Embedded Files
A Monophasic TMS system contains a capacitor which gets charged by a high voltage power supply; then a solid state switch discharges the capacitor into an applicator coil, thus producing a pulsed magnetic field. In early TMS systems, stored energy was dissipated after each pulse.
Biphasic TMS systems function similarly. Capacitive-discharge drives an applicator coil, thus producing a pulsed magnetic field. But a Biphasic design reclaims stored energy, pulse by pulse, by resonant action. In a resonant circuit, forward-polarity pulses are followed by reverse-polarity pulses. Commercial TMS designs tend to deliver both a forward and reverse polarity pulse, within a single sine wave cycle.
Differing from clinical offerings, my reverse-polarity pulse appears slightly delayed (by 1.4 milliseconds). Neglecting this difference, my TMS design might be loosely described as Biphasic. I don't offer power level control: stored energy here ranks low compared to existing TMS systems. No attempt is made to identify Muscle Twitch Threshold, which is a popular power level setting in clinical TMS therapy. Lacking the more popular Butterfly Applicator Coil, broader magnetic field is projected, producing a circular pattern of stimulation. These differences might prove harmful.
In clinical TMS therapy, seizures occur in about .01% of cases. People with metal plates in their skulls, or histories of Epilepsy, cannot be treated. In the clinical setting, TMS technicians are certified in Basic Life Support, which includes CPR. This training can be had for about $100 dollars (check your local Red Cross for class availability). The FDA has only approved TMS for treating untreatable depression. Be forewarned: TMS should only be used as a last resort, and only out of total desperation. Serious side effects may be permanent.
Below we have test data for TMS Unit #7.
High voltage section, theory of operation. (Please refer to the right hand side of the Pulse Driver Board schematic.) Each pulse sequence begins with a 17uF capacitor bank at full voltage. Primary SCR firing dumps stored energy into the applicator coil. One concern arises: High Voltage (HV) Transformer positive output looks like it should be shorted by Primary SCR conduction. This is not so. Primary SCR firing is timed to coincide with negative HV Transformer output. Cap bank voltage also ends up negative during this period, due to resonant action. HV Transformer output voltage remains negative throughout this period (even more negative than cap bank voltage). So the HV Transformer secondary remains unloaded while it is negative, and the high voltage rectifier remains back-biased. Cap bank voltage stays negative for 1.4 milliseconds. Then the Ringback SCR fires. Negative capacitor voltage discharges into the applicator coil. About 100 microseconds later, due to resonant action, the cap bank ends up recharged to 82% of the initial positive voltage. HV Transformer polarity swings positive again. Half-wave trickle-charging continues, cycle by cycle, in preparation for another pulse. SCR switching continues when high voltage turns off. This quickly dissipates stored energy. Caution: in the event of a power failure, stored energy takes a long time to bleed down. Neon lamps are provided to indicate residual voltage. Use an insulated shorting stick to discharge capacitors before servicing.
PULSE DRIVER BD 1037-SH.pdfNot shown in the Explanation of Terms: subsequent capacitor recharging takes a stairstep path, with each primary line cycle boosting capacitor voltage a little more. Capacitor voltage reaches the highest final value with pulse-repetition-rate set for 10 pulses per second (PPS). In this mode six line-cycles recharge the capacitor bank prior to each discharge. Worst-case transient-SCR-holdoff-voltage appears immediately after Primary-SCR-Conduction ends, just as Negative Overshoot first appears.
We also check SCR midpoint voltage, to verify all SCRs are sharing transient holdoff voltage in a balanced manner. Each SCR is rated 1600 volts. Ideally actual transient SCR holdoff voltages would be measured across each device. In practice, only two of four high voltage SCRs are connected to the Driver Deck chassis, where the oscilloscope can be referenced. I define these as the "Lower SCRs." Transient holdoff voltage across the Lower SCRs can be measured directly. We can't measure "Upper SCR" transient holdoff voltage directly. We can subtract Lower SCR holdoff voltage from total SCR holdoff voltage to get this value. We're most interested in peak voltage when Negative Overshoot first appears, with the lowest pulse repetition rate. This presents our worst case scenario.
Looking at our 10-Pulse-Per-Second Test Data: with 127 volts applied to the primary line, we see excessive voltage appearing across the Lower SCRs. Transient Negative Overshoot reaches -3230 volts. This exceeds 3200 volts, which is our combined series-SCR voltage rating. SCR Midpoint Negative Overshoot reaches -1638 volts, exceeding Lower SCR voltage ratings by 38 volts. By subtraction we infer upper SCR transient holdoff voltage reaches -1592 volts, which does fall within the device ratings. SCR manufacturers provide a 50 volt margin of safety. But oscilloscopes are usually only rated 3% accuracy, and only if calibration is current. My calibration is not current. Actual voltages remain unknown. Device survival becomes the real test.
Microwave oven transformers are Inductively Shunted. This limits line current, which makes them good for capacitor charging duty. Transformer characteristics vary considerably. Output impedance affects Peak Capacitor Voltage. Lower impedance produces higher peak voltage. I keep a random selection of high voltage transformers available, which can be helpful for optimizing peak cap bank voltage. GE Spacemaker microwave oven transformers aren't suitable for TMS capacitor charging duty (impedance is too low).
Net capacitor bank value also affects Peak Capacitor Voltage. Increasing net capacitance reduces peak voltage. If peak voltage gets too high, ~1 microfarad high-voltage capacitors may be added to the cap bank, thus increasing net capacitance. (Such capacitors are found in microwave ovens.)
A bench-model Variable Voltage Transformer (VVT) is used to test the Pulse Driver Board. A typical VVT can deliver up to 132 volts AC. I like to test each Pulse Driver Board up to 150 VAC, which requires a small step-up transformer. A bench model VVT can also be used for preliminary high voltage section checkout, but only for applying relatively low primary voltage. A typical VVT won't deliver full-power line-current (<44 amps peak). High peak current is hard on a VVT wiper, and the typical VVT's boost-tap won't uphold 127 volts under such heavy loading.
Safety Considerations:
The Pulse Driver Deck baseplate shall be connected to Earth Ground for initial high voltage testing. This ground-reference shall be provided through a suitable line cord, via a properly grounded power receptacle. The Oscilloscope shall be Grounded via the same power receptacle. High-voltage compensated divider grounding shall be shall be done via the oscilloscope power cord ground. When testing is complete, the Pulse Driver Deck baseplate will be permanently disconnected from earth ground, leaving it completely isolated. This is done prior to the Hipot Test. The unit should always be unplugged for servicing. Neon indicators might flag presence of high voltage. But to be sure, cap bank capacitors should be discharged using a screwdriver.
Initial High Voltage (HV) testing begins with the pulse repetition rate set for 15 PPS. HV transformer input voltage should be reduced. (This might require the use of a long extension cord, which acts like a power resistor in series with primary line voltage.) The 15 PPS repetition rate should be audible, even with minimal applied voltage. If possible, input voltage should be increased gradually, while viewing SCR holdoff-voltage (and if possible, peak output current) using an oscilloscope. We also monitor high voltage transformer primary line voltage and current using suitable meters. Ideally we gradually raise HV transformer primary voltage to 127 volts. Peak SCR holdoff voltage is viewed using a high voltage probe or high voltage 1000:1 compensated divider. Peak output current is viewed using a 1000-amps-per-volt current transformer. Voltage and current measurements shall be logged on a test data sheet (like the one provided above). Primary line voltage is reduced to minimum before switching pulse repetition rates. Data is also taken for 12 PPS and 10 PPS. Care should be taken to avoid applying excessive voltage to the SCRs when lower repetition rates are selected.
Proper high voltage transformer phasing can be verified with low applied line voltage. This is done by viewing the high voltage transformer output using an oscilloscope (right before the high voltage rectifier). Sweep triggering can be done using the current transformer output. A ragged dissymmetrical sine wave will be seen, and positioning of the pulse sequence within the sine wave can be observed. The pulse sequence should appear with the sine wave trending downward. High voltage transformer output waveform should drop from a positive peak to a flat-spot about 100 microseconds wide, which will end in some ringing, then the sine wave continues its downward trend. For incorrect phasing the pulse sequence ends with transformer-output being positive, trending gradually upwards for the remainder of the half cycle.
Another way exists for verifying proper high voltage transformer primary phasing: full power operation can be tested with both primary phasing modes. Transformer primary current should be about 2 amps lower with correct phasing. Counterintuitively, Final Capacitor Voltage may appear slightly higher for incorrect phasing.
Ideally we would vary line-input-voltage throughout the full range, up to and including +10% of nominal. To do this I insert a 30 volt, 1000 watt, boost transformer in series with the primary line, thus producing 135 volts. I feed this into a 240 Volt 30 amp VVT. More normally, only a 10 amp VVT will be available. This could be used to feed a 30-volt boost-transformer primary. Boost transformer primary phasing can be selected for additive (Boost) or subtractive (Buck) duty. Such reversal becomes useful for performing high and low-line-voltage testing. High power test data is taken for 103 VAC, 115 VAC, and 127 VAC input line voltages.
It's always good to discuss weak points in any design. I'm taking a risk by using 660 VAC "motor run" capacitors in this capacitor bank. This might represent the most economical solution—if it proves reliable (and it has, so far). I suspect these might prove adequate, based on the following logic: 660 VAC capacitors are generally rated for 1500 volts DC (though these particular capacitors aren't labeled with a DC rating). In our worst case condition, delivering 10 pulses per second, with primary line voltage set for 127 volts, the capacitor-bank sees 2760 volts peak. This appears to fall within a 3000 volt DC capacitor-bank rating. Why do I consider this risky? 660 volts RMS translates to a nominal value of 1867 volts peak to peak. With two capacitors in series, our nominal peak-to-peak voltage rating becomes 3734 volts. Our test data shows Peak Capacitor Voltage reaching +2760 volts, with Negative Peak Capacitor Voltage hitting -2265 volts. The difference between these excursions gives us 5025 volts peak to peak, far exceeding a 3734 volt nominal peak to peak voltage rating. Will corona destroy internal dielectric materials? Oil impregnation mitigates corona, and fortunately power-line duty is notoriously harsh. Hopefully unspecified manufacturing safety margins will prove sufficient for indefinite capacitor lifespan. Bearing this in mind, we do need to treat these capacitors as nicely as possible. Capacitor cases are necessarily isolated and left floating. These capacitors run a little bit warm, so minor airflow is directed at them.
High voltage-transformer phasing should be set so pulse-delivery occurs while transformer output is negative. Primary line current appears about 2 amps higher if phasing is reversed. This is true because negative capacitor polarity appears across the transformer output while it attempts to go positive, presenting a short circuit condition. This short circuit condition only persists for 1.4 milliseconds, and it only happens after every few line cycles. High transformer leakage inductance limits this peak current. Short term primary phase reversal causes no harm, but it should be avoided anyway.
Pulse Driver Board Theory Of Operation
115 volts AC enters via R1 and CR1, which charges C1 and C2 to roughly 130 VDC. This DC provides basis for push pull output, switched by SCR1 and SCR2. R4 powers LED1, giving a green indication whenever primary line voltage is present. 115 VAC also enters via R2, which develops roughly square-wave 102-volts peak-to-peak, across VZ1 and VZ2. This couples capacitively through C3, C4, C5, and C22; together these form a charge-pump, which begins charging C8, C9, and C6 in a stairstep fashion, synchronous with the line voltage. Charge pumping here is deliberately out of phase with high voltage transformer output. Voltages across C8, C9, and C6 can only "bump up" while high voltage transformer output is negative. C12, T1, and T2 primaries provide an AC ground reference for C8 and C9. After 4, 5, or 6 power-line cycles (selectable at SW2), C8 and C9 charge up to about 83 volts; enough to drive VZ3 and VZ4 into conduction. VZ4 is a Diac, which discharges C6 into the gate of SCR1, triggering the Output Rail to switch HIGH. This discharges C6, C8, and C9 via VZ3, CR2, and SCR1 in preparation for the next timing cycle. With the Output Rail HIGH, R3 applies 75 volts to VZ6, which begins charging C7 via R5. After about 1.4 milliseconds, the voltage on C7 exceeds 32 volts: Diac VZ5 conducts, triggering SCR2 to fire, switching the output rail LOW. CR7 and SCR2 discharge C7, in preparation for the next timing cycle.
Resistors R6 and R7 limit peak SCR gate-drive current. R8 and R9 load both sensitive SCR gates (to prevent spurious firing). R10, R11, C10, and C11 limit the voltage rate of rise (dV/dT), seen across SCR1 and SCR2 (also to prevent spurious firing).
When the high power pulse sequence operates, T1 will be seen charging the high voltage capacitor bank to a value ranging from 2100V to 2700V (final capacitor voltage depends on pulse repetition rate and line voltage). Pulse Sequences, which are also called Pulse Trains, last for about four seconds, followed by an eleven-second cooldown period. See the Power Control Timer Board writeup for a thorough discussion of this function.
Each time the Pulse Driver Board Output Rail switches HIGH, C12 delivers a 25 microsecond pulse to T3 primary through CR6 and R26, triggering the Primary high voltage SCRs to fire. Primary SCR conduction dumps the high-voltage cap-bank charge into the applicator coil. Field strength and peak current reach maximum as the cap-bank-charge depletes completely (at about the 50-microsecond mark). Collapsing magnetic field then reverses coil polarity, thus maintaining current for another 50 microseconds; ongoing SCR conduction pulls cap-bank polarity negative, reaching about 90% of the initial absolute value.
Primary SCR conduction doesn't end instantaneously. Sluggish SCR recovery time allows reverse current to flow, then ceasing rather abruptly. The applicator coil builds up some field strength due to this unwanted reverse-current. Rapid current interruption causes a spike in coil voltage. Our biggest concern is resulting Negative Overshoot. If left unloaded, this would apply excessive voltage to the SCRs. Snubbing components R24, R25, C15, and C16 limit this peak voltage to an acceptable level. There may also be dissimilarities in the recovery times of series-connected SCRs. Snubbing components are positioned across individual SCRs. So regardless of when an SCR becomes nonconductive, voltage excursion across the device will be limited. Dissimilar leakage currents might also be an issue. As configured, the snubbing network also forms a low-impedance AC voltage divider, which appears especially necessary to keep voltages balanced across both Primary SCRs, due to a 100 millisecond inter-pulse period. Each SCR is connected antiparallel to another SCR, so one pair of snubbers protects all four SCRs.
1.4 milliseconds after the initial pulse, the Pulse Driver Board Output Rail switches LOW. C12 now drives T2 primary through CR7 and R26, triggering Ringback high-voltage-SCRs into conduction. Negative cap-bank-charge gets dumped into the applicator coil. This charge gets depleted after 50 microseconds, when current and field strength reach maximum. Field collapse drives ongoing current, accompanied by coil polarity-reversal. In the following 50 microseconds, ongoing Ringback-SCR-conduction drives cap-bank polarity positive again, reaching about 83% of the initial positive voltage. Sluggish Ringback-SCR-recovery causes minor positive overshoot, which is also loaded to acceptable levels by snubbing components R24, R25, C15, and C16. This overshoot isn't as perilous as Negative Overshoot, seen immediately following Primary SCR conduction. This is because the absolute value of Final Capacitor Voltage is somewhat lower than Negative Peak Capacitor Voltage following Primary SCR conduction.
Output section switching is maintained continuously, with or without high voltage applied. This dissipates stored energy promptly upon deactivation. This means Pulse Driver Board power comes directly from the line cord, and is not switched. Bleeder resistors also discharge high voltage capacitors, but self-discharge times would be long. It should be noted: prompt discharge won't happen when primary power gets interrupted in the middle of a pulse train. This leaves the cap bank charged, and bleeding down very slowly. Neon bulbs on the Switch Deck and Capacitor Bank may indicate lingering charge, but proper neon-lamp function cannot be assured. Always deenergize the unit using the power switch. Wait a few minutes before servicing. And discharge both high voltage capacitors with a screwdriver before proceeding.
Pulse Driver Board
Lower copper loop is the Neutral Test Point. The middle copper loop is the Output Rail Test Point.
System Indicators:
Pulse Driver board: green LED indicates line voltage is present.
Power Cycle Timer Board: red LED indicates HV ON timer output. Yellow LED indicates airflow is okay, High Voltage ON drive is provided to transformer control Triac.
High Voltage Transformer: incandescent lamp indicates high voltage AC is present.
Switch Deck: tubular Neon Lamp, indicates presence of high voltage, displays pulse action, and shows transient polarity of capacitor-bank charge.
Capacitor Bank: NE-2 Neon Lamps, indicate presence of voltage and balanced voltage distribution across series-connected capacitors.
The schematic below shows all system parts. This is useful when connecting the power devices (it's a mix between a schematic and a wiring diagram).
WETVAC TMS SYSTEM SCHEMATIC 1035SD.pdfPOWER CONTROL TIMER BD 1034-SH.pdfPower Control Timer Board, Theory of Operation:
When power is switched ON, the blower begins receiving roughly half-power through a diode in Bridge Rectifier BR1. Power supply PS1 is also activated, and it begins producing 5 volts.
Power supply PS1 is a modified cell-phone charger. It was disassembled and the power-input pins were removed. Primary power wires were soldered onto the printed circuit board. An existing "doghouse-door cutout" accommodated all input and output wires. While the case was open, the positive output of the power supply was tied to the primary-power system-Neutral, using a short jumper. A Ty Wrap secures all wires firmly. The case was reassembled. The power supply assembly was strapped to the upper mounting plate.
Upon initial power-up, NE555 timer (U4) gets triggered immediately, because C13 begins fully discharged, which holds Trigger-Input pin-2 momentarily below the trigger threshold. (NE555 Trigger threshold voltage is set to 1/3 of the DC supply voltage internally, which is about 4 volts in this case.) Once triggered, timer output goes HIGH, with PIN 3 driving Q2 into conduction. Q2 delivers -5 volts through R29 to the gate of TR2, thus applying full AC power to the blower. The outward appearance is that the blower gets full power immediately. But there are three steps happening in rapid succession: half power is applied through BR1, power supply PS1 activates, and timer U4 triggers: full power is applied to Blower B1. From then on, full blower power is applied whenever the timer output goes HIGH. Whenever timer output goes LOW, Blower-Control-Triac Gate-drive is lost, so the blower runs at about half power, via the diode in BR1. Current through R28 illuminates RED LED2 whenever the blower is receiving full power.
Bridge Rectifier BR1 is a 3000 volt 10 amp rated assembly. This was chosen for having high voltage isolation between the internal diode(s) and baseplate, which is necessary to confidently isolate the primary line terminals from the Driver Deck Ground. Any 10 amp diode would suffice. But care should be taken to isolate the heatsink from the Driver Deck Chassis for more than 5500 volts peak to peak.
Blower B1 pulls a partial vacuum in the bottom chamber of the TMS unit, thus operating Vacuum-Switch S3 to close its contacts. This delivers +5 volts to the Gate of Q1, which is also lightly loaded by R34. With the timer output LOW, and with the vacuum-switch contacts closed, high voltage will be activated. Or if no Vacuum is sensed, R34 pulls the Gate of Q1 low, thus preventing application of High Voltage.
With Timer U4 triggered, R12 and R13 will begin charging C13. Charging proceeds until Threshold Voltage is reached at PIN-6 (NE555 threshold-voltage is set to 2/3 of the DC supply voltage internally, or about 8 volts in this case). Charging requires about 20 seconds following initial powerup. During this time, high voltage will be inhibited. Once threshold voltage is reached at PIN-6, Timer-Output Pin-3 switches LOW, thus sending a High Voltage enable output. The timer's Discharge-Output Pin-7 also switches LOW, which begins discharging C13 via R13. C13 voltage will drop until Pin-2 falls below the Trigger Threshold voltage, which takes about 4 seconds. Timer-output will again switch HIGH. Discharge-terminal Pin-7 becomes high impedance, which allows C13 to begin charging again. This cycle repeats indefinitely.
R12, R13, and C13 are selected to provide an eleven-second charging period and a 4 second discharge period. R12 may need to be hand selected, as apparently batch variations can be significant among 4.7 megaohm resistors. Many resistors tested more than 20% out of spec, despite being rated 5% tolerance. A tantalum timing capacitor was chosen for temperature stability. Nameplate tolerance isn't assured. Some variation should be anticipated. In another unit, an additional .22uF trim capacitor was added to achieve the required 4-second ON time.
Whenever U4 output, Pin-3, switches LOW, and negative 5 volts will be available for driving the Gate of high voltage control Triac TR2. Assuming there is vacuum sensed by SW3, its contacts will be closed, and +5 volts will be applied to the Gate of Q1 (with respect to the Source terminal), which turns ON Q1, thus delivering negative NE555 output-current to the gate of TR2 via R15. Thus high-voltage is activated.
One might wonder why the positive side of the 5 volt power supply is tied to 115-VAC-System-Neutral. A Triac Application Note suggested negative polarity should be used for driving Triac Gates, which avoids using the least stable quadrant of Triac operation.
My original intent was that the vacuum cleaner blower motor shouldn't run while the unit is pulsing. My reasoning was, both loads are significant, and excessive line current might have resulted with both operating simultaneously. But strobing the vacuum cleaner motor ON and OFF seemed obnoxious, inrush current was high, and intermittent operation might have diminished motor lifespan. High voltage section line current was found to be modest. So a diode was placed across the blower-motor-control Triac. The blower motor runs at about half power while the unit is pulsing. Line current isn't excessive. The unit sounds like a regular vacuum cleaner, being used intermittently.
WETVAC TMS WIRING DIAGRAM 1036WD.pdfShown above: Wiring Diagram
This system is "Double Insulated," a term normally reserved for handheld power tools. All exposed surfaces are plastic. Internal metallic surfaces are isolated from earth ground. Only the primary power circuit presents risk of electrocution, and this is isolated from all secondary circuitry with 7 millimeters of tracking distance. The Driver Deck baseplate is tied to the high voltage transformer secondary, but not to earth ground. All primary line voltage wiring is carefully isolated from the pulse driver deck baseplate. In the test phase, we verify primary power isolation holds-off full pulse voltage (exceeding 5KV peak to peak), for one minute. (This is called a Hipot test). With the Driver Deck properly insulated, any point in the high voltage section could contact earth-ground, and harmless current would flow.
Ideally the end user would sit on an isolated stool, placed on a wood floor, wearing rubber shoes, while also avoiding contact with all high voltage points within the system. In terms of safety, the applicator coil presents the weak spot: a user could conceivably come into contact with high voltage there, but tampering would be required for this to happen. (Fingers should not be allowed to enter the airflow holes.)
Disclaimer: my test data may contain errors. My equipment is out of calibration. Looking at the test data, I see a blatant anomaly, in the 10 Pulses Per Second high power pulse checkout section: high-voltage-transformer primary-current appears higher when operating at 115 VAC than it does when operating at 120 VAC (this represents a very unlikely outcome). The unit is gone. There is no way to retake that data.
Page updated
Google Sites
Report abuse