DIY microphone blimp and dead cat

January 5, 2015

High sensitivity microphones in the field need protection from both shocks and wind. After building my own shock mounts, I wanted better wind protection. A blimp and dead cat delivers good signal levels, fidelity and wind resistance.

Building a blimp

A microphone blimp (or zeppelin) looks like an airship. When covered with fur, it looks like a dead cat. The blimp creates a volume of still air around the microphone. Fur provides a better wind barrier than cloth because the fibres move and absorb wind energy.

I made my blimp from 83 mm diameter PVC plastic pipe and push-on PVC vented caps. These can be purchased from hardware and plumbing stores. Larger diameters and dead air space will better attenuate wind noise but larger, heavier blimps are increasingly difficult to handle. I have also used 55 mm diameter PVC for a smaller microphone, but installation of the mic into that narrow tube was more difficult.

Naked and partly assembled blimp with my Sennheiser ME66 microphone inside.

Naked and partly assembled blimp with my Sennheiser ME66 microphone inside.

I cut the PVC to length and drilled it nearly full of holes with a 16 mm diameter hole saw. Round hole saws are necessary, because large drill bits make a mess in thin plastic. I made a hole guide out of a thin steel sheet.

duide

Hole guide. The large holes are aligned with a row of holes on the PVC tube. The smaller holes guide pilot holes for the next row.

I used another hole saw to drill out the ends of the caps. I also sawed off the base of the caps, leaving about 10 mm to grip the tube. Cutting away more plastic makes the blimp lighter and the tube acoustically transparent. If you put the “unholy” tube to your ear, it sounds like a seashell. As more holes are drilled you will no longer hear any effect.

I left some plastic at the bottom of the tube for the grip. My pistol grip is secured with a wing nut (the bolt is threaded into the tube). I contact-glued some rubber (recycled bicyle inner tube) to the base to prevent slipping and squeaking. See my earlier shock mount post for a simpler, fixed pistol grip handle.

The microphone suspension is made from rubber bands. The bands pass through the tube and are secured with “H” pieces at both ends. I used a permanent marker to mark the tube where the rubber bands pass, for whenever they have to be replaced. The “H” pieces were cut from PVC tube. I contact-glued some rubber (recycled bicycle inner tube) to the inside of the “H” pieces to prevent slipping and squeaking. The microphone is inserted through the gaps between each rubber band, then the “H” pieces are lifted and rotated to twist the rubber bands and secure the microphone.

The microphone cable passes outside the rear cap. It is secured with two rubber bands to stop transmission of shocks through the cable. I have also slotted the pistol grip handle, so that the cable can be tucked away. Underneath the rear of the tube I have made a large hole where I can access the switch on my microphone.

The dead cat skin is made from fake fur that I purchased in a fabric shop. It has a single pile and a stretchy, open-weave backing material. Beware of denser furs, that can strongly attenuate the acoustic signal. It is easy to hand sew the dead cat by turning the fur inside out. Use a thread that contrasts with the fur. The fur will hide bad sewing.

Blimp with fur covering. The forward half is slipped over the blimp and the rear half has a zipper.

Blimp with fur covering. The forward half is slipped over the blimp and the rear half has a zipper.

My home-made blimp and dead cat is lightweight and cost about AUD$30 in materials. Commercial dead cats are heavier and often cost more than the microphone!

Model Length (cm) Diam. (mm) Mass (g) Cost Remarks
My dead cat 44 83 370 AUD 20
Rode blimp 49 125 550 AUD 225 Mass excludes fur.
Rycote Super-Shield 40 100 694 USD 359 Mass includes cable.
Sennheiser SEBS2 USD 926 No detailed specs.
Proaim BMP40R 40 USD 132 No detailed specs.
A comparison of blimps and dead cats that will fit a Sennheiser ME66 microphone. My home made blimp is the lightest and cheapest. The best price was selected from three sites: ebay, BHphotovideo USA, Video and VideoGuys Australia. Shipping costs are not included.

The (optional) cloth is made from leftover material from a light diffuser project. The material was light, stretchy and difficult to sew (this was the first time I’ve done machine sewing). Perhaps there are other acoustically transparent cloths that are easier to work with.

Blimp with cloth covering. It is secured with velcro (hook-and-loop) after stretching the material tight. This is a prototype and the sewing is a mess! An easier solution is to wrap the cloth around the blimp and secure it with rubber bands.

Blimp with cloth covering. It is secured with velcro (hook-and-loop) after stretching the material tight. This is a prototype and the sewing is a mess! An easier solution is to wrap the cloth around the blimp and secure it with rubber bands.

Blimp testing

For all tests, the naked mic is the reference response. A good wind protection system will have a frequency response and signal level close to the naked mic and much lower wind noise levels.

To compare frequency responses, I created broad spectrum noise by crumpling dry sheets of newspaper. The microphone was indoors (no wind), the distance to the source was fixed at about two metres (I didn’t write down the exact distance) and all recordings were made with a fixed manual gain. I made three recordings for each set-up and compared median spectra and third-quartile (Q3) frequencies. Q3 frequency is the frequency at 75% of the power spectrum.

The first graph compares frequency responses. Wind protection set-ups ordered from widest (best) to narrowest (worst) bandwidth are:

  1. Foam windscreen and microphone in shock mount. Q3 frequency = -258 Hz relative to naked mic.
  2. Cloth-covered blimp. Q3 frequency = -345 Hz.
  3. Fur-covered blimp (dead cat). Q3 frequency = -904 Hz.
  4. Fur over foam windscreen and microphone in shock mount. Q3 frequency = -1637 Hz.
Frequency response for different wind protection setups and dry newspaper sheets source. The spectra are aligned at 1–2 kHz to account for different signal levels. This comparison shows that fur coverings more strongly absorb higher frequencies (up to -1637 Hz reduced bandwidth). Measurements were made using Raven Pro 1.3 from 16 bit, 44.1 kHz recordings. The graph is zoomed to 0–11 kHz.

Frequency responses for different wind protection setups and dry newspaper sheets source. The spectra are aligned at 1–2 kHz to account for different signal levels. This comparison shows that fur coverings more strongly absorb higher frequencies. Measurements were made using Raven Pro 1.3 from 16 bit, 44.1 kHz recordings. The graph is zoomed to 0–11 kHz.

To compare signal levels I played back white noise through a small speaker. I made the white noise signal using Audacity 2.0.6. The frequency response spectra were rough, but signal levels were more repeatable than for the newspaper sheets. I compared mean levels averaged over about 60 seconds of playback.

The second graph compares signal levels. A difference of -3 dB corresponds to half the power. Wind protection set-ups ordered from strongest to weakest signal levels are:

  1. Foam windscreen and microphone in shock mount = -1 dB relative to naked mic.
  2. Cloth-covered blimp = -2 dB.
  3. Fur-covered blimp (dead cat) = -3 dB.
  4. Fur over foam windscreen and microphone in shock mount = -4 dB.
Signal levels for white noise source and different wind protection setups. The fur over foam windscreen gave a substantially weaker signal. Measurements were made using Raven Pro 1.3 from 16 bit, 44.1 kHz recordings.

Signal levels for white noise source and different wind protection setups. The fur over foam windscreen gave a substantially weaker signal. Measurements were made using Raven Pro 1.3 from 16 bit, 44.1 kHz recordings.

To compare wind noise levels, I simulated wind using a pedestal fan. I compared mean levels averaged over about 60 seconds of wind noise.

The third graph compares wind noise levels. Wind protection set-ups ordered from weakest wind noise level (best) to strongest (worst) are:

  1. Fur-covered blimp (dead cat) = -25 dB relative to naked mic.
  2. Cloth-covered blimp = -20 dB.
  3. Fur over foam windscreen and microphone in shock mount = -20 dB.
  4. Foam windscreen and microphone in shock mount = -6 dB.
Wind noise waveforms for fan source and different wind protection set-ups. This comparison  shows very poor wind protection from the foam windshield. Waveforms from Audactiy 2.0.6.

Wind noise waveforms for fan source and different wind protection set-ups. This comparison shows very poor wind protection from this foam windshield. Waveforms from Audactiy 2.0.6.

The final graph compares signal-to-noise amplitude ratios from the above two tests. Wind protection set-ups ordered from highest (best) to lowest (worst) are:

  1. Fur-covered blimp (dead cat).
  2. Cloth-covered blimp.
  3. Fur over foam windscreen and microphone in shock mount.
  4. Foam windscreen and microphone in shock mount.
Signal-to-noise ratios for different wind protection set set-ups and the experimental conditions in the preceding two graphs. The fur over foam system performed surprisingly well, however the Q3 frequency was substantially lower than for the cloth-covered blimp (see the first graph).

Signal-to-noise ratios for different wind protection set set-ups and the conditions in the preceding two graphs. The fur over foam system performed surprisingly well, however the Q3 frequency was substantially lower than for the cloth-covered blimp (see the frequency response spectra).

Blimp is best

These tests illustrate the trade-off between signal strength and fidelity versus wind noise reduction. The light foam windshield is best for indoors, but hopeless for wind. The cloth-covered blimp is best for “light breezes”. The fur-covered blimp is best for “strong wind”. For high fidelity recording, I suggest to have all three wind protection systems in the field bag and to select the one appropriate to the conditions.

I always see professionals using the dead cat outdoors. For news recording and like, there is no second chance. For nature recording, we choose to avoid windy conditions and I would recommend lighter wind protection (i.e. cloth-covered blimp) for higher signal levels and wider bandwidth.

I will not use the fur over foam system because the signal level and bandwidth were lower than for other set-ups. Some manufacturers claim that their expensive fur is acoustically transparent. I invite them to send me a sample product (to fit a Sennheiser ME66) so that I can test it myself.

Disadvantages for the blimp/dead cat are the bulkier size and slower microphone installation. Installing the microphone actually only takes a couple of minutes because I leave the rubber bands on the blimp.


DIY 12 Volt LED light strip bar

January 4, 2015

12 Volt LED lights are expensive. 12 Volt LED light strips are cheap. I bought a light strip and made my own custom light bar for camping. The materials cost about AUD$20.

LED light bar design

This project upgrades my previous 12 V LED camping light. The objective was to install a rugged LED light bar in the back of my ute (pick-up).

My home made 12 V LED light bar in action with two LED strips, about 540 Lumens.

My home made 12 V LED light bar in action with two LED strips, about 540 Lumens.

I bought a 5 m white LED (WLED) strip light roll and some switches on ebay. Here are the specifications for the strip:

LED 5630 SMD
Colour temp. (K) 5000–5500
Viewing angle (deg.) 120
LED density (LEDs/m) 60
Intensity (lumens/m) 840–900
Current (A/m) 1
Specifications for ebay 5630 LED strip light. The intensity is more realistically about 290–320 lm/m at this current.

I wasn’t sure that one strip would be sufficient, so I installed two side-by-side. Some basic soldering skills are needed for attaching new leads after cutting the strip. I have actually found one strip = 290 Lumens/m × 0.93 m = 270 Lumens quite sufficent.

Small diameter power cables can be used when currents are small. For my design, 1 A/m × 0.93 m × 2 strips = 1.86 A. The following table can be used for sizing power cables:

AWG 2 A 4 A 8 A 16 A Current
1 m 22 22 18 16
2 m 22 18 16 12
5 m 18 14 12 8
10 m 14 12 8 6
15 m 12 10 6
20 m 12 8 6
Twin-core
American Wire Gauge (AWG) cable sizes for 4% (0.48 V) voltage drop, twin-core cable. Read the cable size at the intersection of current along the top row and cable length on the left column. Calculated from American Wire Gauge data on Wikipedia.

Here’s the same table with ISO (mm2) sizes:

ISO 2 A 4 A 8 A 16 A Current
1 m 0.5 0.5 1 1.5
2 m 0.5 1 1.5 4
5 m 1 2.5 4 10
10 m 2.5 4 10 16
15 m 4 6 16
20 m 4 10 16
Twin-core
Metric (mm2) cable sizes for 4% (0.48 V) voltage drop, twin-core cable. Read the cable size at the intersection of current along the top row and cable length on the left column. Derived from the previous table.

For bigger strip lights there are two important points to consider:

  1. The maximum current rating of LED strips is low (typically 5 A) and maximum strip length is short (typically 5 m).
  2. There could be substantial voltage losses along the strip.

The basic solution for the above problems is to use short LED strips in parallel.

LED light bar assembly

The strip lights I bought are waterproof. However, I wanted better protection from shifting cargo in the back of my vehicle. I made a lighting fixture out of thin timber strips (10 mm deep), with a plywood backing (3 mm) and a clear polycarbonate plastic cover (2 mm). The timber I cut myself from scrap and the plywood was leftover from other projects. Polycarbonate and acrylic (‘perspex’) can be found at a fibreglass supplies shop.

Detail of home made LED light bar with wiring cover removed. This end is thicker to accommodate the diameter of the switch.The plastic cover has a foam gasket. The wiring cover has a rubber gasket (recycled bicyle tube, contact-glued on). I contact-glued aluminium foil inside the fixture to minimise reflection losses.

Detail of home made LED light bar with wiring cover removed. This end is thicker to accommodate the diameter of the switch.The plastic cover has a foam gasket. The wiring cover has a rubber gasket (recycled bicycle tube, contact-glued on). I contact-glued aluminium foil inside the fixture to minimise reflection losses.

The two LED strips in my light are wired in parallel, for equal voltage and equal intensity. Here are two different wiring diagrams including switches.

Wiring diagram for two parallel LED strip lights (only 3 LEDs shown in each strip). Switch 1 (master) controls both strips. The optional switch 2 (dimmer) controls the second strip. For electrical safety, there is an in-line fuse between the supply and switches (2A mini blade fuse in my case).

Wiring diagram for two parallel LED strip lights (only 3 LEDs shown in each strip). Switch 1 (master) controls both strips. The optional switch 2 (dimmer) controls the second strip. For electrical safety, there is an in-line fuse between the supply and switches (2A mini blade fuse in my case).

wiring2

Wiring diagram for two parallel, individually-switched LED strip lights (only 3 LEDs shown in each strip).

The self-adhesive backing on the strip lights did not hold for long. I re-glued the strips with contact glue and secured the strips with four cable ties (zip ties).

The thin, flexible plastic cover did not seal well against the timber. I made a gasket from foam tape to improve dust and insect resistance.

According to my solar regulator, this light bar draws 0.7 A at 12.4 V with one strip (design 0.93 A) and 1.4 A at 12.3 V with two strips (design 1.86 A). The intensity is overrated in the first table above.


Understanding LED strip lights

January 4, 2015

I bought a 12 volt white LED strip light on ebay. After some testing and further research, I figured the luminous intensity was overrated. This post should be helpful for understanding LED strip lights.

LED strip light package

An LED strip is basically a package of Surface-Mount-Device (SMD) LEDs with current limiting resistors. The strips are flexible and easy to install with a self-adhesive backing. Waterproof strips have a clear, silicon-like layer over the electronics.

Section of ebay LED strip light (both sides). The strip is 12 mm wide. There are two groups of three LEDs (yellow). The SMD resistors are black. A splice can be seen in the lower left. The strip can be cut and new leads soldered at the copper tabs. Contrary to the logo on the back, the adhesive back is not genuine 3M quality.

Section of ebay LED strip light (both sides). The strip is 12 mm wide. There are two groups of three LEDs (yellow). The SMD resistors are black. A splice can be seen in the lower left. The strip can be cut and new leads soldered at the copper tabs. Contrary to the logo on the back, the adhesive back is not genuine 3M quality.

Examining my strip light, I see repeated groups of three LEDs, each with one 39 ohm resistor. I guess the wiring is series within-groups and parallel between-groups.

Possible LED strip light wiring. R = resistor.

Possible LED strip light wiring. R = resistor.

An LED strip is a passive device. It contains no electronics to regulate the power and brightness. If the supply voltage drops, then brightness drops and vice versa.

Currents are highest near the supply end. Firstly, the current for all the LEDs passes through the supply end. Second, the copper is thin and there could be substantial voltage losses along the strip, meaning that LEDs nearer the supply operate with higher voltages and currents.

You can feel that the strip is warm at the supply end because more heat is produced at higher currents. Long strips in series can melt.

Strip light specifications

Here are the specifications for the White LED (WLED) strip I bought:

LED 5630 SMD
Colour temp. (K) 5000–5500
Viewing angle (deg.) 120
LED density (LEDs/m) 60
Intensity (lumens/m) 840–900
Current (A/m) 1
Specifications for ebay 5630 LED strip light. The intensity is more realistically about 290–320 lm/m at this current.

The above lumens are overrated. The following table compares the ebay LED with specifications for two other 5630 LEDs from Samsung and Philips. The lumens/milliamp calculated for the ebay LED is impossibly high. The ebay seller seems to have calculated strip values using the typical LED intensity, but the LEDs are operating at a lower current.

ebay Samsung Philips
Intensity (lm) 14.5 20 32
Current (mA) 17 50 100
Voltage (V) 3.8 3 3.1
lm/mA 0.85 0.40 0.32
Comparison of 5630 SMD LEDs (single LEDs). I calculated the ebay specification as follows: current = 1/60 A, intensity = 870/60 lm. Voltage was calculated by subtracting the drop across the current limiting resistor from the supply voltage (12 – 39 × 0.017) and dividing by three. Voltage seems a bit high. Other LED data from Samsung SPMWHT5225D5WAR0S0 and Philips LUXEON 5630 datasheets.

If I assume the Philips lumens/milliamp result, I estimate 0.32 lm/mA × 17 mA = 5.33 lm and 60 × 5.33 lm = 320 lm/m for the ebay strip light. Alternatively, I could assume the Samsung current and then estimate intensity as 14.5×17/50 = 4.83 lm (the datasheet showed that intensity is approximately proportional to current) and 60 × 4.83 lm = 290 lm. Both of these estimates seem reasonable in side-by-side comparisons with my 97 lumens Fenix HL21 headlamp.

Some reasons for running the LEDs at lower currents are: to reduce heat, reduce voltage losses and allow longer strips. Using the Samsung LED for comparison, at 60 LEDs/m the current is 60 × 20 lm = 3 A/m and a 5 A strip would be only 5/3 = 1.67 m long. Lower currents and less heat is safer and easier for home made strip lights.

WLEDs come in different colour temperatures. “Natural” or “daylight” LEDs (5000–5500 K) are effectively brighter than “warm white” (3200 K). This is because human colour vision is more sensitive to wavelengths in the middle of the daylight spectrum. The ebay listing did not include the Colour Rendering Index (CRI). My impression is that the LEDs are rather blue compared to 5000-5500 K daylight.

SMD LEDs cast light across a very wide angle (viewing angle typically 120 degrees). If installed inside fixtures, the fixture should be shallow to achieve a wide floodlight. Avoid vertical mounting (e.g. wall mounting), because up to half of the light will be directed upwards.

It is best to have the SMD LEDs horizontal and facing down

It is best to have the SMD LEDs horizontal and facing down

Saving money

Here are some cost-saving measures to expect from ebay LED strip lights:

  • LEDs from low grade bins and mixed bins (variable intensity, colour, forward voltage).
  • Thin copper (more resistance, more heat, shorter maximum strip length).
  • Spliced strips (possible failure points).
  • Poor quality adhesive backing (strip lights come unstuck after a short time).
  • Low currents (see previous discussion).

Cheap ebay LED strip lights may be false economy if more LEDs are required to achieve satisfactory brightness.


Digital photo random noise and solutions

August 1, 2014

Noise in digital photos is very familiar to “pixel-peepers”. Noise obscures details, reduces dynamic range and looks bad when photos are viewed at large sizes. This post examines how ISO sensitivity, colour and exposure affect random noise levels. Random noise is usually dominant and more difficult to remove than fixed pattern noise (“dark noise”) and banding noise, which are not covered here. This post concludes with some tips for reducing random noise when taking digital photos.

Measuring noise with Raw Therapee and Argyll CMS

Camera profiling targets can be used for measuring digital photo noise. A fair evaluation requires “device RGB” images, without any colour processing, noise reduction and other enhancements. I used Raw Therapee to get raw photo linear RGB images:

  1. Processing Profile: Neutral.
  2. Input Profile: No profile.
  3. Output gamma: linear_g1.0 (this will disable the Output Profile).
  4. Spot White Balance on neutral grey patch.
  5. Crop to target and save as TIFF.

Next, I used Argyll CMS scanin to read the images. My chart was an X-Rite ColorChecker Digital SG:

scanin -v -p -a -dipn IMG_9999.tif ColorCheckerSG.cht ColorCheckerSGD50.cie

Argyll CMS generates a text file including mean patch RGB measurements and standard deviations (SD). I pasted these data into a spreadsheet and then computed coefficients of variation (CV = SD/mean). CVs measure relative noise levels. Engineers often prefer the signal to noise ratio (SNR ~ mean/SD = 1/CV), which is commonly reported in decibels: SNR = 10×log_10×((mean/SD)^2).

High ISO noise

Digital photographers know from experience that noise levels increase at higher ISO settings. I tested this by photographing my CCSG target at different ISO settings. The light source was direct afternoon sunlight, to minimise glare from the darker patches. I stopped the lens down to f/11, metered off a 18% grey card at ISO100 (shutter 1/250 s for this example) and took the first image. I then increased ISO in one stop increments (ISO200, 400 … 1600) and held the exposure value constant by halving the shutter speed for each successive image (1/500, 1/1000 … 1/4000). A constant exposure value is necessary for a fair comparison.

The graph below plots CVs versus reflectance. Dotted lines indicate middle-grey (18%) and 32 dB SNR. Middle-grey (L* = 50) is a good reference because most information in digital photos is usually present in the mid-tones. If middle-grey is noisy, people will notice. Middle-grey has a rather low 18% reflectance because human vision lightness response is non-linear. The SNR reference I have stolen from DxOMark: 24 dB (CV = 0.063 ) is bad SNR; 32 dB is good (CV = 0.025); and 38 dB (CV = 0.013) is excellent.

Coefficients of Variation (CV) for greyscale patches (reflectance = 0.7% to 91%) at different ISO settings. CVs for neutral grey patches computed as means of individual RGB channels. Camera: Canon EOS 400D Digital SLR. Target: X-Rite ColorChecker Digital SG.

Coefficients of Variation (CV) for greyscale patches (reflectance = 0.7% to 91%) at different ISO settings. CVs for neutral grey patches computed as means of individual RGB channels. Camera: Canon EOS 400D Digital SLR. Target: X-Rite ColorChecker Digital SG.

The ISO effect appears as a higher “noise floor” at higher ISO settings. This is random amplifier noise (“read noise”). Read noise has a fixed level at each ISO setting and is additive. ISO800 and ISO1600 for my Canon 400D are noisy, from the shadows to the highlights, even though this test was in good light (Exposure Value 14.9).

Noise also decreases very rapidly with increasing signal levels. This suggests “shot noise”, which results from variations in photon counts at each photosite and follows a Poisson distribution, where the CV is the reciprocal square-root of the mean.

DxOMark has measured raw photo noise data for most digital SLR cameras. They present SNR, like in the next graph. For the Canon EOS 400D, DxOMark suggested a maximum ISO664 for good middle-grey SNR (nearest setting ISO400) and my results agree.

Signal to Noise Ratios (SNR) for greyscale patches (reflectance = 0.7% to 91%) at different ISO settings. Presentation like Full SNR graph from DxOMark. Camera: Canon EOS 400D Digital SLR. Target: X-Rite ColorChecker Digital SG.

Signal to Noise Ratios (SNR) for greyscale patches (reflectance = 0.7% to 91%) at different ISO settings. Presentation like DxOMark’s Full SNR graph. Camera: Canon EOS 400D Digital SLR. Target: X-Rite ColorChecker Digital SG.

Colour noise

Colour noise (or “chroma” noise) should also be familiar to digital photographers, but I think less well understood. To investigate colour noise, I examined colour patches from the ISO100 photograph in afternoon sunlight used above.

CVs for individual RGB channels are compared in the graph below. For blue, the red signal is weak and just as noisy as for black. For red, the blue signal is weak and noisy. Green wavelengths are between blue (shorter wavelengths) and red (longer wavelength). The green channel has similar CVs for all three of these colour patches. Also observe that neutral patches reflect all wavelengths equally. A correct white balance is necessary.

Coefficients of Variation (CV) for red, green and blue patches plus white, middle-grey and black for reference. The dotted line equals 32 dB signal to noise. Camera: Canon EOS 400D Digital SLR. Target: X-Rite ColorChecker Digital SG.

Coefficients of Variation (CV) for red, green and blue patches plus white, middle-grey and black for reference. The dotted line equals 32 dB signal to noise. Camera: Canon EOS 400D Digital SLR. Target: X-Rite ColorChecker Digital SG.

Following the above results, in our photos we expect more red colour noise in blues and more blue colour noise in dark reds. The illuminant is also important. For example, tungsten lights have lots of red and very little blue wavelengths. Photographs in tungsten lighting will have higher blue colour noise compared to daylight, which has a more even spectrum.

Exposure effects

Increasing the signal (photons counted) reduces shot noise (“Poisson noise”). The following graph uses data from my linearity testing post. It demonstrates that over-exposure reduces noise. Shadows, represented by the black patch, are most sensitive to exposure.

Coefficients of Variation (CV) for white, middle-grey and black versus exposure. Camera: Canon EOS 400D Digital SLR at ISO100. Target: X-Rite ColorChecker Digital SG. Base exposure (delta EV = 0) metered using a 18% grey card.

Coefficients of Variation (CV) for white, middle-grey and black versus exposure. Camera: Canon EOS 400D Digital SLR at ISO100. Target: X-Rite ColorChecker Digital SG. Base exposure (delta EV = 0) metered using a 18% grey card.

Expose To The Right” (ETTR) has been promoted to increase resolution in shadows and critics have argued that ETTR is not necessary for newer, higher bit-depth cameras. However, ETTR also reduces noise, as demonstrated above. This is true for faster shutter speeds typical of hand-held photography and not for long exposures, where “dark noise” appears.

Solutions for random noise

There are several photographic techniques to reduce random noise:

  • Use a better camera, which usually means a larger image sensor (reduce shot noise).
  • Use a lower ISO setting (reduce read noise).
  • Expose to the right (reduce shot noise).
  • Supplement the natural light (e.g. flash, reflectors) if it is weak (reduce shot noise).
  • Prefer broad, even spectrum artificial lighting (reduce colour noise).
  • Use image averaging for static scenes and a tripod. This reduces noise levels everywhere.
  • Use High Dynamic Range software to merge an overexposed image (with reduced shot noise in shadows) with a correctly exposed image. Two images may suffice and hopefully HDR effects can be disabled when not wanted.

All of the above techniques reduce noise in the raw photo data. Further software noise removal is often required for the final image. However, software can’t restore details that are lost in noise. And image sharpening tends to amplify noise. Low noise images save processing hassles.

Finally, here is an example where ETTR has reduced digital photo noise at ISO400.

Cape York, Queensland. Photographed in the afternoon with a Canon EOS 400D at ISO400.

Cape York, Queensland. Photographed in the afternoon with a Canon EOS 400D at ISO400.

Shadow noise detail at 200% from the above photo at base exposure (shutter 1/100 s). Raw photo processed in Raw Therapee with no noise reduction and no sharpening. There is subtle chroma noise in the shadows. There was more noise in deeper shadows towards the bottom of the photo, but this would be difficult to see on lower contrast ratio displays and for everyday viewing conditions.

Shadow noise detail at 200% from the above photo at base exposure (shutter 1/100 s). Raw photo processed in Raw Therapee with no noise reduction and no sharpening. There is subtle chroma noise in the shadows.

Shadow noise detail at 200% with ETTR (shutter 1/60 s). Exposure corrected by -0.74 stops (linear ×0.60) in Raw Therapee.

Shadow noise detail at 200% with ETTR (shutter 1/60 s). Exposure corrected by -0.74 stops (linear ×0.60) in Raw Therapee. If you have a decent display, you should be able to see an improvement over the base exposure.

 Raw photo histogram for ETTR image above. ETTR can be recommended for this example, where the raw photo RGB channels are not clipping.

Raw photo histogram in Raw Therapee for ETTR image above. ETTR can be recommended for this example, where the raw photo RGB channels are not clipped.


Simple, low cost solar and battery monitoring

July 22, 2014

Battery power can be stressful. How much capacity is being used? How much is remaining? How much is being recharged? Installing cheap panel ammeters have greatly improved my understanding of my auxiliary (‘aux’) battery system.

monitor

Basic solar charge controller upgraded with a DC ammeter. This is from my old system (3 A solar panel, 5 A meter).

Battery monitoring

Voltage is a poor indication of battery capacity in working solar power systems where voltage varies with charging and loads. The battery would need to  be rested before measuring voltage for estimating state of charge. Then voltage should be corrected for temperature.

It is more informative to measure the current going in to the battery and the current going out. If the solar charging current is higher than the average load, the battery will be storing the difference. If the average load is higher, the battery is supplying the difference.

If I am running out of battery capacity, I can use the ammeters to guide me in taking corrective action. The input ammeter can be used to improve solar panel positioning and to estimate daily charging input. The output ammeter can be used to decide which loads to disconnect or reduce.

Current measurement

I use analog DC panel ammeters because they are simple, do not consume power and are cheap (you can find them on ebay). Low-current ammeters (less than about 50 A) do not require an external ‘shunt resistor’ and are very easy to install (in series). I once tried cheap digital ammeters, with no instructions, and burnt both of them.

Ammeters can be installed on the positive cable (or negative if you prefer) and on both the input and output sides of the aux battery. The ammeters I have are marked with a ‘bar’ symbol at the negative or ‘downstream’ terminal. If the needle moves in the wrong direction, then swap the connections.

I recently built a custom power distribution panel using recycled connectors and plugs. You can buy DC power outlets and some have features like USB charging and maybe a voltmeter – just add an ammeter in series.

Custom power distribution panel with DC ammeter (10 A, which should cover any loads I plan to plug in). It’s made from timber and plywood and easily modified should I need to add more outlets. A voltmeter would be useful, but I already have one in my new charge controller.

Custom power distribution panel with DC ammeter (10 A, which should cover any loads I plan to plug in). It’s made from timber and plywood and easily modified should I need to add more outlets. A voltmeter would be useful, but I already have one in my new charge controller.

Save money

Real battery monitors have microprocessors and can estimate remaining battery capacity at variable discharge rates via Peukert’s equation. Battery monitors are expensive gadgets. A simple volts and amps monitoring system is cheap and simple.

I now have an expensive and unnecessary Morningstar Prostar 30-M solar charge controller, which measures voltage and current. However, Morningstar do not recommend (see Tech Notes) to connect inverters and compressor fridge/freezers to the load terminals (installing a ‘clamping diode’ is a possible workaround). It is cheaper to buy a good basic solar charge controller, separate ammeters to monitor the charge and load currents and perhaps a voltmeter for precise battery voltage.


Fenix HL22 headlamp review

July 21, 2014

I am still happy using my Fenix HL21 headlamp, which I reviewed back in 2013. I recently purchased a Fenix HL22 for a gift. The Fenix HL22 has a wider, more useful beam, but is not as rugged as the HL21 and does not have a diffuser.

Fenix HL22 headlamp.

Fenix HL22 headlamp. My photography skills have improved.

Design and features

  • Single Cree XP-E LED.
  • Single AA battery.
  • 45 grams, excluding battery.
  • Three modes: 120 lumens, 45 lumens and 3 lumens with memory.
  • Smooth reflector, glass lens with anti-reflective coating.
  • Water-resistant (IPX-6) and impact-resistant.

Compared to the HL21, the Fenix HL22 is 4 grams heavier and 23 lumens brighter. Both have the same low 3 lumens mode, which I find most useful. In my samples, the colour of the HL22 LED is more white. The beam spot is about 3 times wider (= 9 times larger in area). This beam pattern is more useful than the HL21. However, the HL21 diffuser accessory creates a more even light for close work and reading. The HL22 does not have a diffuser.

Squeezing the power button for 0.5 seconds switches the Fenix HL22 on. Pressing it momentarily switches modes. These operations are opposite to the HL21. It’s easier to change modes with the HL22.

Build quality and pricing

The Fenix HL22 looks and feels cheaper than the HL21. The plastic hinge does not feel durable. The battery compartment has metal threads and no o-ring. It is designed to seal out rain and dust (IPX-6). The HL21 has plastic threads, an o-ring and is waterproof (IPX-8, 2 m underwater). The headband on the HL22 often needs readjustment, same as the HL21.

Overall, the Fenix HL22 headlamp appears to be satisfactory. Shop around – they retail for around USD35.


AGM and wet deep-cycle batteries compared

May 31, 2014

Many times I have read that Absorbed Glass Mat (AGM) lead-acid batteries are ‘better’ than traditional wet (flooded) lead-acid batteries. This article examines the characteristics and performance of the AGM and wet batteries. The ‘best’ choice depends on the budget and the application. AGM’s are sealed and can charge three times faster. Wet deep-cycle batteries cost less and can last 1.4 times last longer. Where either technology is safe, most users would not notice the performance differences between wet and AGM batteries and cost-savings can be substantial.

Products compared

I downloaded products specifications for Ritar DC series AGM (China), SunStone ML series AGM (China) and Trojan Signature Line wet deep-cycle batteries (U.S.A.).

I compared ratings at 12 V. Many of the Trojan batteries evaluated were 6 V and I then considered two 6 V in series. For two batteries in series the mass and voltage are doubled and the capacity (Ah) is unchanged. And for two batteries in parallel, the mass and capacity are doubled and the voltage is unchanged.

Capacity versus mass

More lead = more capacity, as the following graph shows. The linear regression fitted to the entire dataset can be used to predict mass and identify fraudulent ratings. A battery that plots below the trend (actual mass less than predicted) is over-rated.

Mass versus 10-hour capacity for Ritar DC series AGM, SunStone ML series AGM and Trojan Signature Line flooded lead-acid batteries. Linear model fitted to all data points.

Mass versus 10-hour capacity for Ritar DC series AGM, SunStone ML series AGM and Trojan Signature Line flooded lead-acid batteries. Linear model fitted to all data points.

Charging rate versus capacity

More lead = higher acceptance = faster charging. For the two AGMs evaluated, maximum charging rate is 0.30 times 10-hour capacity. For the wet batteries, maximum charging rate is 0.14 times the 10-hour capacity (Trojan actually specifies 0.10 to 0.13 times the 20-hour capacity).

Maximum charging rates for Ritar DC series AGM, SunStone ML series AGM and Trojan Signature Line flooded lead-acid batteries. Maximum charging rates are commonly summarised as a fraction of capacity (= the gradient of the rate-capacity line).

Maximum charging rates for Ritar DC series AGM, SunStone ML series AGM and Trojan Signature Line flooded lead-acid batteries. Maximum charging rates are commonly summarised as a fraction of capacity (= the gradient of the rate-capacity line).

A second consideration is charging efficiency, because not all of the input current is stored. Some of the energy is lost to heat and gassing. Charging efficiency for AGM batteries is about 95% and greater than 85% for wet batteries. Multiplying charging rates by efficiency increases the differences between AGM and wet batteries. For the two AGMs evaluated, the effective maximum charging rate is 0.29 times 10-hour capacity versus 0.12 times for the wet batteries.

These results agree fairly well with my own testing. My Trojan T105s bulk-charge at about 0.13 times 10-hour capacity. My old Ritar DC12-100 charges at about 0.24 times 10-hour capacity.

Note that maximum charging rates vary between manufacturers, depending on the design, construction and safety margins. For Trojan AGMs (not included in this study), maximum charging rate is about 0.2 times 20-hour capacity.

Cycle life

Cycle life determines lifetime cost. AGM batteries are sealed, maintenance free and tend to have shorter cycle lives than wet deep-cycle batteries, as the following graph shows. The larger Trojan Signature Line batteries are exceptionally rugged and can deliver 50% of rated capacity after 1200 cycles. Next, the Ritar DC series are true deep-cycle AGM batteries and can deliver 60% of rated capacity after 850 cycles. Last, the SunStone ML series are really for standby use and can deliver 60% of their rated capacity after 500 cycles.

life-capacity

Cycles to 60% of rated-capacity for Ritar DC series AGM and SunStone ML series AGM batteries and 50% of rated-capacity for Trojan Signature Line flooded lead-acid batteries. Average depth of discharge was 50% for all three manufacturers. Observe that cycle life is unrelated to capacity, except for the smaller Trojans, which must differ in design and construction to the higher-capacity models.

Be careful when comparing cycle life ratings that the depth of discharge and the percentage reduction in final capacity are the same. Increasing depth of discharge will reduce battery life. Increasing the capacity end-point will reduce battery life (e.g. the Trojan cycle life ratings would be a little bit lower at 60% of rated capacity, rather than 50% as specified).

Summary

I have experience with both Ritar DC series AGM and Trojan Signature Line wet deep-cycle batteries. The following table summarises what I think are the important differences between these two battery technologies.

Main pros and cons of AGM and wet deep-cycle batteries.
AGM Wet
Sealed, spill-proof, no explosive hydrogen gas vented. Spill-resistant with the right caps, but must be allowed to vent gas.
Charging nearly 3 times faster than wet batteries. Charging slower, but less load on the charging system.
Advanced construction and higher cost than wet batteries. Lower cost than AGM.
AGM battery life varies between different designs. Maintenance not possible. Cycle life can be 1.4 times greater than AGM. Maintenance prolongs life, especially in hot climates.

Two Trojan T105s are doing a good job in my dual-battery plus solar system:

  • Excellent value for money (low cost, high capacity, long life).
  • Slower charging is no issue because Australia is a big country and I drive long distances when travelling which allows the batteries plenty of time to charge. When camped, the charging rate is limited by my solar panels, which can deliver a maximum of 12 A.
  • Maintaining the T105s is not a great inconvenience. They don’t use much water.

AGMs are required where there are safety concerns or unusual operating conditions:

  • Inside vehicle, where gases can’t be vented.
  • Marine applications (salt water + battery acid = Chlorine gas).
  • Short drives, where the batteries have to charge quickly.
  • Serious off-road driving, where wet batteries can spill (my Trojans spilled some acid at Cape York, Queensland).