Tuning for max power
Graham Armitage, Sat Jan 17 2015, 10:01PMWell, I finally fired up the new coil today, and after some basic tuning and low power tests, I opened her up. Sadly to say the output was well below expectations. The design should handle 700A primary current easily, but the OCD is set at 450A and only trips with a ground strike. Initial arc length was around 28" max but with beefed up primary connections and finer tuning I got it up to 36". She won't go any higher.
I am running 340v (120v voltage doubler) on the DC bus, with the CM300 IGBT H-Bridge and 0.1875uF (5 strings of 4 0.15uF 2kv caps). Coil is oscillating at 96kHz and the whole system is running really cool, so no heating heating concerns. Wave form on the primary looks great with 100us burst length.
The dimensions/specs of the coil are very similar to other coils on the web producing 4'-5' arcs, so I know it can do better. I got similar results with a much smaller coil before, so not sure what is up. Would welcome any ideas on where to focus my debugging.
One thing I just noticed is that increasing the burst length does not extend the ring-up envelop and hence the current - it just staets adding a second envelop after ring-down. Still without any OCD trip. That seems like a tuning issue correct?
Re: Tuning for max power
Andy, Sat Jan 17 2015, 10:16PM
Sad what thats good ,just invert some inputs and you will invert the ouput.
What current flow throught the double were you achiving.
Based on the last better your chocked for input, which addeds to the one above.
Edit
Andy, Sat Jan 17 2015, 10:16PM
Sad what thats good ,just invert some inputs and you will invert the ouput.
What current flow throught the double were you achiving.
Based on the last better your chocked for input, which addeds to the one above.
Edit
Re: Tuning for max power
Graham Armitage, Sun Jan 18 2015, 03:17AM
I did some more testing. The part that is most confusing is why after 100uS the primary starts to ring down, presumably before the current can rise to it's full potential.
I am starting to suspect something may be happening with the interrupter. The interrupter is designed to reduce the burst length if the duty cycle exceeds 5%. Although I don't believe the duty cycle is that high (I am running at low bps). Not sure what else could be turning off the oscillation if the OCD has not tripped and burst length is set to be over 300uS.
The Total input current is only at 4A, which should be MUCH higher - expecting closer to 8-9A at resonance.
Graham Armitage, Sun Jan 18 2015, 03:17AM
I did some more testing. The part that is most confusing is why after 100uS the primary starts to ring down, presumably before the current can rise to it's full potential.
I am starting to suspect something may be happening with the interrupter. The interrupter is designed to reduce the burst length if the duty cycle exceeds 5%. Although I don't believe the duty cycle is that high (I am running at low bps). Not sure what else could be turning off the oscillation if the OCD has not tripped and burst length is set to be over 300uS.
The Total input current is only at 4A, which should be MUCH higher - expecting closer to 8-9A at resonance.
Re: Tuning for max power
Uspring, Sun Jan 18 2015, 12:14PM
It's not unusual for the primary current to drop after having risen to some value. That can be caused by arc loading. The peak current depends on the tuning. I'd try to increase primary L somewhat in order to lower primary fres. You should be able to find a point, where the OCD is just not hit.
Uspring, Sun Jan 18 2015, 12:14PM
It's not unusual for the primary current to drop after having risen to some value. That can be caused by arc loading. The peak current depends on the tuning. I'd try to increase primary L somewhat in order to lower primary fres. You should be able to find a point, where the OCD is just not hit.
Re: Tuning for max power
Kizmo, Sun Jan 18 2015, 12:27PM
This is interesting subject.
Some coilers seem to be very interested about the notching of primary current, and how to tune the coil so that primary current reaches its highest value at first notch, and then cut the drive before second ring up takes place.
But from what i have seen from my big coil, if I run the coil at high power (say more than 15kW DC power), it will not form this notch no matter what, primary current just stops rising and that point seem to be the spot where driving any further doesnt help much at all. At lower powers, below 10kW or so, i can see some heavy notching depending how i tune the primary.
Right now the coil seem to hit its performance peak at burst lengths of about 150 to 190µs at 55kHz.
Kizmo, Sun Jan 18 2015, 12:27PM
This is interesting subject.
Some coilers seem to be very interested about the notching of primary current, and how to tune the coil so that primary current reaches its highest value at first notch, and then cut the drive before second ring up takes place.
But from what i have seen from my big coil, if I run the coil at high power (say more than 15kW DC power), it will not form this notch no matter what, primary current just stops rising and that point seem to be the spot where driving any further doesnt help much at all. At lower powers, below 10kW or so, i can see some heavy notching depending how i tune the primary.
Right now the coil seem to hit its performance peak at burst lengths of about 150 to 190µs at 55kHz.
Re: Tuning for max power
Antonio, Sun Jan 18 2015, 12:46PM
A primary current that first increases and then decreases is normal and correct. This means that the energy coming from the driver is being transferred to the secondary, and not just accumulating in the primary circuit. If you end the burst at the first minimum of the primary current envelope, the system is working as efficiently as possible. If you are using primary current feedback the sistem tends to fall in this situation, but if the resulting burst is short no much energy is transferred. There are several solutions: a) Decrease the primary circuit impedance (more capacitance, less inductance). b) Increase the duration of the "beat", by lowering the coupling. c) Move the excitation frequency to one of the resonances (feedback will move it back). d) Add a breakout point, so the lowered quality factor smooths out the beats.
Antonio, Sun Jan 18 2015, 12:46PM
A primary current that first increases and then decreases is normal and correct. This means that the energy coming from the driver is being transferred to the secondary, and not just accumulating in the primary circuit. If you end the burst at the first minimum of the primary current envelope, the system is working as efficiently as possible. If you are using primary current feedback the sistem tends to fall in this situation, but if the resulting burst is short no much energy is transferred. There are several solutions: a) Decrease the primary circuit impedance (more capacitance, less inductance). b) Increase the duration of the "beat", by lowering the coupling. c) Move the excitation frequency to one of the resonances (feedback will move it back). d) Add a breakout point, so the lowered quality factor smooths out the beats.
Re: Tuning for max power
Kizmo, Sun Jan 18 2015, 12:52PM
If I remember right, expert i talked to said that disappearing notches at heavy spark loads is to be expected too, i just cant remember what the mechanism and theory behind this was.
Kizmo, Sun Jan 18 2015, 12:52PM
If I remember right, expert i talked to said that disappearing notches at heavy spark loads is to be expected too, i just cant remember what the mechanism and theory behind this was.
Re: Tuning for max power
Graham Armitage, Sun Jan 18 2015, 02:29PM
Thanks for the suggestions. Some are easier to try than others. The part I am not understanding is why the ring-up stops at 150uS regardless of the burst duration (in this case 300uS). Looking at the primary current running at max power (pic below) this is what seems to be happening. I am using primary current feedback and ending the burst at or after the ring-down. Increasing the burst duration does not extend the envelope amplitude, it just distorts it or starts adding another envelope at the end of the burst. Does the ring-up stop because energy is being transferred from the secondary via the arc?
I do have a break-out and the torroid is a standard 4"x17" duct torroid - if that influences anything. I will try reduce the coupling first as this is easiest.
The primary current is around 375A and time div is 50uS.
So I tried reducing the coupling and it did extend the envelope a little, but it just flattened out and did not keep ringing up. Here are the specs on the primary coil.
C - 0.1875uF
L - 12.7uH
Z - 8.23ohm
Graham Armitage, Sun Jan 18 2015, 02:29PM
Thanks for the suggestions. Some are easier to try than others. The part I am not understanding is why the ring-up stops at 150uS regardless of the burst duration (in this case 300uS). Looking at the primary current running at max power (pic below) this is what seems to be happening. I am using primary current feedback and ending the burst at or after the ring-down. Increasing the burst duration does not extend the envelope amplitude, it just distorts it or starts adding another envelope at the end of the burst. Does the ring-up stop because energy is being transferred from the secondary via the arc?
I do have a break-out and the torroid is a standard 4"x17" duct torroid - if that influences anything. I will try reduce the coupling first as this is easiest.
The primary current is around 375A and time div is 50uS.
So I tried reducing the coupling and it did extend the envelope a little, but it just flattened out and did not keep ringing up. Here are the specs on the primary coil.
C - 0.1875uF
L - 12.7uH
Z - 8.23ohm
Re: Tuning for max power
Antonio, Sun Jan 18 2015, 04:42PM
I am not liking the appearance of the primary current. The envelope should be more rounded at the top and almost symmetrical. Look at the voltage waveform at the bridge output.
Antonio, Sun Jan 18 2015, 04:42PM
I am not liking the appearance of the primary current. The envelope should be more rounded at the top and almost symmetrical. Look at the voltage waveform at the bridge output.
Re: Tuning for max power
Graham Armitage, Sun Jan 18 2015, 05:11PM
Antonio, if I increase the burst length it does become more rounded at the top, but no increase in amplitude. I can't run at full power with my isolation transformer (not big enough) and so don't want my scope probe near the bridge output. Here is a pic of the Bridge output and primary current during a lower power test with the isolation transformer and after ZCS tuning. This looks normal to me.
Graham Armitage, Sun Jan 18 2015, 05:11PM
Antonio, if I increase the burst length it does become more rounded at the top, but no increase in amplitude. I can't run at full power with my isolation transformer (not big enough) and so don't want my scope probe near the bridge output. Here is a pic of the Bridge output and primary current during a lower power test with the isolation transformer and after ZCS tuning. This looks normal to me.
Re: Tuning for max power
Uspring, Sun Jan 18 2015, 06:47PM
You're shutting off the bridge at peak current. You can see that by the phase reversal of bridge output at that point.
Uspring, Sun Jan 18 2015, 06:47PM
You're shutting off the bridge at peak current. You can see that by the phase reversal of bridge output at that point.
Re: Tuning for max power
Graham Armitage, Sun Jan 18 2015, 07:02PM
On the low-power test, yes I am turning it off. At full power, I can't see the bridge voltage as I cannot scope it at that voltage (not isolated and too high for my probe). The interrupter I am using has a burst length to 300uS but even at 300uS it still rings down after 150uS. That makes me think the energy is being transferred before it has rung up fully. Does this make sense? If that is true, what would cause that? I am running at low BPS so duty cycle protection in the interrupter should not be not an issue.
Graham Armitage, Sun Jan 18 2015, 07:02PM
On the low-power test, yes I am turning it off. At full power, I can't see the bridge voltage as I cannot scope it at that voltage (not isolated and too high for my probe). The interrupter I am using has a burst length to 300uS but even at 300uS it still rings down after 150uS. That makes me think the energy is being transferred before it has rung up fully. Does this make sense? If that is true, what would cause that? I am running at low BPS so duty cycle protection in the interrupter should not be not an issue.
Re: Tuning for max power
loneoceans, Sun Jan 18 2015, 07:09PM
Is it possible for you to reduce the tank impedance by adding more primary capacitance? Also is it possible for you to find a way to increase the voltage to your bus? Perhaps you might be able to find a 208 or 240V outlet (e.g. from the dryer). Your CM300 bridge should be good for lots more volts.
Oh one thing I noticed - your toroid looks very high off your secondary coil's top winding. I'd drop the toroid lower otherwise you might start to see corona discharge from the top of your secondary's winding.
[edit] I see you're using a half bridge instead of a full bridge (is that correct?). Try to get 500-600VDC on your bus with 208V rectified. You'll be able to get a faster ring up and you'll get significantly longer sparks with a shorter drive time.
loneoceans, Sun Jan 18 2015, 07:09PM
Is it possible for you to reduce the tank impedance by adding more primary capacitance? Also is it possible for you to find a way to increase the voltage to your bus? Perhaps you might be able to find a 208 or 240V outlet (e.g. from the dryer). Your CM300 bridge should be good for lots more volts.
Oh one thing I noticed - your toroid looks very high off your secondary coil's top winding. I'd drop the toroid lower otherwise you might start to see corona discharge from the top of your secondary's winding.
[edit] I see you're using a half bridge instead of a full bridge (is that correct?). Try to get 500-600VDC on your bus with 208V rectified. You'll be able to get a faster ring up and you'll get significantly longer sparks with a shorter drive time.
Re: Tuning for max power
Graham Armitage, Sun Jan 18 2015, 08:13PM
It will be difficult to change the architecture now - not much room for more caps. I am using an H-bridge. Changing bus voltage is tough as all components are rated for the 340v. Few places I go have 220v power, so I wanted to keep it at 120v and try get the 4' I am looking for. The specs are very close to Eric Goodchilds 2400w Twin coils (running at 380v bus) so I believe 4' is a realistic expectation. I can lower the toroid - that would be possible.
I just filmed the coil
with corresponding CT monitoring of the primary as I sweep the interrupter from zero to 300uS. You can see that after 150uSec, the primary rings down (never going over 375A), even though the burst length increases. I am really starting to suspect that the interrupter is shutting down despite what I think it's doing. I have not scoped the gate signal while running so maybe try that next.
Ok - I just scoped the GDT input and the burst length goes to a full 300uSec. So can't blame the interrupter any more. Therefore the primary is ringing down at 150uS on it's own - presumably when all energy is transferred.
Graham Armitage, Sun Jan 18 2015, 08:13PM
loneoceans wrote ...
Is it possible for you to reduce the tank impedance by adding more primary capacitance? Also is it possible for you to find a way to increase the voltage to your bus? Perhaps you might be able to find a 208 or 240V outlet (e.g. from the dryer). Your CM300 bridge should be good for lots more volts.
Oh one thing I noticed - your toroid looks very high off your secondary coil's top winding. I'd drop the toroid lower otherwise you might start to see corona discharge from the top of your secondary's winding.
[edit] I see you're using a half bridge instead of a full bridge (is that correct?). Try to get 500-600VDC on your bus with 208V rectified. You'll be able to get a faster ring up and you'll get significantly longer sparks with a shorter drive time.
Is it possible for you to reduce the tank impedance by adding more primary capacitance? Also is it possible for you to find a way to increase the voltage to your bus? Perhaps you might be able to find a 208 or 240V outlet (e.g. from the dryer). Your CM300 bridge should be good for lots more volts.
Oh one thing I noticed - your toroid looks very high off your secondary coil's top winding. I'd drop the toroid lower otherwise you might start to see corona discharge from the top of your secondary's winding.
[edit] I see you're using a half bridge instead of a full bridge (is that correct?). Try to get 500-600VDC on your bus with 208V rectified. You'll be able to get a faster ring up and you'll get significantly longer sparks with a shorter drive time.
It will be difficult to change the architecture now - not much room for more caps. I am using an H-bridge. Changing bus voltage is tough as all components are rated for the 340v. Few places I go have 220v power, so I wanted to keep it at 120v and try get the 4' I am looking for. The specs are very close to Eric Goodchilds 2400w Twin coils (running at 380v bus) so I believe 4' is a realistic expectation. I can lower the toroid - that would be possible.
I just filmed the coil
with corresponding CT monitoring of the primary as I sweep the interrupter from zero to 300uS. You can see that after 150uSec, the primary rings down (never going over 375A), even though the burst length increases. I am really starting to suspect that the interrupter is shutting down despite what I think it's doing. I have not scoped the gate signal while running so maybe try that next.Ok - I just scoped the GDT input and the burst length goes to a full 300uSec. So can't blame the interrupter any more. Therefore the primary is ringing down at 150uS on it's own - presumably when all energy is transferred.
Re: Tuning for max power
Antonio, Sun Jan 18 2015, 09:05PM
The current at 0:11 in the video looks normal. A longer burst may result in longer streamers, since a quite long streamer is already developed. Are you using some feedback? You can look at both sides of the bridge, to see if the switches are operating correctly.
Antonio, Sun Jan 18 2015, 09:05PM
The current at 0:11 in the video looks normal. A longer burst may result in longer streamers, since a quite long streamer is already developed. Are you using some feedback? You can look at both sides of the bridge, to see if the switches are operating correctly.
Re: Tuning for max power
Kizmo, Sun Jan 18 2015, 09:06PM
I would recommend to double the primary capacitance
Kizmo, Sun Jan 18 2015, 09:06PM
I would recommend to double the primary capacitance
Re: Tuning for max power
Graham Armitage, Sun Jan 18 2015, 09:19PM
I have 5 strings of 4x2000v caps (0.15uF) making up the 0.1875uF. This was designed using the Kaizer Calculator with reliability in mind. I tried to over-engineer everything. If I need more capacitance I can do that, but it's an expensive upgrade, so I need to be sure.
[edit] - I just ran the Kaizer cap calculator again and if I re-configure the MMC to 3 in series and 6 strings, I will have .3uF (vs .1875) and still well below the peak voltage. If safe rule of thumb is 10 x bus voltage I am at 3400v. According to the calculator I should see 3300v peak. This may work without buying more caps
Graham Armitage, Sun Jan 18 2015, 09:19PM
The current at 0:11 in the video looks normal. A longer burst may result in longer streamers, since a quite long streamer is already developed. Are you using some feedback? You can look at both sides of the bridge, to see if the switches are operating correctly.The feedback is a 625:1 CT on the primary between MMC and Primary Coil. When I tested the bridge at lower power I confirmed that both sides are working properly. I think we would see an imbalanced oscillation if it was not.
I would recommend to double the primary capacitanceI just did a test by replacing the 4"x17" toroid with a 6"x19" and got significantly better results. Obviously retuned the primary, but the current rose close to 500A with 46" sparks. My assumption is that more energy can be stored in the secondary system allowing a longer ring-up?
I have 5 strings of 4x2000v caps (0.15uF) making up the 0.1875uF. This was designed using the Kaizer Calculator with reliability in mind. I tried to over-engineer everything. If I need more capacitance I can do that, but it's an expensive upgrade, so I need to be sure.
[edit] - I just ran the Kaizer cap calculator again and if I re-configure the MMC to 3 in series and 6 strings, I will have .3uF (vs .1875) and still well below the peak voltage. If safe rule of thumb is 10 x bus voltage I am at 3400v. According to the calculator I should see 3300v peak. This may work without buying more caps
Re: Tuning for max power
Uspring, Mon Jan 19 2015, 12:37PM
Graham Armitage wrote:
. Without going into the details there: Qpri is independent of Zpri as long as primary resonant frequency remains unchanged and also secondary parameters. That implies, that the effective series resistance R will decrease proportionally to a decrease of Zpri. That is what Kizmo suggested in order to get primary current up.
Another method to increase Qpri and thus get to higher currents is to detune the primary to lower frequencies, i.e. to tap your primary for larger inductance.
Changing secondary Q by adding e.g. more top load also can change Qpri. The effect is not so predictable, since that will also change tuning.
Uspring, Mon Jan 19 2015, 12:37PM
Graham Armitage wrote:
I just filmed the coil with corresponding CT monitoring of the primary as I sweep the interrupter from zero to 300uS. You can see that after 150uSec, the primary rings down (never going over 375A), even though the burst length increases.The video suggests strongly, that arc loading limits primary current at the point where current doesn't increase anymore with longer bursts. The effect of arc loading looks much like a series resistor appearing in the primary tank. Formally it can be described as R = Zpri/Qpri, where Qpri is an effective Q caused by secondary loading. This has been discussed in
. Without going into the details there: Qpri is independent of Zpri as long as primary resonant frequency remains unchanged and also secondary parameters. That implies, that the effective series resistance R will decrease proportionally to a decrease of Zpri. That is what Kizmo suggested in order to get primary current up.Another method to increase Qpri and thus get to higher currents is to detune the primary to lower frequencies, i.e. to tap your primary for larger inductance.
Changing secondary Q by adding e.g. more top load also can change Qpri. The effect is not so predictable, since that will also change tuning.
Re: Tuning for max power
Graham Armitage, Mon Jan 19 2015, 02:25PM
Thanks Uspring, I was just reading that post you referenced last night. Very interesting stuff. One good thing about a coil not working as expected is that you get to learn a lot
I also read a post where Steve Conner spoke about ideal surge impedance being 5 to 10 times inverter impedance - given by (4/pi)*(DC bus voltage/current limit). If I read this correctly, for my bus voltage of 340v and CM300 the Inverter impedance is 0.72. (divided by 2 for the H-bridge). The tank impedance with the increased capacitance is now 6.33ohms or 8.8 times inverter impedance. Assuming my understanding of the equation is correct. It was 13.6x with the previous MMC. So this does sound better. Putting it all back together now - will see how that works.
[edit]
well, I rebuilt the MMC (now .3uF) and while I was about it, re-designed the layout of components, beefed up the connectors etc. Redid the ZCS tuning and did a whole set of measurements before putting all back together. Results were definitely improved by increasing the Cpri. By the time I got it tuned and running properly, she was delivering 48" arcs with a peak current of 500A. The OCD is set at 600 and only trip with ground strikes. I have not tried the bigger toroid, but I am happy with the results. As I said i am looking for super-reliability rather than max spark length. 4ft is manageable for most indoor performance.
Thanks to all for help in resolving this. Hopefully this thread will serve others too running into this issue.
Graham Armitage, Mon Jan 19 2015, 02:25PM
Thanks Uspring, I was just reading that post you referenced last night. Very interesting stuff. One good thing about a coil not working as expected is that you get to learn a lot
I also read a post where Steve Conner spoke about ideal surge impedance being 5 to 10 times inverter impedance - given by (4/pi)*(DC bus voltage/current limit). If I read this correctly, for my bus voltage of 340v and CM300 the Inverter impedance is 0.72. (divided by 2 for the H-bridge). The tank impedance with the increased capacitance is now 6.33ohms or 8.8 times inverter impedance. Assuming my understanding of the equation is correct. It was 13.6x with the previous MMC. So this does sound better. Putting it all back together now - will see how that works.
[edit]
well, I rebuilt the MMC (now .3uF) and while I was about it, re-designed the layout of components, beefed up the connectors etc. Redid the ZCS tuning and did a whole set of measurements before putting all back together. Results were definitely improved by increasing the Cpri. By the time I got it tuned and running properly, she was delivering 48" arcs with a peak current of 500A. The OCD is set at 600 and only trip with ground strikes. I have not tried the bigger toroid, but I am happy with the results. As I said i am looking for super-reliability rather than max spark length. 4ft is manageable for most indoor performance.
Thanks to all for help in resolving this. Hopefully this thread will serve others too running into this issue.
Re: Tuning for max power
loneoceans, Tue Jan 20 2015, 10:50PM
Just a note on reliability - with your primary impedance at 6.3 ohms, at 500Apk, the MMC will see 3165V. You'll see that the CDEs are rated 500VAC per capacitor. Ideally for best reliability I would size the MMC using the AC voltage rating (not DC, and remember the AC rating is for 50/60Hz not tesla frequencies). Sadly this will mean 4x the number of capacitors!
However for hobby use, the MMC you have should be fine. :) Well done on the build and glad it worked out well!
loneoceans, Tue Jan 20 2015, 10:50PM
Just a note on reliability - with your primary impedance at 6.3 ohms, at 500Apk, the MMC will see 3165V. You'll see that the CDEs are rated 500VAC per capacitor. Ideally for best reliability I would size the MMC using the AC voltage rating (not DC, and remember the AC rating is for 50/60Hz not tesla frequencies). Sadly this will mean 4x the number of capacitors!
However for hobby use, the MMC you have should be fine. :) Well done on the build and glad it worked out well!
Re: Tuning for max power
Graham Armitage, Wed Jan 21 2015, 01:02AM
Thanks Loneoceans, I hope to post some pics of the construction tomorrow. I am happy that it works.
I am interested in your comments on the CDE caps - I have always worked off DC ratings for Tesla capacitors and the Kaizer calculator does too. Based on this calc the 3kv rise across the MMC is 1/2 of what I have (assuming DC rating of course). If the true rating is 500v, then wouldn't I have blown them already? They are not even getting warm.
Graham Armitage, Wed Jan 21 2015, 01:02AM
Thanks Loneoceans, I hope to post some pics of the construction tomorrow. I am happy that it works.
I am interested in your comments on the CDE caps - I have always worked off DC ratings for Tesla capacitors and the Kaizer calculator
does too. Based on this calc the 3kv rise across the MMC is 1/2 of what I have (assuming DC rating of course). If the true rating is 500v, then wouldn't I have blown them already? They are not even getting warm.Print this page