Heidelberg Linoscan 1800 lenses for macro

[Beatsy]

The two lenses from a Heidelberg Linoscan 1800 arrived yesterday and I spent a lot of last night having a first tinker with them (till 4:30am). This is just a preliminary report giving specs and mounting details. A suite of image quality and comparison tests to follow.

Using the lenses as loupes, they look great. Super flat-field of view, absolutely no pincushion or barrel distortion apparent (uncannily flat in fact; never seen anything like it). A very wide imaging circle apparent from looking "round the edges". At first I thought I got a couple of duds though. Despite being able to clearly focus the image at a distance of around 60-80mm (with the eye at any distance behind the lens when used as loupes) I needed at least 200mm of extension tubes to be able to get a focused image on the sensor at a working distance measured in something less than feet. I used the extension pictured below (right hand panel) and added a further 100mm to get a mag difference for the FL and F-Stop calculations.


So it seems these are "infinity" lenses. At least, putting them on a tube lens brought the working distances down to a reasonable 80mm and 60mm respectively, so I assume that makes them infinity. I only have adapters for my 135mm tube lenses, and this gives mushy corners. I'm pretty sure the sweet spot will be a 200mm tube lens - but for the upcoming tests, I'll use 135mm tube and APS-C crop mode as that will be close to the FoV of 200mm and FF. Here's the 92/7.1 lens mounted on the rig. Both lenses slotted neatly inside a bit of T2 extension tube with PTFE to snug them in place.


And here's the first (full frame) stack taken with the 115/5.2 lens. I used 140 micron spacing, about 80 images, PMax with no retouching.


First impressions: the 115/5.2 (1.08x on 135mm tube) looks sharper than the 92/7.1 (1,44x on 135mm) and vignettes a little less. Some of the vignetting comes from the tube lens but I'm not sure how much is from the scanner lens being pushed down too low. To be investigated. The longer focal lengths are great for stacking as there's less perspective scaling so far less image is lost to edge-streakies. The lenses don't "see round OOF edges" as much either, so far fewer issues with overlapping features looking semi-transparent after stacking. That should be a bonus with hairy insects.

I have a couple of Raynox lenses somewhere. Isn't the DCR-150 around 200mm? I wasn't too pleased with their image quality when used with Mitties, but perhaps it will be better here. I need extra adapters for that though, and have to read up on spacing of the Raynox from the sensor, etc, but 200mm would take the lenses magnifications to 1.6x and 2.1x respectively. Bang in the range I'm looking for. Sweet

So off to testing next. In the intereim, I'd be interested to hear any comments on the specs so far. Assuming image quality checks out, would these be considered good? They look it to me.

[mawyatt]

Steve,

Interesting. BTW the Raynox 150 is 8 diopter (125mm), the 250 is 4.8 diopter (208.3mm).

Looking forward to what you find out.

Best,

[rjlittlefield]

So it seems these are "infinity" lenses. At least, putting them on a tube lens brought the working distances down to a reasonable 80mm and 60mm respectively, so I assume that makes them infinity.

The lenses are surely not infinity designs, but they may be good approximations to that.

A true infinity design is optimized for the case where light rays on one side of the lens are organized into bundles of parallel rays, one bundle per point on the subject. Think about an ordinary landscape lens, on the front side. Reverse that lens and it acts like an infinity objective.

Lenses that are designed to work at magnifications far away from 1X are close to infinity designs.

The Linoscan 1800 is a flatbed scanner, maximum document size 8.25 in x 11.7 in. IF the scanner uses a small sensor, the optical magnification inside the scanner must be much closer to 0X than it is to 1X. In that case, each lens, reversed, will be close to an infinity design.

So, look at the sensor. How big is it?

--Rik

[enricosavazzi]

So it seems these are "infinity" lenses. At least, putting them on a tube lens brought the working distances down to a reasonable 80mm and 60mm respectively, so I assume that makes them infinity.

The lenses are surely not infinity designs, but they may be good approximations to that.

A true infinity design is optimized for the case where light rays on one side of the lens are organized into bundles of parallel rays, one bundle per point on the subject. Think about an ordinary landscape lens, on the front side. Reverse that lens and it acts like an infinity objective.

Lenses that are designed to work at magnifications far away from 1X are close to infinity designs.

The Linoscan 1800 is a flatbed scanner, maximum document size 8.25 in x 11.7 in. IF the scanner uses a small sensor, the optical magnification inside the scanner must be much closer to 0X than it is to 1X. In that case, each lens, reversed, will be close to an infinity design.

So, look at the sensor. How big is it?

--Rik

The seller auctioned the sensor board separately. The auction picture shows the chip window. After de-skewing the picture in Photoshop, based on a standard leg pitch of 2.54 mm for the sensor IC package, I estimate 84 mm as the length of the active portion of the sensor chip (i.e. the length of the row of sensels, not of the chip or IC package).

[Beatsy]

I was able to fit the Raynox after all but it was rubbish at the edges and corners, so I reverted to a 135mm Vivitar/komine as a tube lens for now. This vignettes on full frame and the corners are mush, so I used APS-C crop mode for these shots. It's pretty much the same FoV as FF with a 200mm tube lens, just with fewer pixels (18mpix instead of 42mpix). I still think a 200mm tube lens would be about optimum here. Test shots here, comparative shots later.

Note: All shots are straight out of PMax with no adjustments or sharpening - just a tweak of levels prior to export.

Heidelberg 92mm f/7.1 on 135mm tube lens, 1.5x, working distance 55mm




Heidelberg 115mm f/5.2 on 135mm tube lens, 1.2x, working distance 80mm

[rjlittlefield]

The Linoscan 1800 is a flatbed scanner, maximum document size 8.25 in x 11.7 in. IF the scanner uses a small sensor, the optical magnification inside the scanner must be much closer to 0X than it is to 1X. In that case, each lens, reversed, will be close to an infinity design.

So, look at the sensor. How big is it?

...I estimate 84 mm as the length of the active portion of the sensor chip (i.e. the length of the row of sensels, not of the chip or IC package).

OK, so then assuming that the sensor runs side to side, we're talking about a lens that images 210 mm paper width down to 84 mm sensor width, magnification about 0.4X.

If I were looking to get the best performance from that lens, I would try it at either 0.4X front forward, or 2.5X reversed. Pairing it with another lens strikes me as inviting trouble from vignetting and aberrations.

--Rik

[harisA]

Sorry for the offtopic
Seeing a lot of threads lately about scanner lenses makes me wonder if every flatbed scanner has a lens inside.I have a chance to buy a microtek A3 1800XL scanner at a tempting price but I want to be sure that I will find a lens inside.

[rjlittlefield]

wonder if every flatbed scanner has a lens inside.

In general, no. Some scanners such as the CanoScan LiDE series use a wide "contact image sensor" that works with an integrated array of microlenses instead of one large lens. This allows the scanner to be quite small and thin. I would expect any scanner with similar form factor to be using the same approach.

Based on the sales literature at http://www.microtek.com.cn/products/loa ... _cover.pdf, the Microtek 1800XL has a more typical box-like design that suggests an ordinary lens. But I don't know anything beyond that.

--Rik

[ChrisR]

As I understand it, which may be wrong, more modern scanners stitch images taken with a smaller sensor and lens field of view. This has become the trend facilitated by processing advances - it makes the machine smaller and cheaper. Therefore there's a "middle aged" period of scanner design which uses wider FOV lenses which are therefore more useful for fixing to a camera.

[Macrero]

I tried quite a few scanner lenses. Some performed really well, others not so well and some were just not worthy. It seems that now there is some kind of scanner lenses rush caused by Robert's review on the Minolta Dimage 5400 lens. By the way, my predictions are being fulfilled right now there is a Minolta DSE 5400 on eBay listed for spares or repair, 8 days remaining, 27 bids, 201 EUR + shipping current highest bid. Hilarious

I mean, all that glitters is not gold. The Minolta lens may be fine lens, but don't expect all the scanner lenses to perform great cause you may experience a number of disappointments.

[RobertOToole]

...By the way, my predictions are being fulfilled right now there is a Minolta DSE 5400 on eBay listed for spares or repair, 8 days remaining, 27 bids, 201 EUR + shipping current highest bid. Hilarious

The Nikon scanner ED lenses are even worse. I can remember when there were the 14 element lenses for sale for $100. I bought 2. And the 7 element? I paid $25 each!

I mean, all that glitters is not gold. The Minolta lens may be fine lens, but don't expect all the scanner lenses to perform great cause you may experience a number of disappointments.

Thats good advice. It really pays to do a little research, only a small percentage of scanners the nominal resolution, or even 50% of the manufacturers specs.

This is a good source for test samples:

http://www.filmscanner.info/Filmscanner ... ichte.html

For example Plustek claims 7200 dpi for the 7200 scanner but really only resolves a something like a 2900 dpi target.

Best,

Robert