Several weeks ago, I posted a survey concerning the HP PhotoSmart slide and film scanner, asking for people to respond as to whether they experienced banding problems with the model they owned. Here is an example of the banding problem:-
 

Sample scan showing the banding

The section at left is about a 1/10" or about 2.5mm square on the actual slide, so it is quite magnified.

I received complete replies in regard to twenty-seven scanners.  In some cases, one person reported about more than one scanner, usually because they had their scanner replaced one or more times, at some point.

Validity of the Survey

Before going into the results, I should mention a bit about the validity of this survey.  First off, obviously a sample of 27 units is a very small percentage of the units sold.  Secondly, there are a lot of variables to be considered.  There is the subjective nature of how critically people see, what media they typically scan from (slides or negatives or prints), and whether they used samples that had enough density to see banding, since the banding problem mainly appears in shadow regions of positive media.  Further, there is a matter of the display used, (monitor contrast and brightness adjustment), resolution the scans were typically done at, magnification levels used to view the image, and so on.  Lastly, there is the matter of the viewing software used, and the firmware within the scanner, of which there are at least three versions in circulation.

However, some of these variables were mediated by reports from people who had gone through several different scanners in an attempt to find one that worked without banding problems.  In these cases, people reported that they purchased a unit which had banding, then exchanged it (sometimes several times) and finally received one that did not band.  In these cases, the operator was a constant, and in most instances, so was the computer system and configuration in use.  The other side of this was that some people used the same scanner on several different SCSI cards, cables or computer systems, and the banding problem remained consistent, regardless of the computer it was connected to.

Observable Trends

In spite of all of this, I was able to establish some trends.  The vast majority of people were using Pentium 266 MMX computer systems with 64 megs of memory, although CPU speed went from 486-66 to P II 450.  Computer processing speed and RAM memory (which went from 32 megs to 128 megs) did not appear to be a factor as to whether the scanner displayed a banding defect.  Also, the type of SCSI card (HP’s supplied ISA or PCI versions, or a secondary type of SCSI card, such as Adaptec) did not appear to be a factor in the banding defect either.  Most people used the supplied SCSI cable, although a few did not, and again, this seemed not to be important.

What did appear to be a factor is the manufacturing run.  HP appears to have two different plants producing the PhotoSmart scanner, one in the US and one in Singapore.  The majority of those reported to me, sixteen, were US made, while eleven were from Singapore.  Within the US group, there were seven different dates of manufacture.  The Singapore models represented five different runs.  A few models with the same manufacturing date had different firmware in them.  This might be due to units being retrofitted with different firmware after being returned to the manufacturer and then redistributed.

One thing that is fairly clear from the survey, is that the product varies greatly in terms of this banding defect.  As mentioned above, there were a number of people who went through two or three units before finding one which did not exhibit this banding problem.  They claim that other variables weren’t altered.  The same system was used, same slides or negs, same software, and same operator, etc.  Therefore, seemingly,  the only changes were the scanner unit and the firmware within it.

I should also mention that there were three or four people who reported the banding problem, but have since returned the scanner for a refund (due to the banding).  These people were not included in the survey results since they were unable to provide information about software or firmware version or manufacturing run.

Eleven people reported banding or grid-like or boxy defects in their scans, and about another five had their scanners replaced due to banding problems to get one which did it less so or did not at all, meaning about half the people within the survey had units at one point, that manifested the problem.  Most people noticed the defect in dark areas of slides, although a few found the problem with negs or prints also.

Three versions of firmware were reported, Ver: .29, .30 and .32.  There was no pattern that I could determine in terms of the firmware and the banding defect.  There were units with each of these firmware versions which did or did not band.  There was mention by a few people that the version .29 has a tendency to make bands in gradients like blue skies, yet others using firmware version .29 reported no banding.

Because the sample is small, it is difficult to make any conclusive statements, however, certain runs seems to show a lower percentage of problems.  US models produced May 1997 (US75) and September 1997 (US79) and Singapore models produced August 1997 (SG78) and November 1997 (SG7B) seems to be the best in the group.

What is conclusive is that a high percentage of these units appear to manifest this defect, but not all of them.  Certain runs seemed to be made either with design changes, different components or better quality control to not manifest the banding defect.  This tells us that HP is capable of making these scanners in a manner to minimize or remove this defect.

HP is aware of this defect and has acknowledge it to a number of people, although to my knowledge, not publicly.  They also have made exchanges for a number of people in an attempt to provide them with better units and have provided purchase cost refunds to people who cannot get satisfaction.   My recommendation is if you are displeased with your scanner, see about returning it for an exchange, and you might ask about the specific runs mentioned above as they appear to be less likely to have this defect.

Diminishing the banding

Some people have success in diminishing the banding defect by either lowering the shadow exposure slider in the HP software prior to final scan, or by using either VueSmart or PScan32 software which each use different scanning algorithms, different methods of processing the scan and allow for a higher bit depth to reduce the defects in the shadow areas. 

Had the number of responses been greater, it is possible more distinct patterns in the data would have appeared. With only 27 replies only very basic trends can be identified, and not fully accurately. I wish to thank all the people who participated in the HP PhotoSmart Scanner survey.

Soapbox Commentary

I want to comment on the issue of quality control in these type of products.  In the author’s opinion,  just because a product is a value-leader in a marketplace doesn’t excuse defective product or design.  No one was particularly happy with the banding defect, although a surprising percentage (about half of the people who had banding problems and still owned their scanner) indicated they were willing to put up with it for the pricing difference between the HP PhotoSmart Scanner and other scanners on the market which cost two to four or more times as much.  A few mentioned they either do not regularly shoot slides (where the defect is most commonly seen) or they will stop shooting slides and use negative film to prevent most of the problem.  Several people who still own the scanner, felt as I do; that HP hasn’t been fully forthright in this matter.  Their advertising and that of retailers who sell this device continues to state that slide media can be effectively scanned and that those scans can then be printed.  My experience is that of very obvious banding/gridding manifests in prints made from slides scanned from my HP PhotoSmart scanner, to the point that it eliminates a wide variety of slides I own from being used in this manner.  Since some models of this device work well, it would appear that HP is using the consumer as their quality control people.  If a person doesn’t use the unit for slides, or isn’t too critical, or is only using the scanner for web graphic resolution, they might be pleased with it.  If the consumer is unhappy, HP appears to be willing to exchange it several times, or refund the purchase price.

Getting hold of HP to discuss this problem can mean long waits on calls on “your dime”, since they no longer offer toll-free access to their technicians.  This is a growing trend in the computer industry which the writer believes needs to be protested by the buying public in loud enough terms that the manufacturers recognize we feel customer service goes beyond providing a web address.  On the other hand, when we are afforded toll-free access to companies, we need to be responsible enough not to abuse it.  It is not unreasonable for a company to expect their clients to read the manual before calling. 

HP’s tag line, in their 1997 annual financial report, (yes, I own HP stock) was “The Voice of the Customer”.  Hopefully, they are indeed listening, even if we do have to pay for the phone call.

Note: all brand names are registered to their respective owners.