Faxitron MX-20 Reverse engineering (with some good results)

[kilohercas]

I started to make my own electronics for Faxitron MX-20 Machine. I only get electronics, X-Ray generator and X-Ray sensor, so I have to put something inside my won box.

I don't have software of any kind, so i have to write my own. First step was to make USB Hub, one channel wil go to sensor, second channel to FT2232H Controller as UART ( and second IC channels goes to stepper drivers for CBCT turning table)


After that, i had to write my own software for Hamamatsu 12x12 X-Ray sensor. That was quite a difficult task, but i was able to make it happen. I was able to do that in Visual Basic, it uses generic Hamamatsu DCAM-API DLL inside system32 (when you install drivers)



Operation is simple. Connect USB to HUB, it detects if FT2232H and C9732DK-11, after that, you wait for 60s warmup ( x ray tube or something, timer is inside x ray controller ) And you just set  time and kV, and press take  a picture. It automatically sends command to enable x ray source, you should (2.5s is ramp-up to 35kV, only at this point  x ray is stable based on sense output on x ray tube)
put into remote mode: (do this first !!!!)
                Dim mx20(5) As Byte '
                mx20(0) = &H21
                mx20(1) = &H4D
                mx20(2) = &H52
                mx20(3) = &HD
                PORT.Write(mx20, 0, 4)


This is how you set exp time and voltage:

            mx20(0) = &H21
            mx20(1) = &H54
            mx20(2) = &H30 + Fix(NumericUpDown1.Value / 100)' time 0.1 to 999.9
            mx20(3) = &H30 + Fix(NumericUpDown1.Value / 10) Mod 10
            mx20(4) = &H30 + Fix(NumericUpDown1.Value) Mod 10
            mx20(5) = &H30 + ((NumericUpDown1.Value * 10) Mod 10)
            mx20(6) = &HD
            UART_Status = FT_Write_Bytes(UART_Handle, mx20(0), 7, BW)
            Sleep(100)
            mx20(0) = &H21
            mx20(1) = &H56
            mx20(2) = &H30 + Fix(NumericUpDown2.Value / 10) ' Voltage 10-35
            mx20(3) = &H30 + NumericUpDown2.Value Mod 10
            mx20(4) = &HD '

Enabling X-Ray source via UART message is also easy, just few special characters: '0x21+0x42+0x0D  and after short delay 0x43
                mx20(0) = &H21
                mx20(1) = &H42
                mx20(2) = &HD
                UART_Status = FT_Write_Bytes(UART_Handle, mx20(0), 3, BW)
                Sleep(50) ' 50ms Delay
                mx20(0) = &H43
                UART_Status = FT_Write_Bytes(UART_Handle, mx20(0), 1, BW)

It was so bad design to put 500W power supply, it even needs load to work. That is madness. All you need is isolated 24V 1A for x ray ( it is uses all interlocks, and is enabled by relay on J7 connector )
12V 1.3A     (main electronics)
-12V 0.05A (opamps +RS232 ?)
5V 0.5A       (MCU electronics)
5V 5A          (Hamamatsu sensor)

Now, controller electronics looks old, and very complicated. Since i will be using computer, it would be better just connect x ray head to my own electronics, so i spend all tay figuring out controller connector to x ray source.

It is IDC20, and it has control, sense and other signals prescient, just have to figure out what is what.

IDC20 pins 1,2,3,4,12,13 are connected to opamp outputs inside x-ray head electronics. I only understand that 12pin is kV sense output to MCU board (1v=10kV), and 3pin is I sense, or it looks like this, with transfer function of:

10kV-> 0.208mA -> 0.61V
15kV-> 0.201mA -> 0.90V
20kV-> 0.192mA -> 1.19V
25kV-> 0.183mA -> 1.46V
30kV-> 0.175mA -> 1.76V
10kV-> 0.165mA -> 2.05V

IDC20 10PIN is also strange, MCU is supplying voltage, with a transfer function of ( every time, not only when x-ray is on)
10kV = 1.330V
20kV = 2.333V
30kV = 3.116V
35kV = 3.548V
Update: It has something to do with focusing. disconnecting yield blurry images. Maybe could be tuned separately for better picture, so must keep on DAC.

For output that change state when x ray is on, i have few of them:
PIN5 is connected to MCU via 330R  resistor ( mcu supplying 5V when enabling x ray )
PIN7 is connected to MCU via 2k2    resistor ( mcu supplying 5V when enabling x ray , another 12V relay for 24V that is main power for HV supply)
PIN9 is connected to DAC that supplying setpoint voltage from 1V (10kV) to 3.5V (35kV ) again, only when x-ray is on.
PIN11 is connected to DAC, it is supplying 2.7V only when x-ray is on (no mater on set voltage or time)

PIN10 looks like getting 10V from x-ray head, it is connected to potentiometer top input on high voltage power supply. still need to figure out why.
PIN15 and PIN17 is unconnected ( hooray:) )

PIN 14, PIN18, PIN19, and PIN20 are ground

Now I have to try to make my own driver on simple PCB, and we will see how it works. This will enable to buy MX-20 systems on ebay, and just ask to send sensor and x ray source, and i will be able to control everything

This post is mainly to remind myself, what i did so i don't loose it

[eliocor]

What software will you use/are you using for CBTC (Cone beam computed tomography) ?

[kilohercas]

MATLAB.

But it also would be fun to write my own in C/vb

[kilohercas]

few weeks of hard work, and i have complete solution for reading data from sensor as well as controlling x ray tube with my own driver that is based on Intel max10 fpga, works like a charm

[jeremy]

Very cool!

[kilohercas]

https://youtu.be/HrUNLFlLzzU

[jeremy]

What was the USB hub for? I couldn’t quite tell. I know you have an FTDI, but what else?

Also, is there any reason you decided to use an FPGA over a micro?

Would love to know more about how the panel outputs data and in what format.

Edit: ah I see now, the detector is USB connected.

[KE5FX]

Awesome work, would love to hear more.

My MX-20 has the older RadEye sensors, so I had to do a lot more work to get data out of them.  I would very much like to make my software talk to the control panel.  I don't understand what you mean here:

                mx20(0) = &H21
                mx20(1) = &H42
                mx20(2) = &HD
                UART_Status = FT_Write_Bytes(UART_Handle, mx20(0), 3, BW)
                Sleep(50) ' 50ms Delay
                mx20(0) = &H43
                UART_Status = FT_Write_Bytes(UART_Handle, mx20(0), 1, BW)

Are these instructions interpreted by your firmware, or passed to the Faxitron's controller?  If they're handed off to the Faxitron and processed there, how do you connect your controller to the Faxitron's CPU?

I don't currently have a UART available via the USB connection to my digitizer, but I could add one easily enough, if I knew where to connect it to the Faxitron.

[kilohercas]

Are these instructions interpreted by your firmware, or passed to the Faxitron's controller?  If they're handed off to the Faxitron and processed there, how do you connect your controller to the Faxitron's CPU?

UART is connected to faxitron controller, this series of commands will trigger x ray source for fixed amount of time

I made project in hackaday.io https://hackaday.io/project/52104-x-raying-rose-in-30kev-in-high-resolution you will be able to find all information about this project, with design files so you can make it yourself

If you can, please share this video tomorrow, since it since, and it kind a looks beautiful Lets make it as viral video for tomorrow

[KE5FX]

I made project in hackaday.io https://hackaday.io/project/52104-x-raying-rose-in-30kev-in-high-resolution you will be able to find all information about this project, with design files so you can make it yourself

Thanks!  Looks promising.   (Edit, having watched the video, it looks fantastic!)

Unfortunately I don't seem to be able to open the .rar files with WinZip -- it says they're invalid.  Can you post ordinary .zip files additionally/instead?

I would prefer to create a remote-control link to the MX-20's existing control panel, but your work on the new PCB is certainly impressive.  Agreed that the power supply they used is ridiculous.  Mine has a large wirewound resistor that does nothing but provide the minimum load and dissipate heat.

[eliocor]

Unfortunately I don't seem to be able to open the .rar files with WinZip -- it says they're invalid.  Can you post ordinary .zip files additionally/instead?

update your WinZip with the latest version...
Otherwise go to https://www.rarlab.com/download.htm and download WinRar: the archives have no problems at all!!!

[Mr. Scram]

I'd love to get my hands on a tool like that. That looks absolutely amazing and very useful.

[KE5FX]

I'd love to get my hands on a tool like that. That looks absolutely amazing and very useful.

That Hamamatsu panel is... enviable.    Some zero-feedback scammer is currently trying to sell 10 of them on eBay for $10,000 apiece.

Although it sounds like it's limited to <2 second exposure times if I understand the video correctly?  With my RadEye sensors I often have to take 20-30 second exposures, but maybe that's not necessary with the C9732DK. 

[mikeselectricstuff]

I'd love to get my hands on a tool like that. That looks absolutely amazing and very useful.

That Hamamatsu panel is... enviable.    Some zero-feedback scammer is currently trying to sell 10 of them on eBay for $10,000 apiece.

Although it sounds like it's limited to <2 second exposure times if I understand the video correctly?  With my RadEye sensors I often have to take 20-30 second exposures, but maybe that's not necessary with the C9732DK.

My Faxitron ( with 50mm Hamamatsu sensor) does up to 15 sec exposures. Not sure if extending would have much benefit due to noise floor. A few more kV would be much more useful.
The tube is rated to 50kV, but the sensor datasheet states 35kV max - would be interesting to know what the actual constraints are, as the limited penetrating power is the biggest disappointment of this system.

Where does the 2 sec limit come from ?

I don't understand the reason for replacing the faxitron control electronics - is there any benefit ?
Are you still using the original cabinet or were you re-packaging to reduce size?

[kilohercas]

Where does the 2 sec limit come from ?

I don't understand the reason for replacing the faxitron control electronics - is there any benefit ?
Are you still using the original cabinet or were you re-packaging to reduce size?

It comes from sensor, it has double value between 1  and 2. higher exposures are just stitched images made from 2s exposures i think. (DLL takes that data)

I only get detector and x  ray tube, that's why i am remaking it myself. Also, i have greater control over tube voltage, have diagnostics about tube voltage/current, so i can decide then to take picture, and stop x ray after that. I will also planing to add mechanical shutter, if i will try to x ray my hands, just to get dose down. Same goes for stepper control and other automation. this is very powerful system now, i even can think about chaining sensor position with other stepper to get different magnification

Also, electronics are old, if they break, i loose everything, now i know half the system, so less chance of failure

It can work with 100kV. In my work, i get visit from Hamamatsu, he says that problem with energy and power is that CsI will get yellow in time, but that's not a problem, since we still have so much data to work with, and we could use higher exposure times to get higher pixel values, so detector will be good for a long time.

[KE5FX]

My Faxitron ( with 50mm Hamamatsu sensor) does up to 15 sec exposures. Not sure if extending would have much benefit due to noise floor. A few more kV would be much more useful.
The tube is rated to 50kV, but the sensor datasheet states 35kV max - would be interesting to know what the actual constraints are, as the limited penetrating power is the biggest disappointment of this system.

I was able to crank the HV trimmer up to get 37.015 kVp at 325 uA out of mine.  The PSU folds back either current or voltage once the total power exceeds 12 watts, which is a good match for the maximum-rating spec of the TruFocus tube.

Where does the 2 sec limit come from ?

Linas mentioned on his Hackaday page that the Hamamatsu sensor has "very limited range, only between 1 and 2 seconds of integration."

I don't understand the reason for replacing the faxitron control electronics - is there any benefit ?

From my own perspective, I'd like to be able to set the voltage and exposure time from the software and trigger it remotely.  I often take bracketed exposures at different settings, looking for the best compromise for a given subject.  It'd be nice if I didn't have to punch a lot of buttons every few seconds, or keep track of the resulting image settings manually. 

The original Faxitron software could also carry out automatic calibration exposures, which I'd like to implement in my own software at some point.

[mikeselectricstuff]

From my own perspective, I'd like to be able to set the voltage and exposure time from the software and trigger it remotely.  I often take bracketed exposures at different settings, looking for the best compromise for a given subject.  It'd be nice if I didn't have to punch a lot of buttons every few seconds, or keep track of the resulting image settings manually. 

You can do that via the serial interface

[KE5FX]

From my own perspective, I'd like to be able to set the voltage and exposure time from the software and trigger it remotely.  I often take bracketed exposures at different settings, looking for the best compromise for a given subject.  It'd be nice if I didn't have to punch a lot of buttons every few seconds, or keep track of the resulting image settings manually. 

You can do that via the serial interface

I didn't exactly see any 9-pin RS232 ports on mine.  

I'm sure the necessary serial lines were brought out to the SCSI-style connector on the rear panel, but that's long gone.  Have you run across any documentation for their command set?

[mikeselectricstuff]

From my own perspective, I'd like to be able to set the voltage and exposure time from the software and trigger it remotely.  I often take bracketed exposures at different settings, looking for the best compromise for a given subject.  It'd be nice if I didn't have to punch a lot of buttons every few seconds, or keep track of the resulting image settings manually. 

You can do that via the serial interface

I didn't exactly see any 9-pin RS232 ports on mine.  

I'm sure the necessary serial lines were brought out to the SCSI-style connector on the rear panel, but that's long gone.  Have you run across any documentation for their command set?

I'm not familar with all the variants - I assumed they all had remote control, but as these units were probably originally designed for film, maybe not.

[KE5FX]

From my own perspective, I'd like to be able to set the voltage and exposure time from the software and trigger it remotely.  I often take bracketed exposures at different settings, looking for the best compromise for a given subject.  It'd be nice if I didn't have to punch a lot of buttons every few seconds, or keep track of the resulting image settings manually. 

You can do that via the serial interface

I didn't exactly see any 9-pin RS232 ports on mine.  

I'm sure the necessary serial lines were brought out to the SCSI-style connector on the rear panel, but that's long gone.  Have you run across any documentation for their command set?

I'm not familar with all the variants - I assumed they all had remote control, but as these units were probably originally designed for film, maybe not.

I'm sure that it can be controlled remotely because this particular unit came with its manuals.  There is a variant of the usual MAX232 interface IC on the controller board, so it wouldn't be hard to trace its pins through the cable that formerly went to the external connector.   But the manuals have no information on how the low-level communication with the PC worked, and disassembling the firmware would consume more time than it'd be worth.

[mikeselectricstuff]

From my own perspective, I'd like to be able to set the voltage and exposure time from the software and trigger it remotely.  I often take bracketed exposures at different settings, looking for the best compromise for a given subject.  It'd be nice if I didn't have to punch a lot of buttons every few seconds, or keep track of the resulting image settings manually. 

You can do that via the serial interface

I didn't exactly see any 9-pin RS232 ports on mine.  

I'm sure the necessary serial lines were brought out to the SCSI-style connector on the rear panel, but that's long gone.  Have you run across any documentation for their command set?

I'm not familar with all the variants - I assumed they all had remote control, but as these units were probably originally designed for film, maybe not.

I'm sure that it can be controlled remotely because this particular unit came with its manuals.  There is a variant of the usual MAX232 interface IC on the controller board, so it wouldn't be hard to trace its pins through the cable that formerly went to the external connector.   But the manuals have no information on how the low-level communication with the PC worked, and disassembling the firmware would consume more time than it'd be worth.

A while ago I grabbed a dump of the serial comms - it was pretty simple & obvious ASCII - will see if I can find it

[KE5FX]

A while ago I grabbed a dump of the serial comms - it was pretty simple & obvious ASCII - will see if I can find it

Thanks, that could come in very handy.  I'm mostly interested in starting/stopping the exposure and setting the kVp.  The ability to set the exposure time would be good as well, but the PC could emulate that if necessary.

[kilohercas]

A while ago I grabbed a dump of the serial comms - it was pretty simple & obvious ASCII - will see if I can find it

Thanks, that could come in very handy.  I'm mostly interested in starting/stopping the exposure and setting the kVp.  The ability to set the exposure time would be good as well, but the PC could emulate that if necessary.

Have you try to look at the first post ? All of this information is detailed. only stop function is missing...

[KE5FX]

Have you try to look at the first post ? All of this information is detailed. only stop function is missing...

That doesn't do me any good if I don't use your hardware, does it?  I already have a working system that I don't want to change to that extent.  I'd just like to be able to send remote commands to the original Faxitron controller.

That's basically why I was asking if your serial code was communicating with the original hardware or with your own controller.

[kilohercas]

Have you try to look at the first post ? All of this information is detailed. only stop function is missing...

That doesn't do me any good if I don't use your hardware, does it?  I already have a working system that I don't want to change to that extent.  I'd just like to be able to send remote commands to the original Faxitron controller.

That's basically why I was asking if your serial code was communicating with the original hardware or with your own controller.

put into remote mode: (do this first !!!!)
                Dim mx20(5) As Byte '
                mx20(0) = &H21
                mx20(1) = &H4D
                mx20(2) = &H52
                mx20(3) = &HD
                PORT.Write(mx20, 0, 4)


This is how you set exp time and voltage:

            mx20(0) = &H21
            mx20(1) = &H54
            mx20(2) = &H30 + Fix(NumericUpDown1.Value / 100)' time 0.1 to 999.9
            mx20(3) = &H30 + Fix(NumericUpDown1.Value / 10) Mod 10
            mx20(4) = &H30 + Fix(NumericUpDown1.Value) Mod 10
            mx20(5) = &H30 + ((NumericUpDown1.Value * 10) Mod 10)
            mx20(6) = &HD
            UART_Status = FT_Write_Bytes(UART_Handle, mx20(0), 7, BW)
            Sleep(100)
            mx20(0) = &H21
            mx20(1) = &H56
            mx20(2) = &H30 + Fix(NumericUpDown2.Value / 10) ' Voltage 10-35
            mx20(3) = &H30 + NumericUpDown2.Value Mod 10
            mx20(4) = &HD '

Enabling X-Ray source via UART message is also easy, just few special characters: '0x21+0x42+0x0D  and after short delay 0x43
                mx20(0) = &H21
                mx20(1) = &H42
                mx20(2) = &HD
                UART_Status = FT_Write_Bytes(UART_Handle, mx20(0), 3, BW)
                Sleep(50) ' 50ms Delay
                mx20(0) = &H43
                UART_Status = FT_Write_Bytes(UART_Handle, mx20(0), 1, BW)

This is serial to original controller

[mikeselectricstuff]

From memory I think it's 9600 baud

[KE5FX]

This is serial to original controller

Can I ask where you were providing this data, and at what baud rate (9600, as Mike says?) 

The MAX202 that's next to the 10-pin port labeled "RS232" doesn't have +5 on its VCC pin by default.  I tried applying +5 externally and sending a couple of commands (!MR<cr> and !T1230<cr>) at rates from 110 to 115200, but nothing happened.

[mikeselectricstuff]

This is serial to original controller

Can I ask where you were providing this data, and at what baud rate (9600, as Mike says?) 

The MAX202 that's next to the 10-pin port labeled "RS232" doesn't have +5 on its VCC pin by default.  I tried applying +5 externally and sending a couple of commands (!MR<cr> and !T1230<cr>) at rates from 110 to 115200, but nothing happened.

Can you post a photo of your controller to see if it's the same as mine - may be some link options or different firmware

[KE5FX]

Can you post a photo of your controller to see if it's the same as mine - may be some link options or different firmware

Here's a link to the photo (too big to post in the thread directly):

http://www.ke5fx.com/fx_ctrl.jpg

The 27C512 EPROM is labeled "V3.1 MOD", with the checksum 317E.  Binary image is here.

There is another chip that looks like a mask-programmed ROM, which is labeled 43805-65202-F SN 4312E-082 MX20 MICROPROS.

[mikeselectricstuff]

Can you post a photo of your controller to see if it's the same as mine - may be some link options or different firmware

Here's a link to the photo (too big to post in the thread directly):

http://www.ke5fx.com/fx_ctrl.jpg

The 27C512 EPROM is labeled "V3.1 MOD", with the checksum 317E.  Binary image is here.

There is another chip that looks like a mask-programmed ROM, which is labeled 43805-65202-F SN 4312E-082 MX20 MICROPROS.

Strange about the unconnected supply to the 232 transceiver - may be worth seeing if you can trace where it goes to get a clue as to why it's disabled ( a fine wire brush can be useful for this).
 

[KE5FX]

Strange about the unconnected supply to the 232 transceiver - may be worth seeing if you can trace where it goes to get a clue as to why it's disabled ( a fine wire brush can be useful for this).

I suspect that it's some kind of factory diagnostic port, given that nothing was originally plugged into it.   Ordinarily I would X-ray it to find out where the pins go, but... 

Contrary to what I said earlier, there actually was a 9-pin connector in the digitizer area, connected to J5 on the controller board, but it doesn't have a MAX202 nearby and it doesn't act like an RS-232 port.  Pins 3, 4, and 6 are pulled up to +5, with all others low.  It appears on the block diagram in the manual as "Cable A." 

[kilohercas]

https://youtu.be/T_qmQ1a4ftw

[Fraser]

KE5FX,

The 9 pin connector (cable A) located low down on the rear of the cabinet is the standard RS232 serial communications link to a PC serial port. I have various versions of MX-20 control software but they all expect to find a camera module connected so sadly they will not work for you

Fraser

[KE5FX]

If it's a standard port, it certainly doesn't act like one.    

The path of least resistance in my case will probably be to run some spare I/O lines from the FPGA board to optocouplers across the up/down and time/voltage keys.  Just seems like unnecessary hackery, that's all.

[Fraser]

I think I have the schematics for the control board. If so I will check the serial port design for you.

It will have to be tomorrow though.

Fraser

[Fraser]

I just had a thought........

The early MX-20 Digital with Bioptics camera array is not as highly integrated with the OC in terms of control.

From memory the PC controls the camera side of things but only provides monitoring of the safety interlocks, X-Ray on condition and initiatedps X-Ray off. It does not need a complex RS232C control set to achieve such limited monitoring and control. It could even be achieved with just the handshake lines !

The later MX-20 design provides much more control over the cabinet from the PC. The PC is able to monitor and set Exposure and Kv as well as control the X-Ray generator. I am uncertain what changes were made to the control PCB but IIRC it had an additional 7 segment LED for the time window for tenths of a second. I both versions of the MX-20 X-Ray head so have the EPROM's if that is all that needs to be changed.

Fraser

[KE5FX]

I think I have the schematics for the control board. If so I will check the serial port design for you.

It will have to be tomorrow though.

Fraser

  That would rock, if you can find schematics.  Even if your board is different from mine (and it probably is, if you have 0.1-second exposure time resolution) it could still come in very handy in the future.

To date, the only connection I've added is a wire from the "X-ray on" LED to an input line on my digitizer board, so that the software can tell when the exposure starts and stops.  If I eventually decide to add remote controls to some of the keypad buttons, a schematic could save some hunting and poking, even if it's not the right version.

[Fraser]

OK, I do have schematics for the control board...... of the original film type MX-20 circa 1997.

Whilst the controller does have a RS232 port it does not seem to be for remote control as the film cabinet had no need for such (having no PC).

Looking at the schematics I can also seen references to a 9 pin AEC sensor D type connector. I will upload the controller schematics but I am wondering........ do you have a Digital era MX-20 or the original film type cabinet that does not have the remote control capability or digital camera in the extended base area ?

The film type cabinet did have the uptight of an auto exposure X-Ray sensor that sat under the cabinet chamber and helped the user achieve good exposures of the film.

A digital era MX-20 would have the camera option detailed on the cabinets ID label on its rear. There would be a "DC" option type shown.

Fraser

[Fraser]

Schematics for the non digital 1997 Faxitron MX-20......

[Fraser]

Continued

[Fraser]

Continued

[Fraser]

Last pictures for now

[KE5FX]

Schematics for the non digital 1997 Faxitron MX-20......

Much appreciated.  Here's a tidied-up .PDF version: Faxitron_MX20_43805_65201D_65202C_schematics.pdf (6.6 MB)

These drawings apply to rev C of the 43805-65202 controller, while my PCB has revision F of the same part #.  So they should be a good match.  U7 is the mystery MAX202 with no Vcc connection, for instance. 

My 9-pin connector goes to the "Medoptics interface" at J5, which is definitely some sort of proprietary interface.  One difference between revs C and F may be pin 3 of this connector.  It's grounded on your rev-C schematic, while it seems to be connected to something else on the rev-F PCB.  That may have been done to add proprietary control support to film-only machines.

[danielbriggs]

This is an interesting thread, and much along the lines of what I'm slowly heading towards with my machine.

Great work kilohercas! 

Do you have any more details on the CBCT MATLAB route? I tried several algorithms myself, but am curious to what you used in the end as your results look really sharp.
[I found while most algorithms worked, some were ultra efficient, and some took a huge amount of computational effort for mixed results].

Even better... care to share you MATLAB .m files // or post an example of what you did?


@KE5FX + Mike...
Attached are my findings from when I was having a sniff around the RS232. It's not an exhaustive list, but contains pretty much all you'd need to run the cabinet from RS232.
[Excuse the notes too, it's just what I noticed the bus doing and scribbled down the findings].

[Manawyrm]

@KE5FX:
Almost 4 years later, I now own one of these beautiful machines myself.
I'm at the same point as you, with the same Rev. F PCB and also missing the +5V supply on the MAX202.

Have you done any further work on your machine? How did it turn out?

@Fraser:
You seemed to have a different/newer firmware. Would you happen to have a backup copy of the EEPROM?
It might be a good idea to start a little collection.
My machine also has the "v3.1 MOD" ROM.

Thanks,
Tobi

[KE5FX]

I haven't had time to do any more tinkering with it, unfortunately.  It works well enough with the keypad, although I'd still prefer to have full remote control if it could be achieved without a lot of modification work. 

One thing I'm still hoping to learn is the nature of the "seasoning" requirement, in which the manual says to perform a 5-minute warmup run at 20 kV before doing anything else.  Exactly how often this should be done, and exactly why it would be necessary at such low voltages, has never been explained.  That's the sort of thing I'd automate if I could, but would much prefer to skip altogether.

[Manawyrm]

Thanks for the reply!

Have you figured out how to remove/access the sensor yet?
It looks like the chassis is slightly different than described in my (slightly too new) 2009 service manual.

[Fraser]

I sold all of my MX-20 machines so regret I cannot help with firmware images.

Fraser

[Manawyrm]

... I'd love to talk more about my reverse engineering effort, should I do it here in this thread or open a new one?

Anyway:
I've taken a closer look at the "3.1 MOD" firmware.
It's pretty straight-forward to read, but there are no references to the serial I/O ports anywhere in the code.
That pretty much means that the 3.1 MOD firmware does not have any serial capabilities at all.

[KE5FX]

... I'd love to talk more about my reverse engineering effort, should I do it here in this thread or open a new one?

Here's good!   

Thanks for the reply!

Have you figured out how to remove/access the sensor yet?
It looks like the chassis is slightly different than described in my (slightly too new) 2009 service manual.

I had to build my own digitizer, since my MX-20 didn't come with its original PC.  I haven't made any modifications to the upper part of the machine, though, except for connecting the 'X-RAY ON' LED to the FPGA board. 

[Manawyrm]

I had to build my own digitizer

Heh, yeah, I'll have to do the same thing. I had to download the pictures manually (seems to be a HTTPS->HTTP thing).
Very nice board!

I haven't checked yet whether my machine uses the RadEye 1 or 2 sensor. It seems to be the DC44 option model.
Will open it soon.

My current plan was to use one of the modern STM32H7 processors, save an image to SDRAM and then either use SPI or something else to get the picture onto my computer.

[Manawyrm]

@KE5FX:
You wouldn't happen to have one or two raw 14-bit images, that I could use for developing my user interface while I'm waiting for my acquisition hardware?
Just any format (binary, csv, etc.) would be totally fine, I can make that work

[KE5FX]

@KE5FX:
You wouldn't happen to have one or two raw 14-bit images, that I could use for developing my user interface while I'm waiting for my acquisition hardware?
Just any format (binary, csv, etc.) would be totally fine, I can make that work

No, my application reads the data into memory and saves it in .png format, so there is never a raw image file anywhere.  It's a giant hack, feel free to snag anything you want out of it.   (Probably need to enable non-https transfers for that link as well.)

[LaserSteve]

Seasoning in rotating anode tubes is done to 1. Getter the vacuum to prevent flash over. 2. Uniformly heat the whole rotating anode to prevent mechanical stress. 3. Clean the surface of the anode. 4 Bring the cooling oil to uniform temperature. 5.. Clean the bus bar tracks and brushes  on the rotating gantry to ensure clean power delivery.

Failing to season the tube prior to imaging
Was a instant dismissal offense.
Seasoning adds greatly to tube lifetime.

In larger systems an anode heat capacity calculator is running in the background, to protect the tube.

There was only one accepted excuse. If you were in the field
working on a system and an accident victim came in,
If the ER physician and the radiology technician so ordered, you could skip calibration and seasoning to try to save a life.

I spent a year training to be a  medical CT repair instructor. The FDA shut down my employer's plant for a year, so I moved on.

Steve

[KE5FX]

Seasoning in rotating anode tubes is done to 1. Getter the vacuum to prevent flash over. 2. Uniformly heat the whole rotating anode to prevent mechanical stress. 3. Clean the surface of the anode. 4 Bring the cooling oil to uniform temperature. 5.. Clean the bus bar tracks and brushes  on the rotating gantry to ensure clean power delivery.

Sure, for a high voltage, high power rotating-anode tube.  But this machine has a fixed-anode 35 kV tube, with dissipation in the 10-watt neighborhood.     I don't get what seasoning does, other than to accelerate tube wear.  Do tubes this wimpy ever suffer from flashover?

[LaserSteve]

Yes, most x-ray  tubes do suffer from random outgassing.
It does not take much gas to transition from e-beam to arc.
Tungsten or other refractory metal  anodes are sintered from powder in a press under extreme pressure and temperature.  For that reason they tend to store tiny pockets of gas. Using the heated cathode as an active getter strikes me as asking for trouble. But that is the designer's problem.

Steve

[LaserSteve]

https://xray.oxinst.com/learning/view/article/x-ray-tube-conditioning-procedure

That settles it.   

Steve

[KE5FX]

That article might "settle it" for some unspecified classes of tubes, but as usual, it fails to acknowledge the massive difference in construction and operating conditions between a 35 kV microfocus tube and a high-power rotary anode tube operating at hundreds of kV.  It's like comparing a 4CX100000 to a 6AQ5.  There is, again as usual, nothing type-specific to go by in this article, which is my whole beef.

The procedure in the article also bears little to no resemblance to the procedure in the docs I got with the MX20.  You can't adjust the tube current on the MX20, for example, only the voltage... and the operating note says to do it every day, not every 3 months.

While I might prefer the 3-month advice given here because it's less trouble than doing it every day, I still don't have any reason to think it applies.  Maybe it's more important to season a low-power microfocus tube frequently.  Dunno.

[LaserSteve]

The page above  referenced two of their micro tube numbers. Perhaps ask either Oxford or X-Ray-Worx via email?

My Googl-Fu is decent, I could find almost no detail on seasoning or conditioning microfocus tubes.

After reading up on them, I'm intrigued.   We have one in a mouse/rat scanner downstairs at work.  One of the few instruments in my building I've never toyed with.  I'll see if I can get my hands on the manual.



Steve

[Fraser]

The Faxitron Microfocus tube is made by TruFocus and their datasheet for the tube includes a warranty comment about treating the tube correctly in accordance with their use guides. It may be worth asking the people who actually manufacture this specific tube ?
If you are lucky, they will provide the tube integration guide that should include operational care procedures.

The tube is a TruFocus 8050 side window model.

http://www.trufocus.com/xray_datasheets/microfocus/8000-SWS-Side%20Window/8050%20SWS/MFX-8050-SWS-V01.pdf

TruFocus support POC: info@trufocus.com

My understanding of the need for seasoning was that if gas purity within the tube. A flashover can be catastrophic for the machine and these cost £60K in the UK so I can understand why the users are told to regularly season the tube. As a basic rule, the MX-20 was to be warmed up each day and seasoned on a specified kV for 300 Seconds. This got everything warmed up nicely ready fir the working day. It would also reveal any serviceability issues in the machine as some of these were mission critical in operating theatre for removed tissue X-Ray inspection. Of note, the kV setting fir the conditioning run varied over different user manual generations. It used to be 22kV, then 20kV and I have seen 18kV hand written over the manuals 20kV in one manual.

My take on the conditioning cycle was that it gave the machine time to warm up a bit and settle down fir my imaging run. 300 Seconds was not very long to wait and the digital camera plate takes up to 30 minutes to completely warm up and settle down, if doing a calibration, so in the big scheme of things, the conditioning cycles not an issue fir me. If I had not used the machine for a few weeks I would go as far as running conditioning cycles of 300 Seconds duration at 10kV, then 15kV, then 20kV and finally 30kV. After that treatment I could be sure that the tube was nice and warm ready for it’s work. The wear on the tube was minimal but the benefit was potentially significant so I chose to treat my MX-20 machines with great care to prolong the tube life. A failed tube was normally a ‘end of life’ event for a MX-20 as the TruFocus tube was horrendously expensive. IIRC it was £5K. Why take a chance of harming the tube by not following the Faxitron tube care guidance ?

Fraser

[BUAROD]

@kilohercas:

Regarding the Hamamatsu sensor C9732DK I found this statement in your post:

It can work with 100kV. In my work, i get visit from Hamamatsu, he says that problem with energy and power is that CsI will get yellow in time, but that's not a problem, since we still have so much data to work with, and we could use higher exposure times to get higher pixel values, so detector will be good for a long time.


I think about replacing the TFX-8035 tube in my MX-20 e.g with a PXS10 system, which allows up to 130kV anode voltage.
Beside necessary additional shielding I am concerned what will happen to the C9732DK? What do you think?
Do you have more information than already written from Hamamatsu?
Remark: the MX-20 is not intended to be used in continuous operation but may be once a month ... 
Many thanks in advance!

[LaserSteve]

As you move to a higher power tube, you might find you need to add filters for spectral correction on the tube output if scanning organics and polymers. Higher acceleration causes spectral broadening at the source. So if it's an organic or polymer target, you can generate far more x-ray photons in a region that is not absorbed, and this could add noise to your detector that obscures  weak signals. This results in a nonlinear gamma curve in organics/polymers.

With a hotter tube, you will also  get much more scatter off the shielding  into the detector.


JUST FYI:


In  living things, of course the goal was to tailor the spectrum to specific adsorption to reduce irradiation in the patient, achieve high contrast based on atoms present in the tissue,  and reduce patient exposure to useless low energy photons. Another major consideration was designing the filter to extend scintillator lifetime. Something to think about if your imaging "soft" materials. 


FyI:

Probably not an issue with a micro-focus tube, but with larger anodes the filters were shaped to correct for the beam pattern at the detectors. Ie get rid of the Hotspot in the center and wasted energy outside the detector area that resulted in scatter that reduced contrast.

Filters in my case were stacks of nylon blocks, then thin copper, aluminum, and beryllium sheets

Steve

[BUAROD]

Thank you Steve for the quick reply!

My area of ​​application is more the analysis of defective parts or assemblies, where I would like to see, for example,
the solder balls under a BGA or the bonding wires in an IC. The 35kV are obviously not nearly enough.
This issue has been raised several times here on the forums, but I haven't found a way to upgrade the MX-20 for this application.

Does anyone have an idea or has this already implemented?

[danielbriggs]

The 35kV are obviously not nearly enough.
This issue has been raised several times here on the forums, but I haven't found a way to upgrade the MX-20 for this application.

Does anyone have an idea or has this already implemented?

I asked this question to MikesElectricStuff last night on the EEVBlog Live stream, and he quite correctly pointed out the Hamamatsu sensor isn't rated past 35kV, so pushing the MX-20 past 35kV would be expensive and quite involved.

[LaserSteve]

Golden Engineering makes some point source systems that might be useful for close in PCB inspection.

Steve

[KE5FX]

Thank you Steve for the quick reply!

My area of ​​application is more the analysis of defective parts or assemblies, where I would like to see, for example,
the solder balls under a BGA or the bonding wires in an IC. The 35kV are obviously not nearly enough.
This issue has been raised several times here on the forums, but I haven't found a way to upgrade the MX-20 for this application.

Does anyone have an idea or has this already implemented?

Whether 35 kV is enough depends on several things, most prominently the atomic number of the material you're shooting through and how much of it there is.  35 kV is enough to obtain photos like the ones in the other thread, many of which are on par with commercial X-ray PCB inspection systems. 

What the commercial inspection systems buy you is the ability to see through more layers of copper.  8 layers of 1-oz copper is about as far as I can push my own MX-20.  That's sufficient for the vast majority of boards I work with.

The MX-20 has incredible resolution, enough to make bond wires look like heavy cables.



Of course you can only tell so much about BGA assembly quality by looking straight down at the ball-PCB interface once the beam has had to travel through the much thicker balls themselves.  Stray blobs of solder will stand out nicely, for sure.

You can 'overclock' the MX20 to about 37 kV, and I do think that's worthwhile, but going all the way to 50 seems inadvisable at first glance even though the tube itself is rated for it.  The need for a careful radiation survey would go without question, and you'd have to back off the anode current substantially to avoid damaging the tube.

[mikeselectricstuff]

You can 'overclock' the MX20 to about 37 kV, and I do think that's worthwhile, but going all the way to 50 seems inadvisable at first glance even though the tube itself is rated for it.  The need for a careful radiation survey would go without question, and you'd have to back off the anode current substantially to avoid damaging the tube.

ISTR reading somewhere that the Hamamtsu sensor was only rated to 35kVp

[LaserSteve]

I realize that all you can control on the Faxitron is voltage and exposure time, and probably some other detector parameters. However this is still worth a read.  I'm used to having time, mAS and KvP controls. In my world not having tube current control would really be depressing. Going from 35 to 37 Kv is only a 6% change.

This page has a really good explanation of some basic relationships.

https://howradiologyworks.com/xraycalc/

Steve.

[KE5FX]

Going from 35 to 37 Kv is only a 6% change.

The article points out that exposure is proportional to the fifth power of the voltage, though.  ("Since x-ray exposure goes approximately as kVp⁵ and we generally don’t think in powers of 5 a well known approximation is typically used called the 15% rule of kVp of radiography. This rule comes from the fact that 1.15⁵ ~2.0. Therefore, for each increase of 15% in kVp the exposure will double. Thus, if the kVp is increased by 15% then the mA must be decreased by a factor of 2 in order to maintain constant exposure on the image receptor. ")

I'll confess I don't understand where this relation could possibly come from.  kVp is the spectral location of the bremsstrahlung peak, and mA is its height, isn't it?  What's the physics behind the fifth-power dependency of exposure on anode voltage?

[BUAROD]

Thanks for the contributions!

I found some interesting Hamamatsu documents on the sensors:

1. Overview (page 3!)

https://www.google.de/url?sa=t&rct=j&q=&esrc=s&source=web&cd=&ved=2ahUKEwj-hqyqhdn1AhXz7rsIHWQLDEs4HhAWegQIFBAB&url=https%3A%2F%2Fwww.hakuto.com.tw%2Ftw%2Fdowfile.php%3Fp%3D72&usg=AOvVaw028KB2JaRtDm3MO6tYX3E5


2. Technologies (pay attention to statement on page 213 top left!)

http://www.studiodentisticovenuti.it/wp-content/uploads/2012/06/x-ray-detectors.pdf


3. Applications

http://notes-application.abcelectronique.com/233/233-48641.pdf

[LaserSteve]

I need to pull a paper out of the library at work on Monday to get a good explanation of that 5th power rule.. It's the whole chain of transfer functions, both source and detector. It's more film oriented, but it does apply to a detector structure that does not apply signal processing to the image.

Monday...

Steve

[LaserSteve]

From:

G.S. Al-Balool, D.L. Newman,
The relationships between kV, mAs and thickness in film-based radiography: 25% and 15% rules. OK?,
Radiography,
Volume 4, Issue 2,
1998,
Pages 129-134,
ISSN 1078-8174,
https://doi.org/10.1016/S1078-8174(98)90009-8.
(https://www.sciencedirect.com/science/article/pii/S1078817498900098)

Abstract: Purpose: To investigate the validity of the 25% and 15% rules in manual exposure selection. Methods: The mAs required to produce a constant film density of 1.5 was determined over the X-ray tube potential range 50–150 kV for different thicknesses of water, under grid and no grid conditions. Results: The mAs change is an exponential function with depth and is dependent upon the attenuation coefficient. The 25% rule assumes a constant attenuation coefficient with a half value layer (HVL) equal to 3 cm. The results indicate that for non-grid work at low X-ray tube potentials and for grid work at higher tube potentials the 25% rule is applicable. The 15% rule implies an inverse power relationship between mAs and tube potential, the power term being 5. The results indicate that for both grid and no grid conditions the 15% rule is obeyed only over the tube potential range 50–100 kV and for thicknesses up to 15 cm. Discussion: The 25% and 15% rules have provided useful practical guidelines for exposure factor selection. It is shown that they are applicable to some but not all radiographic situations.
Keywords: exposure factors; selection; 25% rule; 15% rule


Simplified Conclusion of the paper: 15% rule holds for tissue of less then 13 cm thickness and Voltages less then 100 kV.
For other more drastic conditions the 20$% rule holds.

I quote:

" In fixed kV techniques, the 25% rule is widely
applied [1, 2]. This states that for every I cm
change in patient thickness a 25% mAs change is
required. This rule derives from the exponential
attenuation relationship in that the required mAs
change is proportional to e ^ ±Hx where His the
attenuation coefficient and x the thickness, the sign
depending on whether an increase or decrease in
mAs is required, i.e. whether there is an increase or
decrease in patient thickness. Therefore, as the
thickness changes the mAs needs to be changed
exponentially to maintain constant film blackening.
For the 25% per cm rule to apply over the whole
range of diagnostic tube potentials,/a is assumed to
remain constant and equal to 0.23I c m - ~ i.e. a half
value layer (HVL) for soft tissue of 3 cm. This, of
course is a simplification since 1~and hence HVL are
energy dependent. However, the mAs stations
on many X-ray units today follow this rule in that
each consecutive mAs step represents a change of
25%. Thus, in fixed kV techniques, thickness
changes can be compensated by appropriate step
changes in mAs."

For the 15% rule:

I quote

"In variable kV techniques the situation is not as
straightforward because there are a number of kV
dependent changes that take place when the tube
potential is varied. These include changes in tube
output, X-ray attenuation (absorption and scatter)
and film screen speed. At a given patient thickness,
adjustment of the tube potential requires a compen-
satory adjustment in mAs so that the resultant
exposure produces the same level of film blacken-
ing. The so-called 15% rule [3-6] has been used as
a guide to the change in mAs that is required. This
rule states that to achieve the same film blackening,
a 15% change in kV requires a compensatory
change in mAs of a factor of 2. If an inverse power
relationship exists between mAs and tube poten-
tial, this rule assumes that the power term is 5,
since 1.15s=2, i.e. mAscz kV -5. This rule results
from the relationships between mAs and tube
potential in determining X-ray output, attenuation
and film screen response. With the introduction of
high frequency generators and energy-dependent
intensifying screens it is possible that these rela-
tionships have changed and hence the rules may
now not be applicable or at least need to be
amended"

REFERENCES for above PAPER:

A Kelly
Black or white?—hit or miss?
The Radiographer, 35 (1988), pp. 19-26
View Record in ScopusGoogle Scholar
2.
Why exposure tables with a point system?, Siemens Radiographic World News (1962)
Siemens
Google Scholar
3.
RR Carlton, A Adler
Principles of Radiographic Imaging
Delmar Publishers Inc, Albany (1992), pp. 480-481
Google Scholar
4.
JC Malott, J Fodor
The Art and Science of Medical Radiography
(7th edn.), Mosby Year Book Inc, St Louis (1993), pp. 130-131
Google Scholar
5.
QB Carroll
Radiographic Exposure, Processing and Quality Control
(5th edn.), Charles Thomas, Springfield (1993), pp. 102-106
CrossRefGoogle Scholar
6.
JE Wallace
Radiographic Exposure
Principles and Practice, Davis Co, Philadelphia (1995), pp. 92-96
View Record in ScopusGoogle Scholar
7.
DN Chesney, MO Chesney
Radiographic Photography
(3rd edn.), Blackwell Scientific Publications, Oxford (1971), p. 293
View Record in ScopusGoogle Scholar

[CwazyWabbit]

Found this info on the serial protocol being used, not sure if it adds any more to what you already know though.

[Andrew Seltzman]

@danielbriggs

I wrote a MATLAB package for cone beam CT on the DX-50, it should run the MX-20 since it uses the same protocol.
http://www.rtftechnologies.org/physics/faxitron-DX50-CT-scan.html

[danielbriggs]

Thank you Andrew! Really impressive work. 
What an excellent resource on your site. I'll work my way through the videos, and adapt any code where appropriate for my MX-20.

[MindBender]

I wrote a MATLAB package for cone beam CT on the DX-50, it should run the MX-20 since it uses the same protocol.
http://www.rtftechnologies.org/physics/faxitron-DX50-CT-scan.html

Would you consider sharing the source?

[CwazyWabbit]

Looks like they are on his website, follow the link in his message.

[KaneTW]

I found a Faxitron unit on eBay with a DC44 for a reasonable enough price that I immediately bought it.

Has there been any progress in reverse engineering the interface for it, or is it possible to reasonably replace the imaging sensor?

[KE5FX]

I found a Faxitron unit on eBay with a DC44 for a reasonable enough price that I immediately bought it.

Has there been any progress in reverse engineering the interface for it, or is it possible to reasonably replace the imaging sensor?

This is still a good source of info to start with.  I archived it under docs/Haidekker in the zipfile in reply #52 above, if you haven't seen that yet.

There is probably no point in replacing the Bioptics sensor unless you can find a Hamamatsu panel.

[KaneTW]

I found a Faxitron unit on eBay with a DC44 for a reasonable enough price that I immediately bought it.

Has there been any progress in reverse engineering the interface for it, or is it possible to reasonably replace the imaging sensor?

Small update, the seller had a DC12 unit available and is sending that instead.

[KaneTW]

Unit arrived. Seems in good condition, except:

* Tube current output shows something weird. -0.20V in VDC, 0.06V in VAC, 0.21V in VDC+AC. Voltage is on point. There's an image produced and I have no idea what's causing the weird current readout.
* Rear USB-A to USB-A adapter didn't work, I had to remove it and connect the cable directly for my PC to detect the sensor.

Apart from the weird tube current issue everything looks ok.

[Fraser]

SR Software and Technical manual link sent today. Hopefully that will assist you in testing your unit.

Fraser

[KaneTW]

I queried Gamma High Voltage and the current monitoring is a 1kOhm resistor + capacitor + 2 Diodes placed across GND and HV return. So a negative voltage is expected. This is also confirmed by the schematic having an inverting amplifier (-10x) on mA feedback.

I believe the oscillation is actually occuring in that case and not just a readout bug. Once I have the chance I'll probe the current setpoint and see if it's oscillating as well.

[KaneTW]

I've probed the current monitor with my scope and the it looks like EMI to me. I've debugged further and noticed a peculiarity.

There's a reed relay (Fraser, can I post the schematic?) K1 that ultimately switches the L2-Current an L3-Focus outputs. It does so via switching the wiper output from the L2/L3 pot to an opamp.

The voltage before the relay is nominal. The voltage after the relay is much higher -- I've measured:
Wiper: 1.31V, node after relay (R5/R3) 3.3V
then 1.47V -> 5.26V, then 1.56V -> 6.08V

The inverting output is at most 2.3V, so the opamp goes to its max possible output of 10.5V pretty quick, which corresponds to about 9.2V on the BJT emitter that is upconverted to the tube control voltage.

I need to take out the gun board and probe some more, but it's a huge pain to reach the bottom srcrews...

[Fraser]

No objection to showing the schematics.

I will look in the service manual to see if there is anything in there to assist you.

Fraser

[KaneTW]

Here's the set of links:
Gun board: https://i.kane.cx/dFTeBu https://i.kane.cx/8JmAtu https://i.kane.cx/xvgpbQ
Controller board: https://i.kane.cx/H7T6TI https://i.kane.cx/GmY9hV https://i.kane.cx/VGPG0v https://i.kane.cx/TGXgZR

Gun board annotated with measurements:


Thermal cam shot:
 

The component getting hot is the reed relay K1. The temperature did not noticeably change whether it was on or off.

Some scope shots of the mA current output:
10kV: https://i.kane.cx/Mh6Z1i https://i.kane.cx/qQl6TF
20kV: https://i.kane.cx/FQfNOj https://i.kane.cx/3cXARo
35kV: https://i.kane.cx/s8l59J https://i.kane.cx/gcR71o

The high frequency content is always present and I'm pretty sure it's just EMI/bad probing (the current output port only fits a multimeter lead). The ~25kHz content however is increasing with increased kV, so something's weird there. Note that the average remains roughly constant at -250mV.

[KaneTW]

After some debugging I noticed that L2Q was set to a way higher value than the manual specified, around -280V. It does claim that you need to follow the tube test sheet but I adjusted it to -200 or so and got closer to the 0.3mA of current that it should have. Still can't go above it (so really bad SNR at lower voltages) but that's good enough for me since I image at max voltage.


Some images of a 10-layer board attached.