[ §1 · control ]

SSTC that will not start: how feedback takes over at start-up

SSTC

A current-feedback driver cannot start from the signal it is meant to produce. Getting from that blind start to closed-loop running has its own settings and failures, and they look just like weak feedback.

A current feedback loop cannot start itself. It runs on the primary current, and when the interrupter says go, that current is zero. The first cycles of every bang are therefore driven by something that is not measuring the coil, and then the driver has to switch over to the current transformer. That switch has its own settings and its own ways of failing, and the UD2.7, the UD+ and the UD3 each do it differently.

When it fails, the coil will not start, or it starts on some bangs and not others. A loop with too little gain looks the same, so the first job is to work out which of the two you have. This article covers how each of those boards makes the switch, the settings it gives you, and how to diagnose a coil that will not start.

What it is and how it works

An oscillator made of an amplifier and a sensor is a loop with a sign on it. The amplifier drives a load, the sensor watches what the load does, and the sensor's output decides when the amplifier switches next. Nothing in the circuit sets the frequency. The load does, and the circuit follows. That is why the arrangement gets used wherever the load's frequency is not known in advance or will not stay still.

Nothing in that description starts, though. Before the amplifier has driven anything there is no output to watch, so the sensor says nothing and the amplifier is never told to switch. The loop is complete and correct, and it sits there doing nothing.

Schematic titled Micro SSTC, Steve's High Voltage, Steve Ward, rev 1.0, 11/25/2003. A single IRFP460 MOSFET switches a three turn primary off a rectified 25 VAC supply, with a 0.001 microfarad 2 kV capacitor across it and the secondary beside the primary. Its gate is driven through 10 ohms by a TC4420, whose input is an antenna clamped by two 1N60 diodes. A push button marked Start Oscillation connects that input to the supply through 100k. The notes include: reverse primary wires if coil fails to oscillate.
The problem at its smallest, on Steve Ward's Micro SSTC: the loop is an antenna into a gate driver, and the way it is woken is a push button marked Start Oscillation. The note under it is the loop's sign: reverse the primary wires if it fails to oscillate. Schematic by Steve Ward, from his Micro SSTC page, used with his permission.

There are three ways out of that. One is to let noise grow: with enough gain round the loop and enough Q in the load, whatever is on the input gets amplified round and round until it dominates. Another is to drive the load from a local oscillator that measures nothing, until it is moving enough for the sensor to see it. The third is to rebuild the sensing element so that it swings on its own when nothing is arriving, so the loop never goes quiet in the first place. The second has to be designed, because at some point the local oscillator has to stop and hand over to the sensor.

For that handover to work, three things have to be right. The circuit needs a criterion, some test it applies to decide that what the sensor says now is real. The local oscillator has to be close enough to the load's own frequency that the load responds to it at all. And both ends of the loop need a defined idle state. The last of the three is what gets left out.

The idle state matters because the first push is a change. The output stage is commanded from one state to the other, and that change is what disturbs the load. If the output stage is already sitting in the commanded state, nothing moves, the sensor sees nothing, and a circuit with every component working does nothing. The same holds at the sensor's input. A comparator with nothing driving it idles wherever its input happens to sit, and if nothing defines where that is, it idles somewhere different from one attempt to the next.

What it does in a coil, and your design choices

In a solid state coil the load is the primary and its capacitor, the sensor is the feedback current transformer with its burden resistor, and the output stage is the bridge. The local oscillator sits on the driver board and drives the gates directly. The four parts above are four real components, and the decisions below involve all four.

A photograph of an analogue scope screen showing one burst: a sine wave that starts small at the left, grows cycle by cycle over about eight cycles to its largest swing, then stops and settles to a flat line across the right half of the screen.
One bang on Steve Ward's DRSSTC-1, from the scope captures on his page: the burst grows from nothing, which is the window the start has to get through. Scope capture by Steve Ward, from his DRSSTC-1 page, used with his permission.

The third way out, a sensor that swings on its own, is decided before any of this. On the forums it is called self-oscillation. A driver modified to self-oscillate never stops measuring, so there is no local oscillator to stop and no handover to get wrong. davekni suggests it in two of the three threads quoted below and calls it his favourite way to fix start-up issues. For the UD2.7 he gives the modification: a resistor of about 50 kilohms from the comparator's positive output back to its inverting input, R7 removed, and C33 adjusted until the free-running frequency is reasonable.

This is a different thing from the noise route. The added resistor and C33 make the comparator swing by itself at a frequency the capacitor sets. The same change is described under converting a DRSSTC driver, where it "turns the comparator into an oscillator: it starts swinging on its own instead of waiting for the current transformer". davekni attaches a condition to it, "This works only when jumper SV1 is inserted", and SV1 is the burden jumper that the builder in the 2020 thread below had taken out. Everything from here on is about the boards that do have a handover. The noise route turns up once more under "What to watch out for", uninvited, as a scope probe.

Where the first cycles come from

The two boards that document this describe the same arrangement. Phillip Slawinski's UD+:

The startup oscillator generates the signal used to drive the gates when a new interrupter pulse is received. The startup oscillator will be used for the user-specified number of cycles, and then if valid feedback is detected, the gates will be driven by feedback from the coil.

The UD3B spec sheet by Steve Ward and Phillip Slawinski, dated 30 November 2015, puts the requirement first:

This implementation requires the driver "knows" the proper frequency to start driving the system at before feedback is present. The user must program a "Start Frequency" and "Start Cycles". At the start of a burst, the driver will output pulses at the Start Frequency for the programmed number of Start Cycles before switching over to a drive signal that is phase-locked with the feedback input.

That makes two settings, start frequency and start cycles. Set them badly and the coil does not start at all.

How each board decides when to hand over to feedback

The three designs differ here, and which one you have decides what a failed start means.

Timeline of one bang. The primary current builds from nothing along the axis, and a shaded band at the start marks the window in which the board is driven blind by its start oscillator. Three numbered marks sit along the axis at increasing times. One, the analogue UD2 line: no criterion at all, it follows whatever appears at its input the instant it appears and locks up if nothing does. Two, the UD plus: it counts the programmed cycles and then rejects the feedback if a pulse runs past one and a half times the start oscillator's half period. Three, the UD3: it counts and also waits for the current to pass a threshold its firmware calls min underscore fb underscore current. A note beside the band says a QCW's window is longer, because its bus starts near nothing, so the current builds slowly and whichever criterion the board uses is satisfied later.

The three marks are about how long each board keeps driving blind. The analogue line never asks, and the other two decide it in different ways.

  • The UD+ counts, then checks the length of what arrives. After the programmed cycles it looks at the feedback and rejects it if the pulses are too long. Slawinski: "Setting the improper startup frequency can cause the coil to stop switching after the startup cycles have elapsed. This happens when the driver detects pulses longer than 1.5 times the half cycle period for the startup oscillator." A pulse here is a half cycle, so the test trips when the feedback frequency has fallen below two thirds of the start frequency [derived].
  • The UD3 counts and also measures the current. Its firmware (Netzpfuscher/UD3, Jens Kerrinnes, building on Steve Ward) has a parameter min_fb_current, described in the source as "Current at which to switch to feedback". The same routine, configure_ZCD_to_PWM in ZCDtoPWM.c, loads the start cycle counter and writes that threshold into a DAC, a digital to analogue converter.
  • The analogue UD2 line has no criterion. The comparator follows whatever appears at its input the moment it appears, and does nothing if nothing appears. Gao Guangyan's walkthrough of Steve Ward's UD2.7 says that with no input signal in the feedback jacks the driver will not see any feedback, causing the drive to lock up. Nothing in that path checks whether the signal is real, so the idle state of the input has to be fixed by hand. On the UD2.7 it is: R3 and R4 with C6 divide the five volt rail down to a standing offset of about 1.6 V, the feedback is coupled on top of it, and the comparator's own reference is set to the same 1.6 V.

The idle state of the comparator input

AstRii reported the failure on a UD+ in June 2024:

sometimes the driver skips an interrupter pulse. It tries to begin pulse but quickly triggers protection of the driver and sets Fault high.

davekni explained it in the first reply as a design difference between the two boards:

Unlike UD2.7, UD+ has no DC bias on FB comparitor input. State between bursts is indeterminant.

The DC bias he means appears to be the 1.6 V offset above. Gao documents it on the UD2.7 and davekni says the UD+ lacks it, but connecting his sentence to Gao's walkthrough is our reading, not something either of them said. The key word is indeterminate. Nothing is broken and the loop is not mistuned. Between bangs the comparator's input sits nowhere in particular, so its output is high on some bangs and low on others, and on the bangs where it is already high the first push has nothing to change. He offered a patch and, in the same post, said he would not recommend it:

Not what I'd recommend, but you could patch the issue with a weak pull-down (~100k) on + input of comparitor, forcing idle state low.

AstRii tried it and reported that it "works as a charm". The resistor to ground carries no signal, changes no threshold and does not enter the loop's arithmetic anywhere. All it does is make sure the circuit knows which way it is pointing before the bang begins. Why davekni still did not recommend it is the last warning under "What to watch out for", below.

The idle state of the bridge output

The same thing happens at the bridge end, and davekni has pointed it out on three different coils, in 2020, 2023 and 2024. The first was a oneTesla TS half bridge rebuilt around a UD2.7, in a December 2020 thread started by a builder posting as David:

One likely issue is that the initial bridge output voltage is already at the same VBus rail that the initial pulse drives it to. Adding a bleed resistor across the opposite IGBT will fix that issue.

That thread is about more than the handover. David's coil would not start with the 51 ohm SV1 burden jumper fitted, and the paragraph this sentence sits in begins "Concerning feedback, I suggest replacing the jumper and then figuring out why there isn't enough signal." The reply itself opens on something else again, IGBT failures under hard switching and the parasitic inductance that causes the voltage spikes behind them. The bridge idle state comes up as one likely issue inside an investigation of loop gain, and both kinds of failure are live on the same coil at the same time. The self-oscillation recipe given earlier is the next paragraph of that same reply, and that is where its condition bites. The modification needs SV1 in, and with SV1 in this coil would not start.

On RoamingD's UD2.7 in September 2023 the mechanism is leakage rather than charge:

Without scope load, if upper IGBT happens to have slightly higher leakage current, output may drift high. Initial start pulse tries to pull it high. If output is already high, no initial change, so no feedback.

In AstRii's thread he puts it as a design rule: "H-bridge output needs to start centered (0V differential) so that either polarity of initial half-cycle starts oscillation." The bridge page puts a bleeder across the inverter output for this reason. A stuck output stops the machine outright instead of making it misfire.

davekni also gives a trade between the two fixes in the same thread. If the driver generates more than one transition on the gate signals before feedback starts, no bleed resistor is needed across the bridge outputs, though he adds there is no harm in fitting one beyond a little wasted power. Our reading is that the second transition goes the other way, so the output cannot already be sitting in the state both of them command. Start cycles and an output bleeder partly substitute for each other, and extra start cycles are the harsher choice, because they are hard switched.

AstRii's own bleeder did not fix his fault, which is useful to know. He fitted 22 kilohms across the bridge output against a 50 nF tank, giving what he calls a 1.1 ms time constant, at about 100 bangs per second, and the skipping continued. At that rate the resistor gets roughly nine time constants between bangs [derived], plenty of time to do its job. That left the comparator input, and that is where the fix landed.

Setting the start cycles and the start frequency

The published guidance is one line of the UD3B spec sheet: "The start cycles can in some cases be set to 0 or 1, but most setups work best with at least 2 to 4 start cycles for DRSSTC and up to 8 cycles for QCW." The UD+ sets the same number on four DIP switches. Its table runs from 1 to 15 cycles, with zero reserved for a test mode.

We have not found a published reason why a ramp needs more of them. Our own reading is that a QCW's bus starts near nothing, so the primary current builds slowly and whichever criterion the board uses is met later. The blind window is longer because what it waits for arrives later.

For the frequency, Slawinski gives a bench procedure rather than a calculation. Set the start oscillator as close as possible to the coil's operating frequency, find that by probing primary current and a bridge output and looking for the lowest overshoot, and expect to adjust it once the bridge is live. The spec sheet explains why the setting has some slack, in its note on bench testing a bridge without feedback: the further the drive frequency is from resonance, the slower the primary current builds.

Close to resonance the current arrives fast and the handover happens early, but that is also where an error in either direction matters most. Away from resonance the current builds slowly, which is useful on the bench and the long way round on a coil.

AstRii also went after his fault with this setting. Tuning the start-up frequency helped a little, he reported, but even with it matched to the resonant frequency some pulses were still skipped. Check the idle state first and the frequency second.

The numbers

UD3 firmware defaults, read from cli_common.c. The descriptions of start_freq and start_cycles come from the project's own parameter list. Those for min_fb_current and max_tr_pw are the CLI (command line) help strings in the source, the first because the wiki does not list that parameter at all and the second because the wiki words it differently:

start_freq       63.0 kHz   "The start oscillator resonant frequency in [kHz]"
start_cycles     3          "Number of start cycles with the internal oscillator"
min_fb_current   25 A       "Current at which to switch to feedback"
ct1_ratio        600 turns
ct1_burden       3.3 ohm    2 W, per the UD3B spec sheet
max_tr_pw        1000 µs    "Maximum TR PW [uSec]"
max_fb_errors    0          off

To size the blind window: start cycles are full cycles on both boards. The UD+ page says so, and the UD3 multiplies start_cycles by two before writing it into the zero crossing counter, since a full cycle has two zero crossings. So the blind window is a count of cycles of the bang, and its share of the bang is cycles over cycles, which needs no frequency at all:

3 start cycles at 63.0 kHz                47.6 µs        [derived]
the longest bang the firmware passes
  by default, max_tr_pw 1000 µs,
  at that same 63.0 kHz                   63 cycles      [derived]
3 of those 63                             4.8 per cent   [derived]

63 cycles is the longest bang the firmware allows, not a typical one, so 4.8 per cent is the best case. The fraction rises in proportion as the bang gets shorter. AstRii's own test is 170 µs at about a hundred bangs a second. Later in the same thread he models a primary matched to his coil and finds it resonant between 125 and 135 kHz, which puts that bang at 21 to 23 cycles [derived]. Three start cycles would be 13 to 14 per cent of it [derived]. The UD+ dial goes to fifteen, which on a bang that size would be roughly two thirds of it [derived].

For a QCW the spec sheet suggests up to eight start cycles, as guidance rather than a default. Against a ramp that is negligible. Our reference QCW runs a 25 ms ramp while its frequency slides from 519.1 down to 463.3 kHz, so the ramp holds somewhere between 11,500 and 13,000 cycles [derived], and eight of them is under a tenth of a per cent of it [derived]. But the UD3 will not accept 519.1 kHz. start_freq is declared with a range of 0 to 5000 and a display divisor of ten, so the CLI ceiling is 500.0 kHz and set start_freq 519.1 comes back E:Range. The spec sheet says the same in prose twice, once as experience, "The UD3 has been tested at frequencies of 30khz to 500khz", and once as advice, "This driver is not recommended for applications over 500khz due to this timing resolution limitation." So the QCW figure above is arithmetic about a ramp, not about a UD3 driving one.

The handover threshold in volts, from the expression in ZCDtoPWM.c:

25 A / 600 turns x 3.3 ohm  =  0.1375 V at the burden   [derived]

The setting is in amps of primary current, and the firmware converts it using the ratio and burden you have configured. Change either in the configuration and the volts move while the amps stay put. What moves the amps is a change the firmware is not told about. Swap the ring or the burden resistor on the board without updating the configuration, and the handover current moves by the same factor.

The resolution is coarser than the setting suggests. One step of the threshold DAC is 0.016 V, so at that ratio and burden one step is 2.9 A [derived]. The firmware rounds the current limit thresholds but truncates this one, so a min_fb_current of 25 works out at 8.59 steps, is cast to 8, and gives 0.128 V, a real threshold of 23.3 A [derived]. The setting is an integer number of amps on a 2.9 A grid. The number you typed is not the number it uses, and the error is always downward.

min_fb_current is not on the project wiki's parameter list, which was last edited in August 2022. It is in the firmware and the CLI, so a builder working from the documentation will not learn that the handover has a setting at all.

The UD3 senses through one ring. The spec sheet says the lead "is set in software as there is no analog filtering to provide phase lead", so nothing hangs off the feedback transformer. It also says what that transformer is for, "CT: Current transformer input for gate drive synchronization and current limiting", and rules out the other input: the feedback CT is CT1, and CT2 is for the hall effect sensor and may be omitted. One ring does both jobs.

The usual reason given for two rings is that the lead network loads the first stage, and this board has no lead network. Two rings, not one covers the drawback of sharing, which is that it forces the sensitivity down. It also has a builder who did it on purpose: Steve Conner shared one ring deliberately in 2006 and published the circuit, two back-to-back diodes in series with the burden so the feedback tap sees two diode drops more than the protection tap.

Burden over ratio is what these pages call feedback strength, |Z| over N, volts per amp of primary current. At the UD3 defaults it is 3.3 over 600, or 5.5 mV per amp [derived]. The band given for that quantity appears five times across three articles as "Working boards sit around 0.02 to 0.04 V/A", and the UD3 default is 3.6 times below its floor [derived]. That page now limits the band to rings that carry feedback and nothing else. Conner's two shared rings sit at 0.010 and 0.0033 V/A [derived], on either side of the UD3. That spread shows how much sensitivity a shared ring loses.

Our explanation is that the band is a two ring number and the UD3 is a one ring board. The spec sheet gives the current limit comparator a hardware limit of 4.08 V and a maximum current setting of "1.23A*CT-ratio", which at 600 turns is 738 A [derived]. A burden network at 0.02 V/A would reach the same 4.08 V at 204 A [derived]. A ring that carries the protection as well as the phase therefore cannot have a feedback strength anywhere near the band, or the ceiling would sit below the working current. Where the band is still stated flat, with no qualifier, one of the two statements has to change, and we have not decided which.

What to watch out for

Read these before the first start. Three are about the opening cycles of the first bang. The fourth is about what to do after one of them has bitten you.

One more point is about testing. Feeding a signal generator into the current transformer input tests the loop but skips the handover entirely, because the signal is there before the bang is. The UD3 wiki describes this test, with the caution that the input has to be left floating. It is a good test of everything downstream of the sensor and tells you nothing about the start.

What goes wrong

The last two entries are not handover failures. They are listed because their symptom looks the same as a handover fault.

  • Bangs are skipped at random, and the fault sets at the very start of the pulse. The comparator's idle state is undefined, so on some bangs the first push commands a state the output is already in. This is AstRii's report, and it comes from a design difference between the UD+ and the UD2.7 rather than a defect. A weak pull-down defines it.
  • It never starts at all, and the bridge output is sitting at a rail. Charge or device leakage has left the output where the first pulse would put it. A bleeder across the output, or more start cycles.
  • It starts only while a scope probe touches the comparator input. The probe is an antenna. Nothing about the coil is fixed.
  • Tuning the start frequency helps a little and does not cure it. The frequency is not what is wrong. An idle state does not improve gradually.
  • The coil runs, and the running frequency is exactly the number you typed. The handover never happened. On a UD3 that is silent by default: the threshold may be above anything the first cycles reach, or the transformer may be dead or reversed.
  • It stops switching cleanly, just after the start cycles end. On a UD+ that is the validity test rejecting what arrived, because the pulses coming back are longer than 1.5 times the start oscillator's half cycle. Slawinski names this one directly, and the cure is the start frequency.
  • Faults during a frequency sweep, which then stop. Expected. The UD3's tune command deliberately turns the handover off for the duration: the source sets start_cycles and max_fb_errors to zero with the comment "set configuration of fb current detector to ignore missing feedback", and raises min_fb_current to the maximum transient current so feedback never engages. The sweep walks 128 frequency steps, so at the default of one sample each it is 128 open loop shots on purpose [derived from ROWS and autotune_s].
  • It will not start, and the signal at the comparator is small. Not the handover. This is loop gain, and it is what davekni went after first in the 2020 thread above, with the burden jumper back in place. The question there is why there is not enough signal, and the answers are the transformer, its ratio and its burden rather than any of the start settings above.
  • A clone will not start on settings copied from a UD2.7 thread. Also loop gain: the feedback diodes were changed and the signal has to be bigger before anything oscillates. That is on the UD boards page.

There is also an arithmetic trap in that sweep, which we found in the source rather than in anybody's report. min_fb_current is declared as an eight bit value and max_tr_current as sixteen, so the assignment that raises the threshold wraps. At the default 400 A it stores 144 [derived: 400 minus 256]. That is still far above anything a 50 µs test pulse reaches at a sane bus, so the sweep still does what it is meant to. If it ever does not, this is how it will look: a max_tr_current of 256 would store zero, and the sweep would hand over on the first cycle of every shot.

Where next

  • Inside the driver, the four jobs this one is the first few microseconds of.
  • Two rings, not one, the sensor that has nothing to sense until the current is there, and where the feedback strength band is discussed.
  • The bridge and what kills it, where the output bleeder lives.
  • Converting a DRSSTC driver, the other way round the problem: a driver that self-oscillates never has a handover, because it was never waiting.
  • The two poles, because the frequency the start oscillator has to be near is not the one you measured on the secondary.
  • Self-oscillation frequency gives the free-running frequency the board starts at, the one number the loop cannot get from the machine itself.

The start path has no box of its own on the department diagram. It sits inside the driver, between the interrupter's edge and the first signal from the current transformer. The figures above are quoted from the firmware and its documentation or marked as derived with their inputs given, and our own reasoning is marked as ours.

more in SSTC →