You can also find me on qrz.com.

This is a design for a 13.8 V power supply that can charge a sealed lead battery and simultaneously power equipment with a 13.8 V nominal supply voltage. Maximum current is 1.5 A and this can be delivered continuously, even when the output is short-circuited. Output current will be limited further when the regulators run too hot.

The main circuit consists of the mains transformer with a secundary 20 V – 2 A rating, rectifiers, buffer capacitor and regulators. The bridge rectifier must be suitable for at least 30 V and 2 A; I used one with a higher spec, always good to overdimension a bit. The first regulator U1 acts as a current limiter. The maximum current is determined by R2, R3, R5, R10 and R11; I combined resistors from my junk box to get to 1.5 A as close as possible. A single 0.82 Ω with a 5 W rating will get you close enough. Resistors R12…R16 constitute a circuit to measure the actual current and offer protection against accidental short circuiting of the measuring terminals; it is designed to yield 0.5 V/A with a 327 Ω internal resistance. Cooling of the regulators is important of course, get a cooling block of 2 K/W; if cooling is inadequate, the regulators will protect themselves from overheating by decreasing their output current.

The second regulator U2 functions as the voltage regulator for the 13.8 V output voltage. R1, R2 and R26 determine the output voltage; R1 should be 120 Ω, adjusting the output voltage should be done by varying R2 and R26. I don’t use potmeters or trimmers here because they can easily have bad contacts or be misadjusted, potentially damaging the equipment connected to the output. That’s why I adjusted the output by trying different resistor values. Nominally the value for R2 is 1205 Ω, but tolerances will almost surely necessitate to adjust this. For good output regulation it is absolutely essential that the connection between R1 to U2’s output pin is not shared with a wire that carries the output current; I have mounted a separate wire from R1 directly on the output pin.

The circuit for the power indicator LED looks a bit odd, but showing the presence of the AC voltage of the transformer is the only way to indicate that power is delivered to the output. When power is off and a battery is connected to the output, the whole DC part is powered by the battery due to the protection diodes around the regulators.

The second circuit provides temperature compensation for the output voltage. This whole circuit can be omitted if you don’t want or need this temperature compensation. Ideally a sealed lead battery should get a standby charging voltage of 13.8 V at 20º C, decreasing with 18 mV/K above that. This is done by using a small diode as a sensor at the outside of the cabinet because the temperature inside the cabinet can be much higher. The diode’s forward voltage has a temperature coefficient of -2.2. mV/K and this is amplified and converted to a current that draws from the adjust output of U2, decreasing the output voltage by around 18 mV/K.

Trimmer R32 should be set at 20º C such that the voltage over R31 is very low, say 50 mV. That way the temperature compensation starts to be effective above 20º C. Any diode or base-emitter of a transistor can be used as a sensor, just what is convenient. No harm is done when the diode’s leads are short-circuited or left unconnected. There are simpler ways to provide temperature compensation, but this circuit is safe and affects the output voltage in a predictable way, even when there’s a malfunction somewhere in this circuit.

Usage is very simple: connect the equipment to the 13.8 V output and connect the battery in parallel. When the equipment draws less than 1.5 A, the battery is charged with the remaining current and the output voltage drops to the battery voltage. As most 13.8 V equipment is intended to be useable with a battery, the 12 V of an almost empty battery under charge will adequately power the equipment.

First the loop amplifier was built, intended to be used near the loop.

This is a fairly simple affair, consisting of an input stage with high input impedance to avoid loading the tuned-loop circuit, followed by a buffer stage to provide a 50 Ω output. The loop consists of two turns of unshielded wire, with a center tap connected to ground, resulting in a balanced loop; the input amplifier’s load is negligible. I’m not sure if C11 and C15 are really necessary but I just let them sit there for the time being. Earlier I added these capacitors in an effort to balance the loop, but that didn’t work very well, using the centre tap of the loop is much more effective in suppressing QRM.

The tuning range is 500-1800 kHz, spanning the entire medium-wave band. This is of course critically dependent on the loop dimensions; extra varicaps can be added if needed, and if the tuning frequency doesn’t come high enough, you can experiment with decreasing C11 and C15 or leaving them out altogether. Supply voltage can range between 10 and 15 V, current consumption is around 25 mA. The tuning voltage should be between 0 and 12 V (the maximum rating of the BB112 varicaps); there’s no point in raising the voltage above 9 V as the capacitance hardly decreases above that voltage. If you look at the BB112 datasheet, you’ll see that its capacitance is only shown for a minimum voltage of at least 0.3 V, depending on which manufacturer’s datasheet you use. I’ve found that applying a tuning voltage of 0 V isn’t a problem, reducing the number of varicaps required; this works because the loop has only two turns and is not very large, so the signal on the loop and the varicaps usually stays below 10 mV at my location (no large signals from nearby stations).

Next you need a control box to generate the tuning voltage and power for the loop amplifier. Here I’ve been profoundly lazy and built the simplest setup I could think of.

The power comes from 10 AA NiMH batteries and fully charged these will last for a few days. The tuning voltage is not stabilized; if that bothers you, it’s easy enough to add a 78L09 regulator for powering the tuning potmeter. Inspired by the design of the AOR LA400 antenna I use UTP cable for power and tuning voltage, the output signal is fed to the radio with RG-58 coax cable.

The loop consists of 2 turns of 40 x 33 cm.

In the upside down view the connectors are visible. The plastic box is supposed to protect the bottom with the connectors from rain, but I’m afraid the plastic box will be too shallow for that, have to replace it by a deeper one. Alternatively, some plastic foil can be wrapped around. The black wire is the loop’s centre tab, the two wires left and right of it are the loop ends that carry the signal. And finally, the rope serves to hang the antenna off a fence in my garden.

All connections of the amplifier box are placed at the bottom so that water does not run easily into the connectors. Lower left BNC connector is the 50 Ω output, the RJ-45 connector is for power and tuning voltage and the two Cinch ones are for the loop; the centre tap of the loop goes to one of the ground terminals of the Cinch connectors.

The loop amplifier circuit board is bolted to the bottom plate of the aluminium box and all connectors are directly mounted on the circuit board, avoiding wiring. Upper left shows the 30 varicaps that I ended up with, not as much as I had anticipated beforehand. Being able to get the circuit board out by just loosening 6 screws of the box is really convenient for modifications and experimenting.

Tuning knob at the left, power switch in the middle and battery voltage test point at the right. The battery block contains 10 AA batteries for powering 12 V stuff and an additional 3 packs of 8 AA batteries each, because the battery block also serves to power my home-built radio.

Left round hole has the RJ-45 connector for power and tuning, middle connector is 12 V power and the left hole is empty (the box was re-used from another project, which I even can’t remember). The battery block has a 6-pole DIN plug as power connector for both battery packs, this has become sort of my standard power interface. The black and red terminals are for charging the 24 AA battery pack, the 10 AA battery holder is simply taken out for charging.

And finally here’s a quick-and-dirty video I made of the antenna in operation, connected to my home-built receiver.

There isn’t much to be had in the way of native 28mm lenses for Sony FE cameras. Of course, Sony offers the FE 2/28 and I’ve tried that one out several times but couldn’t bring myself to buy it. I just don’t like its rendering, the pictures are dull and lifeless to my eyes. That only leaves legacy lenses as prime 28mm options.

For some time now I’ve been looking around for a legacy 28mm to serve as a wide-angle for landscapes and cityscapes. I like this focal length: it’s really wide-angle without the extreme characteristics of 21mm or even 24mm. First place I looked is in the Minolta SR-mount range because I happen to collect them, but alas, there is no top-of-the-bill model there: all of them suffer too much from field curvature. The plain MD 2/28 is the best I have, showing the least amount of field curvature and nice contrast and colors.

On to other brands then. The last version of the Leica Elmarit-R 2.8/28mm has an excellent reputation but its price tag is way out of my comfort zone. I once had a Contax Distagon 2.8/28 but even though it was sharp I didn’t like its rendering, cold and harsh tones and inky black shadows. That leaves two other well-reputed lenses: the Olympus OM Zuiko 3.5/28mm and the SMC Pentax 3.5/28mm, which I acquired only recently; it’s not for sale very often and if it is, prices can be outrageous. I also have the compact SMC Pentax-M 3.5/28 about which I wrote in an earlier post, and as a reference the Minolta MD 2/28 will join the ranks in this comparison.

The contenders

Test results

This time I’m not going to show specific examples of center or corner sharpness of the various lenses because I couldn’t come to a clear-cut way to compare the results. All test shots are available here in full-resolution.

There are differences between the lenses but they’re not huge and I think they are a mix of variations in field curvature, asymmetries and object distances. I never use test cards on a wall because precise alignment is very difficult and a perfectly flat subject is hardly representative of actual photographic scenes. And last but not least, I own only one sample of each lens model so it’s impossible to say if one of them is exceptionally good or just average; I think there are no real lemons between the 4 samples here.

Observations

• The Pentax and the Olympus share a very similar color rendering, the Minolta is markedly cooler.
• The Pentax-M 3.5/28 suffers from appreciable field curvature, yielding unsharper corners than the other ones. The Minolta 2/28 has some field curvature, the Olympus and Pentax 3.5/28’s score best here. I evaluated this at f/8 because that’s my preferred landscape aperture.
• None of the lenses is particularly good wide-open although center sharpness is reasonable. The Pentax-M 3.5/28 has the least amount of purple fringing.

Epilogue

The Minolta MD 2/28 is the odd one out here: it’s a much faster lens and its color signature is different. Still it acquits itself competently as a landscape lens and I feel it’s the only Minolta 28mm lens up to this task.

The Pentax-M 3.5/28 suffers from too much curvature of field to my taste and that leaves the Pentax and Olympus 3.5/28 lenses as the best options. This is borne out by my experience with them: both are capable of sharp pictures across the frame and I can’t really pick a favorite. The Pentax is best stopped down to f/11 for that last bit of sharpness everywhere, the Olympus doesn’t benefit too much from stopping down beyond f/8.

Recently a Leitz Summicron-R 50mm 1:2 fell into my lap; it was my father’s and I’d completely forgotten about it until my mother asked me what to do with it.

As the Zeiss Loxia 50mm 1:2 is one of my favorites, I set out to compare the Leitz 50 with that treasure.

Unfortunately the Leitz doesn’t attain infinity focus on the Sony A7R2 with my Novoflex NEX/LER adapter, so I couldn’t use my usual landscape-type shot and did a bookshelf instead. I don’t think the Novoflex adapter is off, it’s more likely that my father adjusted the focus during his adaptation of a Practice B200 with a Leica-R bayonet mount; yup, he liked to tinker around. He would have had the time of his life in this era of adapting everything to mirrorless cameras!

Tests

Anyway, back to the comparison. I haven’t made a Lightroom profile yet for the Summicron so I disabled the profile for the Loxia to get a more level playing field, although I don’t know how much of the built-in profile still is applied to the Loxia raws. All test shots were made with the Sony A7R2 and processed in Lightroom.

The Loxia has more contrast and looks sharper, the Summicron shows halo and clear purple fringing, which showed up in other shots as well.

More or less the same differences: the Summicron is hazy, the Loxia also but less. Sharpness is reasonable for both lenses. I checked between center and corner for sharpness variations but nothing special happened. I definitely prefer the Loxia here because of its better rendering. I’m showing only one of the corners here, fortunately none of the lenses is plagued by symmetry problems.

Still blue tinges around the highlights with the Summicron, sharpness impression is much improved and at least as good, if not a little better, than the Loxia, which has higher contrast though.

The Loxia has the edge here, not as much halo and equal sharpness. No surprises when looking halfway between center and corner.

Both lenses perform very well by now, the Summicron looks a bit sharper and the haze is gone.

Here it shows that the Loxia’s field curvature increases, the Summicron is clearly sharper and the haze in the corners is gone.

At f/4 the Loxia has a ring of lesser sharpness between center and corners, clearly shown in this sample. The Summicron has much less field curvature.

Center performance isn’t shown for f/5.6, there isn’t much between them. The Loxia is getting a bit sharper now, but the Summicron looks really good.

The Loxia improves again because depth-of-field covers up for field curvature but the Summicron still is better.

Epilogue

Of course there are a lot of characteristics that I didn’t test here. Geometric distortion is not a concern of mine, almost always a custom-made Lightroom profile eliminates it to a very high degree. Ghosting and flare remain to be seen for the Summicron, the Loxia is very good in this respect. Bokeh quality? Not tested either.

Both lenses have a reputation, and it shows. The Leitz Summicron 2/50 really comes to life at f/4 and beyond with high sharpness across the frame; I don’t like the rendering at f/2 and f/2.8, too hazy and too much purple fringing, especially wide-open.

The Zeiss Loxia Planar 2/50 is a different beast: it’s already quite good wide-open, even better at f/2.8 but suffers from field curvature at f/4 and f/5.6, preventing a perfectly sharp picture across all of the frame. At f/8 depth of field covers up the field curvature, yielding adequate sharpness across the frame. I use the Loxia a lot and indeed, when I want everything sharp I stop down to f/8; for details near corners and edges f/4 is excellent and f/2.8 will do in a pinch.

Testing is nice but using a lens is a different matter. I much appreciate the fact that the Loxia is transferring its lens data into the EXIF so that Lightroom automatically applies the appropriate profile. Using legacy lenses means editing the EXIF data in order to have Lightroom apply the corresponding lens profile automatically. And that’s a pain, in spite of whatever tricks can be used to remember or record what you’ve been doing in the field. And the Summicron is really good for landscapes I guess, but it needs a repair to regain infinity focus… and at the other end of the price scale there’s my very cheap Minolta MD 2/50 which is really sharp at f/8 across the frame, so I’d simply take that one if the Loxia’s edge and corner sharpness wouldn’t be up to snuff. That said, I did landscape shots with the Loxia and they turned out fine, the Loxia simply delivers beautiful pictures.

It appeared for sale on a Dutch auction site. The seller readily accepted my offer and today it arrived. No box, but fortunately the manual was included, one A5-ish sheet with English on one side and Japanese on the other.

Operation of the panorama head appears to be quite simple in practice. I followed the manual’s advice to avoid anything shorter than 50mm and used the Sony A7R2 with the Zeiss Loxia 2/50. I have standardized on Arca-Swiss heads and mounts and the Minolta head features the standard 1/4″ thread, so I had to use a mounting plate for the panohead and the camera was used by itself, i.e. without the usual L-plate. That’s a pity because in this setup it’s not possible to mount the camera vertically. The spirit level in the panorama head is leveled by adjusting the tripod’s ball head and then it’s just a matter of making a picture and move the panohead to the next click stop, 15 times in total.

Off to Lightroom with the 15 pictures. No dice, Lightroom refused to merge the files into any sort of panorama. I had better luck in Photoshop, the resulting file is almost 6 GB! Here’s the 360º pano of my backyard.

And there’s my problem with pano’s: the excessive width/height ratio makes it almost impossible to show it appropriately on a monitor, you’d have to print it really big to see what’s going on. It is visible here that the pano head didn’t do a proper job of leveling the camera, the scene’s verticals are skewed. I noticed in use that there is some play in the head and I think that’s by construction, the pano head doesn’t look overly worn to me.

If I really wanted to do panorama’s, I’d go for a Really Right Stuff panorama head that is Arca-Swiss compatible and will probably perform better too. Costly business, not interested enough in panorama pictures to spend that kind of money.

Found this lens on an open-air fair in The Netherlands last Sunday. It looks like an enlarging lens to me with its M39 mount and the fact that the aperture numbers are readable with the mount upward like in an enlarger. It’s beautifully made and the aperture ring runs smoothly. I made some test shots with it on the Sony A7 and the Minolta Auto Bellows III and it performed quite well, allowing for large amounts of tilt and shift, which supports the theory that it’s an enlarging lens for large formats like 4×5″.

Update: Holger Bargen answered me on dpreview, pointing me to an eBay auction where a very similar set of lenses is advertised. It’s thus very likely that JML Optical is (one of) the suppliers. I have saved the pictures from the eBay auction, but can’t show them here because they are copyrighted.

The lenses

Minolta has used three different optical designs for the 28mm 1:3.5 lenses in the SR-mount line-up. The first one appeared in 1963 as an Auto-Rokkor model and was updated to a very similar MC-variant in 1966 to make use of the SR-T 101’s TTL-metering. It contained 7 elements in 7 groups. These lenses are readily recognized by their 67mm filter threads, distinghuishing them from all later models. I used the MC model for the comparison.

The second design appeared in 1968 and featured 7 elements in 7 groups like its predecessor but of course in a different configuration. Its size is markedly reduced, leading to a 55mm filter thread size. All lenses bear the MC W.Rokkor-SG designation. For the test I used the second incarnation of this optical design, the one with the hills-and-dales focussing ring.

Finally the third and last optical design came out in 1975 as an MC-mount lens and came with 5 elements in 5 groups, so again no cemented elements, and with a much smaller front lens diameter. Minolta dropped the SG lettering suffix denoting the number of lens elements and groups. This design lasted until the end of the SR-mount lens range and was produced in four different versions, one MC model and three MD models; the filter thread size of the first two models was 55mm, which was reduced to 49mm for the last two. I guess the most common version is the third one, the MD W.Rokkor variant with 49mm filter thread, which is why I used it for the test.

Test results

Full scene

My usual test scene, revealing sharpness across the frame.

Showing all three here, to my eyes the youngest lens clearly shows better colours and contrast, generally the most pleasing image. All photos were processed in Lightroom with the same settings for white balance, contrast, whites and blacks. Only the exposure was corrected a bit because the camera compensated noticeably for the vignetting at the largest apertures.

Centre at f/3.5

The oldest lens clearly has a veiling flare and doesn’t look really sharp. The A7R2’s 42 MP are merciless! #2 is already better, shows more contrast and less flare and is sharper. The most modern lens is the best one here, again note the more saturated and warmer colours.

Corner at f/3.5

This is the extreme upper-right corner of the A7R2’s frame. Not unexpectedly all are fairly bad here. Again the most modern lens performs best, it’s already fairly good except for the very extreme upper-right part in the crop, and contrast and colours are better.

Centre at f/8

At f/8 differences become less, even the oldest lens performs quite well here. Remember, 42 MP! The newest lens still has an edge on sharpness, contrast and colours.

Corner at f/8

And again the youngest lens shows the best performance. The older two don’t differ a whole lot, except for the sharp decrease in sharpness in the extreme corner for #2.

Epilogue

Choosing between the three lenses is easy: the youngest lens is the clear winner, better sharpness, colours and contrast, suitable for landscape pictures at f/8 or f/11. The older lenses have a flatter look and are less sharp, which opens opportunities for getting a vintage look.

Specification

The data pertain to the version shown above.

History

Due to its popularity there have been quite some model variations based on the same optical design of 7 elements in 7 groups. It appeared first in 1975 in the form of a MC W.Rokkor.

Minolta came with a budget version for the North-American market in the form of a Celtic variant.

In 1977 the lens was updated to an MD W.Rokkor version to service the XD7’s dual exposure automation.

Again a Celtic model appeared for the North-American market.

Soon enough the MD W.Rokkor got a cosmetic update and lost some weight to conform with the trend to more compact lenses dictated by the Olympus OM cameras. That is the version featuring in this post, repeated here for the sake of clarity and completeness.

Also this more compact lens was offered as a Celtic model.

I have seen several samples of a MD W.Rokkor variant with the looks of the Celtic lens but with the Rokkor designation up-front. Some people state that these are fakes; I don’t have enough information to confirm or deny that.

The last guise in which the optical design appeared was the plain MD version without Rokkor designation, introduced in 1981.

Later a successor with a simpler optical design appeared with the same 28mm 1:2.8 specification.

Image quality

All test shots are made with the Sony A7R2.

Here is the usual test scene.

To me this image quality wide-open is quite respectable, except for the extreme corners. Stopping down to f/4 though delivers more contrast and the veiling flare is gone.

Centre sharpness is excellent but the upper left corner is a bit disappointing. From what I see it’s coming from curvature of field that becomes more prominent on stopping down; the lower right corner crop is sharper while it is much closer.

Epilogue

The Minolta MD W.Rokkor 28mm 1:2.8 is an affordable introduction into the world of wide-angle legacy lenses, offering very decent image quality in a small and light package; its biggest weakness is curvature of field. I actually would prefer this lens for landscape use over the FE 3.5-5.6/28-70mm which isn’t all that great at 28mm in edges and corners. However, in my experience the Minolta 28mm is outperformed by the Sony FE 2/28, 4/24-70 and 4/16-35 lenses at their corresponding focal lengths, no contest.