Video Cameras characterization

Stuff I am working on
Post Reply
Daniel Wee
Site Admin
Posts: 2449
Joined: Wed 25 Feb 25 2009 8:00 pm

Video Cameras characterization

Post by Daniel Wee »

In this test, the pickup element was a helical antenna tuned to the 70cm (433MHz) band so you can expect extra sensitivity at these frequencies. The markers are placed at 432.1MHz (UHF LRS band) and 1.575GHz (GPS L1 band). These markers are labelled "1" and "2" respectively (in red).

Camera: KPC VSN505NH(H-RES), SN555, KX-550
Link:
Mode: NTSC
Nominal operating voltage: 5V
Minimum operating voltage: ~3.6V @170mA
Current draw: 140mA @5V (700mW)
TVL: 550
p-p (75-ohms): 1.11V (AC-coupled)
RFI profile:
KPC VSN505NH(H-RES)
KPC VSN505NH(H-RES)
Cameras 004.png (6.58 KiB) Viewed 49625 times
Camera: KPC VSN500NH(H-RES), SN777
Link:
Mode: NTSC
Nominal operating voltage: 12V
Minimum operating voltage: ~6.2V @80mA
Current draw: 50mA @12V (600mW)
p-p (75-ohms): 1.11V (AC-coupled)
TVL: 550
RFI profile:
KPC VSN500NH(H-RES)
KPC VSN500NH(H-RES)
Cameras 003.png (7.6 KiB) Viewed 49625 times
Camera: FH-25HD
Link: http://www.foxtechfpv.com/fh25hd-550tvl ... p-184.html
Mode: PAL
Nominal operating voltage: 9-12V
Minimum operating voltage: ~6.0V @70mA
Current draw: 50mA @11V (550mW)
p-p (75-ohms): 1.14V (AC-coupled)
TVL: 550
RFI profile:
FH-25HD
FH-25HD
Cameras 005.png (6.74 KiB) Viewed 49625 times
Camera: FH-20C, CCD Killer
Link: http://www.foxtechfpv.com/fh20c-mini-cm ... p-185.html
Mode: PAL
Nominal operating voltage: 5V (cable has embedded regulator)
Minimum operating voltage: ~4.3V
Current draw: 60mA @5V (300mW)
p-p (75-ohms): 0.76V (DC-coupled)
TVL: 420
RFI profile:
CMOS
CMOS
Cameras 006.png (5.94 KiB) Viewed 49625 times
Camera: FH??
Link:
Mode: NTSC
Nominal operating voltage: 12V
Minimum operating voltage: ~7.0V @60mA
Current draw: 50mA @12V (600mW)
p-p (75-ohms): 1.19V (AC-coupled)
TVL: 550
RFI profile:
FH??
FH??
Cameras 009.png (6.47 KiB) Viewed 49625 times
Camera: IF-FLD01CDN, KX-171
Link: http://www.foxtechfpv.com/foxtech-kx171 ... p-190.html
Mode: PAL
Nominal operating voltage: 12V
Minimum operating voltage: ~6.0V @90mA
Current draw: 90mA @12V (1.08W)
p-p (75-ohms): 0.93V (AC-coupled)
TVL: 420
RFI profile: (tested with two separate cameras of same model for verification)
KX-171
KX-171
Cameras 007.png (5.29 KiB) Viewed 49625 times
Camera: CM-3134BEX (B/W - appears to be IR sensitive)
Link: n/a
Mode: PAL
Nominal operating voltage: 12V
Minimum operating voltage: ~10V @90mA
Current draw: 100mA @12V (1.2W)
TVL: 420
RFI profile:
CM-3134BEX
CM-3134BEX
Cameras 008.png (5.97 KiB) Viewed 49625 times
12 August 2010
Daniel
Daniel Wee
Site Admin
Posts: 2449
Joined: Wed 25 Feb 25 2009 8:00 pm

Re: Video Cameras characterization

Post by Daniel Wee »

Of the seven cameras tested above, the four that are most similar are the:-
  • VSN500
    VSN505
    FH-25HD
    ?? (unknown model)
Cameras 016a.JPG
Cameras 016a.JPG (52.05 KiB) Viewed 49625 times
Apart from the VSN505 that runs on 5V, the others all run on 12V and appear to have similar specifications. These similarities aren't just restricted to the specifications, size and profile of the cameras. They extend to the insides as well - the boards all seem to be built around the same standard design:-
Cameras 020a.JPG
Cameras 020a.JPG (81.47 KiB) Viewed 49625 times
Cameras 026a.JPG
Cameras 026a.JPG (72.89 KiB) Viewed 49625 times
Having said that, the cameras clearly aren't from the same manufacturer. The reference design is pretty much the same though - the NextChip NVP2010/NVP2014 chipset (see attached datasheet) and the AD9943 CCD signal processor.

As can be seen from the spectrum sweeps above, these cameras produce a lot of wideband noise. In the two key frequencies that we're looking at - the UHF and GPS bands, the noise is pretty strong and will therefore be detrimental to performance of devices using those bands. As was reported, the VSN505 does exhibit less wideband noise (above 800MHz) compared to the other three and will not impact GPS bands in the same negative way. Unfortunately, the UHF noise is about the same and you will still have problems at UHF with the VSN505.

Grounding the casing seems to do little to improve the situation with the noise. It is likely that the noise is emanating from the supply lines themselves (and/or the video line). To test this, I wrapped the feed cable with one loop around a ferrite choke and re-tested the noise levels (using the unknown model camera) (yellow trace without filter, purple trace with filter):-
Impact of choke on feed cable
Impact of choke on feed cable
Cameras 028.png (7.65 KiB) Viewed 49623 times
The purple trace is from the test with the ferrite choke and the yellow trace, without it. It can be seen that the noise level is clearly lower with the choke on the feed cable, and this is with only one loop. We can, therefore, surmise that with better feed line filtering, the noise levels can be brought down considerably. This same scheme may also work to improve noise performance in the other cameras as well.

I then decided to make a simple filter to see if this would work better than the ferrite choke. The choke was common-mode and should behave similarly to the ferrite choke. The results on the VSN500 were as follows (purple trace without filter, yellow trace with filter):-
VSN500 with and without in-line filter (low inductance)
VSN500 with and without in-line filter (low inductance)
cameras 029.png (8.16 KiB) Viewed 49619 times
The filter didn't work as well as I had hoped but it did work. The thing I observed, though, was that shielding is important. The provided cable is unshielded and I think using a properly shielded cable could make a significant difference (especially at the higher frequencies - not so much in the sub-800MHz band) when used in conjunction with an in-line filter. I should try higher inductance values as well, to see if those will improve the situation.

Daniel
Attachments
AD-AD9943.pdf
(485.15 KiB) Downloaded 2015 times
NextChip CCD Image Signal Processor.pdf
(413.41 KiB) Downloaded 2191 times
Daniel Wee
Site Admin
Posts: 2449
Joined: Wed 25 Feb 25 2009 8:00 pm

Summary

Post by Daniel Wee »

A number of useful conclusions can be reached with the results from the tests above.

1. The choice of camera can determine your UHF and GPS performance. Some cameras, namely the NextChip reference design based ones, generate high levels of RFI in the UHF through to the GPS bands. In some cases, the noise reaches even 2.3GHz but at much lower levels. GPS modules typically operate around -140dBm levels and what we are seeing would completely wipe that signal out. It is possible that this noise may be coming from the internal switching power supplies (yes, plural) which may be interacting to produce harmonics and mixing products through the band. It is, perhaps, for this reason that the VSN505 works a little better than the VSN500, for the lack of an additional switching power supply.

2. Cameras with linear regulators (such as the KX171) exhibit markedly lower levels of RFI, but also consume more power (lower power efficiency) when run at higher voltages. This inefficiency can be mitigated, to some extent, by running the camera at lower operating voltages as allowed.

3. RFI noise from these cameras emanate largely from the unshielded and poorly filtered feed cables. Using shielded cables, clip-on ferrite chokes, and in-line filtering and mitigate noise levels to some extent. I have not explored the full extent of the effectiveness of such measures but the preliminary tests show significant improvement when the measures are properly employed. Because the noise is being carried in the feed cable, traditional advise on equipment separation may not work quite as well. Separation is still a good idea but even better if the line was properly filtered and shielded.

4. Video output levels should not be expected to conform to CVBS standards (1V p-p into a 75-ohm load.) It seems that there is a wide range of possible output levels, and some (typically CMOS types) tend to be DC-coupled, whereas the majority appears to be AC-coupled. These voltage level variations can affect how well the camera works with certain OSDs, the vibrancy of the colours produced, and the available dynamic range (eg. handling of bright scenes). Many receivers will adjust for level variations but not all do.

5. Many cameras use the same reference designs, or are sometimes simply re-labelled. In many cases, while very similar, there exists some qualitative differences in the boards produced.

6. In the metal cased cameras, the casing is usually painted on both the inside and the outside. This makes for poor grounding of the casing, diminishing their effectiveness as shields. Scraping away the paint can help improve case grounding, but the die cast aluminium oxidizes quickly and the oxide layer makes for poor conduction.

14 August 2010
Daniel
Daniel Wee
Site Admin
Posts: 2449
Joined: Wed 25 Feb 25 2009 8:00 pm

Re: Video Cameras characterization

Post by Daniel Wee »

Camera: KPC-DNR700NHB, WDR700
Link:
Mode: NTSC
Nominal operating voltage: 12V
Minimum operating voltage: ~7.4V @130mA
Current draw: 100mA @12V (1.2W)
TVL: 550
p-p (75-ohms): 1.2V (AC-coupled)
RFI profile:
KPC-DNR600 010.png
KPC-DNR600 010.png (8.09 KiB) Viewed 49305 times
KPC-DNR600 012a.PNG
KPC-DNR600 012a.PNG (37.12 KiB) Viewed 49305 times
As can be seen, the camera, like many other high performance cameras, produces quite a bit of RFI in the UHF433 bands, and to a lesser degree in the GPS L1 band. It is better than the Nextchip cameras (VSN-series for example), though, in terms of RFI emissions in these bands. The output signal is a little on the high side but other than that, this seems to be a good camera. At 7.4V the RFI is about 5dB lower than it is at 12V. Power dissipation is also lower at 7.4V indicating a better efficiency at the lower operating voltage. Anything below 7.4V yields video artefacts and eventually a complete loss of image. As a result, this might not be the best camera to run off 2S voltages unless you have a step-up (to about 8-10V would be ideal.)

I was told that the camera uses the new SONY HAD II CCD sensor chips but was unable to verify this. As can be seen in the next photo, this is another Nextchip processor based camera (NVP2170) and it might be said that Nextchip designs seem to be notorious for generating RFI.
KPC-DNR600 016a.JPG
KPC-DNR600 016a.JPG (53.6 KiB) Viewed 49305 times
The dynamic noise-reduction seems to work in retrieving blown out highlights, although the colour reproduction of the recovered parts are not as good. The user can adjust this using the built in OSD menu, along with quite a number of other parameters.
SnapShot.JPG
SnapShot.JPG (46.94 KiB) Viewed 49305 times
SnapShot(0).JPG
SnapShot(0).JPG (60.75 KiB) Viewed 49305 times
SnapShot(1).JPG
SnapShot(1).JPG (63.82 KiB) Viewed 49305 times
Overall, the camera was very sharp, had a lot of user configurable options through the convenient on-screen menu. The AGC response time is about 1-second and is very quick when it corrects for different lighting. DNR seems to work. Given the RFI though, you'll probably want to make sure that you have good separation between the camera and your UHF receiver and GPS. Having a choke or a suitable in-line filter on the cable where it comes out of the camera is also not a bad idea if you want the best performance out of the UHF system. Interestingly, this particular camera has quite a bit of RFI in the 2.4GHz band as well so users of 2.4GHz control equipment should be careful to test for the impact on their control range.

update: The WDR700 has some magenta colour shift that could be indicative of an insufficient IR (hot) filtering.

Daniel
14 Sep 2010
Daniel Wee
Site Admin
Posts: 2449
Joined: Wed 25 Feb 25 2009 8:00 pm

Re: Video Cameras characterization

Post by Daniel Wee »

Camera: KX-171LV-P
Link: http://www.nghobbies.com/cart/index.php ... cts_id=549
Mode: PAL
Nominal operating voltage: 5V
Minimum operating voltage: ~4.4V @150mA
Current draw: 120mA @5V (600mW)
p-p (75-ohms): 0.93V (AC-coupled)
TVL: 420
RFI profile:
KX171 001.png
KX171 001.png (7.14 KiB) Viewed 49302 times
You can see from the spectrum sweep that this camera puts out a little more RFI compared to it's 12V counterpart (sweep shown below):-
KX171 002.png
KX171 002.png (7.13 KiB) Viewed 49302 times
It still draws 120mA but is a lot more efficient than the 12V camera. Resolution looks to be exactly the same - no surprise there. The only change has been the power supply circuitry, which has raised the RFI a bit though at such low levels, it is nothing to be concerned about. For this type of testing, the difference in RFI generated could be due to the image in the camera and I would consider the noise performance of these two cameras to be identical for all intents and purposes.

Daniel
14 Sep 2010
Daniel Wee
Site Admin
Posts: 2449
Joined: Wed 25 Feb 25 2009 8:00 pm

KX-171 Variants

Post by Daniel Wee »

KX-171 Variants

The KX-171 from Cameray, apparently comes in a number of variants. The datasheet suggests at least twelve possible variants, of which three (six if you count PAL and NTSC separately) major ones are now available in the FPV market. These are:-

1. KX-171 (traditional), runs from about less than 7V to 12V @90mA
2. KX-171LVP (NGHobbies), runs from 4.4V to 5V @150mA
3. KX-171 G2 (RangeVideo), runs from about 8V to 12V @60mA
KX171-group 003a.JPG
KX171-group 003a.JPG (52.21 KiB) Viewed 48749 times
Physically, all three cameras look identical except for text on the backplate. The first two are very similar except for their power supply specifications, but the third one, the G2, is a different beast altogether - utilizing a different CCD as well as power supply system. In order to get the low-down on all these three and to see if the G2 with improved resolution, retains the low supply noise of its predecessors, I ordered all three cameras to run these tests.

Methodology

As we are doing some relative noise testing, all three cameras are re-tested using the same configuration with a 433MHz helically tuned antenna as the primary sensor. What this means is that it will pick-up noise better in the UHF band as this is the band of our primary interest. Markers are set for 435MHz (UHF), 1.575GHz (GPS), and 1.11GHz which is somewhere in between the two. The total span of the test is from 150MHz to 3GHz. We could have gone quite a bit higher but are constraining the test to a more relevant portion of the RF spectrum. The cameras were tested at the rated voltage, as well as the minimum operating voltage - which is defined as the voltage below which the picture starts to degrade noticeably.

RF Emissions Profile

The KX-171 (traditional) yielded the following spectrum profile, which we will use as the baseline for testing the other two cameras:-
KX171-Tests 008.png
KX171-Tests 008.png (7.62 KiB) Viewed 48761 times
As can be observed, there is some RFI, concentrated mostly in the UHF and below range. In practice, this level is only noticeable because of the sensitivity of the spectrum analyzer and the EMI/RFI sensing setup.

Then comes the KX-171LVP:-
KX171-Tests 010.png
KX171-Tests 010.png (7.45 KiB) Viewed 48761 times
The RFI levels appear to be slightly lower but in the same sort of range. One can surmise that for all intents and purposes, these two cameras are in the same class as far as RFI emissions go.

Compare this with the KX-171 G2:-
KX171-Tests 006.png
KX171-Tests 006.png (7.84 KiB) Viewed 48761 times
Just a cursory glance reveals a completely different RFI profile compared to the previous two cameras. Not only are the levels different, but the RFI covers a much broader spectrum, all the way up to nearly 3GHz. Testing at the minimum operating voltage usually show a slightly lower RFI level (no more than 5dB).

It is interesting that while the G2 has a lot more RFI, the actual increase at 435MHz is actually quite small, lower in fact that the first two cameras. Where there is a real increase would be in the GPS L1 band where it was previously completely clean but now has a significant level of noise. It is likely that this camera won't be a problem for UHF LRS systems but whether it will be a problem for GPS reception is something we will have to investigate. The current draw profile suggests that these cameras utilize linear regulators so much of the noise is coming directly from the video circuitry itself but propagated through the supply lines in particular.

We will next test the image quality that these cameras put out to see if the IQ gains (if any) warrant putting up with the higher noise levels of the G2.

Sensor Details

The sensor for the G2 is purportedly better than the other two. To start with, the CCD itself has a higher resolution of 752(H) x 582(V) for PAL and 768(H) x 494(V) for NTSC. This is a significant improvement over the other cameras' 500(H) x 582(V) for PAL and 510(H) x 492(V) for NTSC. That improvement, however, is confined more to the horizontal resolution with vertical resolution remaining virtually the same. In theory, this should yield better clarity in the horizontal (about 50% better) and should be noticeable in images. The reason vertical resolution wasn't improved much is probably due to the vertical rate limitation of PAL and NTSC. Even so, the G2 gives a 480TVL (NTSC) and 520TVL(PAL) so you should also see a corresponding increase in resolution in the vertical as a result of this. In this case, the PAL cameras will be far better in terms of resolution, compared to NTSC ones. Where you are likely to see the enhanced resolution of the G2 is in diagonal lines which are rendered more smoothly, compared to the "stepped" or "jagged" look of the lower resolution variants.

Resolution isn't the only difference though. The SONY Ex-View 1/3-inch CCD sensor also has different light handling. The following three images demonstrate the widely different low-light handling capabilites of the three cameras:-
KX-171 (standard)
KX-171 (standard)
KX171.JPG (49.47 KiB) Viewed 48758 times
KX-171LVP
KX-171LVP
KX171LVP.JPG (43.35 KiB) Viewed 48758 times
KX-171G2
KX-171G2
KX171G2b.JPG (63.65 KiB) Viewed 48758 times
It is quite apparent that the G2 is out performing the other two cameras and that the LVP is better with low-light than the original KX171. There were at least two dead-pixels on the G2 sensor, though, but this is not uncommon for this type of sensor and would be unnoticeable in daylight. If was going to fly at night, the G2 would represent a pretty good choice judging from the test images. It should be said that the LVP isn't doing too badly as it does look closer to what the actual scene looks like, whereas the G2 looks enhanced in brightness.

*In the test, I did not refocus the optics for the best infinite focal-length, except for the original KX-171 which had pre-focussed. This could account for some differences in the far field detail in the captured images.

Daylight Performance

The follow are daylight tests of the three cameras. The pictures should give you some idea of the key differences, both in colour/saturation and resolution:-
KX171 (outdoor daylight)
KX171 (outdoor daylight)
KX1712.JPG (42.02 KiB) Viewed 48750 times
KX171LVP (outdoor daylight)
KX171LVP (outdoor daylight)
KX171LVP3.JPG (37.29 KiB) Viewed 48750 times
KX171 G2 (outdoor daylight)
KX171 G2 (outdoor daylight)
KX171G23.JPG (38.19 KiB) Viewed 48750 times
KX171 (indoor daylight)
KX171 (indoor daylight)
KX1714.JPG (53.15 KiB) Viewed 48750 times
KX171LVP (indoor daylight)
KX171LVP (indoor daylight)
KX171LVP4.JPG (51.55 KiB) Viewed 48750 times
KX171 G2 (indoor daylight)
KX171 G2 (indoor daylight)
KX171G24.JPG (54.69 KiB) Viewed 48750 times
Surprisingly, the first thing that struck me wasn't the added resolution of the G2 but the somewhat over-saturated colours. In this respect, the LVP actually performed very well, giving the closest colours to the actual scene if somewhat muted. The G2 colours looked artificial by contrast, while the original KX171 was clearly lagging behind. The improved resolution of the G2 comes into its own with the indoor shot where you have a lot more details and where you see less colour fringing and moire compared to the original KX171. I would have to give this one to the G2 although the LVP held its own quite respectably.

RFI impact testing

See next post.

Conclusions

If I were to pick one of these cameras, it would be a toss-up between the LVP and the G2. The older KX171 is clearly, well, old. The LVP stands out in low RF noise performance, while still holding its own in image quality. The 5V supply can be an important feature for some users too, although the current draw is on the high-side (150mA@5V) compared to the lower current draw of the G2 (60mA@7V). The low light performance is not that great either. The G2, on the other hand, is a fantastic low light performer, has much better overall resolution and good indoor colours. The outdoor colours are a little bit to strong but this may be correctable in post-editing. There was a patch of blur-ness in the lower right of the image which could be due to the optics (lense) or sensor alignment. More significantly, the G2 puts out a more noise than the LVP and across much bigger spectrum (although slightly less noise in the UHF), including the GPS L1 band. Another important consideration is that the LVP costs only USD65 compared to the G2's USD95 (see links at the end of the post). If I were going to do night flying, I would get the G2 without question, but short of that - the LVP looks like a more attractive solution.

Daniel Wee
2 November 2010

http://www.rangevideo.com/index.php?mai ... cts_id=207
http://www.nghobbies.com/cart/index.php ... cts_id=549
Daniel Wee
Site Admin
Posts: 2449
Joined: Wed 25 Feb 25 2009 8:00 pm

Camera RFI impact on GPS

Post by Daniel Wee »

Camera RFI impact on GPS

It isn't yet common knowledge that the RFI from cameras can adversely affect GPS performance. The following test demonstrates an example of this:-
Before powering up the VSN500
Before powering up the VSN500
SnapShot(29).JPG (55.77 KiB) Viewed 49981 times
After powering up the VSN500
After powering up the VSN500
SnapShot(31).JPG (55.45 KiB) Viewed 49981 times
The second screen capture was taken within seconds of powering up the camera and it is obvious how the GPS reception has degraded dramatically. At this distance, the RFI is being radiated directly more so than it is coming across the wiring.

Of interest to me is the impact of the various types of KX171, and in particular the RFI performance of the G2 compared to that of the original KX171 and the LVP model (both of which have similar RFI profiles). In our test, the KX171 (original) and the LVP have negligible or no measurable impact on GPS performance, even when place within 1cm to the GPS module (LS20031). The G2, on the other hand, degrades GPS performance up to 5cm away from the GPS module, though not nearly as bad as the VSN500 (and presumably cameras of the same design).
Before powering up the KX171 G2
Before powering up the KX171 G2
SnapShot(19).JPG (55.26 KiB) Viewed 49981 times
After powering up the KX171 G2
After powering up the KX171 G2
SnapShot(20).JPG (54.96 KiB) Viewed 49981 times
Conclusions

Other cameras tested, such as the WDR700 also show degradation of GPS performance when in close proximity to the GPS module. In this test, most of the RFI would be direct radiation but in an FPV setup, it is highly possible that RFI is transmitted across the wiring and power supply as well. It was also found that different GPS modules performed differently in terms of how much they are impacted by the RFI. Some modules were still able to retain enough signal to lock even through RFI but the general accuracy is degraded.

In the light of these findings, the following recommendations are made:-

1. Keep as much separation as possible between the camera and the GPS module. Even 5cm is not enough separation in many cases. You probably need at least 12cm or more.
2. Where separation is not possible, use a camera with low or no RFI in the GPS L1 band.
3. Where and if possible, place an RF choke (ferrite toroids) on the camera cable as near the camera as possible - for cameras that are noisy (RFI-wise.)
4. Where and if possible, place, place an RF choke (ferrite toroids) on the GPS cable as near the GPS module as possible.*
5. Not all GPS modules perform equally. The one you are using may fare better or worse than another module or design.

*There are some people advising that the choke should not be at the GPS end and that is incorrect and wrong information. The choke should be as close to the GPS as possible.

Daniel Wee
3 November 2010
Daniel Wee
Site Admin
Posts: 2449
Joined: Wed 25 Feb 25 2009 8:00 pm

Resolution comparisons

Post by Daniel Wee »

Actual scene
Actual scene
Res-test 005a.JPG (78.51 KiB) Viewed 49966 times
KX171 original
KX171 original
SnapShot(37).JPG (48.84 KiB) Viewed 49967 times
KX171 LVP
KX171 LVP
SnapShot(34).JPG (47.66 KiB) Viewed 49967 times
KX171 G2
KX171 G2
SnapShot(33).JPG (50.92 KiB) Viewed 49967 times
FH-25HD
FH-25HD
SnapShot(35).JPG (58.72 KiB) Viewed 49967 times
WDR700
WDR700
SnapShot(39).JPG (46.4 KiB) Viewed 49967 times
Daniel
Post Reply