cut-off for
shorter wavelengths than 400 nm and so do the CCDs. Meteoric
continuum emissions and some air-glow spectral lines beyond 800 nm
are also attenuated by the coatings, but the CCD's performance is
the dominating factor there. Ignoring NIR continuum emission, the
BBAR-window still beats the soda-lime by 5% for meteoric emissions
and similar increase in meteor rates is possible by the almost 0.1
mag improvement in limiting magnitude.With a WAT-902H(2 ULTIMATE)
camera the increase is expected to be almost 250 meteors annually.
The idea to use CCD camera with much longer MTBF, say 50 000 h, and price <15% of a second generation intensifier, hopefully makes sense, even considering the lower numbers of meteors detected. Actually, these CCD cameras (hopefully) become obsolete some time in the future due to CCD chip evolution with increase of sensitivity, thus limiting camera's practical operating years. Unfortunaltey the disappearing B&W CCTV surveillance market may no longer be the driving force for product development - nothing new has come out since the ICX429ALL, which is replaced by Sony with ICX829ALA, an EXview 2 chip, 752 x 580 effective pixels (CCIR), 8.6 um x 8.3 um.
The total investment for CCD camera & lens + a PC with the Matrox 2 grabber card, is about 1000...1500 €. On top of that we must not forget the voluntary work by S. Molau on his free software to amateur use. Professional user license for MetRec is 250€. Comparing the operating costs of the Dedal 41-based intensified cameras at 2 €/h vs. CCD camera's operating cost 0.06 €/h, the difference is stunning. Even if the Gen. 2 MCP would reach it's theoretical 3000 h life-time, it would still cost 1€/h. Thinking this issue in €/meteors, the gap gets smaller, even if the intensifier works for those 3000 h. Placing bets on fragile intensifiers suits fine, if the funding can be arranged to replace this expensive camera component - way before it has reached it's expected life-time. The CCD cameras are much more robust, cheaper and have fewer failure modes. The intensified camera is only better in totally dark site and when there is no moonlight. For city, suburbs and moonlight nights the EXview CCD camera picks up just as many meteors as a 10-times more expensive intensified camera does.
Caution: Do not point the CCD camera to
the sun or let sun drift over the field-of-view with lens
shutter wide open (use auto-iris lens & timer). This
barbequing may or will damage the CCD chip.
Updated: May. 17. 2021
Super HAD CCD, EXview HAD CCD, EXview 2,
are trademarks (TM)of Sony
Impactron is a trademark of: Texas Instruments™.
Luxar is a trademark™ of Glas Trösch AG.
Link to CCD
Link to intensified (MCP) Meteor video camera
Link to skylight polarisation during twilight and it's reduction by filters
A few more external links with info on:
Effect of Integration
CCD camera's interlacing scan
Sony's B&W video CCD chips
Back to OH5IY's main page with near real-time NightSkyCam
Copyright 2002...2021, I. Yrjölä