X-Ray imaging - The Gendex GXS-700 X-Ray dental imaging array.

[Fraser]

As those who have been following my X-Ray images thread will know, I have purchased a Gendex GXS-700 dental X-Ray imaging sensor that is in need of repair. Well it arrived from the USA today.

These sensors are used in place of dental X--Ray film and are composed of an electronics package that converts X-Ray energy into images that are passed to the host PC via USB. The Gendex GXS-700 is actually a re-badged DEXIS Platinum sensor. The design is advanced in that all of the electronics are located in the head and USB plug, with no need for a large module that normally sits between the two. It is a highly integrated electronics package.

So how does it work ?

In the head you will find a scintillation plate that produces light when illuminated by X-Ray energy. Behind this there is a special fibre optic micro-channel plate to control light dispersion. The micro-channel plate passes the collected light to the CMOS imaging sensor. This is a physically large CMOS imaging sensor that is similar to that found in conventional digital cameras such as a Digital SLR. The CMOS imaging sensor converts the light into a voltage that may be read out by the support electronics and passed to the analogue to digital converters. The digital signal is then passed to processing electronics, and then to the USB interface.

The software on the Host PC processes the image information and presents it to the user for viewing and manipulation.

The sensor head detects when X-Ray energy is present and triggers its capture electronics to start the image exposure. The exposure is automatically stopped when the required ADU target has been achieved. The user does not need to setup the exposure, just the X-Ray Generator output duration, as with a film exposure. Very simple.

These sensors have a hard life and failure is common. They are in daily use and rely on a fragile umbilical cable. The actual head electronics are quite robust and the case is hermetically sealed against fluid ingress. The regular flexing of the cable tends to lead to failure, even if a cable protector sleeve is used. The companies that provide these imaging sensors usually offer service plans that repair or replace the sensors whenever they fail, which is often. As a result the support contracts are expensive. For Gendex it is $4000 per year ! Some owners elect to scrap older sensors as a result. Hence my purchase of a faulty one.

The GX-700 sensor that I have is the larger Size 2 and that normally retails at around $7000  Used ones sell for anywhere between $1200 (ebay) to $4000 (OEM refurbished).

These sensors are NOT the easiest items to repair. The head electronics are a total nightmare to work on and opening the head case tends to ruin the hermetic seal making them unsuitable for dental use. For hobby applications a 'compromised' head is still usable however.

The sensor is capable of producing high resolution images, albeit small ones, of PCB's and components. This is the use that I have in mind for my unit, if I can repair it (which is by no means certain)

Upon receiving my GXS-700, I was pleased to see physical damage to the USB connector. Why was I pleased? Well physical damage to such a connector is usually repairable and far less serious than a failure in the head electronics. The USB plug has been crushed and bent. This commonly occurs when a draw is closed on the plug or a wheeled chair driven over it  Yes people do ruin $7000 sensors through carelessness 

I shall X-Ray my sensor head, cable and USB plug to see if anything obvious is visible before I dismantle the required parts.

I will update this thread as I progress with the investigation. I am very busy at the moment, but watch this space 

Finally, a warning to potential purchasers of these types of imaging sensors...... they can suffer all manner of failures, plugs, cables, electronics and also firmware corruption. It is not unknown for sensors to be scrapped as they are beyond economic repair, which considering their value is surprising. Firmware corruption is not uncommon and requires the OEM's bespoke configuration and programming software. Not something they share with anyone except a service agent. Buyer beware. It is very easy to waste your money on these units.

On the software front, Gendex kindly provide some driver software on their web site for free download. Users will need the essential unique Calibration and defect file that accompanies their sensor though. A sensor without a calibration file is effectively useless as the software will not run without such being present (serial number locked).

Fraser

[Mashpriborintorg]

I have already played with similar sensors at work, Kodak brand, but exact same pricetag. I guess they are stupidly overpriced....

[Melt-O-Tronic]

Another awesome follow-worthy thread!         

[mikeselectricstuff]

 A long time ago I picked up an old dental sensor - surprised to see nothing inside but a huge slab of silicon and a scintillator - not even so much as a decoupling cap.
The only descrenable text on the die says "SUNI" and 1997
You can see why these things are so expensive though with a lump of silicon that big (27x38mm)
I gather that larger/(later?) ones use a process similar to TFT LCDs to make the large area sensors.

[Fraser]

@Mike,

That sensor head is the earlier type that uses a separate 'Dongle' interface to process the raw output from the imaging array. Those systems tend to suffer cable failures as the copper conductors are very thin to allow the use of a thin cable. The USB versions only have 4 conductors so can use slightly larger cross section copper cables.

The processing electronics that normally reside in the processing 'dongle' have been crammed into the Gendex head. Hence my comment on the high level of integration used. The down side of this is the challenge of repairing such miniaturised electronics and the need to open the hermetically sealed head housing, as opposed to a normal dongle housing that is screwed together.

The anatomy of the Gendex/Dexis imaging head is detailed here:

http://www.dexis.com/sensor-anatomy

In the attached image the layers are as follows (top down)

1. Front cover
2. Scintillator plate
3. Fibre optic micro-channel plate
4. Large CMOS detector array
5. Read-out and processing electronics
6. Rear cover with integral X-Ray attenuator
7. 4 conductor screened cable exits rear of sensor case via a small bulge.

Fraser

[KE5FX]

Interesting summary, thanks.  I actually bought a couple of these broken sensors recently (one adult-size and one child-size).  They are currently in the junk box awaiting some time to investigate repairing them.

Firmware corruption is not uncommon and requires the OEM's bespoke configuration and programming software. Not something they share with anyone except a service agent. Buyer beware. It is very easy to waste your money on these units.

I did notice that someone has reverse-engineered the 8051 firmware for the Cypress USB chip in these sensors (https://github.com/JohnDMcMaster/uvscada/tree/master/gxs700), or at least captured it for replay.  This should go a long way toward removing the dependence on the Gendex calibration files and/or proprietary software.  While I don't currently have much X-ray experience, I do have a good working knowledge of that particular chip, and I also have a friend in the area with wire-bonding and hybrid repair capabilities.  It's unlikely that anything could possibly go wrong with these things that we couldn't handle, given enough time to play with them... but that's the catch, as usual.   If you're able to post some detailed X-ray images of these things, it will help immensely if/when I have to open them up.

[Fraser]

For those readers wondering how good the images produced are when using these X-Ray sensor heads. I attach an image grabbed form a Dexis platinum ebay auction (not my sensor head but the same model).

Not bad quality.

Sensor head size 31mm x 42mm
True resolution is 1324 x 1842 pixels (with corners cut off)
Scintillator is Csl
Pixel size is 19.5um
Total pixels 2.4MP

Fraser

[Fraser]

@KE5FX,

Good to hear that you have some of these units as well.

For info, I have the OEM firmware reprogramming software used by the service agents when the unit suffers serial number corruption 

I have reshaped the USB plug and plugged it into my laptop. The good news is that the laptop now sees the sensor. I will need to install the driver software to test it further.

Fraser

[tautech]

Hell, need to subscribe to this thread too.

Thanks Fraser for sharing all these goodies.

[tautech]

For those readers wondering how good the images produced are when using these X-Ray sensor heads. I attach an image grabbed form a Dexis platinum ebay auction (not my sensor head but the same model).

Not bad quality.

Sensor head size 31mm x 42mm
True resolution is 1324 x 1842 pixels (with corners cut off)
Scintillator is Csl
Pixel size is 19.5um
Total pixels 2.4MP

Fraser

Would the SMD components that seem missing be ceramic caps and those seen resistors? (top right)

[Fraser]

Hi Tautech,

As you likely know, it is the density of the SMD component substrate that dictates the level of transparency when imaged using X-Rays. In my experience SMD resistors are normally pretty opaque when using lower levels of KVP on the tube. You can see through them by increasing the kVp and adjusting the ADU viewing window in the SR software. Common SMD capacitors do appear more transparent when imaged at the same KVP but I do advise caution. Not all capacitor types will appear the same. Tantalum's come to mind as they have higher density and it is possible to mistake an SMD ferrite for a resistor if you are not careful.

X-Ray imaging can be very useful in reverse engineering work as you can image the DUT and decide how best to proceed without causing damage to the PCB or components. Sadly X-Ray imaging is not very helpful when it comes s to identifying component characteristics. Visual inspection and component analysis is still needed.

Where an X-Ray image might excellent us in the disassembly or repair of a potted package or even an IC. Take for example a very rare Dallas RTC with internal potted battery that coming to the end of its life. The X-Ray would show exactly where to drill or cut in order to intercept the battery connections without damaging the electronics package. Very useful and a task I have carried out on several occasions using an X-Ray to guide me.

Fraser

[Fraser]

Mr McMaster is a clever chap.

There might be something of interest for you on this page Mike. He built a linear stage X-Ray scanner

http://uvicrec.blogspot.co.uk

At the bottom of the page you will see that he has released an independent GX-700 driver. Job done. Sweet.

Fraser

[Fraser]

Just for interest, the following page shows the many different models of Dental X-Ray sensor that are available.

http://dentalproducts.dentalproductsreport.com/s/dental-digital-radiography/products/filter/?category=dental-digital-x-ray-sensors

Fraser

[Fraser]

This is a good site that discusses the sensors and there is even an article on common hardware faults.

http://sodiumdental.com/digital-x-ray-review-part-1/

http://sodiumdental.com/category/sensor-repair/

Fraser

[marshallh]

Mr McMaster is a clever chap.

There might be something of interest for you on this page Mike. He built a linear stage X-Ray scanner

http://uvicrec.blogspot.co.uk

At the bottom of the page you will see that he has released an independent GX-700 driver. Job done. Sweet.

Fraser

Yup, he is. I gave a presentation with him this summer in Canada about chip decapping. If you are ever in the States you should see about visiting his garage of doom

[Fraser]

Very little time to 'play' with the GXS-700 but after reshaping the USB connector I have been able to get the sensor recognised and its firmware revision listed in the GXpicture driver. This is a good start as it suggests the USB comms channel and firmware in the head are OK. I will need to find a way to activate a capture mode in order to test the unit with live X-Ray but that will need to wait until another day. I have already loaded a trial version of the Gendex Twain interface so should be able to use that with generic capture software.

Pictures of GXpicture windows attached. It is just a driver interface and has no capture capability that I can find. The Gendex status indicator (top left corner of status box screen grab) is GREEN indicating that the sensor is ready to take an image and no problems have been found. The fact that the serial number is being shown indicates that the Calibration file is the correct one for the sensor    I wasn't sure that such would be the case or whether the serial number in the head would be corrupted.

It may be wishful thinking but the fault with this unit could just have been a damaged connector ? We shall see.

Fraser

[Macbeth]

Interesting. I take it you require an external xray source with these? They act like the old fashioned film plates the dentist shoves in your mouth before placing the xray device near your cheek then leaving the room and zapping you?

[Fraser]

Yes this is just the detector side of the system. A source of X-Ray energy is needed to illuminate the sensor array. The spec states a 60 to 70 KVp dental X-Ray generator. I have a Gendex 65kVp X-Ray generator so should have the required beam intensity for this sensor.

Fraser

[amyk]

Are we going to see some pictures of your teeth soon?

[Macbeth]

Are we going to see some pictures of your teeth soon?

Ooh, you don't want that... after all he's British (and ex secret service!) 

[tautech]

Are we going to see some pictures of your teeth soon?

Ooh, you don't want that... after all he's British (and ex secret service!) 

Nice one Macbeth. 
Real goood first laugh of the morning. 

[SeanB]

So they wil look like an old picket fence then?

[Fraser]

The Gendex GXS-700 contains a metal X-Ray scatter shield behind the PCB so I was not expecting it to be possible to image the PCB using my low energy Faxitron units. With this in mind I did some searching to see if anyone had tried to X-Ray the sensor with higher powered sources. I struck lucky.......

http://xrayfilmvsdigital.com/wp-content/uploads/2015/03/Dental-XRay-Radiography-study-s.pdf

I have extracted the image from the document. Not a great deal to see but the X-Ray does show some BGA footprints and the cable connections etc. The scatter shield is not that effective against X-Ray penetration, as noted in the document.

Fraser

[cex]

Mr McMaster kindly provided teardown images at SilconPr0n:

http://siliconpr0n.org/wiki/doku.php?id=gendex:gxs700

You may find these useful.

Also the API of his Linux driver allows triggering image sensor without X-rays that may also be useful to check if electronics are OK.

Regards,
Cex.

[Fraser]

@Cex,

Many thanks for the information, much appreciated.

Fraser

[KE5FX]

Mr McMaster kindly provided teardown images at SilconPr0n:

http://siliconpr0n.org/wiki/doku.php?id=gendex:gxs700

Doesn't seem to come up for me, anyone else?

[Fraser]

It's a weird web site format.

You need to search on GXS700 using the search box (top right)

Go to the page that it finds and select 'sensor' sub page.

Fraser

[Macbeth]

I had the same trouble. Perhaps this URL will work instead?

http://siliconpr0n.org/wiki/doku.php?id=uvscada:gxs700

[Fraser]

That works fine

[cex]

Oooops!

Sorry, surely the site was restructured after I bookmarked it, and I didn't test before posting the URL.

[Fraser]

Yes he appears to have restructured the site from when I last visited it. I had missed the teardown pictures though. Thank you for leading me back to the site to capture those.

The sensor is, as expected, highly integrated with the use of BGA chips and exposed bond wires between the PCB and imaging array. Not a very friendly repair proposition unless you have the correct rework equipment. It also contains an ASIC so failures in that IC would likely render the unit scrap.

I really need to set up my dental X-Ray generator to test mine but I have no time at the moment. After Christmas maybe.

Fraser

[tommyhn]

I need a favor, if you can send me the file to calibrate the gendex gxs 700 sensor, I lost my installation disk, my computer is broken and there is no way to recover the files

[Fraser]

Calibration file is unique to each sensor. I rehomed my Gendex sensor so no longer have the file set.

[Andrew Seltzman]

I have a few of these sensors and am trying to get them to work.

I have been talking to some Chinese sellers on ebay that are selling gendex sensors. They claim to have "third party calibration files" but say that although they allow the sensor to work, the image is not perfect, though they will send the third party files along with the sensor. The pictures they post that the sensor took look resonable, though not perfect. I strongly suspect that what they've done is found a way to either change the serial number on the calibration file or the serial number on the sensor to allow them to use a calibration file for a different sensor with the one they are selling.

There's a tool to reprogram the sensor eeprom(if corrupted) and serial number, though I'm not using it except as a last resort to get a sensor working.
http://justinshafer.blogspot.com/2014/11/dexis-fusion-tool.html

Could it be as simple as changing the serial number on the calibration file name to match that of the sensor in order to fool the software to work (I know the cal files won't be correct and the image won't be as clear, but just to test the sensor with the OEM gendex software)?
Does anyone have cal files for any gendex size 1 or 2 sensor that they could post so I could have a look at them?
They should be named (where the numbers are the unique sn of the sensor):

    012345678_dark.flf
    012345678_dfm.dfm
    012345678_flat.flf

[DogP]

Has anyone opened one of these sensors up, and if so, any tips for opening non-destructively?  I recently picked up one of these sensors, but it doesn't enumerate, and the cable looks a bit lumpy at the sensor end.  I'm hoping I just need to open it up, cut the last few inches of cable, and re-terminate the cable.  Unfortunately, the sensor is sealed (as you'd expect for a dental device), so I'm looking for some tips before I just start busting it open.

Am I correct in assuming all the electronics are in the sensor end, and there's no electronics in the USB plug end?  I wish I had access to an X-ray so I could X-ray this thing.

Thanks,
Pat

[DogP]

Just to follow up... I ended up getting this open, though it wasn't exactly a success.  So, I'll post my thoughts/advice/pics for anyone else (or the future me, if I come across another really cheap sensor).

To get into the sensor, you can see a flat spot on the side where you'd expect the two halves to join together.  In this case, it's sealed, so you need to carefully cut around the edges and sorta pry it apart (I used a sharp new X-acto knife blade).  It's small, so you'll want some magnification (I did it under a microscope).

The mistake I made was cutting closer to the top of the flat spot (side that the cable goes into) instead of bottom (side that says "GENDEX").  After opening it, I can see that the bottom goes inside the top (i.e. there's a lip, which you can cut to, without the knife blade going into the electronics).  Also, specifically be VERY careful on the end where the cable points out of... the bond wires to the image sensor are RIGHT behind the plastic and completely unprotected.  Some of mine look to have gotten mangled (not sure if I put the knife through them, pushed plastic into them, etc).

I originally suspected a bad USB cable, but once I got into it, I was able to check the cable, and it was fine (cable is soldered to a flex circuit on the top, which comes down to the connector on the board).  There are no electronics in the USB plug... it's straight USB wires into the sensor.  When the sensor is plugged into the PC, it doesn't enumerate, so something on my board must be broken (in addition to the now broken bond wires).

It looks like they used a two-part gap filler, which generally don't reuse well... I dunno how hot these sensors get, but if you open it, you might want to replace it with a new gap pad or something.

Also, if you really suspect just a bad USB cable, the easiest/safest way to fix it is probably to cut the plastic lump off the top where the cable goes in, and re-terminate the cable there.  That way you don't risk damaging any other parts by opening the sensor portion.

Couple pics attached...

Pat

[drgnanam]

hi,
i have one of sensor gendex gxs 700 size 2, does any body calibration file(i know each have unique one), just to test the idea changing serial number to fool the software, i sensor, with serial number 2133632432, sensor gets identified in gx picture  with all , just need original cal file to test it, i bought it from ebay cheaply,if any body have  califration file for different one  , any other way in windows os, please chime in,

[frogg]

Wow, I'm surprised that it uses the MCP with a CCD.

1997 was kind of the turning point for amorphous silicon x-ray detectors, coupled with a scintillator. GE and EG&G were the people that made it happen.

I always figured filmless x-ray imagers worked with ASi or ASe directly behind a scintillator, but it looks like other technologies are still in use. Cheaper to produce?

[Ben321]

That Dexis sensor is tiny, and doesn't even require an external interface box to connect it to the PC. My first guess is that means it's as cheap as any ordinary USB webcam (maybe $100 at most), only this one sees X-Rays instead of visible light. However I then realize it is a specialized device intended for medical use, and so the price is probably many times my first guessed price, kinda like thermal imagers (or maybe even more expensive like like SWIR imagers).

So does anybody here know the cost for one? Their website https://www.kavo.com/en-us/imaging-solutions/dexis-platinum-digital-intraoral-sensors doesn't show any prices. From my past experience, this usually mean that the device is VERY expensive, so expensive they are afraid that showing the price will scare off any potential customers.

[Fraser]

Ben321,

Indeed they are very expensive !

Regarding the parts cost, it is the large CCD FPA that is expensive. Apart from that the electronics are nothing very special compared to modern digital camera electronics.

The price of the units is high because they are for medical use and that tends to add a zero or two in many cases ! These items are also more specialist in nature than, say, a web cam. That results in a premium on the cost. Only a few companies make these sensors for dental work and most operate in a aggressive manner in terms of initial purchase and cost of ownership Let me expand on that to explain what I mean..........

1. You buy the sensor(s) that you require for your dental practice. They replace your film based system and are mission critical assets that are also expensive to replace if they fail.

2. You buy the software to use with the dental sensors. This is bespoke software and is very expensive. Some may be an outright purchase of a licence whilst others are a subscription based software model.

3. The seller of the sensors will warn you of the risk of sensor failure and loss of the capability in your practice until replaced at great expense. They will offer you a form of support/insurance contract which provides same day or next day delivery of a new sensor if one of yours fails. This contract is comprehensive in its coverage but very expensive ! It will continue to replace failed sensors every time one fails until the contract ends and is not renewed. For a dental practice it can be a necessary evil however. With the contract comes ongoing hardware support in addition to the software support.


The sensors that appear on eBay are often the units that have failed in service and been scrapped. Common failures are cables as they get a lot of bending at end points and even bitten by patients ! Other failures include loss of identity due to a problem in the flash memory or physical damage to the module casing or CCD FPA from being dropped. These things are fragile.

So yes, they are expensive, artificially so, but that is the nature of the beast and we all know how much private Dentistry costs ! An alternative to these active sensors is the CR plate that is reuseable but captures an X-Ray that is then read by a special scanner and eraser unit. The plate looks similar to a standard film type plate but it costs a lot more ! It saves money on film and developing chemicals over the years of use though and is relatively robust compared to the active sensors.

Fraser

[Ben321]

Ben321,

Indeed they are very expensive !

Regarding the parts cost, it is the large CCD FPA that is expensive. Apart from that the electronics are nothing very special compared to modern digital camera electronics.

The price of the units is high because they are for medical use and that tends to add a zero or two in many cases ! These items are also more specialist in nature than, say, a web cam. That results in a premium on the cost. Only a few companies make these sensors for dental work and most operate in a aggressive manner in terms of initial purchase and cost of ownership Let me expand on that to explain what I mean..........

1. You buy the sensor(s) that you require for your dental practice. They replace your film based system and are mission critical assets that are also expensive to replace if they fail.

2. You buy the software to use with the dental sensors. This is bespoke software and is very expensive. Some may be an outright purchase of a licence whilst others are a subscription based software model.

3. The seller of the sensors will warn you of the risk of sensor failure and loss of the capability in your practice until replaced at great expense. They will offer you a form of support/insurance contract which provides same day or next day delivery of a new sensor if one of yours fails. This contract is comprehensive in its coverage but very expensive ! It will continue to replace failed sensors every time one fails until the contract ends and is not renewed. For a dental practice it can be a necessary evil however. With the contract comes ongoing hardware support in addition to the software support.


The sensors that appear on eBay are often the units that have failed in service and been scrapped. Common failures are cables as they get a lot of bending at end points and even bitten by patients ! Other failures include loss of identity due to a problem in the flash memory or physical damage to the module casing or CCD FPA from being dropped. These things are fragile.

So yes, they are expensive, artificially so, but that is the nature of the beast and we all know how much private Dentistry costs ! An alternative to these active sensors is the CR plate that is reuseable but captures an X-Ray that is then read by a special scanner and eraser unit. The plate looks similar to a standard film type plate but it costs a lot more ! It saves money on film and developing chemicals over the years of use though and is relatively robust compared to the active sensors.

Fraser

I wonder what an actual high resolution CCD panel would normally cost, if not used in medical equipment?

[ycrcb]

I managed to find some calibration files on an obscure dental website. GXS-700 is also called Dexis Platinum
https://www.atlasresell.com/sensordrivers
The site have instructions.

It is also possible to install the calibration files, and use GxPicture 3.8.1 with VixWin platinum 3.3 in demo mode on Windows 10.
http://www.medtec.pl/download.html

[devilmastah]

I think i need to follow this thread   (working on a diy X-ray setup, i am currently running it with a mammography screen + digital camera setup which works surprisingly well)
I orderd one of these sensors for 50 euros from china (knowing it would be a gamble). As expected its broken. But the failure mode is quite weird to me.
Seems that multiple of the bonding wires are corroded, looks like 5 are gone. I am now looking for a hobbyist that has (access to) a bonding machine in my country 

I wonder if this is a common failure on these.

[kgh]

Could anyone please reupload the calibration files again or dm me.
the link is not working anymore.
Thanks a lol in advance <3

[KE5FX]

Could anyone please reupload the calibration files again or dm me.
the link is not working anymore.
Thanks a lol in advance <3

Seems that the Github repository was moved here (according to the readme file in the old repository).  Is that what you were looking for?