Re: SSTC feedback and general design
Dr. Dark Current, Tue Feb 17 2015, 07:46AM


Then 100% go for the 4046. If you know what you're doing, you can design the circuit for proper operation on first light (after a quick tuning by one pot). The Shmitt trigger feedback method usually has problems with losing signal during ground arcs, which often results in popped transistors or the coil going out of oscillation in the better case.

For a normal SSTC operating at a higher frequency (say >200 kHz), you should strongly consider MOSFETs above IGBTs.

Re: SSTC feedback and general design
Gregary Boyles, Tue Feb 17 2015, 03:10PM

I decided to make my own schmitt trigger inverter out of dual comparator (LM393) for the convenience of being able to power it from the same Vcc that powers my 555s etc (16V). And LM339 is a smaller IC as well.

I found a calculator on line and just plugged in the trigger voltages specified for 15V supply in the datasheet for 74HC14. I used trim pots to get the exact resistance values as specified by the online calculator.

I am yet to try it out but I can't see any reason why it would not work as well as the 74HC14.

Re: SSTC feedback and general design
TwirlyWhirly555, Tue Feb 17 2015, 07:20PM

I did use the 4046 for my SSTC , they work well when setup properly .

How much power do you want to push into the secondary , I warped a plastic secondary after 3 minuets CW with about 900W in from rectified smoothed DC mains at 230V AC , but my secondary was only 4" by 5.5"

Re: SSTC feedback and general design
Sigurthr, Wed Feb 18 2015, 02:20AM

Some basics about the 4046 PLL:
1) The lock range is the range it will stay locked on once it has already locked on previously, a.k.a. the acceptable delta-f range.

2) The capture range is the range of frequencies it will gain an initial lock on.

3) There are two phase comparators to choose from, each with a different operational mode and circuit requirements.

4) the only data on how to select the tuning range is an old low resolution datasheet with curves for 10v and 15v, so you have to interpolate by eye to reach a normal 12V value. No formula for computing R1, R2, or C is given, nor is any information on designing a proper loop filter for the phase comparator used.

5) PC1 induces a 90deg phase shift intrinsically when at the center of the capture range, with a varying phase shift based on where in the range it is set. PC1 allows for a loop filter to determine the lock range in addition to the capture range. It has a tendency to lock on to harmonics of the center frequency of the capture range if the lock range is made large enough to allow for that though.

6) PC2 does not have a lock range based on the loop filter, and only uses the loop filter for capture range. It does however, lock the phase angle of the two signals with no offset.

7) There is a long thread here on 4HV where extremely experienced users (far more than I) have watched as the 4046 VCOin slams from rail to rail when it can't get a lock. Depending on your capture range this certainly can pop a bridge as the maximum frequency of the chip is around 2MHz, and most IC type gate drive barely tolerates 1MHz operation with even only "easy" TO size switches. If I can find the thread I'll link it.


Practical design considerations (sizing R and C properly) aside, which phase comparator you use and which feedback method you use are important to consider because you'll have a phase angle offset to deal with: Secondary topload feedback (antenna) is in phase and needs to be directly duplicated in the drive. This means you either use PC2 or induce a correcting phase shift when using PC1. Remember that the phase shift in PC1 varies on where in the range the lock is. Secondary base current feedback (CT) will have a phase shift compared to the topload voltage, so this needs to be compensated for somewhere in the drive circuitry.



Schmitt Trigger feedback only pops switches on loss of feedback when poor construction techniques or bridge design is used. If you shield your feedback circuit from noise and lock down your bridge with protection diodes and snubbing capacitors the worst that will happen is the arc will go out and oscillation will stop because the secondary was shorted out and there isn't enough gain in the feedback circuit to meet the trigger threshold. Even this only really happens if you use antenna or feedback coil feedback systems. If you use secondary base CT then your feedback circuit is watching the current flow, not the voltage output, and current flow is at maxima during a ground arc, so oscillation will continue.

The main drawback to using a CT for feedback is that oscillation is not self starting because the dc conductivity of the burden resistor pulls the output to ground when no current is flowing through the secondary, so you have to start oscillation by pinging the bridge. Many gate drive chips will do this, but some require a transition on their enable input to send the ping. This can often be applied intrinsically upon power on, but then the power on sequence has to be that the bridge is powered before the driver otherwise the single ping will be missed. If you are operating in interrupted mode you don't encounter this problem as pings are repeated. The exact chain of events resulting in a failure to self start oscillation is this: burden resistance pulls driver input to ground, driver hard pulls gate drive chip input to ground, chip thinks it should remain output low, no ping is sent. With a floating driver input the gate drive chip input is left floating, and thus a ping gets sent. Even if the gates were totally uncharged at power off and thus inputs were at ground potential, having them floating will allow leakage currents and noise received to cause a transition eventually resulting in a ping.

If a bridge pops during a ground arc and it was still oscillating at resonance the cause is usually that the designer didn't take into consideration the ground arc peak current level.

I would like to point out that I moved away from using the 74HC14 because I achieved much better performance and easier implementation with alternatives which accept the normal 12V Vcc level.

Ohh, and I agree with Dr. Dark Current about using FETs instead of IGBTs, but I would move that transitional boundary down to 100KHz, as I've never got IGBTs to run well above there.

Remember that you need to decide upon a power level requirement too, as it makes a huge difference.

Re: SSTC feedback and general design
GrantX, Fri Feb 20 2015, 06:46AM

Thanks for the input everyone, it's been extremely helpful. I've definitely got some research on 4046-based builds to do now. Even though I'm aiming for a reliable system at the end I'm happy to experiment with different topologies at reduced power in the meantime. This is a learning experience, so the coil will probably go through several iterations before I decide to move on.

As for the maximum power levels, I think 500W CW for very short runs is enough and I was hoping to squeeze in more during half-wave and interrupted modes (obviously this requires a change in primary inductance otherwise a half wave input will simply be 50% the CW value). I was planning on making a tapped primary so I can easily adjust the input current, as well as using a 2A 230V variac for finer control (until the system is running smoothly straight from mains). As TwirlyWhirly555 pointed out, I don't want to melt my PVC coil form. I was wondering if the spreading the primary out along the length of the secondary a little bit could help reduce heating, or is it the rather large RF currents in the bottom few windings that are the main problem? I remember seeing scope shots that showed many amps on the ground line, but I can't remember what sort of coil it was.

I'm almost certainly going for a feedback CT, and it makes sense that the signal strength will rise during ground arcs. I could include a momentary switch on the enable pins of the driver IC's to kick-start oscillation. I'll just have to figure out a good way to route the switch wiring so that it remains adequately shielded and doesn't add too much parasitic inductance.

As for IGBT's versus MOSFET's, I can can order some FET's if the performance gain is worth it (I'll probably go for SOT-227 devices in that case. There's no such thing as "overbuilt"). The only reason I chose these particular IGBT's is that Loneoceans used them in his 'ramped' SSTC with great results and my resonator is only 10 or 20 kilohertz higher. The IGBT's were cheap and very fast (137ns total typical delays), so I'll end up using them in another project if they're unsuitable here.

I want all the logic to be run from a single 12V rail, so If I do try out the Schmitt trigger option I'll probably look for something other than the 74HC14. Sigurthr, which IC did you switch to that performed better?

My bridge design includes beefy screw-terminal snubber capacitors across the DC rails and TVS's across both switches. Connections will be very short and a ground plane will be glued underneath the bridge with decoupling capacitors, plus a grounded aluminium case for the logic. The heatsinks for the bridge switches aren't massive, but they are almost 3 times larger than those used in Loneoceans' R-SSTC, and I plan to have a very high airflow over them (my friend found some small industrial blowers and commented that I should make a "twin-turbo-style cooling system," which could look rather nice, haha). Hopefully I'll end up with a bridge that is slightly over-spec for the intended power level.

I think I'll have to order a few 4046 IC's to experiment with. I'm understanding the theory (somewhat, hah) but I probably need to see them in action to get a feel for their behaviour and practical issues. If I can get them to play nice I'll definitely switch from simple feedback to PLL. At least then I have some peace of mind that the coil is less likely to explode in my face.

One more thing, regarding the DC input. I wanted to include the un-smoothed half wave supply as a sort of "poor man's sword spark" option. I figured that the smooth voltage ramp combined with the rather high operating frequency could produce something similar to the straight sparks seen in VTTC builds. I don't expect to achieve the ridiculously long sparks produced by QCW's, but at least 150mm (6") discharges will be satisfactory. :)

Re: SSTC feedback and general design
Sigurthr, Fri Feb 20 2015, 07:43AM

Regarding secondary heating; there are two main causes...
1) resistive losses (iiR) from secondary current
2) eddy currents induced by the primary

Primary heating has the same two sources of heat, but instead of secondary current it is the primary current.

500W CW can be achieved with unlimited duty cycle without too much trouble. I have my dual mode coil (feedback, and VCO) which runs at ~450W CW unlimited duty with only passive cooling, and it is hard switched. For 350-600KHz expect about ~5cm CW bush maximum for 500W output. Streamer length of interrupted drive will of course be longer, but 15cm may be unattainable at such a high f0. It has been a long time since I've run my dual mode coil (350KHz 450W) in interrupted mode, but I think I got about 10-12cm streamers at optimum.

I use the CD40106BEE for my schmitt trigger. Note that choice of input coupling capacitor weighs heavily on operation stability.

I'm all for anyone diving full on in to working out the kinks on the 4046 for SSTC use, as a solid solution would benefit us all. Practical implementation issues aside, the main "failure" modes I've witnessed first hand are the following:

1) PLL won't lock, slams rail to rail trying to lock
2) PLL won't lock, doesn't even try and just sits at the frequency set by trimmer no matter the input signal, or lack there of.
3) PLL locks on to harmonics during ground strike
4) PLL locks initially, then loses lock during ground strike and then can't lock again. It then follows #2.
5) nonlinear tuning range; example: first 40% of resistance value of pot tunes linearly from 450KHz to 580KHz, next 10% of pot the frequency jumps up to 700KHz and VCO increases exponentially with voltage increment from pot up to over 2MHz.

#5 was a circuit someone asked me to debug and not of my own design.
#1 was reported by several here on the forum, but I couldn't replicate the issue

Re: SSTC feedback and general design
GrantX, Tue Jul 14 2015, 05:21PM

Updates! With photos!

Sorry for the long hiatus. I moved house, broke a bone, damaged a ligament and started a new job. Obviously it hasn't turned out to be the best year for hobby projects.

Anyway, I've decided to simplify the first prototype, as it'll make life a lot easier while I'm trying out different switching devices, DC bus input and interrupters (basically, a pile of old parts that need to be used). When they finally get delivered, I'll be using the good ol' 74HC14 with the usual pair of UCC drivers and a feedback CT. I'm still planning to buy some 4046 ICs to play with, as well as investigating other methods.

I've taken the advise regarding MOSFETs onboard and dug out a pair of FCA47N60F FETs. 600V 47A, made by Fairchild. They look a little bit slower than the FGH40N60SMD IGBTs on paper, but hopefully they'll be good enough for lower power operation. The half bridge itself was built entirely from stuff laying around, so if the smoke comes out I'm not too worried. It'll be interesting to see the difference between these FETs and IGBTs.

Hopefully I'll be able to get it up and running in the next few weeks, depending on how many free weekends I can snatch. Anyway, a few pics of the recent progress:


150mm x 38mm aluminium toroid


The whole resonator. I still need to replace the ground line with a heavier cable, and cover up that hideous glue connection between the topload support and coil.


The unfinished 555 interrupter and linear-regulated DC supply.



The 47N60F MOSFET half bridge. The heatsinks are a little smaller than I would have liked, but that's all I had laying around. A fan will be mounted above them to help improve the situation. The DC blocking caps are 800VDC 5uF PP snubbers, there's TVSs and ultrafast diodes across each FET and back-to-back zeners across the gate/source. The gate resistors are 12 Ohm, with ultrafast diodes for fast turn-off.

And finally, just for fun, I dug out EVR's old SSTC 1.0 driver and hooked it up to the resonator:


The little purple flame is around 2cm long.

Re: SSTC feedback and general design
Wolfram, Wed Jul 15 2015, 05:15PM


I've played a lot with the 4046 (74HC4046) for my induction heater work. There are some traps, and I think I've fallen into every one of them. Along the way, I figured out how to make a stable design, the three crucial points to making a stable design as follows:

1: Phase comparator 2 is nice in that it locks with zero phase offset over the whole frequency range, but it's edge triggered, which is bad news around any noisy power electronics. It can be made to work, but I wouldn't trust it for reliability. Phase comparator 1 with an integrator in the loop will lock with a constant (adjustable, around 90 degrees) phase offset over the whole frequency range. Capture range doesn't make much sense to talk about, since the PLL is essentially trying to lock to itself. The frequency it gets fed with is always the frequency the VCO is running at, only the phase is different.

2: The lock range needs to be limited by limiting the VCO input voltage range. Some 4046es behave strangely around the ends of the VCO input voltage range. Most designs try to choose an appropriate lock range by changing R1/R2/C, but this doesn't help when the VCO does strange things at the ends of the range. This is especially bad with the 74Hc4046, and was the reason for a lot of strangeness in my designs, until I limited the VCO voltage range.

3: The feedback needs to be right. The signal going into pin 14 of the 4046 needs to be strong enough (usually > 100 mV) over the whole frequency range that the VCO can run at, and it needs to be nicely behaved with regards to the phase. Some phase shift difference over the VCO frequency range is acceptable, but if the phase does crazy things then you might get trouble. This ties together with point 2, limiting the lock range makes it easier to design a feedback network that works over the entire range.

Since phase comparator 1 is the best option, you have to live with the 90 degree phase shift. This means that your feedback has to be either tank cap voltage (in a double resonant coil) or tank current shifted by 90 degrees. For a regular SSTC, antenna feedback should work, but I'm not sure how predictable the phase shift is.

If you have trouble with the PLL sticking, not locking or similar, troubleshooting can be done in the following way. Disconnect the output of the loop filter from the VCO input, connect a potentiometer to the VCO input. Disconnect the second phase comparator input (pin 3) from the VCO output and ground it. Use an oscilloscope to observe the output of phase comparator 1 (pin 2) and the VCO output (pin 4). With the resonator connected, adjust the VCO voltage over the range that the loop filter can control it. The phase comparator output and the VCO output should be more or less in phase when above Fres and out of phase below. If the phase shift does strange things or there are glitches, you know the problem. Limit the lock range or improve the feedback network until you get a clean square wave out of the phase comparator. Make sure to run the bridge at a voltage where you won't damage anything when it's out of tune during this procedure.

Re: SSTC feedback and general design
GrantX, Sat Jul 18 2015, 06:47AM

Thanks for the info, Sigurthr and Wolfram. Once I've ordered the 4046 ICs this will be very helpful.

I need a sanity check regarding the SSTC's grounding system. I over analyzed it and now I can't decide which option is best. Here's a rough block diagram of the whole system:

The parts I'm fussing over at the moment are the LV DC supply, driver and interrupter, with regard to grounding.
Would it be better to:
- 1) Connect the common negative LV DC bus directly to the grounded enclosure(s)
- 2) Connect the LV DC bus to ground via decoupling caps
- 3) Leave the LV DC supply floating (along with the driver and interrupter)

It makes sense to leave it all floating (which appears to be the case in Steve Ward's and Eric Goodchild's Mini and Interrupted SSTCs), but tying the DC rails to mains earth via decoupling caps (just like the 325V DC bus) also seems viable. I'm worried that connecting the negative to earth, or even just adding the decoupling caps, will create a path for interference.

Originally, I planned for the interrupter to be handheld and connected to the coil via a shielded RCA cable, with a 12V wall wart for power. I'm now going to power the interrupter from the same 12V supply powering the driver in the coil. Having two different 12V supplies separated by a few meters and both connected to the coil sounded like a recipe for hard-to-eliminate interference (especially since the 12V wall wart's lead is about 2.5 meters long and will act as an antenna).

Does any of this look correct? If I can get the layout, shielding and grounding correct first, I can hopefully play with different bridges, drivers and feedback methods without trying to debug strange behavior.

Re: SSTC feedback and general design
Sigurthr, Sun Jul 19 2015, 03:54AM

FWIW I always leave my LV supply floating.

My rationale is this: neg rail is floating in respect to earth so that chip inputs are not ground referenced. Thus being in the E field of the TC does not impose a high voltage on the components when no shielding is used. As the voltage imposed on a point in the E field is a result of the radius between the origin of the field and the location of that point, two points only see the difference in radius. Given that the distances between sensitive nodes inside the driver are small, there will be negligible field gradient change and thus negligible voltages between all sensitive points. If you tie negative rail to earth you lose this.

Re: SSTC feedback and general design
GrantX, Sun Jul 19 2015, 04:27PM


Thanks, leaving it floating does make the most sense.

I think I'm going to simply install the interrupter into the main driver's grounded enclosure in the base of the coil, since I won't be dodging any huge sparks. I'd need a multi-core shielded cable, or two RCA AV cables (or other coax connectors), to connect the handheld interrupter to the coil: for 12V power, interrupter output, and protective earth for the metal case. Don't really feel confident about relying on RCA cables for protective earthing in a mains-voltage environment. It makes more sense to put the money and effort towards making the coil's enclosure look a bit nicer.

Quick question, has anyone seen affordable digital panel meters for DC voltage and current, something around 1000VDC and 50A absolute maximum. Power measurement would be nice, but it's not important.
I found these for $20 USD, but having both voltage and current displayed at the same time would be perfect:

Re: SSTC feedback and general design
TwirlyWhirly555, Sun Jul 19 2015, 04:44PM

Why not use fibreoptic between the interrupter and SSTC driver ? , Just power the interrupter with a battery .

Re: SSTC feedback and general design
GrantX, Mon Jul 20 2015, 04:01PM

I've been considering it. At first it seemed a bit extravagant for such a small coil, but it's likely I'll end up building something larger too. I've got a big spool of green enameled 23 AWG wire and another 150mm x 38mm aluminium toroid here already. Lengths of PVC up to 250mm in diameter aren't very expensive here, may as well start winding! Having a slightly fancier interrupter with battery power, more features and a fiber optic link is becoming more appealing now that I'm making plans for bigger coils in the future. More research is certainly needed, I've yet to use fiber optic gear in any projects, but after some quick googling it seems simple enough.

I've almost finished off the half-bridge and DC supply, and will now tackle the driver and feedback CT assemblies. Out of curiosity, does anyone have an estimate to how much power these FCA47N60F MOSFETs and their small heatsinks will handle at 420-460 kHz? They have rather long total switching delays, so I'm not expecting much.

Re: SSTC feedback and general design
Ash Small, Mon Jul 20 2015, 05:54PM


Is it normal to use digital meters in such a 'noisy' environment? Moving coil meters may be more reliable, and can be biased to read any range you want with the appropriate resistors.

Moving coil meters are fairly immune to noise, due to their inherent inductance.

Re: SSTC feedback and general design
GrantX, Tue Jul 21 2015, 04:09PM

I was a bit concerned about noise, but for DC voltage measurement only I think a digital meter should be work. It would be measuring directly across the smoothing capacitor, with MKP decoupling caps. The meter's supply would be isolated and decoupled directly at the input to the meter itself. But DC current and power measurements might be more susceptible to interference. Now it's starting to sound cheaper and less risky to go with reliable analog panel meters. If I could find some fancy illuminated meters that aren't crazily expensive I'd be very happy.

Re: SSTC feedback and general design
Ash Small, Tue Jul 21 2015, 09:44PM


You'll probably have to make your own scales for them, so make them translucent and put some LED's behind, or wherever. Paint the needles with luminous paint, or whatever. Make them as fancy as you like. Shine UV on them if you want, and get the scale and needle fluoresce if you want.

It won't cost much and won't take much time. The bit that will take the time is the scales, but you can print those out on a printer, after using a CAD program, or something (or getting someone else to supply a DXF file, or whatever).