Where EEs Navigate A Changing World.
Globe

Astra 1220P Color Scanner Teardown

Creative Commons License Michael Stanley & EE HomePage.com
This report is licensed under a Creative Commons Attribution 3.0 Unported License.

Figure 1: Intact Scanner

Introduction

When I searched for the term "scanner teardown" while writing this report, Google returned only two relevent hits. The first report costs almost $2000 to download, and the second had precious little details. In this report, we're going to tear my trusty UMAX scanner down to the bones.
Figure 2: Under the Cover

This model is more than a few years old, so we can expect that the electronics are further consolidated in newer models. It contains three PCB boards. Two are a mix of surface mount and through hole devices. The third is simple, with single sided, thru-hole only construction. My biggest interest in the design is from a mechatronics perspective.

The 1220P scanner is a classic desktop scanner (shown in Figure 1), in which the item to be scanned in placed flat on a glass panel (Figure 2) under which a scanning mechanism (shown in Figure 3) operates. The scanner communicates with the PC via a parallel port interface (similar models supported USB). Power is supplied from an external wall-wart, which is not shown. The technical specs for this scanner (available on the UMAX support site) list a hardware resolution of 600 dpi x 1200 dpi. I also found the service manual for this scanner at http://www.eserviceinfo.com.

Figure 3: After Removing the Top Housing

Before proceeding with the teardown, I should note that more detailed versions of most of the photos in this teardown can be viewed simply by clicking your mouse on the photo in question. In several cases, the detailed photos include labels of critical components in support of the text.

Teardown

Major Components

The detailed version of Figure 3 is labeled to show major components of the scanner: A key point to be made at this juncture is that pixels in the X-dimension are electronically scanned, one row at a time, by the CCD; scanning in the Y-dimension is done mechanically via the motor assembly and timing belt.
Figure 4: Top View of the CCD Chassis

The CCD Chassis

Let's start by looking at the CCD chassis in more detail. The detailed version of Figure 4 shows a closeup of just the chassis. Notice that one side of the chassis rides a guide rod. The other side has a simple roller assembly. The holes through which the guide rod pass are fitted with metal brushings - presumably to improve wear and decrease friction. The lamp assembly is labeled. Most of the optics are hidden from this view, although you can see some of the mirror clips and the lens mount. The CCD sensor itself is hidden from view, and is mounted on the chassis side of the CCD PCB.

Figure 5: Rear of the CCD Chassis after removal of the CCD PCB

Figure 5 shows the rear of the chassis after removing the CCD PCB. You can now see the lens mounted inside the chassis.

Figure 6: Exposed view of the CCD PCB

Figure 6 shows the front side of the CCD PCB once it was removed from the chassis. There doesn't appear to be a lot of intelligence on this board: the two small ICs on this PCB appear to be an HC04A hex inverter and a HCF4052B analog multiplexor. I believe the later is used to multiplex the CCD outputs into a single channel prior to passing them back to the main board). In the upper right, you can see a photo sensor which is probably used as an indicator that the chassis is at one extreme of its range. Any item passing between the left and right extensions interrupt a light beam. A protrusion on the bottom of the housing should do the job. Unfortunately, I couldn't confirm this, as the housing had already been discarded before I noticed the photo sensor.

There is also a pressure-fit connector for the ribbon connector at the bottom of the PCB, which passes key signals back to the main board for further processing.

Figure 7: The CCD side of the PCB normally never sees the light of day

Figure 7 shows the flip side of the same PCB. The sole visible component is the CCD, which the service manual identifies as a 2556D CCD. This sensor has three rows (one each for blue, green and red light) of 5340 charge-coupled devices. All three rows will be exposed simultaneously to the same source, and then the analog results scanned out of the CCD. The analog signals are converted to digital values by the CCD Sensor Processor elsewhere in the system.

Assuming 600 DPI, the sensor can scan an 8.9 inch wide sheet. The device shown has 32 pins, whereas the only datasheet we could find was for a 22 pin device. But micro-photographs of the sensor (Figures 8 and 9) only show 16 signals bonded to the device, The rest will be no-connects.

Figure 8: Signals are bonded to the 2556D die on each end.
Figure 9: The other end

Figure 10: Simplified Optical Path
Now would seem to be a good time to talk more about the optical path. Key components here are the lamp, which illuminates the document to be scanned, the lens and the CCD sensor. These are shown in rough form in Figure 10. The cross section of the CCD chassis is roughly 2X3 inches, and it should be apparent that the optical path cannot fit into the confined area of the chassis without modification. The solution is to "fold" the optical path into a smaller volume using a series of mirrors - in this case, four. If we take the optical path in Figure 10, turn it on its side, and "fold" it a number of times, we end up with Figure 11. This is the approach taken by the designers of the 1220P scanner. Figures 10 and 11 clearly demonstate that the optical path is a key limiter in terms of reducing the volume of the scanner.
Figure 11: Folded Optical Path

Figure 12 shows the four mirrors extracted from the interior of the chassis.

Figure 12: Mirrors used to fold the optical path into a smaller volume

Pulling back a bit, Figure 13 shows the chassis again, but with the Housing for the lamp circuit peeled back (it was attached only with adhesive) to show the lamp circuit PCB. This same PCB is also visible in the rear view of the chassis shown in Figure 5. Both lamp and support circuit are easily removed from the chassis, are are shown in Figure 14.

Figure 13: The lamp support circuitry is revealed here
Figure 14: Lamp, lamp shade and support circuitry

Drive Train

Figure 15: Top view of the motor assembly
Figure 16: Bottom view of the motor assembly

Figure 17: Gear Chain
Figures 15 and 16 show the motor assembly, which includes a Mitsumi M35SP-7 stepper motor and several gears. Figure 17 and the accompanying table have all the information necessary to calculate scanner step resolution in the Y-direction:

12.5 teeth/inch X 90/25 X 80/18 X 48 steps/16 teeth = 600 steps/inch. Given that the scanner specs call for 1200 DPI resolution, it seems reasonable to expect that a half-stepping or microstepping algorithm must be used to drive the stepper with smaller step sizes. The gear assembly does have some backlash component, but I assume this is not applicable when the assembly is driving consistently in one direction.

Element Quantity Units Comments
Mitsumi M35SP-7 stepper motor 7.5 Step size in degrees Equates to 48 steps per revolution
Gear A 16 teeth/revolution attached directly to stepper shaft
Gear B 80 teeth/revolution Driven by "A". Locked to "C".
Gear C 18 teeth/revolution Locked to "B". Drives "D".
Gear D 90 teeth/revolution Driven by "C". Locked to "E".
Gear E 25 teeth/revolution Locked to "D". Drives timing belt.
Timing belt 12.5 teeth/inch Driven by "E". Attached directly to CCD chassis.

Figure 18: Motor-Timing Belt
Figure 19: Return path for timing belt

The figures above show both ends of the timing belt in situ. Figure 19 shows that timing belt slack is taken up via a spring mechanism located near the front of the scanner.

EMI Suppression

An item often overlooked in teardown reports is shown in Figure 16 above: the ferrite ring core. In this case, the harness for the stepper has a single winding around the core. Passing all conductors for the motor through the core helps to ensure that the ferrite material does not saturate (because DC currents in opposite directions effectively cancel one another). High frequency components are blocked.
Figure 20:

Figure 20 is a closeup of the specialized ferrite core (also visible in Figure 3) used to suppress high frequencies on the flat cable connecting the main and CCD PCBs. I've included a couple of references on the use of ferrite materials for EMI suppression in the "To Learn More" section.

Main Board

Figure 21: The main PCB
The main PCB has a mixture of thru-hole and surface mount devices, and is shown in Figure 21. By today's standards, there is nothing nothing terribly exciting here. The Umax Service Manual lists the primary components:
  1. LM9811 CCD Sensor Processor
  2. ULN2003 Darlington Array
  3. Astra 1220P ASIC
  4. Shuttle Technologies (now part of SCM Microsystems) EP1284-01
  5. 74HCT125 quad bus buffer
  6. M5M418128AJ 256K X 8 DRAM
  7. LM7805 Regulator
  8. 12 MHz crystal
  9. 24 MHz crystal
  10. UTA (Umax Transparency Adapter) Connector - a connector to an optional adapter for scanning transparencies.
  11. Parallel port connections for the PC and printer.
The detailed version of Figure 21 (click on the figure here) has each of the above labeled. A block diagram of the main board, as well as a graphic of the silk screen is included in the service manual mentioned earlier.

Summary

The key to a successful teardown is not to just rip something apart, but to try to determine why the original designers built it the way they did. What were their primary design goals? What tradeoffs did they make? What were the technology limitations? How can I apply these ideas to projects I'm working on? I always learn something new whenever I do a teardown, and this was no exception. I hope you found it useful as well.

I took a LOT of photos during the teardown, and only some of them made it into the report above. The remainder can be viewed in the Addendum to this report.

To Learn More

You can improve your learning experience by investigating the following resources:


Re-Use

Creative Commons License Michael Stanley & EE HomePage.com, December 2007

The original publication of this report resides at http://www.eehomepage.com/report.php?report=20071220. You can download a zipped archive of this report here. The zip file is around 40MB.

This report is licensed under a Creative Commons Attribution 3.0 Unported License.

We ask that you include a credit similar to the following:

You can use the following HTML for the credit:

Provided courtesy of <a href="http://www.eehomepage.com">EE HomePage.com</a>