DC Sputtering Source Reference and Guide
DC Sputtering Power Supply (100 W): Requirements, Constraints and Build Plan
Status: draft v0.1, 2026-11-13. Owner: Alacrity R&D. Partner: Hacker Fab (Waterloo / CMU).
Scope: a 100 W DC magnetron sputtering supply for the Hacker Fab DC sputtering machine, powered only from a Mean Well UHP-200-24.
Warning: the output of this supply is lethal (up to 1.5 kV open circuit, up to 0.4 A in operation). This page is an engineering specification, not a safety approval. Nothing here replaces a review by a qualified person before first power-up.
How to read this page
- Reference keys: H Hacker Fab documentation, P papers and conference material, E equipment datasheets, X patents, S standards. The full list is at the end.
- Levels: MUST = required for v1. SHOULD = design goal, may slip with a written reason. MAY = optional hook for later.
- "(derived)" marks a number we calculated from the cited sources instead of reading it from them.
- The standards are paywalled. Clause numbers and limits come from secondary sources and from the 2010 edition of IEC 61010-1, and must be re-checked against the real text before the design review.
1. Starting point
1.1 The Waterloo machine today H1
- Source: planar magnetron, target just over 30 mm, unbalanced, mild steel pole piece, ground shield with a lip over the target edge. No cooling yet.
- Chamber: borosilicate bell jar on an aluminium baseplate, home-made feedthroughs (threaded rod, PTFE spacer, Viton gasket), roughing pump only (according to wiki, in reality, during out DIscord meet they mentioned they got their hands on a turbomolecular one). Base pressure 5 to 10 mTorr.
- Process: stable plasma from 10 to 200 mTorr, aluminium sputtered at 200 to 300 mTorr of argon, copper films partly oxidised.
- Power: about 20W sustained, because the magnets overheat. One magnet set was demagnetised while sputtering aluminium.
- Supply: a second-hand electrophoresis supply (Fisher Biotech FB600). Its ground-leak and arc-fault protections made it very hard to strike and hold a plasma. Their conclusion led to this collab, namely to build a dedicated supply and cover the missing protections with other engineering controls.
- Planned: turbopump (Edwards EXT70H), mass flow controllers, a cooled source, RF in the long term.
1.2 Why an electrophoresis supply fails on a plasma
Electrophoresis supplies are sold with no-load detection, sudden-load-change detection, ground-leak detection and arc detection as safety features E9. Each one is triggered by normal plasma behaviour.
| Protection in the electrophoresis supply |
Why a sputtering plasma trips it |
What this supply does instead |
| No-load detection |
Before ignition the chamber is an open circuit P1 |
Holds the strike voltage into an open circuit, with only a configurable ignition timeout (IGN-2) |
| Sudden load change detection |
Ignition takes the load from open circuit to a few kOhms in ms, an arc takes it to near short circuit P8 |
Current-source output, soft strike, controlled ramp (IGN-1, IGN-4, IGN-5) |
| Ground-leak detection |
The anode is the earthed chamber, so the return current flows in earthed metal by design P1 |
Positive output bonded to protective earth inside the supply. Safety comes from enclosure, interlocks and earthing (OUT-1, SAF-7, SAF-8) |
| Arc detection with latched shutdown |
Micro-arcs are a normal part of sputtering, above all on new or oxidised targets P8 |
Quench in microseconds, wait, recover automatically, count. Latch only on persistent arcing (ARC-1 to ARC-7) |
2. The load: what the literature says the supply has to drive
2.1 Operating window
| Quantity |
DC diode glow (no magnets) P1 |
DC magnetron, typical [P1]P3 |
Hacker Fab today H1 |
Commercial 1 to 1.3 inch sources [E6][E7]E8 |
| Argon pressure |
2 to 4 Pa (15 to 30 mTorr) |
0.1 to 1.5 Pa (0.75 to 11 mTorr) |
10 to 300 mTorr |
0.5 to 600 mTorr (MAK 1.3), 2 to 50 mTorr (ONYX-1), about 1 mTorr to 1 Torr (MTI 1 inch) |
| Cathode voltage |
2 to 5 kV |
300 to 700 V |
not documented yet |
200 to 1000 V (MAK, MTI), 100 to 1000 V (ONYX-1) |
| Current |
about 1 mA/cm² |
4 to 60 mA/cm² |
not documented yet |
up to 1 A (ONYX-1), 2 A (MAK), 3 A (MTI) |
| Power |
not given |
several tens of W/cm² |
about 20 W, 2.8 W/cm² (derived) |
75 W indirect cooled (ONYX-1), 250 W (MTI), 350 W (MAK), all water cooled |
| Magnetic field at target |
none |
about 20 to 50 mT |
not documented yet |
NdFeB magnets |
Supplier guidance for target loading is about 16 W/cm² with indirect cooling and about 39 W/cm² with direct water cooling E7.
2.2 Electrical behaviour that shapes the design
- The load is not a resistor. A DC magnetron follows roughly
I = k * V^n with n typically 4 to 10 P4 (5 to 15 in another report P6). The exponent reflects how well the electrons are confined P4, and the curve moves with gas heating and rarefaction in front of the target P5. Consequence (derived): at 400 V and 0.25 A the small-signal resistance is V / (n * I), about 160 to 400 Ω, while V/I is 1.6 kΩ. A stiff voltage source cannot set the operating point, and 2% of voltage ripple becomes 8 to 20% of current ripple. In short it kind of behaves like an oversized Zener.
- The voltage depends on the target. The minimum sustaining voltage scales as
V_min = Ec / (beta * gamma_eff), with Ec about 20V per ion pair in argon and gamma_eff the effective secondary electron yield of the target surface [P1]P2. Measured 1/V increases with the secondary electron yield of the target material P7. Copper, aluminium and an oxidised aluminium surface therefore run at different voltages for the same power, and the supply needs a wide full-power voltage range.
- Ignition needs much more voltage than running. Breakdown follows the Paschen relation
VB = B*p*d / (ln(A*p*d) - ln(ln(1 + 1/gamma))). With the argon constants of P1 (A = 0.0862 per cm Pa, B = 1.32 V per cm Pa) and gamma = 0.1, the minimum is about 100 V at p*d of about 76 Pa cm (0.57 Torr cm), and the expression has no solution below about 28 Pa cm (0.21 Torr cm) (derived). For a 5 cm path that boundary sits near 42 mTorr. Today, at 50 to 300 mTorr, ignition is easy. After the turbopump, at 1 to 10 mTorr, the gap is far to the left of the minimum: ignition then relies on the magnetic trap and on the open-circuit voltage available, which is why commercial magnetron drives provide 800 to 1500 V for striking E3. The constants are fitted for E/p between 0.75 and 4.5 V per cm Pa only, so treat the left branch as a trend, not as a number.
- Arcs are a second discharge mode, not a fault of the supply. A magnetron plasma is a magnetically enhanced glow: electrons are released one ion impact at a time, so it needs several hundred volts. An arc emits electrons collectively from micrometre-size cathode spots. The voltage collapses to less than 100 V while the current is limited only by the supply. Spot ignition takes less than a microsecond and a spot lives for about a microsecond P8.
- What starts arcs. Dielectric inclusions and oxide layers charged by ion bombardment, mostly at the edge of the racetrack. Aluminium ignites arcs far more easily than a refractory metal such as tantalum, and a new target carries an oxide skin that has to be sputtered away first [P8]P22. This matters here: aluminium is one of the two Hacker Fab materials.
- What arcs cost. Each arc throws droplets of target material. Droplet size grows with the energy fed into the arc, droplet count with the arc rate [P9]P10. The supply controls the first directly and the second indirectly.
2.3 What 100 W means on a 30 mm target (derived)
- Target area 7.1 cm². 100 W is 14 W/cm², at the low end of the power densities typical of magnetrons P1 and just under the 16 W/cm² indirect-cooling guideline of E7 (113 W for this target). 100 W is therefore a sensible ceiling for a cooled 30 mm source, and five times what the uncooled source survives today H1.
- Electrical corners at 100 W: 250 V and 0.40 A (625 Ω) to 800 V and 0.125 A (6.4 kΩ). Current density 18 to 57 mA/cm² averaged over the target face, inside the 4 to 60 mA/cm² range of P1.
- The low-voltage, high-current corner matters today (high pressure). The high-voltage, low-current corner and the strike voltage matter after the turbopump.
3. What professional supplies do
|
AE MDX 500 E2 |
AE MDX 1 kW / 1.5 kW E3 |
AE Pinnacle family E4 |
TRUMPF TruPlasma DC 3000 (G2) E5 |
This supply (target) |
| Rated power |
500 W |
1 kW / 1.5 kW |
3 kW and up |
10 to 300 kW |
100 W |
| Run voltage |
0 to 600 V at 1 A, or 0 to 1200 V at 0.5 A |
1000 or 1300 V / 500 or 750 V |
370 to 1000 V at full power (3000), 325 to 800 V loads (Plus+) |
1000 V, full power over a 1:4 impedance range |
250 to 800 V at full power, 1000 V max |
| Strike voltage |
1200 V range |
1500 V "soft" (standard Z), 800 V (low Z); 800 or 1200 V on the 1.5 kW |
programmable strike voltage limit |
n/s |
600 to 1200 V programmable |
| Regulation |
power, current, voltage |
power, current, voltage |
power, current, voltage |
n/s |
power, current, voltage |
| Ripple |
2% p-p at 100 kHz, 1% p-p line |
5% at 50 kHz |
line-related below 2% RMS (Plus+), 5% RMS (3000) |
n/s |
5% p-p |
| Accuracy |
display within 2% of reading or 0.2% of full scale |
regulation 0.2% |
repeatability 0.1% |
0.5%, repeatability 0.2% (Series 4000) |
2% of reading or 0.5% of full scale |
| Energy into an arc |
"orders of magnitude" below older designs, no figure |
low stored energy, no figure |
below 1 mJ per kW (series), below 6 mJ per kW (3000) |
below 0.3 mJ per kW |
5 mJ absolute max, goal 1 mJ |
| Arc detection |
senses a change in current (Arc-Out) |
Arc-Out |
0.5 to 3 µs (series), 3 µs (3000), below 25 µs (Plus+) |
below 100 ns, adjustable U and I thresholds |
2 µs |
| Arc interruption |
n/s |
adjustable suppression time |
10 µs (3000), below 200 µs (Plus+); trip level, delay and shutdown time programmable |
n/s |
converter off within 10 µs, off-time 0.1 to 100 ms |
| Ramp |
0.1, 1 or 10 s |
1 to 10 s or 1 to 10 min |
ramp and run recipes |
n/s |
0.1 s to 10 min |
| Interlocks and remote |
25-pin analog/digital port, 0 to 10 V, three interlock pins (vacuum, water, auxiliary) |
front panel and user port |
AE Bus serial, Profibus, DeviceNet option |
n/s |
hardware interlock loop, isolated serial |
| Output connection |
UHF (N or SHV optional), RG-8 coax |
UHF or terminal block |
n/s |
n/s |
SHV, RG-59 coax |
| Standards claimed |
EN 50081-2, EN 50082-2, EN 50178 |
n/s |
IEC 61010-1 and IEC 61326-1 (Pinnacle); EN 55011, EN 61000-6-2, EN 50178 (Plus+) |
n/s |
design targets IEC 61010-1, IEC 61326-1 |
n/s = not stated in the documents we checked.
Takeaways:
- The AE units regulate power, current or voltage, and strike at a voltage well above the run voltage.
- The arc figure of merit improved by about three orders of magnitude in twenty years: 10 to 100 J per arc with SCR supplies before 1983, about 100 mJ/kW with the first switch-mode units, about 10 mJ/kW in 1990, below 2 mJ/kW in 1995, below 0.2 mJ/kW in 2004 P9. The gains came from removing output capacitance and reacting in microseconds.
- Per-kW figures do not transfer to 100 W. What the film sees is the absolute energy of each arc [P9]P10. A 3 kW Pinnacle 3000 may put up to 18 mJ into an arc and a 20 kW TruPlasma up to 6 mJ (derived from [E4]E5). A 100 W supply that stays below 5 mJ is already in that class, and 1 mJ is reachable.
- A published 5 kW laboratory design reached about 110 mJ per arc in DC mode and 9 mJ in pulsed mode P11, a realistic picture of what a first academic design achieves.
4. Requirements
| ID |
Requirement |
Level |
Basis |
| IN-1 |
Runs from one Mean Well UHP-200-24 (24 V, 8.4 A, 201.6 W). No other power input. |
MUST |
Project decision (availability) E1 |
| IN-2 |
Full performance from 22.8 to 25.2 V (the UHP trim range). No damage from 0 to 32 V, which covers the UHP over-voltage window of 26.4 to 31.2 V. |
MUST |
E1 |
| IN-3 |
Input current at most 6.0 A in every operating and fault condition, enforced in hardware. The UHP goes into hiccup at 110 to 140% of rated power and tolerates 150% for 100 ms only. |
MUST |
E1 |
| IN-4 |
Efficiency at least 75% at 100 W, goal 85%. At 85% the supply draws 118 W, 4.9 A, 58% of the UHP rating (derived). |
SHOULD |
E1; Pinnacle 3000 is above 90% E4 |
| IN-5 |
Waits for a stable 24 V (under-voltage lockout near 20 V with hysteresis), tolerates the UHP rise time of 80 ms and setup time of up to 3 s, and soft-starts so that inrush stays inside IN-3. |
MUST |
E1 |
| IN-6 |
When the arc handler stops the converter, the 24 V bus overshoot stays below 26.4 V. The UHP over-voltage protection latches and needs a power cycle. |
MUST |
E1 |
| IN-7 |
Fuse and reverse-polarity protection on the 24 V input. |
MUST |
Single-fault thinking of S1 |
| IN-8 |
The UHP is mounted as its datasheet assumes (aluminium plate or chassis wall of about 300 x 300 x 3 mm, 5 cm free space), or derated. |
SHOULD |
E1 |
4.2 Output envelope (OUT)
| ID |
Requirement |
Level |
Basis |
| OUT-1 |
Negative output to the cathode. Positive output bonded to chassis and protective earth. The anode is the earthed chamber. |
MUST |
P1; ground-fault lesson of H1 |
| OUT-2 |
Regulated power 5 to 100 W continuous. Hard cap at 100 W in hardware and firmware. |
MUST |
Project decision; section 2.3 |
| OUT-3 |
Full 100 W from 250 V to 800 V (625 Ω to 6.4 kΩ). |
MUST |
300 to 700 V typical P1 plus margin for high-pressure operation H1 and for different targets P7. Commercial units give full power over a 1:2 to 1:3 voltage range [E4]E5 |
| OUT-4 |
Run voltage limit 1000 V. Operation down to 150 V at reduced power. |
MUST |
Small sources are rated 100 or 200 to 1000 V [E6][E7]E8; MDX 500 ranges E2 |
| OUT-5 |
Output current 0 to 400 mA, goal 500 mA. |
MUST / SHOULD |
100 W at 250 V (derived); ONYX-1 0.1 to 1 A E7 |
| OUT-6 |
Stable down to 10 mA for conditioning and very thin films. |
SHOULD |
Glow regime extends to low current P1 |
| OUT-7 |
Voltage ripple at most 5% p-p at 100 W and 400 V or more into a resistive load. Current ripple into the plasma at most 10% p-p. |
MUST |
MDX 500 2% p-p, MDX 1 kW 5% [E2]E3; ripple is multiplied by n in the plasma (derived from P4) |
| OUT-8 |
Current-source output: high output impedance at the switching frequency. Total capacitance across the output, filter plus cable, at most 6.8 nF; no µF-class bulk capacitors; ripple filtered by series inductance. |
MUST |
Steep I-V curve P4; a current source gives fast arc detection and low arc energy P13; voltage sources with output capacitors are the classic arc problem [X5]X7 |
4.3 Ignition (IGN)
| ID |
Requirement |
Level |
Basis |
| IGN-1 |
Open-circuit strike voltage programmable from 600 to 1200 V, goal 1500 V, current limited ("soft"). |
MUST / SHOULD |
MDX 500 1200 V range E2; MDX 1 kW 1500 V soft strike, 800 V low Z; MDX 1.5 kW 1200 V open circuit E3; Paschen behaviour P1 |
| IGN-2 |
Holds the strike voltage into an open circuit without any fault. A configurable ignition timeout (default 5 s) ends the attempt with a "no ignition" status, not a latched error. |
MUST |
H1; E9 |
| IGN-3 |
Ignition is recognised from current rise together with a voltage that settles above the arc threshold. A "plasma present" status is reported. The arc detector uses its own thresholds during the strike. |
MUST |
MDX reports "plasma is present" E2 |
| IGN-4 |
After ignition the output ramps to the setpoint in a programmable time, 0.1 s to 10 min. |
MUST |
MDX 500: 0.1, 1 or 10 s; MDX 1 kW: 1 to 10 s or 1 to 10 min [E2]E3; new targets need gentle conditioning P8 |
| IGN-5 |
The ignition transient does not trip any protection and does not overshoot the current limit by more than 50% for longer than 100 µs. |
MUST |
H1; E9 |
| IGN-6 |
Digital output that can request a short gas pressure burst during the strike. |
MAY |
Our addition, follows from the Paschen curve P1 |
4.4 Regulation and metering (REG)
| ID |
Requirement |
Level |
Basis |
| REG-1 |
Regulation modes: power (default), current, voltage, selectable at run time. |
MUST |
MDX and Pinnacle offer all three [E2][E3]E4 |
| REG-2 |
The inner loop always regulates current. Power and voltage are outer loops. |
MUST |
Derived from the I-V steepness P4 |
| REG-3 |
Setpoint accuracy 2% of reading or 0.5% of full scale, whichever is larger. Repeatability 1%. |
MUST |
MDX 500 display accuracy E2; TruPlasma 0.5% E5; Pinnacle 0.1% repeatability E4 |
| REG-4 |
Readback of V, I and P at 1 kS/s or faster, plus arc count and arc rate. |
MUST |
Arc rate and count are process diagnostics P9 |
| REG-5 |
Programmable limits for power, current, voltage and strike voltage, stored per machine profile in non-volatile memory. |
MUST |
Pinnacle: programmable output, strike and process limits, non-volatile settings E4 |
| REG-6 |
Energy mode: stop after a set delivered energy in joules. |
SHOULD |
Pinnacle Joule mode E4 |
| REG-7 |
Built-in I-V sweep that logs the curve and fits the exponent n. |
SHOULD |
n indicates confinement quality P4; 1/V follows the secondary electron yield of the target surface P7 |
4.5 Arc handling (ARC)
| ID |
Requirement |
Level |
Basis |
| ARC-1 |
Arcs are quenched and the process resumes automatically. The supply never latches off on a first arc. |
MUST |
H1; Arc-Out E2; P8 |
| ARC-2 |
Energy delivered into an arc after its onset: at most 5 mJ anywhere in the run envelope, goal 1 mJ at 500 V. Total output capacitance including cable at most 6.8 nF, goal 4.7 nF. |
MUST / SHOULD |
Stored-energy history P9; commercial figures [E4]E5; droplet size follows arc energy [P9]P10; budget in section 6.2 |
| ARC-3 |
Detection by voltage collapse (absolute threshold 50 to 900 V, or relative 30 to 70% of the 1 ms average, default 50%), optionally combined with dV/dt and over-current. Detection time 2 µs or less. |
MUST |
Arc voltage is far below glow voltage, and the voltage criterion is the least ambiguous P8; Pinnacle 0.5 to 3 µs E4; TruPlasma adjustable U and I thresholds E5; voltage and dV/dt sensing X1 |
| ARC-4 |
Gate drive inhibited, bridge switches off, within 2 us of arc detection, through a hardware path with no firmware in the loop. Energy delivered by the converter after inhibit is at most 1 mJ, measured with the arc simulator. The resulting C_out overshoot at restart stays within PRT-1. |
MUST |
About 10 µs was the state of the art in 2006 P8; Pinnacle 3000 interrupts for 10 µs E4 |
| ARC-5 |
Off-time programmable from 100 µs to 100 ms, default 1 ms. Recovery by a controlled ramp, programmable 0.1 to 10 ms. If the plasma does not re-establish, the supply falls back to the strike state. |
MUST |
Programmable trip level, delay and shutdown time E4; shutdown of the order of milliseconds, longer off-times disturb the process more P9; hot spots must cool before power returns E2; the plasma needs time to re-establish P14 |
| ARC-6 |
Progressive response: if the arc is still there at restart, the off-time is multiplied up to a limit, then a "hard arc" fault is latched. |
MUST |
Progressive arc shutdown X3; micro-arcs against hard arcs P9 |
| ARC-7 |
Conditioning mode: the power limit is reduced automatically while the arc rate is above a threshold. |
SHOULD |
Target conditioning mode of Pinnacle Plus+ E4; oxide skin on new targets P8 |
| ARC-8 |
Survives a continuous output short circuit and at least 10 000 arcs at full power without damage or drift. |
MUST |
MDX runs into any load mismatch and is protected against short and open circuits E3 |
| ARC-9 |
Series inductance of the order of 1 mH between the output capacitor and the connector, so that arc current rises slowly and falls to zero by itself. |
SHOULD |
Less output inductance means higher peak current, longer arcs and more energy P12; ring-out and passive arc control [X2]X4 |
| ARC-10 |
Mechanical and firmware provision for a later pulsed-DC stage: reverse voltage of 5 to 20% for 1 to 10 µs, at 10 to 250 kHz. |
MAY |
Pulsed DC removes charge from insulating layers [P1][P14][P15]X6 |
4.6 Protection (PRT)
| ID |
Requirement |
Level |
Basis |
| PRT-1 |
Output over-voltage: hardware limit at 110% of the programmed strike voltage and an absolute hardware shutdown at 1.6 kV, both independent of firmware. |
MUST |
MDX: internal protection against over-voltage, over-current and over-power E3 |
| PRT-2 |
Over-current and over-power: cycle-by-cycle limit on the primary side, firmware cap, absolute hardware cap. |
MUST |
E3; IN-3 |
| PRT-3 |
Over-temperature sensing on switches, transformer and rectifiers: fold back first, then shut down. |
MUST |
S1 clause 10 |
| PRT-4 |
Loss of the controller (watchdog, brown-out) disables the gate drive in hardware. |
MUST |
S1 single-fault condition |
| PRT-5 |
Plausibility checks on the sense chain (open divider, open shunt) lead to shutdown. |
MUST |
S1 single-fault condition |
| PRT-6 |
Every fault ends in the same safe state: converter off, output bled down, status latched and readable. |
MUST |
S1 |
4.7 Safety (SAF)
| ID |
Requirement |
Level |
Basis |
| SAF-1 |
No hazardous-live part is accessible. Above 60 V DC a part counts as hazardous live (120 V DC under a single fault) unless current and charge are limited to 2 mA DC and 45 µC. This output exceeds those limits by orders of magnitude. |
MUST |
S1 clause 6.3 |
| SAF-2 |
The output leaves the enclosure only through an SHV connector (5 kV, 5 A class, recessed contact, ground mates first) on solid-dielectric RG-59 coax. No banana jacks, no MHV. |
MUST |
E10; MHV not accepted for live connections E11; MDX offers SHV E2 |
| SAF-3 |
Output below 60 V within 1 s after HV-off or loss of power, through a passive bleeder that needs no firmware. Design value about 0.1 s with 4.7 MΩ (derived). |
MUST |
60 V within 5 s, 1 s for plug pins S2; hazardous-live limit S1 |
| SAF-4 |
Stored charge at the highest voltage below 45 µC. With ARC-2 it is about 10 µC (derived). |
SHOULD |
S1 clause 6.3 |
| SAF-5 |
Reinforced insulation between the HV secondary and everything else (24 V, control, interfaces), rated for 1.5 kV working voltage and verified by a voltage test. |
MUST |
S1 clauses 6.7 and 6.8; S3 |
| SAF-6 |
PCB spacings at least those of IPC-2221B for 1.5 kV: 2.75 mm internal, 7.5 mm external uncoated, 3.85 mm external coated (derived). IPC gives functional spacing only: the HV to low-voltage barrier follows IEC 60664-1. |
MUST |
[S4]S3 |
| SAF-7 |
Protective earth stud on the chassis, positive output bonded to the chassis, a dedicated terminal for an earth strap to the chamber baseplate. |
MUST |
S1 protective bonding; the MDX manual forbids defeating the ground connection E3 |
| SAF-8 |
Hardware interlock loop of series contacts, fail-safe: at least vacuum OK, bell jar and HV shroud in place, cooling or over-temperature, emergency stop. Opening the loop removes the gate-drive enable. Restart needs a deliberate re-arm. |
MUST |
SEMI S2 sections 11 and 12 S5; SEMI S22 S6; MDX has vacuum, water and auxiliary interlock pins E2 |
| SAF-9 |
"HV ON" indicator visible from the operator position, key or arming switch, hazard labels. |
MUST |
S1 clause 5; S5 |
| SAF-10 |
No single component failure leaves the output energised with the interlock open. |
MUST |
S1 clause 4.4 |
| SAF-11 |
Internal measurements only through built-in dividers and test points at safe levels. Cover removable only with a tool, with a discharge-time label. |
MUST |
S1 |
4.8 Interfaces (IF)
| ID |
Requirement |
Level |
Basis |
| IF-1 |
Galvanically isolated digital interface (USB CDC and/or RS-485) with an open, documented protocol: on/off, mode, setpoints, limits, ramp, arc parameters, readback, status, faults. |
MUST |
Functions of the MDX user port E2; serial and fieldbus on larger units E4 |
| IF-2 |
Status bits: interlock OK, output on, plasma present, at setpoint, ramping, arc occurred, fault. |
MUST |
MDX status list [E2]E3 |
| IF-3 |
Analog programming and monitoring, 0 to 10 V, plus enable, as an optional board. |
MAY |
MDX 0 to 10 V E2 |
| IF-4 |
Local controls: arming switch and HV indicator as a minimum, display optional. |
SHOULD |
E2 |
| IF-5 |
Input for a temperature sensor on the magnetron body, with programmable power fold-back. |
SHOULD |
Magnets lost to overheating H1 |
4.9 EMC and robustness (EMC)
| ID |
Requirement |
Level |
Basis |
| EMC-1 |
Design targets: emissions to CISPR 11 (EN 55011) Group 1 Class A, immunity to IEC 61326-1 industrial levels. |
SHOULD |
Pinnacle: IEC 61326-1; Pinnacle Plus+: EN 55011 and EN 61000-6-2 E4 |
| EMC-2 |
The supply and its host link survive 10 000 arcs at full power without reset, lock-up or corrupted setpoints. |
MUST |
Arcs are sub-microsecond events with very fast current and voltage edges P8 |
| EMC-3 |
Coaxial, shielded HV path from connector to feedthrough, with a single-point bond. |
MUST |
MDX output on RG-8 coax E2 |
| EMC-4 |
The complete assembly (UHP plus supply plus cables) is re-checked for EMC. The UHP is a component supply whose compliance does not carry over to the final equipment. |
SHOULD |
E1 |
4.10 Environment and mechanics (ENV)
| ID |
Requirement |
Level |
Basis |
| ENV-1 |
Indoor use, 5 to 40 °C, up to 2000 m, pollution degree 2. |
MUST |
Normal conditions of S1; MDX operates from 0 to 40 °C E2 |
| ENV-2 |
Closed metal enclosure, finger-safe, no openings above HV parts. |
MUST |
S1 clause 6.2 |
| ENV-3 |
Up to 18 W of loss removed by conduction to the enclosure, with a small fan as a fallback. |
SHOULD |
IN-4 (derived) |
4.11 Replicability (REP)
| ID |
Requirement |
Level |
Basis |
| REP-1 |
Every part orderable from mainstream distributors, with a second source for everything except the magnetics. |
MUST |
Same reasoning as the choice of the UHP-200-24: availability for everyone |
| REP-2 |
The HV transformer and the output inductor can be wound by hand on catalogue cores and bobbins, from a winding sheet with acceptance measurements (leakage inductance, winding capacitance, voltage test). |
MUST |
Proposed, follows from REP-1 |
| REP-3 |
Schematic and PCB sources in an open format (KiCad proposed), firmware, bill of materials with sources and prices, and the test procedure are published. |
MUST |
Hacker Fab practice of publishing the BOM H1 |
| REP-4 |
Assembly with an iron and hot air, no BGA parts. Board manufacturable at standard low-cost fabs. |
SHOULD |
Proposed, follows from REP-1 |
| REP-5 |
A commissioning checklist that a builder without HV experience can follow, with clear stop points. |
MUST |
Section 5.2 |
5. Constraints
5.1 Design constraints
| ID |
Constraint |
Why |
| C-1 |
One UHP-200-24 feeds the supply. Budget: at most 6 A of its 8.4 A, which leaves about 58 W for other 24 V loads of the machine. |
IN-1, IN-3 E1 |
| C-2 |
Output power never exceeds 100 W. The cap exists twice, in hardware and in firmware. |
Section 2.3 |
| C-3 |
DC only in v1: conductive targets, argon. Reactive sputtering of insulating oxides, RF and HiPIMS are out of scope. |
Steady DC on insulating layers arcs P14; RF is a later Hacker Fab goal H1 |
| C-4 |
Low stored energy has priority over low ripple. No bulk capacitor on the output, ever, not even "temporarily for testing with plasma". |
ARC-2 [P9]X7 |
| C-5 |
The arc, interlock and over-voltage shutdown paths are hardware. Firmware supervises, counts and decides on recovery, but is never the only thing between a fault and the gate drive. |
ARC-4, SAF-8 S5 |
| C-6 |
Everything on the HV side is rated for 1.5 kV working voltage, with components derated to 2 kV or more. |
IGN-1, SAF-5 |
| C-7 |
Parts and processes must stay within REP-1 to REP-4. A better part that cannot be bought by anyone is not an option. |
Availability goal |
5.2 Electrical safety constraints
| ID |
Constraint |
Why |
| C-8 |
The output is treated as lethal. At up to 0.4 A it is far outside what HV practice calls an output-limited supply (5 mA below 5 kV). Never connect or disconnect the HV cable with the supply armed. |
E11; S1 clause 6.3 |
| C-9 |
Earthing is a star at the supply chassis: mains protective earth, UHP frame ground, chamber baseplate strap, coax shield. The positive output joins the chassis at one point, through the current shunt. |
OUT-1, SAF-7; all cathode current is then measured whatever its return path |
| C-10 |
The 24 V return is bonded to protective earth at one point. A floating 24 V bus could be lifted to cathode potential by a transformer insulation fault, and the UHP output is only tested to 1.25 kV AC against its frame. |
E1; S1 |
| C-11 |
There is deliberately no ground-fault trip on the output. The protection against shock is enclosure, earthing, interlocks and discharge, not leakage detection. This is the opposite of the electrophoresis supply and must be stated in the manual. |
Section 1.2 H1 |
| C-12 |
The cathode feedthrough on the air side of the baseplate is hazardous live. It needs an earthed or insulating shroud whose removal opens the interlock loop. |
S1 clause 6.3; the present feedthrough is a threaded rod through the baseplate H1 |
| C-13 |
HV may only be armed below a pressure threshold and with the bell jar in place (vacuum interlock). |
SAF-8; MDX vacuum interlock pin E2 |
5.3 Machine and process constraints
| ID |
Constraint |
Why |
| C-14 |
Default machine profile limits power to 20 W. Higher power is unlocked only for a source with validated cooling and a closed cooling interlock. |
Magnets overheat above about 20 W and have already been demagnetised once H1; all commercial sources of this size are water cooled [E6][E7]E8 |
| C-15 |
Power density on the target stays below about 16 W/cm² with indirect cooling (113 W on 30 mm, 81 W on 1 inch). |
E7 (derived) |
| C-16 |
Gaps where no plasma is wanted (cathode body to ground shield) keep p*d well below 28 Pa cm at the highest operating pressure: under 7 mm at 300 mTorr, in practice a few millimetres. |
Paschen left branch P1 (derived) |
| C-17 |
HV cable no longer than 3 m, one piece, no splices. Typical RG-59 adds about 70 pF/m, which counts toward ARC-2. |
ARC-2; cable energy feeds the arc [X2]P20 |
| C-18 |
New and oxidised targets, aluminium above all, are started in conditioning mode with a slow ramp. |
[P8]P22; ARC-7 |
6. Reference architecture (starting point, not binding)
6.1 Block diagram
mains -> [UHP-200-24] -> 24 V -> [fuse, filter, UVLO, input current limit]
-> [resonant inverter, 100 to 300 kHz]
-> [HV transformer, split secondaries]
-> [rectifiers, series stacked]
-> [C_out, a few nF] -> [L_out, about 1 to 2 mH]
-> [bleeder + V divider] -> SHV -> RG-59 -> cathode (negative)
chamber baseplate (anode, positive) -> coax shield + earth strap -> chassis / PE
rectifier positive node -> [current shunt] -> chassis (single point)
[fast comparators: arc, over-voltage, over-current] -> [PWM inhibit, hardware]
[interlock loop] -> [gate-drive enable, hardware]
[MCU: state machine, loops, telemetry] <-> [isolated USB / RS-485]
Notes:
- Converter family. A resonant stage that behaves as a current source fits the load. A series resonant converter in discontinuous mode gives arc protection without extra circuits because of its current-source behaviour P16. LCC and LC-LC tanks absorb the leakage inductance and winding capacitance of an HV transformer and can deliver constant current at any load; a 2 kV, 100 mA prototype is published P17, with design methods in P21. The series-parallel family works from no load to full load P18, which is what strike and run demand. A hard-switched bridge with an LC output filter is the wrong choice: the filter capacitor empties into every arc X7.
- Transformer and rectifier. Split the secondary into sections with their own rectifiers and stack them in series. This lowers the voltage per winding and per diode. Stacked rectifiers and the insulation between windings are discussed in P16.
- Sensing. Compensated HV divider and return-path shunt, both with at least 1 MHz bandwidth for the arc comparators.
- Control. A microcontroller whose PWM timers have hardware fault inputs fed by analog comparators (for example the STM32G4 family), so that ARC-4 and C-5 hold by construction.
- State machine. OFF -> ARMED -> STRIKE -> RAMP -> RUN <-> ARC_QUENCH -> RECOVER, with FAULT (latched) and DISCHARGE reachable from every state.
6.2 Arc energy budget (derived)
E_arc <= 1/2 * C_out * V^2 + 1/2 * L_out * I^2 + P * t_resp
| Case |
C_out at V |
L_out at I |
P * t_resp |
Total |
| Limit case: 6.8 nF, 1000 V, 0.1 A, 10 µs |
3.4 mJ |
0.01 mJ |
1.0 mJ |
4.4 mJ |
| Typical: 6.8 nF, 500 V, 0.2 A, 10 µs |
0.85 mJ |
0.04 mJ |
1.0 mJ |
1.9 mJ |
| Goal: 4.7 nF, 500 V, 0.2 A, 3 µs |
0.59 mJ |
0.04 mJ |
0.3 mJ |
0.9 mJ |
- L_out is taken as 2.2 mH.
P * t_resp is an upper bound: a current-source stage pushes far less than its rated power into a 30 V arc.
- The response time weighs as much as the capacitor. That is the reason for the hardware shutdown path.
- Without series inductance, 6.8 nF at 500 V discharges into an arc through about 1 µH of wiring with a peak near 40 A for a fraction of a microsecond. With 2.2 mH the extra current stays under 1 A and rings back to zero within some tens of microseconds, where the rectifiers block and the arc can go out by itself [P12][X2]X4.
- The inductor sees the full output voltage across its terminals during an arc and must be insulated for it.
- In per-kW terms the limit case is about 44 mJ/kW, between the 1983 and 1990 commercial generations, and the goal about 9 mJ/kW, the 1990 level P9. In absolute terms it is below what a multi-kW production supply delivers per arc (section 3).
7. Build plan
No fucking clue (low fucking clue to be precise), need to check with Alex and Tudor.
Verification matrix for step 8
| Test |
Method |
Pass criterion |
| Output envelope |
Dummy loads at the corners, 30 min each |
OUT-2 to OUT-5, REG-3 |
| Ripple |
Compensated HV probe at the connector, current probe in the return |
OUT-7 |
| Efficiency and thermal |
Input and output power at 25, 50, 100 W; soak at 40 °C |
IN-4, ENV-3, PRT-3 |
| UHP interaction |
Start-up, load steps, arc shutdown, short circuit, with the real UHP |
IN-3, IN-5, IN-6 |
| Strike |
Open circuit hold, then a load switched in to emulate ignition |
IGN-1 to IGN-5 |
| Arc energy |
Arc simulator at 250, 500, 800, 1000 V; integrate v*i |
ARC-2 |
| Arc timing |
Scope: arc onset to comparator, to gate-off, to current zero |
ARC-3, ARC-4 |
| Arc logic |
Single arcs, bursts, a persistent short |
ARC-1, ARC-5, ARC-6, ARC-7 |
| Endurance |
10 000 simulated arcs at 100 W, then repeat the accuracy test |
ARC-8, EMC-2 |
| Faults |
Open divider, open shunt, controller reset, over-voltage injection, input brown-out |
PRT-1 to PRT-6 |
| Discharge |
Scope on the output after HV-off and after pulling the 24 V |
SAF-3 |
8. Open questions for Hacker Fab
- Which power can the cooled source take, and when will it exist? Until then the 20 W profile of C-14 applies.
- Measured voltage and current today at 10, 50, 200 and 300 mTorr for Cu and Al (step 2). If the discharge sits below 250 V at high pressure, OUT-3 has to move.
- Target pressure range after the turbopump, and whether a gas burst for ignition is acceptable.
- Target materials planned beyond Cu and Al. Magnetic targets and intentional oxides change the requirements.
- Geometry: cathode to shield gap, cathode to baseplate distance, feedthrough voltage rating, and who builds the HV shroud of C-12.
- Which controller talks to the supply, and over which interface (USB, RS-485, analog).
- Which interlock signals exist or can be added: vacuum gauge relay, bell jar switch, coolant flow, emergency stop.
- Cable length between supply and chamber.
- Cost ceiling for the bill of materials, excluding the UHP.
- Is the pulsed-DC option (ARC-10) wanted soon enough to reserve board space for it in rev A?
9. References
Hacker Fab
Papers and conference material
- P1 J. T. Gudmundsson, A. Hecimovic, "Foundations of DC plasma sources", Plasma Sources Sci. Technol. 26, 123001 (2017). Author manuscript: https://pure.mpg.de/pubman/item/item_2506681_5/component/file_2537497/Gudmundsson_Foundations.pdf
- P2 J. A. Thornton, "Magnetron sputtering: basic physics and application to cylindrical magnetrons", J. Vac. Sci. Technol. 15 (2), 171-177 (1978). Used as summarised in P1.
- P3 R. K. Waits, "Planar magnetron sputtering", J. Vac. Sci. Technol. 15 (2), 179-187 (1978). Used as summarised in P1.
- P4 G. Y. Yeom, J. A. Thornton, M. J. Kushner, "Cylindrical magnetron discharges. I. Current-voltage characteristics for dc- and rf-driven discharge sources", J. Appl. Phys. 65 (10), 3816-3824 (1989).
- P5 S. M. Rossnagel, H. R. Kaufman, "Current-voltage relations in magnetrons", J. Vac. Sci. Technol. A 6, 223 (1988). https://researchweb.draco.res.ibm.com/publications/current-voltage-relations-in-magnetrons
- P6 Proceedings of PAC07, paper MOPAN026 (DC magnetron I-V characteristics with n between 5 and 15). https://jacow.org/p07/papers/MOPAN026.pdf
- P7 D. Depla, S. Mahieu, R. De Gryse, "Magnetron sputter deposition: Linking discharge voltage with target properties", Thin Solid Films 517 (9), 2825-2839 (2009). Used as summarised in P1.
- P8 A. Anders, "Physics of arcing, and implications to sputter deposition", Thin Solid Films 502, 22-28 (2006). Manuscript LBNL-54220: https://escholarship.org/uc/item/2sq5t35m
- P9 D. J. Christie, D. Carter (Advanced Energy), "The impact of power supply arc response on production yield and field reliability", 49th SVC Technical Conference, 2006. https://nccavs-usergroups.avs.org/wp-content/uploads/TFUG2006/TFUG_2006_08_Christie.pdf
- P10 D. Christie, "Magnetron Arcing: Considerations for Large-Area Coating", Advanced Energy, 2014, https://www.advancedenergy.com/zh-cn/about/news/blog/magnetron-arcing-considerations-for-large-area-coating/; D. Carter, H. Walde, "Arc Energy in Large Scale Magnetron Sputtering", AVS 57th Symposium, 2010, https://www2.avs.org/symposium2010/Papers/Paper_EN+PS-MoM11.html
- P11 N. S. Sochugov, V. O. Oskirko, R. E. Spirin, "A power supply for magnetron sputtering systems", Instrum. Exp. Tech. 56, 178-184 (2013). https://link.springer.com/article/10.1134/S0020441213010302
- P12 V. O. Oskirko et al., "Power supply output inductance effect on arc parameters of high-power impulse magnetron sputtering" (2022), and "Arc energy minimization in high-power impulse magnetron sputtering", Vacuum 202, 111213 (2022). https://doi.org/10.1016/j.vacuum.2022.111213
- P13 "High Power Processes: Power Delivery and Discharge Management", SVC TechCon 2017 (abstract). https://asm.confex.com/asm/svc2017/webprogram/Paper42902.html
- P14 A. Belkind, A. Freilich, R. Scholl, "Using pulsed direct current power for reactive sputtering of Al2O3", J. Vac. Sci. Technol. A 17 (4), 1934-1940 (1999), https://doi.org/10.1116/1.581706; A. Belkind et al., "Pulse duration effects in pulse-power reactive sputtering of Al2O3", Surf. Coat. Technol. 109, 558-563 (1998).
- P15 J. Sellers, "Asymmetric bipolar pulsed DC: the enabling technology for reactive PVD", Surf. Coat. Technol. 98, 1245-1250 (1998); S. Schiller et al., "Pulsed magnetron sputter technology", Surf. Coat. Technol. 61, 331-337 (1993). Used as summarised in P1 and P8.
- P16 S. R. Jang, H. J. Ryoo, J. S. Kim, S. H. Ahn, "Design and analysis of series resonant converter for 30-kW industrial magnetron", IEEE Trans. Ind. Electron. 59 (3), 2012. Note: a microwave magnetron supply (14 kV, 3 A); cited for the topology argument only. https://doi.org/10.1109/TIE.2011.2163915
- P17 M. Borage, K. V. Nagesh, M. S. Bhatia, S. Tiwari, "Design of LCL-T resonant converter including the effect of transformer winding capacitance", IEEE Trans. Ind. Electron. 56 (5), 1420-1427 (2009). https://doi.org/10.1109/TIE.2009.2012417
- P18 R. L. Steigerwald, "A comparison of half-bridge resonant converter topologies", IEEE Trans. Power Electron. 3 (2), 174-182 (1988). https://doi.org/10.1109/63.4347
- P19 "Design and construction of a bipolar pulsed-dc power supply for magnetron sputtering systems". https://www.academia.edu/99060311/Design_and_construction_of_a_bipolar_pulsed_dc_power_supply_for_magnetron_sputtering_systems
- P20 P. Ozimek et al., "HIPIMS Power Supply Requirements: Parameters and Breakthrough Arc Management Solution", SVC Technical Conference, 2009. https://www.svc.org/clientuploads/directory/resource_library/09_285.pdf
- P21 M. P. Foster et al., "Methodologies for the design of LCC voltage-output resonant converters". https://core.ac.uk/download/54931.pdf
- P22 C. E. Wickersham Jr. et al., J. Vac. Sci. Technol. A 19 (6), 2741 and 2767 (2001); 20 (3), 833 (2002). Arc initiation by dielectric inclusions in aluminium targets, as discussed in P8.
Further reading, not used for specific numbers: J. T. Gudmundsson, "Physics and technology of magnetron sputtering discharges", Plasma Sources Sci. Technol. 29, 113001 (2020), https://doi.org/10.1088/1361-6595/abb7bd; P. J. Kelly, R. D. Arnell, "Magnetron sputtering: a review of recent developments and applications", Vacuum 56, 159-172 (2000); R. Scholl, "Power Supplies for Pulsed Plasma Technologies: State-of-the-Art and Outlook", Advanced Energy white paper.
Equipment datasheets
- E1 Mean Well, UHP-200 series datasheet (24 V model). https://www.mouser.com/pdfDocs/MeanWell_UHP-Series.pdf Later revisions add the 150% peak load for 100 ms and list IEC/UL 62368-1: https://www.trcelectronics.com/products/mean-well-uhp-200-24
- E2 Advanced Energy, MDX 500 W data sheet. https://www.advancedenergy.com/globalassets/resources-root/data-sheets/mdx-series-500-w-data-sheet.pdf
- E3 Advanced Energy, MDX 1 kW and 1.5 kW brochure, https://web.pa.msu.edu/people/edmunds/MDX-1K_Magnetron_Supply/mdx_1kw_1.5kw_brocher.pdf; MDX 1.5K manual, https://mfc.engineering.arizona.edu/sites/default/files/2024-01/AdvancedEnergy-MDX1.5K-PowerSupply-manual.pdf
- E4 Advanced Energy, Pinnacle series, https://www.advancedenergy.com/globalassets/resources-root/data-sheets/en-ppg-pinnacle-series-data-sheet.pdf; Pinnacle 3000, https://www.advancedenergy.com/globalassets/resources-root/data-sheets/pinnacle-3000-data-sheet.pdf; Pinnacle Plus+, https://www.advancedenergy.com/globalassets/resources-root/data-sheets/en-ppg-pinnacle-plus-series-data-sheet.pdf
- E5 TRUMPF Hüttinger, TruPlasma DC Series 3000 (G2) and 4000 (G2) brochure. https://www.trumpf.com/filestorage/TRUMPF_Master/Products/Power_Electronics/DC_plasma_excitations/TruPlasma_DC_Series_3000_G2/TRUMPF_TruPlasma_DC_Serie_3000_G2-DE.pdf
- E6 Ferrotec MeiVac, 1.3 inch MAK sputter source. https://meivac.ferrotec.com/?p=6891
- E7 Angstrom Sciences, ONYX-1 IC Mag II specifications, https://angstromsciences.com/wp-content/uploads/2024/10/ONYX-1-IC-Mag-II-US.pdf; power and cooling reference, https://angstromsciences.com/power-cooling-reference
- E8 MTI, 1 inch magnetron sputter source HVMSS-SPC-1. https://mtixtl.com/products/hvmss-spc-1
- E9 Fisher Scientific, FisherBiotech electrophoresis power supplies (safety features), https://www.fishersci.com/shop/products/fisherbiotech-electrophoresis-power-supplies-9/fb300q; Fisherbrand power supplies brochure, https://origin-eu-com.cloudprod.eu.fishersci.com/content/dam/fssite/eu/brands/f/22640---fisherbrand/EN_Power_Supplies.pdf
- E10 CeramTec, coaxial connectors and feedthroughs (SHV, MHV, N, HN ratings), https://www.ceramtec.com/ceramaseal/coaxial-connectors/; SHV connector overview, https://www.wikiwand.com/en/articles/SHV_connector
- E11 Laboratory for Laser Energetics, University of Rochester, "High Voltage Diagnostics, LOTO Guidelines", S-SA-M-060. https://www.lle.rochester.edu/media/safety/electrical/S-SA-M-060_B.pdf
Patents
- X1 US 5,427,669, "Thin film DC plasma processing system" (arc sensing by voltage and rate of change, voltage reversal). https://patents.google.com/patent/US5427669A
- X2 US 6,943,317, "Apparatus and method for fast arc extinction with early shunting of arc current in plasma" (stored energy in cable and output filter, ring-out). https://patents.google.com/patent/US6943317
- X3 US 7,247,221, "System and apparatus for control of sputter deposition process" (progressive arc shutdown). https://image-ppubs.uspto.gov/dirsearch-public/print/downloadPdf/7247221
- X4 US 6,524,455, passive arc control with a resonant network and reverse voltage clamp. https://patentimages.storage.googleapis.com/41/b0/f3/2d6ac1e268e4f8/US6524455.pdf
- X5 US 5,777,863, "Low-frequency modulated current mode power supply for magnetron sputtering cathodes". https://image-ppubs.uspto.gov/dirsearch-public/print/downloadPdf/5777863
- X6 EP 0 692 138, "Reactive DC sputtering system" (reversal for 1 to 10 µs at 5 to 20% of the DC voltage). https://data.epo.org/publication-server/rest/v1.2/patents/EP0692138NWB1/document.pdf
- X7 US 7,081,598, "DC-DC converter with over-voltage protection circuit" (energy of the output LC filter in an arc). https://image-ppubs.uspto.gov/dirsearch-public/print/downloadPdf/7081598
Standards
- S1 IEC 61010-1:2010, Safety requirements for electrical equipment for measurement, control, and laboratory use, Part 1. Limits of clause 6.3 as summarised in https://www.designworldonline.com/selv-pelv-en-61010-ul-61010-regulations-and-power-supplies/ and https://cdn.intertek.com/www-intertek-com/dms-legacy/UL-61010-1-CSA-C22-2-No-61010-1-SUN-Rev-11-16-2018-and-11-21-2018-ED-1-1-2022.pdf
- S2 IEC 60204-1, Safety of machinery, Electrical equipment of machines (residual voltage). Summary: https://www.electronicsonline.net.au/content/business/sponsored/iec-60204-1-machine-safety-540830386
- S3 IEC 60664-1, Insulation coordination for equipment within low-voltage supply systems (clearance and creepage).
- S4 IPC-2221B, Generic Standard on Printed Board Design, Table 6-1. Summaries: https://help.altair.com/pollex/topics/pollex/verification/dfe_ipc_material_r.htm, https://pcbsync.com/ipc-2221/
- S5 SEMI S2, Environmental, Health, and Safety Guideline for Semiconductor Manufacturing Equipment (sections 11 to 13). Overviews: https://www.intertek.com/semiconductor-equipment/semi-s2/, https://r6.ieee.org/scv-pses/wp-content/uploads/sites/17/2019/11/IEEE-PSES-Slides-SEMI-Steve-Baldwin-oct19.pdf
- S6 SEMI S22, Safety Guideline for the Electrical Design of Semiconductor Manufacturing Equipment. https://store-us.semi.org/products/s02200-semi-s22-safety-guideline-for-the-electrical-design-of-semiconductor-manufacturing-equipment
- S7 IEC 61326-1 (EMC for measurement, control and laboratory equipment) and CISPR 11 / EN 55011, as claimed by the supplies in E4.