Teardown of Magstim 200 Transcranial Magnetic Stimulator

[dexters_lab]

Teardown of a Magstim 200, a 1980s TMS base unit. I have two of these, one has released the magic smoke and the other is seemingly working.

Unit discharges a large high voltage capacitor into a coil placed on the patients head, the magnetic field produced induces electrical currents into the brain.

https://en.wikipedia.org/wiki/Transcranial_magnetic_stimulation

Part 1


Part 2

[dexters_lab]

small update on what i know since the part 2 video...

i found a document referencing the Magstim 200, it details full charge of the capacitor is 2800 volts

so this should allow a few things to fall into place... the first potted brick, which i am calling the charging control is in series with the 120/240v power feed to the HV transformer, the output of the transformer runs into this same brick. So i think i am making a pretty good guess that the brick will switch power to  the transformer, rectify the output and when the voltage is correct for the power set on the front panel, it turns off.

[Andy Watson]

If you have access to academic journals try searching the medical physics papers for the profs I.L.Freeston and A.T.Barker - it was their baby in the the early 1980s. 

[dexters_lab]

If you have access to academic journals try searching the medical physics papers for the profs I.L.Freeston and A.T.Barker - it was their baby in the the early 1980s.

thanks Andy, it was through searching some of the academic papers that i found this...

http://www.psicomag.com/biblioteca/Guia%20MagSrtim/mguide.pdf

i dont have access to any journals, but i have some friends that do so i am going to see if they can get the original text

[dexters_lab]

made a little progress

I started looking at the front panel connector now, as i'm getting a handle on how this works i need to figure this out to hopefully disable the 'Replace Coil' error that seems to be locking everything out.

First off i popped out the pins from the charging brick that connect to the primary of the HV transformer so that should disable any HV charging and make it a bit safer for me.

So i mapped out the front panel connector, 19 pins total, 10 for the main coil power and 9 low voltage lines for sensing etc. These 9 all run to the front panel logic board.

I found ground, +5 & +12v on four of the pins. Pin 9 is connected to the logic board but doesn't seem to go anywhere. Leaving pins 2, 3 and 8 unaccounted for. 2 & 8 have no voltage on them at all so I took a wild stab and took a 10k resistor and probed that between the +5v pin to pin 8 and the "Coil Temp" light came on so that'll be the coil temperature sensor. When doing the same to pin 2 it lit up 7 segments of  the bargraph on the front panel, switching to an 18k resistor it lit up 6. One thing that did make me LOL is the fact the bargraph has 11 segments i wonder if they had been watching too much Spinal Tap when making this?

I am wondering if the different coils have different maximum power levels, they could place a fixed resistor in the coil body to feedback a basic power rating to the base unit. The output power maybe scaled depending on the coil that is attached. A user guide would probably help here! 

Finally pin 3, i found an odd -166mV here so left this alone. I may trace this pin on the logic board to see if that helps.

[dexters_lab]

need some help guys... i have been looking at this device in more detail trying to figure out how to get this 'replace coil' error/lockout turned off and to be honest i am not making much progress

what i would like to achieve is to keep the overall workings original with an aim to replace the front coil connector with some connection of my own but retain all the original power adjustment controls, triggering etc. I plan to make my own coil to their spec to get this working thankfully they detail the coil turns and inductance so i can make something up.

i have tried to trace the circuit for the mystery pin 3 which i believe is some kind of sense to detect faulty coils but it's proving difficult for me, mostly because i have never worked with opamps before which is compounded by the board being 4 layers, having various bodge wires and bodge boards on.

this Pin 3 runs from the main connector into two LM324AN quad opamps through a 33k resistor, on the fist 324 it runs into two of the 4 opamps the negative input of one and the positive input of another, the output of one of these i know traces to a driver IC that in turn switches the 'replace coil' led indicator so i know i am on the right track here.

i do know there is another interlock on the coil head itself, just a simple switch that has to be held before you can trigger it, this is also something i have not figured out yet

of course if i had a coil i could see how it was constructed but i cant find anything online about them and magstim's patents on coil design dont reveal that much detail to help
One thing that puzzles me is how they are detecting a good coil over a bad one using this low voltage connection, i think knowing this would help.

any thoughts on how they could check to see if a a coil is good or bad?

 

[SeanB]

2 opamps says a window comparator, so they likely are detecting the resistance of something in the head, like a thermistor for over temperature, or they have a fusible link inside that melts open after the requisite number of pulses and then the next pulse is dissipated by a resistor to trip a thermal fuse and a series resistor.  Look at the other pins of the opamp, and see the voltages there, one will be higher than the other, giving you the window voltage they expect so if you put in something in this range it will fool it into thinking coil good and allow triggering.

[dexters_lab]

Thanks Sean, i'll take some readings when its powered on

there is a separate temperature input which is a basic thermistor, reading some of the manuals they expect the user to have more than one coil to hand so they can be swapped out when they get too hot

but i hadn't thought about using something like you suggest to permanently disable a coil, i guess they could do in house testing, work out the typical lifespan of a coil and they design in a safe failure mode that happens before the actual coil fails... designed obsolescence? But i guess its fair enough considering someone is holding a 3000volt 2 tesla coil next to an innocent patient!

[SeanB]

I learned that from a toner unit used in a Konica copier, which had a small PCB on it connected by a set of 5 contacts. Common ground ( shared with the static ground), a variable resistor used to set the photoconductive drum voltages ( used in the power supply to set the corona voltages per drum), 2 used for the red LED and phototransistor for low toner detection and the last was connected to common with a 50mA wire in fuse. They would typically do 2000 pages per cartridge, and you could refill them easily just by popping the end cap off and putting in new toner, but they would check the fuse and blow it after around 200 copies after toner low. So I wired in a 20A wire link and refilled them till the drum wore out after around 5 refills. Toner was free as the Sharp copier next to it had a refillable toner box and they dropped off a bottle of toner a month, when the copier would do 2 on it. Toner was compatible, and as it was not on a maintenance contract ( was a lot cheaper to call out once a year if needed, plus we were not buying the consumables) it ran till it wore out.

[dexters_lab]

interesting story

it's a damd shame i dont have one of the coils, but the coils from this machine are compatible with the newer Magstim 200² (basically a 2nd generation unit dating from the 2000s) so wherever these came from i guess they saved them to use

[dexters_lab]

bah 

well i did make some progress, but not in the right direction

turns out i am pretty sure now those opamps are for the temperature sensing, while probing with my dmm between certain pins i could get the coil temp error leds to enable/disable but noting happened with the replace coil leds.

So headed over to a ULN2803 transistor array that turns the four main status leds on and off, traced the pin that turns on the replace coil leds and it's controlled from a PAL16P8ACN device, it's quite a simple affair 8 inputs and 8 outputs with programmable AND/INVERT logic between them.

So my thought is to pull it out and rig up a counter to the 8 inputs and just run through all the 256 combinations of inputs to find the right input state to unlock it. Hopefully this is the last piece of logic that controls the lockout and not just some poxy logic for the led indicators 

  where is my breadboard?

[dexters_lab]

 

well that seems to have worked, simply counted through all the logic possibilities (brute force style) on the inputs to the PAL device with a binary counter, compared with the Magstim and found i only needed to pull one pin high on the PAL device.

replace coil error gone

can arm it

HV comes on and charges the capacitor

just need to see what happens when i try and trigger it

[SeanB]

So the coil has a fusible link on it with a pull down.

[dexters_lab]

So the coil has a fusible link on it with a pull down.

i dont know, i gave up trying to figure out the analog side, as i mentioned in my other post i think the opamps are for the temperature sensor, so went for the logic instead which proved to be much easier

[dexters_lab]

video update

[dexters_lab]

well finally got this thing working, turned out i only needed to ground one of the coil plug pins

still no sign of charging to the 2800v i expected, so for the moment it's limited to a total of 1700v. It is very controllable though so i can go from about 0.25J to 275J in 1% steps but it is non-linear so the power goes up quicker as you increase the %, if i can figure out how to unlock the higher power mode i can get to 745J, that is if i am correct in believing it is capable of it.

i have some interesting ideas for future uses including combining with my air-gap flash unit for some interesting photography but for the moment i need to find something more interesting to blow up. I have tried can crushing but 275 Joules isn't enough 

[Andy Watson]

I have tried can crushing but 275 Joules isn't enough

If you are going down that route I would urge you to watch what happens to the coils on the can and coin crushing videos. One of the engineering challenges that faced Magstim was packaging the coils such that they did not self-destruct - violently. Be safe!

[dexters_lab]

thanks for the tip Andy, i am careful around these things 

sounds like you know a little about magstim?

[opusensemble]

Hi Dexter and thank you for the very interesting thread. If possible for you to measure, I would be interested in knowing:

1) The value of the 8 parallel resistors in the board in series with the transformer.
2) Examples of the currents on these resistors for different capacitor voltages (oscilloscope plot eventually)
3) Discharge pulse oscilloscope plot for a given measured coil inductance.

I'm not sure how to take the readings 3), and specially 2), due to the safety concerns regarding mains earth. 
A differential probe would probably be the only way I suppose? Or reading it through a HV probe using the oscilloscope disconnected from mains earth?

[dexters_lab]

Hi Dexter and thank you for the very interesting thread. If possible for you to measure, I would be interested in knowing:

1) The value of the 8 parallel resistors in the board in series with the transformer.
2) Examples of the currents on these resistors for different capacitor voltages (oscilloscope plot eventually)
3) Discharge pulse oscilloscope plot for a given measured coil inductance.

I'm not sure how to take the readings 3), and specially 2), due to the safety concerns regarding mains earth. 
A differential probe would probably be the only way I suppose? Or reading it through a HV probe using the oscilloscope disconnected from mains earth?

Hi there!

Point 1:
Can certainly do this, there are also additional resistors in the charging brick, there are some pictures on my blog but i will draw out a schematic for you. The charging circuit is very simple, just the transformer, rectifier and current limiting resistors. I am interested to know how often the transformer is switched on as it is switched by two SSRs.

Point 2
This is probably not necessary given the simple design, would imageine you can work out the currents from theory.

Point 3
I have been getting some traces using inductive coupling, simply placing the ground cable around the wire to the coil and back to the scope tip makes a simple single turn current transformer and means the scope is isolated. I have made a few coils but need an lcr meter to measure their inductance.

I am away until tomorrow so will follow up when i get a chance

[opusensemble]

Hi there!

Point 1:
Can certainly do this, there are also additional resistors in the charging brick, there are some pictures on my blog but i will draw out a schematic for you. The charging circuit is very simple, just the transformer, rectifier and current limiting resistors. I am interested to know how often the transformer is switched on as it is switched by two SSRs.

I am away until tomorrow so will follow up when i get a chance

Hi again Dexter.

It would be great that you could create a schematic for it.

I have just used tinycad (downloadable here http://sourceforge.net/projects/tinycad/) to create a brief sketch of what I understood of the circuit so far based on your videos.
I'm attaching the schematic file in case you want to edit your findings there.

In particular, I would like to have learn more information on the following aspects, in descending order of importance:

1) Grounding of the device.
Is the device connected to mains earth? If so, to which components does the ground line exactly connect inside the device? Only to the box chassis (in case it has a metallic chassis inside?). I am currently assuming that the ground line does not connect to either the transformer and the capacitor, (Kindly check this), as I'm guessing the secondary of transformer and all connected components should be floating, to avoid accidental closed loop through the patient.

2) The relative placement of the series diodes and the SCRs.
In my tentative schematic, I have placed them how I would expect them to be theoretically, but I'm not sure if this is correct or not. I have placed the diodes in anti-parallel with the SCRs as their function would be to conduct on the negative cycle of the LC tank, i.e, while the magnetic field collapses. (I have attached a drawing I made to describe the 4 quadrants on a LC tank). According to the specs of Magstim 200 the pulse is monophasic which would mean that the diodes would conduct on the second quadrant, reversing the voltage of the capacitor.
Theoretically it would make no sense to interrupt the current at the end of quadrant 1, considering the law of continuity of the magnetic field (guess it's related with Gauss' Integral Law of Magnetic Flux), so I assume the 2 IXYS diodes are actually placed to conduct the reversing current while the field collapses and not to block it. Kindly check how this is actually connected to confirm or refute this, reflecting any necessary corrections in the schematic. (checking the discharge current in the scope through your current transformer should shine a light on this too)
I believe Ben Krasnow used diodes in anti-parallel in his experiment, although I was never able to understand from his scope readings how exactly the current was extinguished.
Maybe in Magstim 200 they simply cut out the magnetic field while it's at its maximum, without allowing it to collapse?  (Now my big doubt here: could a magnetic field on a coil be extinguished without letting a reverse current to be established? This is currently a paradox to me)

3) The charging module.
If possible, try to replace the charging module (the black box) in my schematic with the elements you found inside and the connections between them.
You mention a rectifier in the charging circuit. Is that rectifier inside the charging module or is it apart not shown in your videos? I got very curious too on how that charging module is working as seems to be the tricky part (wondering if they have a patent on this part of PSU) You mention the SSRs are switching the transformer, so if that's the case, my tentative schematic will have to be quite modified. Please mess up with it as needed.

4) The trigger module.
This is the less important, but I'm guessing if they use pulse/isolation transformers there or other components (like opto coupler maybe? too sophisticated for a 90s machine maybe?), am wondering if it has other functions there, like reading the capacitor charge level, or other functions, but if you can figure out the connections and the components types (assuming that you won't be probably able to retrieve most of their values), would be a nice to have.

Sorry if I'm giving you too much work. Thanks so much for your help.

[dexters_lab]

1) Grounding of the device.
Is the device connected to mains earth? If so, to which components does the ground line exactly connect inside the device? Only to the box chassis (in case it has a metallic chassis inside?). I am currently assuming that the ground line does not connect to either the transformer and the capacitor, (Kindly check this), as I'm guessing the secondary of transformer and all connected components should be floating, to avoid accidental closed loop through the patient.

The unit is earthed, the case is all plastic the only metal parts that connect the inside to the outside are the trigger input/output and expansion connectors on the back which are well isolated from the HV side.

There is a star earthing point inside, which connects to mains earth. This connects to both the HV transformer case and to a shield connection on the low voltage toroidal transformer, earth also connects to the metal enclosure encasing the HV potted devices and to the external trigger & expansion ports.

The negative side of the capacitor is connected to mains earth through a 100 Ohm resistor. - I dont recall seeing this in my original teardown which was a much older unit.

2) The relative placement of the series diodes and the SCRs.

See somewhat simplified schematic, it should be correct!

I have included the diodes of the MDD95-16N1B module which connect across the capacitor through two parallel PCB modules with thick copper tracks located in the fan cooled section. Best i can measure these to is 0.15ohms each. You can see these PCBs at 23:00 in my original video. The consensus was these were resistive elements so i showed them as resistors on the schematic.

I have tried to be as complete as possible in some areas, like showing the two connections for the SCR gate drive and including the discharge circuit (which i have slightly simplified)

I have shown the capacitor as 190uF at 1700v because that is what i am measuring on my unit, though i do suspect it's rated much higher and mine is stuck in some low power mode. I have quotes from papers written by the magstim inventors stating the voltage should be 2800v and the coils needing at least 500J. My magstim is only outputting just over half of that. Also i have found a almost perfect match for the capacitor; the General Atomics 39504 pulse capacitor, the dimensions are spot on with the exception of the weight. Mine is showing 4.6kg but it's listed as 5.0kg
http://www.ga.com/series-dp-general-purpose-pulse-capacitors

3) The charging module.
If possible, try to replace the charging module (the black box) in my schematic with the elements you found inside and the connections between them.
You mention a rectifier in the charging circuit. Is that rectifier inside the charging module or is it apart not shown in your videos? I got very curious too on how that charging module is working as seems to be the tricky part (wondering if they have a patent on this part of PSU) You mention the SSRs are switching the transformer, so if that's the case, my tentative schematic will have to be quite modified. Please mess up with it as needed.

The charging module contains two SSRs (Z1 & Z2) that switch L & N to the HV transformer. One output of the transformer runs to the fan cooled area where it passes through 8 parallel power resistors (Shown as R1) making 5kohms. From there it then enters the charging module to meet with it's other connection from the HV transformer and passes into a bridge rectifier made up of 8 discrete diodes (D1-D8), each of the 4 legs of the rectifier then pass through 4 33ohm power resistors (R5-R7) which then make the HV DC output which then exits the module to pass onto the trigger module.

4) The trigger module.

This is the less important, but I'm guessing if they use pulse/isolation transformers there or other components (like opto coupler maybe? too sophisticated for a 90s machine maybe?), am wondering if it has other functions there, like reading the capacitor charge level, or other functions, but if you can figure out the connections and the components types (assuming that you won't be probably able to retrieve most of their values), would be a nice to have.

I only know a little about this part as i have not been able to depot it enough, given i now have a working unit i wasn't intending to look at it further but the details on the schematic are as much as i know.

there are some pictures of the charging and triggering modules on my blog:
http://dexterslab2013.blogspot.co.uk/2015/08/magstim-200-hv-charging-control.html


let me know if you spot any errors or things that dont make sense and i will clarify & correct

[dexters_lab]

additional:

two traces attached, i should qualify these by saying:

output coil is a 70mm coil i made, is 15 turns of 2.5sq/mm cable

pickup is simply attaching the ground clip to a regular 1X scope probe around one of the wires to the coil

magstim set to 95% (equates to about 1600v, or 243 Joules)

Would welcome any comments about other ways to capture the waveform... joeqsmith, i think you are the guy to speak to about HV scope probes!?

[opusensemble]

Thank you for your great work on the schematic Dexter.

I'm trying to make sense of it and these are some of the thoughts I'm having.
Probably some of the things on this list are self-evident but I'm including them here, for the case you have some inaccuracy in the schematic, and in case you don't, if you can comment on why it is done that way.

1) Ground connects to the HV transformer case. Does the transformer have two terminals on the secondary or just an HV terminal like this http://www.hvtesla.com/img/tesla%20coil%20mot.jpg ? In case it has only one terminal the other HV terminal would be the transformer case, right?

2) The 100 Ohm connecting the C negative terminal to ground is quite intriguing. My first thoughts is that it is hazardous as it seems to facilitate this scenario during discharge http://sub.allaboutcircuits.com/images/00056.png . Or maybe it prevents it since it also connects the coil to ground?
I've seen a TMS circuit example where they do that as well (http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3697055/figure/F3/ ), but am wondering which case is the safest.
In your older Magstim have you noticed if the negative of the capacitor was connecting directly to ground without the 100 Ohm resistor?

I think this issue is really critical and wanted to dig further on it with you if possible. In this reference https://books.google.pt/books?id=IgTSBQAAQBAJ it's mentioned that not only the coil but the entire circuit of the stimulator should have a floating voltage isolated from earth ground: https://drive.google.com/file/d/0B_pa-g27qiWIbWptZ1VEZWczanM/view?usp=sharing

In this reference https://books.google.pt/books?id=KRImUIVWLCUC the same is said https://drive.google.com/file/d/0B_pa-g27qiWIZmJ1YlA3UXNSOW8/view?usp=sharing, where they refer that noise artifacts may need to be bypassed to ground with a big resistor or small capacitor, which is nothing what we have there in Magstim. Are you sure the resistor is 100 Ohm only?

3) Is the switching of the primary through SSRs probably a way of avoiding the use of more expensive devices (IGBTs) on the secondary? What is your take on that?

4) What on earth do you think the 33 Ohm resistors are for if there's a 5k Ohm in series with it? Seems a bit irrelevant but quite puzzling.

5) What is really intriguing is the discharge curve. I could never understand why it behaves like that if they mention that the curve is monophasic, which means that the diodes should be blocking the collapsing current of the field, but I don't see how that can happen in that schematic. The curve looks a damped biphasic curve and is similar to Ben Krasnow's which I believe used diodes in anti-parallel with the SCR (image attached), as mentioned in the literature, which explains the negative current.
But in this schematic are you sure the diodes are not across the coil instead of across the capacitor? If it's like what you have in that schematic I really have no explanation for the negative current on the pulse.
I've attached a diagram trying to depict this. Maybe there's something wrong in the schematic here?

6) What are you measuring in the rising curve in your second trace?

[opusensemble]

I'm speculating the following, regarding the possible coil currents curves, in the attached image:

Circuit A generates a biphasic curve
Circuit B generates a monophasic curve (with damped negative current)
Circuit C (in your previous schematic).. I can't understand how this works. It would discharge the capacitor, and somehow an overshoot current bypass the blockage of the SCRs below 0A ? Seems a bit impossible to me, and in this case I don't see the purpose of the diodes... but maybe I'm interpreting this circuit wrongly and it works through a different mechanism? Really puzzled about this.