Category Archives: Hardware

Stuff

The new Kindle – Part II

Hardware

Second try.

Kindle, iPad 1, MacBook Air.

I last wrote about the predecessor to this device a couple of years ago, having returned mine because the UI was about as bad as it gets. The keyboard was indistinguishable from a bilge pump. Both suck, and the page turn buttons on the long sides were way undersprung, so I was always turning pages unintentionally. It’s the only thing I recall returning – OK there was that awful 20mm Canon EF lens – in ages.

Well, the new one – mine is the bottom of the line $79 model – dispenses with 3G, being wi-fi only, and deletes the physical keyboard in exchange for a button-activated virtual on-screen version. An improvement from the execrable to the awful. This makes the device smaller and lighter, as it’s now down to 6 ounces, and the only time you really have to use the clunky keyboard is when first signing in to wi-fi. UK residents pay $138, 75% more, which will serve to remind them not to try and tax our tea ever again. Call it war reparations.

In the US, Amazon Prime members ($80 annually for 2 day UPS on almost everything, plus free movie rentals – recommended) can also get free access to all the Harry Potter books which is enough to make you want to cancel your Prime membership. Still, you don’t have to read this tripe and J K Rowling has no need of your filthy American money. They have enough worthless paper there already, not least in her interminable, glutinous prose-laden novels.

Such a deal.

This is actually a lending library, so I thought I would get off to a good start and borrowed Michael Lewis’s ‘The Big Short‘ because though I am tall I am also short, if you get my drift. Woo hoo, $9.99 saved already. (Lewis attended my alma mater Salomon Brothers. No, it’s not a Catholic school).

Why am I subjecting myself to this torture again? I want to spend some quality time with our son on the beach and as my earlier piece illustrates, the iPad is useless in direct sun, whereas the Kindle is useless everywhere else. Bright sun is a reasonable bet on California’s beaches, and the prospect of sitting there on me bum like a dope doing nothing would mean early insanity. Swimming? Excuse me, I would rather like to return home with at least two legs attached, rather than with one stashed in the belly of the local Jaws. We are talking Pacific Ocean, here, and those boys gotta eat. Now all I have to do is avoid the discarded hypodermic needles on the beach. So Kindle it is.

The device remains useless for viewing pictures. It automatically downloads software updates unprompted, and did so in my first 30 minutes of use. Nice, if accompanied by more flashing than your local member of Congress in the men’s room at the airport. Unlike your MacBook Pro, there is no telltale odor announcing the fact that you have just fricasséed your gonads, as the Kindle remains perfectly cool.

The weight is so low that it goes unnoticed, one hand operation is a non-event (page turn buttons are on both uprights) and the buttons need a far firmer push on this model, preventing accidental page turning. As before, it comes pre-registered to your customer name and account at Amazon – very nice. Best of all the gadget fits in the back pocket of your Levi 501 button-front jeans, the type Real Men wear, allowing you to do it justice when you next sit down in a forgetful moment.

Finally, not having email etc. means there is no distraction from the old guilt feeling when using an iPad that you have not checked email in the last five minutes or the markets in the last two seconds. That’s not all good, though. Germany loses at soccer, their Chancellor goes berserk and next thing you know the market is up 20% because she was seen making off to the Führer Bunker in the Chancellery with a loaded pistol and a cyanide capsule clenched in her teeth. Well, you will miss that joyous event in real-time if the Kindle is the only gadget you have with you at the time. I suppose there’s always the iPhone and a towel over your head for a quick trade or two while the waves roll in as you twiddle your toes in the sand and broken bourbon bottles.

The regular model is $109 and comes without home page ads; you can pay the extra $30 to remove these from the $79 version I bought, at any time. Smart pricing. These ads are referred to as ‘Special Offers’ which is rather like saying the IRS invites you to file an annual tax return. Your freedom of choice is identical. In practice the ads appear on the home page only and are unobtrusive. The first one I got was for diapers …. maybe condoms will be next? Battery life is claimed to be one month which is Kindlespeak for a week, I suppose. You recharge from any USB socket using the provided micro USB cable. Too bad it’s not mini USB – yet another thing to pack. Mine came 40% charged and was up to 100% in an hour using the MacBook Air as a power source, meaning it should recharge in about 3.5 seconds on a desktop machine. Switching wi-fi off is the best bet for battery life and trust me, only true masochists will keep it on for web surfing.

The Kindle continues to suffer from inept synchronization design. Amazon makes much of how a book read on one device, say an iPad, will be synced to the same location on any other, say a Kindle. They call it ‘Whispersync’ and I call it ‘Bollocks’. In practice if you choose to re-read a finished book, no matter which page you leave it at, all devices will think you are at the last page, and there is no ‘Reset’ button. Gee, how hard is that to do? The simplest fix I have found is described here and it works well. I reset my finished book on the MacBook Air, and then all other devices thought I was at the beginning, or whatever location I set it to on the MBA. No biggie, but really, Amazon!

And don’t ever dare hit a footnote reference because it’s back to Bollocksync and all your devices will once again be at the back of the book. Reminds me of my 10 years dealing with US Immigration. (OK Immigration and Naturalization, though there was nothing Natural about the methadone cases behind the counter). Any time you made an insignificant error on one of their interminable forms, it was back to the end of the line. About that time I recall entertaining serious thoughts about joining the Communist Party and lying on my application. I imagine they were running a pool to see who could accomplish this sadistic ‘back of the line’ trick most often. “Yo, Phil, I did sixteen today!” Once such victim in the line in front of me had a fully fledged nervous breakdown when this treatment was visited on her. Needless to add, she was French. I suspect the software designers at Kindle must be the same ones who work on Fuji cameras or INS forms. Neither ever uses what they create and sure as hell none of them are Americans.

There’s also a ‘Shop in Kindle Store’ option which is rather like asking an amputee to use chopsticks. Yup, that keyboard again. Buy your books using a Mac or iPad, then sync the device. Life is too short for the alternative. I challenge you to find the fiction work you want from the catalog of 504,779 titles and counting. This year’s US Budget will make it 504,780.

One special feature is the ability to rotate the screen orientation whereupon, in landscape mode, the device’s buttons take on all the utility value of a eunuch looking for a spot of fun in a harem. Don’t go there. (The landscape mode; the harem’s fine). However, in flipped portrait mode it works well, moving the button pad to the top. It’s not used when reading and the long, side-mounted page turn buttons fall more easily under the fingers. However owing to another bit of coding genius from the chaps at Amazon, the home page advertisements do not invert which is either wrong or inspired. Probably inspired, as the first ad I saw upside down was for AT&T proclaiming the might of its 4G cellular telephony. Upside down it looks more like 1/4G, which is about right.

Finally, there’s a web browser under the mysterious menu choice of ‘Experimental’. After personal discussions with Jeff Bezos, I can disclose here for the first time that Bezos took my comments to heart and that the choice is being renamed to “I just crapped my pants”. I know this as I just looked up this site using the Kindle after a few hours of virtual keyboard input, and it looked like s**t.

Use with cellular wi-fi? No problemo. My iPad has Verizon cellular (the working alternative to AT&T in the US; AT&T is like government – you pay and nothing happens); fire it up, switch on the personal hotspot on the iPad and login the Kindle after seventeen attempts using the virtual keyboard. Works perfectly. You even get that little symbol on the Kindle telling you that magic is happening and this is not your grandfather’s wifi. You know, the kind in every Dell computer.

Hemingway on the Kindle in the sun.
The ability to change font sizes is welcome.

More when I have got through some more Hemingway. But take it from me, this is very much a single purpose device, designed for reading in bright light only. No idea if it does Chinese. Fair bet it does. Anything else, fughedaboutit.

First field test results:

On the beach at Half Moon Bay

Battery life – update July 23, 2012:

I kept wifi off except for downloads, read seven full length books since I bought mine on June 28, and the battery was showing dead with wifi on and maybe 10% left with wifi off. So Amazon’s ‘four week battery life’ is more like three weeks as long as you leave wifi off, which is not so easy, as after switching it on for a download, you have to dig into the Settings menu to place the device in ‘Airplane Mode’ which is Amazonspeak for ‘wifi off’. A full recharge using the USB cable took 2 hours.

HP100 – keeping the i7 cool

Hackintosh

The optimum setting.

Computer builder FU Steve writes about overclocking, heat management and power supply selection:

* * * * *

Having had a few days to experiment with the Core i7-2600 Intel CPU in the HP100, replacing the excellent i5, I set about varying overclock settings to see where the sweet point lies for Thomas’s HP100 Hackintosh.

The willingness of these CPUs to be overclocked falls, like most populations, on a Bell curve. Most will go to some decent OC, some will not, some will be superstars.

From much of what I have read the stock 3.4GHz speed is easily increased to 4.4GHz (+29%) simply by changing one setting in BIOS, the frequency multiplier. Thereafter things get tricky.

Here are my findings with settings for a stable running CPU; Geekbench results are shown also:


Intel Core i7-2600K overclocking table.

As is clear, the law of diminishing returns kicks in rapidly beyond 4.4GHz with Thomas’s sample of the i7. Go from 4.4 to 4.5GHz, a barely noticeable 3% Geekbench increase, and the idle temperature rises 8F. Go to 4.6GHz, a further 3% Geekbench improvement, and the temperature jumps another 11F, or 19F warmer than at 4.4GHz. After that the temperature rise:speed trade-off worsens quickly. You have to start juicing the core voltages significantly, Bclck is maxed out at 1005, and heat skyrockets. This is not consonant with a long, hard-working life for the CPU which, at $280, is the costliest part in the computer case.

Vcore will be set by the BIOS to 1.36 volts if you leave that setting (in “Advanced Frequency Settings”) at ‘Auto’. However, for maximum stability, override the Auto setting and set it to 1.385 volts. The maximum permitted by Intel is 1.50 volts, so it’s not like you are over-stressing anything.

Optimal setting? 4.4gHz, Vcore 1.385 volts, all other variables at default, failsafe BIOS settings, aftermarket CPU cooler.

In high stress scenarios, such as ripping and compressing a movie file, you can add 70F to the core temperature, so at 4.4GHz the temperature rises to 169F. Intel publishes exhaustive technical data on its CPUs, and you can find the following table in its 2nd Gen Intel® Core™ Processor, LGA1155 Socket: Thermal Guide which makes for fascinating reading:

Thermal table for the 2nd generation Intel i5 and i7 Sandy Bridge CPUs.

The maximum power consumption of the desktop Core i7 is 95W, at which point Intel specifies a case temperature of 72.6C (162.7F). Case temperature (the outside of the CPU) is 22F lower than CPU core temperature (the inside), which is where the temperature sensors used by apps like Bresink’s Temperature Monitor (free download) reside. So Intel’s maximum of 162.7F becomes 184.7F as a safe limit temperature as reported by Temperature Monitor.

In the following 24 hour chart I show temperature for all the cores. The small spikes half way across are for Carbon Copy Cloner running overnight SSD and HDD back-up jobs at 2am and 3am. The peak here is 129F – no biggie. The massive spike at the end is where I used HandBrake to rip a 4gB DVD movie and compress the result to H264 format for playing on an AppleTV2. The process took all of 11 minutes, by the way, and as you can see from the data, all eight cores of the i7 were working hard. There were several LR4 and PS CS5 sessions in the 24 hour period illustrated, but they barely register any heat increase. This is a mighty testimony to the suitability of the HP100 for still picture processing.

24 hrs, concluding with a movie rip/compress.

Ambient temperature was 75F when ripping the movie. The Core temperature rises to 169F, 15.7F below the Intel recommended core maximum of 184.7F. That’s slim headroom, but very safe, owing to the variable speed fan fitted to the large Coolermaster 212 CPU cooler – all of $28 and the best money you can spend on your Hack. Ripping movies with the ineffectual stock Intel cooler is a bad idea. For that matter, ripping movies on any iMac or, worse, MacBook Pro or MacMini, where cooling is severely compromised in the interest of sexy looks on the sales floor, is a positively rotten idea. Hey, it’s your money but be warned. Other than the MacPro, Apple’s hardware is simply unsuited to prolonged movie processing.

Here then is how my Hackintosh HP100 is set with regard to CPU speed and cooling. I use an Antec Sonata III case with the stock 500 watt power supply and two 120mm case fans – these can be manually switched to H/M/L.

  • Rear case exhaust fan – L.
  • HDD case fan – M.
  • CPU 120mm fan – variable – attached to Coolermaster 212 CPU radiator. This fan spools up instantly when the CPU heats up.
  • Intel Core i7-2600K CPU – overclocked to 4.4GHz.
  • BIOS CPU warning set at 176F – this will flash the temperature display from Temperature Monitor in the menu bar.

If overclocking you should experiment with frequency multipliers. Assuming you are using an efficient aftermarket CPU cooler, start by taking the stock frequency multiplier setting of 34x to 38x for the i7, the maximum supported by Intel’s warranty. Then take it up in steps of 2x, doing a Geekbench and Cinebench GPU run each time. The former will report CPU speed, the latter measures GPU speed and is also a good stability test. Once you get a Kernel Panic (grey screen with warnings) in OS X you know you have reached the simple adjustment limit and need to revert to the last stable setting, and you are done.

A note on ambient temperature. My measurements disclose that a 2F rise in room temperature results in a 1F rise in idle CPU core temperature. The above movie rip/compress was at an ambient temperature of 75F. At 95F ambient (we get that a few days a year in the Bay Area, and have no air conditioning) that means the CPU will be 10F warmer, meaning a peak temperature of 179F when ripping/compressing a movie, still within the 184.7F safe limit.

Preliminary measurements on the new MacBook Pro with Retina Display disclose that this is an exceptionally hot running machine, probably owing to the over-spec’d display, something no one needs in a small 15″ screen. And it doesn’t even have a DVD drive. I dread to think what its temperature runs once you connect an external DVD burner for ripping movies.

Determining power supply needs:

It’s easy to overlook the need to correctly specify the power supply when building a high performance Hackintosh. There’s an excellent free calculator at Outervision Extreme where you simply input your components to determine the wattage needed. Not all watts are the same. A cheap PSU with light windings and chintzy transformer iron cores will overheat and compound heat management and stability issues. Good brands include Antec and Thermaltake. I have used both extensively with no issues and many builders to whom I have recommended these remain happy. Here’s my output after using that calculator:

The result suggests that HP100 is bumping up against its limits with its 500 watt PSU, and that a beefier power supply may be called for in the interests of cool running and stability. Mercifully, the upgrade involves a few screws and a couple of connectors, and less than $100. There is no beating stock PC parts for price and reliability.

* * * * *

Thank you, FU, for that timely update, coming on the heels of your sterling work (FU is English, after all!) on the recent HP100 CPU upgrade.

Microsoft Surface recycled

Hardware

More of the same.

Microsoft Surface was a visionary product from Redmond, a decade ago. A huge table-style touch screen allowed icons and elements to be moved around by touch, and the whole thing was just so tomorrow.

Of course it was so good that it had to die; Microsoft has long known how to kill a good thing. Though there are many brilliant, innovative thinkers at Microsoft, neither skill has been in abundant supply in the corner office, occupied by a doofus whose only claim to fame is that not even he has managed to screw up one of the world’s great annuities – Windows + Office. The only other product which has sold in volume in the Ballmer years is the Xbox360/Kinect, emphatically not a Microsoft invention, and one whose profitability is a rounding error, so brutal is price competition in the game console segment.

For a long while I have refrained from knocking Microsoft in this journal as it has just become sort of assumed that anything from Redmond would be a costly time sink, and why mock the afflicted? But Microsoft’s latest just begs for a thorough thrashing, which it is my pleasure to administer.

Doofus with his latest disaster.

I mean, goodness, must he really insist on demonstrating something he is clearly clueless about? Has no one at MSFT the courage to tell this emperor that he has no clothes? Imagine the management culture at a company which prohibits criticism.

You see, Surface is the name for MSFT’s new line of tablets. Yes, the same tablet which Doofus was knocking (in this case, knocking on) two years ago. That was just when MSFT’s previous tablet effort was scrapped. And there’s little reason to think that the new one will last any longer. Goodness, they couldn’t even make up a new name for the product.

Each iPad comes with bullet proof software, iOS, robustly interlinked Mail, Scheduling, Address Book, etc. through iCloud, and a vast ecosystem of hundreds of thousands of applications developed over many thousands of man-years.

Now look at MSFT’s offering. First they are offering the tablet with two completely different CPUs – ARM and Intel. Eh what? Lucky developers having to craft apps for that. Then they come with plain, touch or type keyboard covers. There are two sizes – 32 and 64gB. And while they are mum on cellular, you can bet they will have to offer at least two carriers in the US, in addition to wi-fi only. And of course they come with a stylus, presumably because the touch interface is so poor there is no alternative.

And this from a manufacturer whose experience in hardware is largely limited to making mice and has little skill in complex hardware supply chain management.

Oops, did I mention applications? Ummm, no, there are none as of the time of writing.

Worst of all, the Surface Mk.2 runs Windows. Two different versions, no less. That spells DOA to me. And by the time these things hit the stores, if they ever do, iPad4 will have been released, instantly obsoleting whatever claims to currency Surface2 may have made.

So thanks, Microsoft, for reinforcing my disgust in you and confirming my decade old decision never to touch one of your foul products again. Apple may make mediocre, overpriced PC and laptop hardware, but OS X and iOS are robust as they come, the iPad is insanely great, and the whole ecosystem linking the two hardware platforms is not something that the dysfunctional corner suite in Redmond is remotely capable of disciplining into a functional whole.

Kudos to Bill Gates, however. He got out at the top and is doing truly wonderful things with his fortune. I would prefer to see a fool like Ballmer running MSFT into the ground than have Gates return and leave behind his groundbreaking philanthropic work. The only thing which mystifies is that Gates allows this Boob to continue running the business.

A little more speed for the HP100

Hackintosh

A little tweak ….

The other day the Hackintosh HP100 got a nice performance boost when the boot+applications SSD was upgraded from SATA2 to SATA3. Fast disk I/O is essential for best Lightroom and Photoshop performance. Now it’s the CPU’s turn.

Geekbench is a test of CPU speed. It’s a simple and quick comparator of great use to photographers as apps like Photoshop and Lightroom are far more dependent on CPU speed than on the latest in GPUs. Little is to be gained, data suggest, from using a high-end gaming GPU.

Cinebench framing rates are a measure of GPU speed. My Hackintosh HP100 (Sandy Bridge Core i5, 16gB RAM) uses a three year old, low power draw, Nvidia 9800GTX+ GPU, yet returns a very high Cinebench framing rate.

One of the beauties of the Sandy Bridge and later Ivy Bridge CPUs is that overclocking is trivially simple, unless you go crazy. Clock speed is a near-linear indicator of effective speed for like CPUs. Double the clock speed and you should see an almost identical change in the Geebench score. In summary, the Sandy Bridge i5-2500K overclocks from 3.3gHz stock to 4.4gHz with one key entry in the BIOS. The i7-2600K goes from 3.4gHz to 4.5gHz for the same effort. As long as you dispense with the inept stock Intel fan and fit a Coolermaster 212 ($27) or similar, you will be thermally protected. Further, the BIOS has many failsafes to turn things off if heat rises too much.

I have been running my Sandy Bridge i5 at 4.0gHz since inception, or 21% over the 3.3gHz stock, but the new Geekbench Ivy Bridge data for the just released MacBook Pros spurred me to action. Here are those data:

Meanwhile, HP100, perking along at 4.0gHz, records the following in Geekbench 64:

Actual speed with several apps running. 4.0gHz is correct, GB states it incorrectly.

Hmmm. Not good enough, even if my environment measures ‘real world’ results with Mail, Finder, Safari, Firefox, etc. running. You can also bet that the above MacPro data are in an ideal setting with no other apps running. That’s how Apple does data.

So I hopped into the BIOS on the Gigabyte motherboard, changed the ‘Frequency Multiplier’ from 40x to 44x, meaning the clock speed is now 4.4gHz, and restarted. Two minutes later I had the following result:

Core i5 Sandy Bridge at 4.4gHz. 42% faster than stock.

That’s more like it. A 10% clock frequency increase realizes a 9.4% CPU speed gain, and equalling the fastest, latest and greatest from Apple, at no incremental cost to me.

Heat, that bugbear of all computers, remains unchanged.

Temperature graph at 4.4gHz.

The above graph reports the temperature of the four CPU cores from restart. The usual start-up spike quickly disappears to settle at 109F, indistinguishable from the reading at 4.0gHz. The CPU cooler is set in BIOS as a variable speed device, meaning it cranks up only when needed. It sounds just a little louder than at 4.0, meaning it’s working harder but just as effectively. On the other hand, when I was running this test, ambient temperature was a high 85F (we have no air conditioning as it rarely gets that warm in the SF Bay Area) so there’s little to worry about. Things can only get cooler on regular days. The spike toward the right results from starting Lightroom 4. Starting Photoshop CS5 does not make any discernible difference. I have had no stability issues so far.

The Cinebench tests for GPU speed are outstanding. Brown (#5) is for HP100 at 4.0gHz, Orange (#4) is at 4.4gHz (not 4.0 as shown) – 13.3% faster. The highest reading here (#1) is for a Xeon equipped machine with a high end gaming GPU – meaning $1,400 more for the CPU and $1,300 more for the GPU – for a 25% speed increase. Goodness, the all in cost of HP100 is less than one of those components! And PS and LR do a very poor job of multi-threading so a 12 thread CPU is money wasted. Those economics do not solve for me nor does any photographer need to spend that sort of money. #3 is for HP100 running at stock GPU speed but with all other apps closed – hardly realistic, but impressive if you are a marketer. Marketing, after all, is lying for a living.

GPU results from Cinebench.

And when Apple gets faster, you can bet on one thing. With a tweak or two, HP100 will be right there at very little or zero cost.

Looking forward:

If you accept that CPU speed increases are leveling off, and that the focus will increasingly be on lowering power consumption, then simply dropping an i7 in place of the i5 will yield a 25-30% speed increase, for a net upgrade cost of maybe $150 after reselling your i5. I doubt Intel will be able to increase its CPUs’ speeds by more than 5% annually henceforth.

GPUs are already so far ahead of anything photographers need that spending lavishly here makes little sense. Any conceivable pixel density is already supported.

In the case of both, Adobe’s software is a long way behind what the hardware can do. Poor use of multi-core, multi-threading technologies means that far greater gains are to be had from software design than from hardware upgrades. Lightroom, in particular, is showing massive code bloat, with no improvement in operating speed. LR4 is some ten times the size of LR3.

The next frontier is peripheral I/O, where Intel’s LightPeak (we keep hearing that Apple’s Thunderbolt is the latest invention from Cupertino, when in reality it is simply LightPeak on which Apple’s 1 year wasted exclusive has now expired). LightPeak promises disk read/writes ten time faster than USB2, maybe three times faster than USB3. Whether it succeeds like USB2 did, or fails like Firewire has, remains to be seen. Very slow adoption is not encouraging, and I suspect it’s simply not a mass-market selling point. External drives are hardly the norm in the average home. If it does succeed, you can bet cheap PCIe cards will become available and that photographers’ Hackintoshes will be adding these for a few dollars.

Update:

As I’m not about to be beaten by Apple’s poor hardware, I set about adding a little more fire to the pot by tweaking the i5’s frequency multiplier from 44x to 45x, for a CPU frequency of 4.5gHz, and increased the VCore voltage to 1.385volts for stability. Nothing else was changed. System ambient temperature remains at 113F (45C) and is stable. At stock VCore it kernel panics. Intel specifies the maximum safe VCore at 1.52 volts so it’s not like I’m really puashing it here. This is the result – faster than the Core i7 in the fastest MacBook Pro:

Intel Core i5 – 2500K at 4.5gHz CPU speed.
1.4% faster than the fastest MacBook Pro, with more to come.

Cinebench GPU data remain unchanged.

I have shutdown failsafes in the BIOS set at a CPU temperature of 176F (190F is the danger point for the i5 2500K) so everything remains conservatively specified.
These data suggest that a modestly overclocked i7 – 2600K Sandy Bridge should be good for 16,000 or more. But you do need proper cooling to do this sort of thing, not Intel’s stock cooler.

The Z68 chipset on the motherboard does not work happily with OS 10.7.4, and while there are workarounds, it’s not worth the effort. (It slows to a crawl). H67 and P67 chipsets have no issues with 10.7.4. So on the HP100 I’m sticking with 10.7.3 for now. Only P and Z motherboards support overclocking.

Another 25% in speed?

Sure. Get a Core i7-2600k. Look here.

100mm, f/1.4

Micro Four-Thirds, Nikon lenses

Nikkor MF lenses on the Panasonic MFT bodies.

This piece will finally join the heretofore parallel lines for the Nikon D700 and Panasonic G3 systems I use. Absent the one in the iPhone 4S and an old Panasonic Lumix LX-1, I have no other cameras.

Adapters and their limitations:

Adapters, most around $25, are available to use Nikon and Canon and a host of other manufacturers’ lenses on MFT bodies made by Panasonic and Olympus. But just because you can do that, does it make sense?

For the most part the answer is a resounding ‘No’.

You have no autofocus, auto-exposure is aperture-priority only, and Canon EF and Nikon ‘G’ lenses require specialized adapters to control the aperture. Otherwise you are restricted to full aperture only as those lenses lack a manual aperture ring. Except for Olympus MFT bodies which have anti-shake built into the body, a Panny user loses that feature also. Any VR/IS in a Canon or Nikon lens is lost. The sheer bulk of most full frame lenses destroys the compact concept of the MFT body’s design and the whole idea has a rather Rube Goldberg aspect to it. Cool to tinker, useless in practice.

Still, I plonked down $23 for one of these the other day and just received it. It adds some value in specialized applications and works with Nikon pre-Ai, Ai’d, Ai, Ai-S and AF-D (manual focus) lenses. If you want to adapt a G series AF-S lens as well as all older Nikkors, buy the costlier adpater with a mechanical aperture control ring. Read on.

Click the picture to go to Amazon US. I get no click-through payment.

Adapter quality:

I opted for the Rainbow Imaging version as user reviews suggested it has a better release catch for Nikon lenses than other cheap ones. Manufacturing quality is very high, the interior is semi-matte but that’s unlikely to have any effect on image quality as the reflectivity is low. Fit of both the Nikon end and the Panasonic end is excellent. Novoflex makes adapters for $300. Save your money. The cheap ones are fine. You can see the full range of Rainbow Imaging adapters by clicking here. There are 30 adapters for MFT alone, including such odd ducks as Alpa (a superb Swiss 35mm film SLR whose quality of engineering puts Leitz to shame), movie C-mount, Contax/Yashica, Retina Reflex (!), Exacta/Topcon, Zeiss Ikon Contax rangefinder (!!), and many others. Fotodiox makes an inexpensive adapter for Hasselblad lenses to MFT.

Checking the flange-to-flange dimensions with a micrometer I found a maximum-to-minimum variation of 0.0001″ (0.0025mm), right at the limit of accuracy of the measuring tool. That would be tough to beat at any price. The grinding of the front flange, which mates with the Nikon lens of choice, is to a very high standard. The body of the adapter is made of very thick alloy and not about to flex, regardless of the lens fitted. The serrations on the barrel provide a decent grip for installation and removal on the camera. A small set screw on the rear flange provides adjustment of tightness of fit on the camera. Springs permit adjustment of the tightness of the front mount. Both front and rear on mine were set just right on receipt, but it’s nice to know that adjustments can be made in the event of wear.

Best lenses:

So which lenses make sense? The MFT sensor is one quarter the size of a full frame one, meaning that you are using only the center of the image projected by a full frame lens. Thus a 50mm lens becomes a 100mm. However, the depth of field remains that of a 50mm lens. Depth of field is solely a function of focal length. A 50mm lens on a 4″ x 5″ plate camera will have the same DOF at any given aperture as a 50mm lens on medium format, full frame, APS-C, MFT, you name it.

That pretty much means wide angle lenses from full frame bodies are a waste of time. Even a super wide 17mm, with all its associated bulk, becomes a semi-wide 34mm on MFT. You are far better off using the kit zoom with all its automation, than using a gargantuan FF wide. It just gets worse the wider you go. A monster 14mm Nikon or Canon is a not so wide 28mm on MFT. Silly. If you want really wide, use something like Panny’s 7-14mm or Oly’s 9-18mm. I use the latter and it’s an outstanding optic.

Likewise, modest aperture standard or medium long lenses make little sense. The Panny kit zoom – 14-42 or 14-45 – meaning 28-90 equivalent on FF, has you covered. And if you want something really long, using a monster FF telephoto on MFT bodies makes little sense unless you need a very fast aperture. But then why bother with an MFT body when FF will deliver superior results with little aggregate change to weight and bulk? The superb Panny 45-200mm (=90-400mm) has decent apertures fully open and built-in anti-shake, making it perfectly useable at the long end hand-held. And it’s tiny compared to anything from a full frame body.

That leaves fast FF lenses and special purpose ones.

50mm f/1.4 Nikkor-S on my Panasonic G3 body.

The fast 50mm makes for a fine portrait lens and permits limited DOF effects, if you can handle manual focus.

Winston. One 60 watt bulb for lighting. Nikkor-S 50mm f/1.4 at full aperture, Panasonic G3, ISO 1600.

As you can see from the snap, DOF is extremely limited fully open and close-up.

In use on the Panny G3:

You switch the body to Custom->Use Without Lens (go figure; I saved this to C2-2 – the G3’s custom settings allow one on C1 but three on C2, the latter selectable using the LCD rather than the top dial) to enable control of the adapted lens and here’s where one of the great advantages of the electronic viewfinder in selected MFT bodies kicks in. With the camera set to aperture priority automation, as you stop the lens down the finder brightness remains unchanged. It’s as if you were using a standard auto-aperture MFT lens! The EVF adapts as the FF lens’s aperture changes, only the perceived depth of field changes. If only the D700 came with an EVF ….

So aperture automation is not an issue, though the finder will report the aperture as 0.0 regardless of how set. You have to check the lens to see which aperture you are using. With aperture-priority automation, the shutter speed is correctly displayed in the EVF.

As for focus, Panny has another trick up its sleeve. By depressing the control wheel into the body, with the G3 you get a 10x magnified center rectangle (the magnification is variable at will), picture-in-picture, which makes manual focus trivially simple and dead accurate. (Panny’s MFT bodies do not have a focus confirmation LED). Far easier than using MF on the FF D700! Press again or touch the shutter release and the EVF returns to normal display. (In the earlier G1 the whole finder image is magnified, but the functionality is near identical). Thus, with a 50mm lens you are getting the focus accuracy of a 500mm, and even at smaller apertures the magnified image snaps in and out of focus sharply, leaving little room for doubt.

Picture-in-picture 10x focus tool in use on the G3.

For my purposes there are just a few lenses in my extensive Nikkor MF collection which make sense to use on the G3. They include the 50mm f/1.4 and 85mm f/1.8 for their fast apertures and shallow DOF when fully open (one of the banes of MFT is too much DOF with just about any lens), the 100mm f/4 Micro-Nikkor for its close focusing ability, and the 300mm and 500mm Nikkors for extreme reach. The 300mm is sort of silly as it’s large, heavy and hard to hold at the best of times, but the 500mm (1000mm equivalent) is a real surprise. This mirror lens, with its slow f/8 fixed aperture. is an absolute pig to focus on the D700. The focus LED indicator is at the very limit of its capability (it starts checking out much below f/5.6) and the finder image is dark. With the G3, the finder image is bright as can be and focusing is a joy. No need for the 10x focus feature. The unmagnified image is easy to focus in any light. And the 500mm Reflex Nikkor, once you get the hang of it, is really a special lens – positively a midget for that focal length and sharp as can be when properly handled. Balance on the small G3 body is excellent.

500mm Reflex Nikkor on the G3.

Neighbor’s backyard test target. 500mm Reflex Nikkor, 1600 ISO, G3, 1/1000.

The above was snapped hand held through a dirty window, the ‘target’ is some 100 yards away.

So the FF->MFT adapter has its uses, even if they are somewhat limited. However, a mirror reflex on the G3 is a joy and a pleasant surprise. It’s almost as if the Reflex had to wait all these years for a body capable of doing it justice.

Using the adapter with the Nikon Micro-Nikkor 105mm f/4 makes for a powerful combination. At closest focus you get 1:1 reproduction, compared with 1:2 on an FF body. Despite the small maximum aperture, critical focsuing is very easy thanks to the EVF, and the outfit balances nicely in the hand.

An even better body for use with really long lenses would be the recently released Olympus OM-D MFT SLR, which has in-body image stabilization, though I do not know whether the IS in that camera works with adapted lenses. However, at $1,000, this overpriced body currently costs twice as much as the G3.

A note on CPUs, processing and EXIF data:

If you have installed CPUs in your Nikon MF lenses, as I have, these do not interfere with the adapter. EXIF data in LR or whatever you use for processing will be missing any lens information, as the camera has no way of knowing the focal length used. Thus if you want to apply a lens correction profile, it will have to be selected manually. As only the central part of the image is being used, the need for lens correction profiles is lower than with FF sensors.

The 16mp G3 sensor figures to the equivalent of 4mP on a four times large FF sensor for same-sized prints. That’s perfectly adequate for 18″ x 24″ prints, as the walls around me testify, provided your technique is up to it.