Fig 1. OMAX V434B stereomicroscope, C$450
MicroscopeNet.com is a distributor based in Canada of generic, no-name microscopes and accessories. It's similar to the California-based Amscope; both apparently import unbranded products from China.
Amscope implies that you'll be getting branded products but without the brand label: "This microscope is made by the same technicians and on the same production line as optical instruments for Leica, Zeiss, Nikon and Olympus". "Its retail value is about $3,000", says Amscope of a scope they sell for $699.
So the prices are good, but how well do these generic scopes perform? Amscope says things like "This microscope offers high resolution and good depth within a broad field of view. It provides crystal clear sharp stereo images" but neither Amscope nor MicroscopeNet offer specifications, such as line pairs per millimeter for resolution or numerical aperture for the objectives.
Both offer a 7-day evaluation period, but shipping cost is on you, both ways, if you decide to return it.
After exchanging emails with MicroscopeNet and getting good impressions, and talking around town, I decided to give MicroscopeNet a try; I like that they are in Canada (as am I), eliminating potential customs/border issues which remain despite a free trade agreement (NAFTA) with the US. I ordered an OMAX V434B trinocular stereomicroscope, pictured in figure 1, priced at about C$450 (C$ ≅ US$ in 2009). Two days later, it arrived in a well-packaged box.
Generally the V434B was as described. It's not a multi-thousand-dollar instrument so I expected some 'rough edges', and I found a few, including:
The lighting controls are smooth and work well, though of course there is a colour shift as the halogen bulb brightness is adjusted.
The stand and components are solid and aligned. I like that the post mount allows for considerable vertical range, and that the optical unit rests in a ring, thus it can be rotated or removed and used independent of the stand.
Fig 2. The convergence angle is ~12.1°. (Here the objective shroud is removed.)
The convergence angle is approximately 12.2° as determined by the measured distances in figure 2 (the lens diameter is 13.9 mm, with 20.6 mm between lens centers; the distance to the stage from the lens center surface is approximately 96.4 mm at 2.5 zoom). Thus each objective is viewing the subject plane at an angle of half that, ie., approximately 6.1°.
Viewing from an angle causes keystone distortion, and means only a narrow strip (plus/minus the depth of field, and perpendicular to the plane of the objectives) can be in focus for each objective at any instant. Those strips can only align if the focus brought to the center (which of course is typical).
The phototube uses only one objective (the left one), so flat subjects, if they are to be perpendicular to the sight line, must be tilted 6.1°.
Subjectively, the V434B delivers images of quality comparable to an Olympus SZ61, Meiji EMZ-TR, and Acuter 700, and better than a Wild M4A. Even at 4.5x zoom with 20x eyepieces, there is less chromatic aberration than observed in more expensive equipment. Alignment seems fine; I haven't noticed any eye strain, etc.
Resolution (based on the Rayleigh Criterion) of the microscope with three different cameras was measured using three different methods (see appendix A, below) at selected zoom settings, obtaining the following results:
* Resolution at 0.7x predicted by extrapolation, not measured
The table below shows the field of view as observed through the provided 20x eyepiece at various zooms, and also through the phototube with a 10x Wild eyepiece (not provided with the package), as photographed with Canon A570. The column 'Min pixels wide' is a calculation based on the measured resolution (from the table above) and the width of an inset no-vignetting camera frame (4:3 aspect ratio), with a pixel sampling factor of four (for details, see How many pixels does a microscope camera need?). The fields of view are of course affected by the eyepiece used. If the camera has insufficient pixels, it will not be able to capture all the detail provided by the microscope (the image will be fuzzier than when viewed via the eyepiece), especially at lower magnifications (higher field of view).
|Field of view (millimeters)|
|20x eyepiece||10x phototube||Camera needs:|
|Zoom||Diameter||Diameter||Camera frame||Min pixels wide|
Adding a 2x Barlow lens from MicroscopeNet created significant, off-putting chromatic aberration (blue fringes).
Chromatic aberration is discussed further in 'Photomicrography' below.
MicroscopeNet offers adapters for the V434B's phototube. With a MicroscopeNet Canon DSLR 2x adapter (SKU A14CanonSLR) and a Canon EOS 600D / Rebel T3i camera, the images show some chromatic aberration. This was investigated using the LoCA Focus Analyser, which measures longitudinal chromatic aberration by looking at the degree of edge spread in a photomicrograph of a sharp straight edge (eg., a carbon-soot-coated razor blade). The test procedure and results are described at Stereoscope: OMAX V434B.
The degree and nature of the chromatic aberration varies with zoom level, but are similar to the result below for 1.5x zoom:
Fig 3. LoCAte analysis for a straight-edge target (razor blade) inclined at 13°. The best-focus distance for the green colour channel is about 12 microns closer than that of the red, and about 27 microns closer than that of the blue channels. The blue channel resolution is not as good as that of the red and green.
Ideally the curves for the three colours would overlap. Instead, the red and blue channels are coming to their best-focus points a few tens of microns further from the lens. Also, even at their points of best focus, the red and blue channels are not coming to focus as well as the green channel.
For test results at other zoom levels, please refer to Stereoscope: OMAX V434B.
These are not great results, and it means that when pushed to its limit, photomicrographs from this set-up (microscope + adapter for the camera) will not be sharp, and will have colour fringing. For example, have a look at the photomicrographs in the Appendices, eg., the top panels of figures B1 and B2, where even at the points of best focus there is colour fringing and lack of clarity.
(Results from an earlier investigation into chromatic aberration, using a different method, are presented in Appendix B, below.)
In practice, I have found that I can usually get better photos using a macro lens (eg., Stereogram of Navajo/Page Sandstone sand grains) or even hand lens (eg., figure 3 in Adapters for afocal optical coupling) than using this stereomicroscope. So I tend to use this stereomicroscope for visual work, not photomicrography.
I haven't tried to determine how much the camera adapter's optics contribute to the result because aberrations are visible even when looking through the microscope's eyepieces, either using eyes or using an afocally-coupled camera. I expect the stereomicroscope is contributing most of the aberrations.
Despite some shortcomings, when compared with name-brand stereomicroscopes at similar prices, the V434B is good value, ie., good performance and utility per dollar. In that ratio, the main benefit is in the denominator, ie., the price; quality is seemingly comparable to some higher-priced name-brand models. Chromatic aberration is better or comparable to many name-brand scopes, and is reasonable when using the eyepieces; however, 'colour fringing' is present when using the phototube for photomicrography, especially at high zoom; how much of this is due to the MicroscopeNet camera adapter vs the microscope is uncertain.
Caveat: I'm not a professional microscopist. If you have suggestions or information that would improve this page, I'd be pleased to receive comments via this web form.
Measuring the resolution of a microscope
Characteristics of the Canon A75, Canon A570 IS, and Canon EOS 600D / Rebel T3i digital cameras
Adapters for afocal optical coupling (mounting a camera on a microscope).
Three methods were used to measure resolution of the V434B microscope: Airy pattern, FFT of images, and MTF of a slanted edge. The results of each are described in the sections that follow:
See Measuring resolution for information about the procedure used here.
The numerical aperture of a microscope can be obtained by measuring the Airy pattern produced. In the photomicrograph below (figure 1) of an integrated circuit, a few of the surfaces happen to strongly reflect light into the stereomicroscope objective, creating strong diffraction patterns that are visible where they spill into adjacent dark areas. The inset image in figure 1 shows the diffraction pattern at the pixel level. Green lines were placed by eye to mark the minima.
Fig 1. Integrated circuit, photographed with Canon A75 on a V434B stereomicroscope at 4.5x
Figure 1 was obtained with the V434B at 4.5x zoom via the phototube with a 10x Wild M12 eyepiece and a Canon A75 camera focused at infinity at maximum zoom (3x, or 16.2 mm). The camera produces JPEG images with in-camera image processing (eg., sharpening).
In this configuration, each sensor pixel corresponds to 1.030 μm in the subject plane (determined by photographing a 1 mm scale with the same microscope set-up). Assuming an average wavelength of λ = 550 nm (yellow-green) for the observing light, we get:
NA = 550 nm / (2π 1.40) = 0.063
Fig 2. Airy rings, using a Canon A570
on a V434B stereomicroscope at 4.5x
To check this result, I repeated the procedure with an Canon A570 IS, which has a pixel pitch of 1.87 μm. Figure 2, at the right, shows an Airy pattern photographed with the A570 at full 4x zoom (23.2 mm) (only the green channel intensity is shown). Based on those rings, using the same method as above, the estimated NA is 0.058 .
Fig 3. Airy rings, Canon Rebel XT
on a V434B stereomicroscope at 4.5x
To determine whether the camera lenses or built-in imaging processing or the phototube eyepiece were influencing the result, I tried projecting the image directly onto the sensor of a Canon Rebel XT with no camera lens or phototube eyepiece, just a plano skylight filter on the body for dust protection. The pixel pitch of the XT is 6.42 μm. Raw images were used, with no sharpening or other post-processing in their conversion to TIFF. To obtain magnification sufficiently high to resolve the diffraction bands, the camera body was mounted on a tripod with approximately 20 cm of space between the top of the phototube and the flange of the camera. Figure 3, left, shows an Airy pattern (green channel intensity). Based on those rings, using the same method as above, the estimated NA is 0.064 .
The resolution as measured by the Rayleigh Criterion is the Airy radius (the distance to the first minimum), then using the figure 1 result:
rAiry = 0.61 λ / NA = 0.61 λ / 0.063 = 5.25 μmsubjPlane (at 4.5x zoom)
By the Rayleigh Criterion, two points of light on the subject plane can be resolved if separated by more than the Airy distance, rAiry. But information is still available from subject plane objects that are smaller than rAiry, especially if they are isolated, or too dim to generate visible rings and a broad central peak, as can be assessed by looking at figure 1; note the 5 μm scale bar in the lower-left corner.
Figure 1 is a 100% crop. Figure 1's inset and figures 2 and 3 were created by zooming the camera-produced images in Photoshop to make individual pixels visible (outlined by thin white lines in figure 1 inset), then doing a screen capture and analysing the captured image.
See Measuring resolution for information about the procedure used here.
Two targets were used: a straight edge razor blade coated with carbon (this was the MTF target, discussed in the next section), and a piece of metal with 'random' scratches and stains (see figure 4). Both targets are expected to have higher spatial frequency content than can be passed by the microscope. The targets were photographed at four zoom levels: 1.5x, 2.5x, 3.5x, and 4.5x. The results presented are from a Canon A75, and were checked with a Canon A570 and Rebel XT with no lens, both of which yielded similar results.
A 512 x 512 pixel well-focused area of each photo was selected and then processed with ImageJ. First contrast was maximized (to normalize the FFT) using Process->Enhance Contrast, and then a FFT generated (Process->FFT->FFT).
The length of the perpendicular line in the FFT, caused by the edge in the subject, reflects the maximum spatial frequency passed by the microscope and camera. Having the subject edge slanted separates it visually from the energy at the axes that are artifacts of the FFT process.
Similarly, the radius of the fuzzy ball created by the random scratches on metal reflects the maximum spatial frequency passed.
|Fig 4. Straight edge (left) and scratched metal (right), with FFT's below|
The maximum spatial frequency declines with zoom strength. The estimated FFT values are given in the spreadsheet fragment in figure 5, below. Of course the estimates are qualitative, somewhat arbitrarily based on where the 'fuzzy white area' (which is probably dependent on display factors) seems to fade out. MTF (discussed in the next section) is based on FFT analysis and provides more quantitative results.
Fig 5. Resolution, derived from maximum spatial frequency in FFT
The values in columns B and C are from ImageJ. The scale values in column E were obtained by photographing a millimeter scale. Column G is simply column B / column E; similarly for H. The resolution figures are thus in subject plane units of length. The NA figures are calculated as NA = 0.61 λ / resolution, using λ = 550 nm.
The results in figure 4 also tell us that Canon A75 is a reasonable match for the microscope in this zoom range. If the 'FFT cloud' were to extend to the edges of the plane, as it nearly does at 1.5x zoom, there is a risk of aliasing or, if it goes beyond 2 px/cycle (the Nyquist frequency), losing information. But likely at 0.7x zoom there would be too much spatial frequency information for the A75 to capture.
See Measuring resolution for information about the procedure used here.
The MTF (Modulation Transfer Function) of an optical system is a measure of how well the system transmits contrast, as a function of spatial frequency. An MTF of 1 is perfect transmission, and of 0 is no transmission. Typically MTF varies inversely with frequency. Points of interest are MTF(50), the frequency at which contrast has been reduced by half (50%), MTF(30), where contrast has been reduced to 30%, and MTF(9), which is the frequency at which contrast is as at the Rayleigh Criterion of resolution.
MTF is computed from an image of a contrast edge slanted about 5% with respect to the camera sensor rows. This was done at four zoom levels of the V434B stereomicroscope, once using a Canon A75 and again with a Canon A570, mounted on the trinocular port via a 10x eyepiece. The cameras were zoomed to their maximum (3x and 4x, respectively), focused at infinity, and with maximum aperture. The results were processed with two MTF software packages, Imatest and QuickMtf. The results are shown below.
Fig 6. Resolution, derived from MTF(9), at various zoom level. Canon A75
All figures with micron units refer to length on the subject plane. Scale was determined by photographing a millimeter scale. Resolution results for MTF(9) are highlighted, as those are the Rayleigh criterion resolutions.
Here are the results using a Canon A570:
Fig 7. Resolution, derived from MTF(9), at various zoom level. Canon A570
Ideally, a sharp edge slanted at 5 degrees would make the transition from low to high ('rise distance', from 10% to 90%) on the camera's sensor over a distance of 1.25 pixels (not zero pixels, due to the slant). Non-ideal images, blurred by the loss of high frequency components, have a longer transition. Thus the rise distance is a measure of resolution. These measurements are recorded in columns P and Q in the tables above, both in sensor pixels and subject plane microns.
To quantify longitudinal chromatic aberration (wavelength-dependent differences in focal distance), a straight edge razor blade was photomicrographed from one the side of the blade, with the blade inclined (by propping up one end of the blade) at an angle of 18° to the optical axis (the angle is somewhat arbitrary; the steeper the angle, the more rapid the rate of defocusing at either side of the focus point). Using an inclined target ensures that best focus occurs somewhere along the blade. Any difference between red, green, and blue channels becomes visible in terms of differing locations of the point of best focus along the inclined target when examining each colour channel individually. A Canon EOS 600D / Rebel T3i DSLR was used, with sharpening off (set to zero). LED lighting.
Using ImageJ, the photo was convolved with a 3x3 matrix [0 0 0, -1 2 0, 0 -1 0] to emphasis the transition from gray to black at the razor edge (ImageJ->Process->Filters->Convolve, also available in Photoshop as a 'custom filter'). To make the result more visible, contrast was expanded (ImageJ->Process->Enhance contrast). The convolved image was separated into colour channels (ImageJ->Image->Color->Split channels) and converted to a stack (ImageJ->Image->Stacks->Images to stack). Flipping through the three images in a stack makes it obvious if the point of best focus varies with colour channel, eg., using ImageJ->Image->Stacks->Animation. Point of best focus was estimated by subjectively, by eye.
Horizontal displacements were measured using Photoshop, and converted from pixels to length by using the known scale. The target was at a known angle to the view of the microscope so the corresponding vertical displacement can be calculated using trigonometry.
The razor blade has three surfaces per side: Two bevels and the body. The body has marks parallel to the edge, the shallow bevel has marked perpendicular to the edge, and the final bevel appears black in the lighting used here (see caption of figure 4). The blade is 32 μm thick, so the edge is 16 μm further from the microscope than the blade's body surface.
At a zoom of 4.5x, there is significant chromatic aberration (figure B1). The blue channel comes to focus 90 μm higher than the red, and the green channel focus is 37 μm below. The depth of field is about 90 μm, so unfortunately there is no point at which all three channels come into focus. Looking at the RGB panel (top-most) of figure B1, the best choice may be between the focus points of red and green.
Fig B1. Photomicrograph of razor blade edge at 4.5x zoom, inclined at 18° (left side highest). The top image is the normal RGB image; the background is grey, the beveled razor edge shows as black, with the secondary bevel showing as vertical scratches. Red, green, blue channels are shown after a convolution filter to emphasis edge focus. Estimated points of best focus are indicated by arrows.
At 0.7x zoom, the red and green channels are pretty well aligned, but the blue channel comes to focus roughly 1.11 mm higher than the red/green. The red and green depth of field is about 673 μm, so again there is really no point at which all three channels come into focus. However, perhaps the breadth of the blue range makes the problem less severe, as suggested by less colour-fringing in the RGB view (top panel, figure B2).
Fig B2. Photomicrograph of razor blade edge at 0.7x zoom, inclined at 18° (left side highest). The top image is the normal RGB image; the background is grey, the beveled razor edge shows as black, with the secondary bevel showing as vertical scratches. Red, green, blue channels are shown after convolution filter to emphasis edge focus. Estimated points of best focus are indicated by arrows.
2010-Nov: First issue
2015-Dec: Longitudinal chromatic aberration sections added.
2016-Nov: LoCAte focus analysis results added.