What is a "Drum " scanner

[frank_a._bridges]

How is it different from a conventional dedicated scanner such as a

minolta dual III.

[marshall]

Though there are important other differences in practice, the key one is that 'conventional' scanners use CCD's and drum scanners use photo-multiplier tubes. PMT's are far more sensitive and less prone to noise. They're also far more expensive and somewhat more difficult to maintain. Typically, drum scanners also allow wet-mounting.

[walterh]

I would have called a drum scanner "conventional" but perhaps by now they really are "pre-conventional" :-)

 

 

The major optical advantage is that images are mounted on a drum , a rotating (while scanning - not while mounting) cylinder. images are scanned by a single point digiting sensor. the drum is rotated and the sensor measures the image points along the periphery of the cylinder. while the drum rotates the scanner head moves parallel to the axis of the cylinder , thus scanning the image along a helix. this way the influence of neighboring areas is minimized and there is plenty of space to accomodate a high sensitivity, high accuracy and low noise detector. these drum scanners reach about the technical data claimed by the advertisements of recent "conventional" film scanners :-)- in other words: better but more expensive to buy, maintain and operate.

 

one of the disadvantages is obviously also the single point measurement, since scanning takes longer times for scanning than with parallel array sensors. also the mechanical construction is expensive since high precision is needed.

[bruce watson]

Walter's description is about right. I'm going to add a bit:

 

A drum scanner is called a drum scanner because of the drum. The drum is acrylic, machined and polished to be optically clear, and a nearly perfect cylinder. The film is usually mounted on the cylinder emulsion down, so that the emulsion is in a fixed, known location (the face of the drum). In operation, the drum spins in the range of 600 rpm up to around 1500 rpm depending on the scanner.

 

The optical system shines a light though the film. The optical system picks up the transmitted light on the other side and directs it to the photo-multiplier tubes (PMTs) via mirrors and lenses. There are three or four PMTs, one each for Red, Green, Blue, and sometimes IR. The optical system splits the light into the channels prior to the PMTs.

 

The point is, there are only three or four PMTs, and they have a fixed location in the scanner. What moves is the optical system that transmits light through the film/drum and part of the mirror system that directs this light to the PMTs.

 

The film is read one "spot" or pixel at a time. As the film flies by, a "line" of pixels is read. After the film has rotated beyond reading range, a micro-stepper motor moves the optical system one "line" deeper into the film. When the film rotates back into view, the next "line" is read. And so on...

 

With a CCD flat bed scanner (a "dedicated film scanner" like your minolta can be thought of as a flat bed scanner on it's side - the principle of operation is the same), the film is placed on or just above a flat piece of glass. A CCD array (the number of CCD elements in the array ultimately determines the optical resolution the scanner can produce) traverses the glass, and reads the film one line at a time (as opposed to one pixel at a time). The entire "line" is processed in parallel. Where the drum scanner has one PMT for each channel, the flat bed scanner typically has three CCDs for each pixel in the "line." In other words, thousands of CCDs in the scanner array.

 

CCDs are cheap and easy to make, and tiny. PMTs are expensive and bulky. PMTs are capable of great sharpness and dynamic range. Flat bed scanners are often seen as optically "soft" and "noisy" in comparison.

 

Finally, there is the topic of fluid mounting. You can mount film on a drum using a specially designed fluid (typically a derivative of naphtha). The fluid goes over and under the film. The film and fluid are covered by an optically clear acrylic sheet. The sheet is pulled tight against the drum and taped on all four sides. The fluid fills in the spaces and irregularities in the drum and film (scratches) and eliminates Newton's rings. The effect on the final image file should not be discounted; it makes a large and very positive visible improvement.

 

I've heard that you can get fluid mount equipment for some of the film scanners. You'll have to check around and see for the particular scanner you have in mind.

 

You can also fluid mount onto a flat bed's glass, but you can't pull the negative tight to the glass as you can with the curved surface of the drum.

 

That's how it works, more or less.

 

The physical differences are that a flat bed scanner (or dedicated film scanner) can be small and light and sit on your desktop. It's fast (comparatively), quiet in operation, and energy efficient. A dedicated film scanner is limited to 35mm or 120/220 sizes. A large flat bed scanner can handle a wide range of sizes, but you probably won't want it on your desktop if it's good for 11x17 inch (tabloid?) originals.

 

A drum scanner, even a "desktop" scanner, typically takes up a space of around .5 x 1.0 meters (2x4 feet) and weights between 70-100 Kg (150-220 lbs). Drum scanners are slow, noisy, and generate lots of waste heat (900W is typical). Drum scanners can handle a wide range of film, from Minox up to the size of the drum. My drum scanner can handle 11x14 inch film, for example.

 

IF you want the ultimate quality from your scan, it has to be a drum scan. Whether it's worth it of not for your particular purposes, only you can decide. YMMV.

[the_macman]

They're also archaic. Current high-end flatbeds beat the crap out of any scanner... be it slide or drum. Creo for instance makes some amazing machines which can also work under isorefractive oil.

 

http://creo.com/global/products/scanning_systems/color_scanners/default

[andrew_rodney1]

The Imacon is a drum scanner (the film is wrapped around a drum) but uses a CCD!

 

The ScanMate 3000 was a drum scanner that didn't use a PMT.

[sunando_sen]

One correction: the scanner head does not move at all. The drum moves from side to side (or up-down, in case of vertical drum scanners).

 

When people say, "drum scanner" they usually mean a PMT drum scanner. The Scanmate 3000 most certianly used a PMT.

 

Another point:

The Macman , jul 29, 2004; 04:13 p.m. wrote:

They're also archaic. Current high-end flatbeds beat the crap out of any scanner... be it slide or drum. Creo for instance makes some amazing machines which can also work under isorefractive oil.

 

This is highly debatable, to say the least. There are some very good CCD scanners out there, particularly from Creo. But nothing so far has been able to touch a good drum scanner, overall.

[scott_eaton]

While I have a lot of respect for CCD based machines like the Imacon, PMT based drums have proven time and time again they can handle density extremes better than CCD.

[roger krueger]

Just to be clear, the "high-end flatbeds" Macman refers to are $20,000-$30,000 items, a

far, far cry from even the nicest $400 Epson flatbed.

 

While the specs certainly look good, I'm especially dubious about the D-Max numbers. I've

compared scans from an earlier generation of Imacon with decent D-Max specs to scans

of the same (rather thick) neg from my drum scanner (a Screen 1045) and the Imacon had

a great deal more highlight noise. Maybe the Creo's multisampling and cooled CCD

overcome this and it really does have a clean 4.1 or 4.3 D-Max, but there's been so much

misleading BS relating to D-Max on CCD scanners I'll believe it when I see it.

 

Sunando: On my Screen 1045 the scan head is what moves left/right. The drum spins but is otherwise stationary.

[sunando_sen]

Roger Krueger: "On my Screen 1045 the scan head is what moves left/right. The drum spins but is otherwise stationary."

 

That's interesting. On the Howtek, the drum definitely moves to and fro. I think I saw Hell scanners work like that too. So I assumed this to be the case in all drum scanners.

 

-------------------------------------

 

Roger Krueger: "Just to be clear, the "high-end flatbeds" Macman refers to are $20,000-$30,000 items, a far, far cry from even the nicest $400 Epson flatbed..."

 

-------------------------------------

 

Interestingly, the highest quality CCD scanners end up costing as much, if not more, than drum scanners. And many of these scanners' vendors recommend oil mounting for best results. The reason I mention this is that some people consider oil mounting an inconvenience associated with drum scanning.

[walterh]

Hogarth, thanks fo the addition :-)

 

I tried to hide that i do not really know about current models of drum scanners so I just gave a description with what i know from about 20 years ago when I built my own drum scanner. Based on what i had seen from commercial units at the time I designed a scanner for scientific BW film to digitize electron micrographs to digitally analyze such images. Interestingly at the time it was no real problem to obtain a small focused spot to obtain maximal resolution. The major problems were the light scattering from the silver grains. This interfered with the measurement of the "density" of the measured spot. Correcting the scattering by adding an additional sensor for the scattering helped for some images but small differences in film emulsion batch and development conditions resulted in "complications". This seems still true for scanning todays BW films based on silver grains, for both drum scanners and scanners using array detectors, like the popular film scanners such as nikons line. Statements that scanners with more than one illuminated spot will do as well as those with just one spot make me smile. Without "seeing" I do not believe that it is possible to eliminate "crosstalk" of neighboring areas of the image , at least in cases of images with extreme contrast , be it color contrast or density.

[david arnold eugene, oreg]

Phew! More than a dozen replies, and one of them more than a dozen paragraphs long. And all this time I thought a drum scanner was simply one that did a bang-up job.

[mark_reynolds1]

In this article a manufacturer - ICG - discusses comparisons between flatbed and drum scanners. Although it's on a drum scanner producer's website it still seems to be a reasonable view. Maybe this can shed more light... <br>

 

<a href="http://www.icg.ltd.uk/icg/whydrum.htm">http://www.icg.ltd.uk/icg/whydrum.htm</a>

[rick_tuttle]

I just found photo.net and this forum. Very interesting. I've had a Screen 1045 since

1996 and also have run a high end flatbed Heidelberg Nexscan F4100 for the past 3 years

and the Nexcan can hold it's own up to about 2000dpi and then it becomes very evident

it's a CCD. The 1045's original price was $90K and the Nexscan $45k which was new in

2001. Roger if you have any questions about the Screen let me know.

[raphael_bustin]

Regarding the relative merits of drum vs. high end CCD scanners, you might want to check out my page of scan samples at

 

<http://www.terrapinphoto.com/jmdavis>

 

After hearing about the benefits of drum scanners for these past several years, I've finally acquired one, a very old but still functional ScanMate 5000. Preliminary conclusion: when properly focused, there's a very modest improvement in resolution over the Nikon LS-8000 (a nice CCD scanner for MF film) but not really worth the extra effort in and of itself. When used in 16-bit mode, there is a noticeable improvement in dynamic range, and for some applications this may be significant.

 

To really get the most from either scanner, you'd probably want to wet mount your originals. I've seen several instances of inconsistent focus -- over the same negative -- with dry mounting on the ScanMate. And of course the Nikon LS-8000 is rather famous for its limited DOF.

 

I compared a 4x5 chrome scanned on a Microtek 2500, versus the same chrome scanned on the ScanMate, but dry mounted. Result: both scans had regions of excellent focus and regions of poor focus. Conclusion: can't really compare without wet mounting.

 

One last point... I really miss having ICE on the ScanMate, but I understand that proper wet mounting can minimize problems of dust and scratches. IMO, digital ICE is a must-have.

[upscan]

There is a dramatic improvement in tonal range and brilliance of any slide when wet mounted. The optics of wet mounting are completely different to those in dry mounting. Dry mounting is so common because manufacturers do not wish to make their scaners appear more difficult to use and because it easy. Wwith all the talk about ICE it is amazing that few people know that wet mounting can be tremendously effective in removing scratches in colour as well as where ICE can't, i.e. B&W silver negs. Wet mounting ensures a fixed plane of focus that is even throughout, eliminates Newton rings etc. etc. For more on wet mounting look at the yahoo group http://groups.yahoo.com/group/SCANMAX/ OR take a look at the file http://www3.sympatico.ca/gluemax/ScanMax/scanmaxbrochure6.pdf

[jake_purches]

The Screen 1045 has a maximum D-max of 3.90. This is a genuine figure and not made

up like so many manufacturers claim.

Its not physically possible to get the same sensitivity to PMTs with CCDs unless the ccd

array is cryogenically cooled with liquid nitrogen.

PMTs are electron tubes and run hot. I have a few from my old scrapped drum scanner -

they look like radio valves. Archaic PMTS may be but they are very much in use in the

scientific world.

[jake_purches]

I am suprised that only IMG as far as I know are still making drum scanners. CCD flat bed

scanners are market led - convenience, compromise and cost cutting.

Hence the rather lack lustre quality of repro these days (and digital camera output too).

What is interesting in my view is that the remaining drum scanners are being used for the

photographic market - not the colour separation trade which has now died out. Drum

scanners are now being used for RGB scanning which was unusual 10 years ago

- it was always CMYK. Fortunately the 1045 scanner does an excellent RGB scan.