High Voltage Forum

Tesla coils => Vacuum Tube Tesla Coils (VTTC) => Topic started by: kyledellaquila on January 24, 2021, 07:09:52 PM

Title: 833A Cathode Modulation
Post by: kyledellaquila on January 24, 2021, 07:09:52 PM
Hello Team!
I want to create a new thread on simple amplitude modulation techniques for an 833A VTTC oscillator.
I want to be able to create QCW sword streamers off of a 3kV DC supply.

The classical method is to feed the plate with high voltage half wave rectified waves and interrupt the conduction cycle at various points throughout the stroke via Staccato interrupter which has an on and off state with no variability in-between. The number of firing cycles are constrained to 50 or 60hz (depending where you live).

Why in the world would I want to do this with a DC supply? : I want (complete) control over the number of ramps per second.
Here is my proposed schematic at the moment:
(https://i.imgur.com/VTT8DRp.jpg)
The hopes here in what I cobbled together in this schematic is:

Cathode modulation is roughly based on the discussions here:
http://amfone.net/Amforum/index.php?topic=43982.0 (http://amfone.net/Amforum/index.php?topic=43982.0)
and here:
http://amfone.net/Amforum/index.php?topic=6051.0 (http://amfone.net/Amforum/index.php?topic=6051.0)

Has anyone done this with success? Will I be exceeding the MOSFET drain to source ratings?

Thank you in advance!
-Kyle Dell'Aquila
Title: Re: 833A Cathode Modulation
Post by: Mads Barnkob on January 25, 2021, 08:56:02 AM
It is a shame that the old website of Dr. Spark is gone, maybe you can find some hints to the VTTC audio modulation from the archive, but it is very sparse with information it seems: https://web.archive.org/web/20090315111936/http://drspark.org/index_1024.htm

There have been two builders that really went into perfecting VTTCs. Drspark.org and audiotesla.org, which both seems to be sites now :( You could try to reach out to the owner of Audiotesla on facebook : https://www.facebook.com/hank.steiffen

There is not even any youtube channels left with all their stuff, I wonder why they removed everything?
Title: Re: 833A Cathode Modulation
Post by: SalinsLV on January 25, 2021, 05:07:56 PM
It seems that we are on a similar path. I'm currently building audio modulated GU5B VTTC with a similar idea that You have. The main difference is that i want to use fiber optics for distant control, hence the received signal is a square wave (PWM) and not quite audio yet. I have a plan to modulate the square wave according to the music, kind of how the class D audio amplifiers work.
Title: Re: 833A Cathode Modulation
Post by: Weston on January 25, 2021, 09:16:22 PM
This all makes sense conceptually. You should probably try simulating it. I think there are some vacuum tube models in SPICE.

The only important change I see would be the bypassing of the filament transformer. You need capacitors going to the center tap of the filament transformer, or else the only path to ground is through the filament transformer itself.

The grid leak network creates a negative grid bias through the action of the grid conducting with positive bias. The RC time constant of the grid leak network might impact your audio modulation.  You might be able to replace it with just an adjustable bias supply. If you move the "833A bias/idle current control" to the return of the grid network I think it should accomplish that.

The voltage across the FET will depend on the bias voltage to drive the tube into cutoff. I believe this will be impacted by your grid leak network and the number of windings on the feedback coil. This is something that simulation may help with.

Title: Re: 833A Cathode Modulation
Post by: kyledellaquila on January 26, 2021, 06:43:43 AM
(https://i.imgur.com/nIS3rBp.jpg)
@Weston: I redrew the diagram with the variable bias supply hooked up to the feedback network. I took the resistor to ground out of the feedback circuit. I figured that resistor wouldn't be doing much here. The power supply and the FET won't need to work so hard in this configuration! Thanks :)

Also I redrew the bypass capacitors at the cathode and hooked up the center tap to where the capacitors meet. What is the danger of having DC currents (opposed) flowing through the 10v transformer? I imagine it should be fine?
What voltage do you think the FET will experience when it is cut off?

I am very out of practice with tools like SPCIE.. All I really know how to use is the Falstdad circuit simulator... which doesn't have the best component simulation libraries.. but it gets the point across to my self in various situations. I have LTSpice but it is like pulling teeth with that software.

@SalinsLV: I am curious to see how you handled your situation! D class amps and fiber optic cables sounds like a high component count to me. I have a light aversion to complicated assemblies. (hence my preference to use a little single MOSFET)

@Mads Barnkob: Its a shame that the wayback machine can't capture those images. Looks like a nice goldmine of schematics and photos! (I assume)
Title: Re: 833A Cathode Modulation
Post by: kyledellaquila on January 26, 2021, 07:04:38 AM
(https://i.imgur.com/x6VfF9N.jpg)
Is the above assumption right? I have already constructed the 833a structure to have a single capacitor underneath. It is probably better this way right?
-Kyle Dell'Aquila
Title: Re: 833A Cathode Modulation
Post by: kyledellaquila on January 27, 2021, 05:43:19 AM
(https://i.imgur.com/cbocFzK.jpg)
I redrew it again with the heater cathode circuit simplified.
I also threw in a SiC C3M0065090D in there as well. It can withstand high DS voltages and high currents.
Title: Re: 833A Cathode Modulation
Post by: Weston on January 28, 2021, 06:00:16 AM
I was not thinking about the DC current flowing through the filament transformer, only the RF. The last scheme should work. The two capacitors to the center-tap should also work.

You might want some sort of LC filter between the cathode and the FET, beyond just the capacitor.


Given the typical currents and possible voltages required for modulation you would probably be better off with a silicon fet (bypassed with a TVS diode if you are worried about over-voltage). The FET is operating in linear mode and could be dissipating a decent amount of power at some modulation points. Silicon parts typically have a bigger die and are better for linear operation (you can check out the safe operating area curves in the datasheets of each part if you want). Something like the IRFP460 would probably work.

You should also consider adding an opamp feedback network around the MOSFET gate so you can have the audio source directly modulate the cathode current.
Title: Re: 833A Cathode Modulation
Post by: kyledellaquila on January 29, 2021, 06:29:52 AM
Weston, thank you for sharing your thoughts and tips..

Quote
Given the typical currents and possible voltages required for modulation you would probably be better off with a silicon fet (bypassed with a TVS diode if you are worried about over-voltage). The FET is operating in linear mode and could be dissipating a decent amount of power at some modulation points. Silicon parts typically have a bigger die and are better for linear operation (you can check out the safe operating area curves in the datasheets of each part if you want). Something like the IRFP460 would probably work.

I came across this FET : https://ixapps.ixys.com/DataSheet/DS100080(IXTK-TX90N25L2).pdf (https://ixapps.ixys.com/DataSheet/DS100080(IXTK-TX90N25L2).pdf)
Which was recommended as a linear operation solution. In the datasheet it has a DC rating for the forward-bias safe operating area.

My question is how to calculate the voltage of DS when the MOSFET is shut off and the 833A above it has 3kv at the plate and no conduction. I don't want to exceed the DS rating of course.

Quote
You might want some sort of LC filter between the cathode and the FET, beyond just the capacitor.
I assume for parasitic oscillation reasons? What values would you select?

Quote
You should also consider adding an opamp feedback network around the MOSFET gate so you can have the audio source directly modulate the cathode current.
Would something like this work?
kerrywong.com/2016/10/08/linear-mosfet-and-its-use-in-electronic-load/ (http://kerrywong.com/2016/10/08/linear-mosfet-and-its-use-in-electronic-load/)
(https://i.imgur.com/Itig0AG.jpg)

and for visualization, and by no means rigorously accurate:
https://tinyurl.com/y2r2cqcj (https://tinyurl.com/y2r2cqcj)
Title: Re: 833A Cathode Modulation
Post by: Weston on January 30, 2021, 09:01:22 PM
A filter network on the cathode is going to be to protect the FET from any RF. Any capacitance on the DS of the FET is going to lead to lower modulation bandwidth, so you would probably want something like a damped CLC (capacitor, inductor, capacitor) filter. If you don't want to bother modeling that, the single capacitor will probably work acceptably.

Linear FETs would of course be ideal, but are typically a bit pricy. Most of the cheap DC loads use FETs not rated for linear operation though, so you can probably get away with out them.

The voltage across the drain of the FET translates to a negative bias on the grid. You can get an idea about the grid voltage swing you will need from the datasheet http://www.ur4ll.net/833a.pdf . I don't think this is something you will be able to easily figure out without modeling it and your grid coil though. You can probably just assume its a few hundred volts and put a TVS across your FET.

That schematic is ~ what I was talking about. But you are missing the shunt resistor on the source of the FET.
Title: Re: 833A Cathode Modulation
Post by: kyledellaquila on January 31, 2021, 07:02:22 AM
(https://i.imgur.com/X6SayXd.jpg)
https://i.imgur.com/YmO31WN.jpg (https://i.imgur.com/YmO31WN.jpg)
I redrew the diagram and spent some time to make a side by side comparison of the different MOSFETs.
Quote
A filter network on the cathode is going to be to protect the FET from any RF. Any capacitance on the DS of the FET is going to lead to lower modulation bandwidth, so you would probably want something like a damped CLC (capacitor, inductor, capacitor) filter. If you don't want to bother modeling that, the single capacitor will probably work acceptably.
I like the simplicity of sprinkling capacitors in the cathode path... How would this lower the modulation bandwidth?

Quote
The voltage across the drain of the FET translates to a negative bias on the grid. You can get an idea about the grid voltage swing you will need from the datasheet http://www.ur4ll.net/833a.pdf . I don't think this is something you will be able to easily figure out without modeling it and your grid coil though. You can probably just assume its a few hundred volts and put a TVS across your FET.
Ah! this is clarifying. It will be tough to model without knowing the voltage swing of the feedback coil.

How can I assure complete amplitude control of the op-amp signal to MOSFET gate? It would be nice to have complete 0-100% volume control over the MOSFET (even thru saturation).
Title: Re: 833A Cathode Modulation
Post by: kyledellaquila on February 02, 2021, 07:08:56 AM
(https://i.imgur.com/fJp7dU3.jpg)
So I have here two methods proposed:
https://tinyurl.com/y2vnymot (https://tinyurl.com/y2vnymot) – represents more of a staccato method of control (where the oscillations are either on or off)
https://tinyurl.com/y66wsdql (https://tinyurl.com/y66wsdql) – demonstrates more of a PWM level of control. Where the width of the pulses also vary.
Title: Re: 833A Cathode Modulation
Post by: Weston on February 02, 2021, 08:55:51 AM
The capacitance on the cathode lowers the modulation bandwidth because the FET also has to discharge it.

In the schematic you showed in the earlier post the resistance is not enough. I assume you want to modulate the cathode current, and the resistor that you have as 1.5m Ohm is a shunt resistor. The feedback loop regulates the current flowing through the FET to be the input voltage / the resistance. I assume you want a max cathode current of ~ 0.5 amps or something. Scale the resistor depending on your expected audio input amplitude. The 1.5m Ohm resistor is way to low in value.

Of the two recent schematics, the first one the opamp is acting as a comparator. It would function like the staccato circuit as you said.

The second circuit you have the feedback loop set up to regulate the voltage across the FET. This is the same as modulating the grid bias, but modulating the grid bias would burn less power.

You probably want to modulate the cathode current, as that is going to be more proportional to the spark size and provide more linear audio.

This system is pretty complicated. You should see if you can get a simulation set up or something. I looked online and could not find a 833A simulation model, but it seems some exist for the 811A. I have been wanting to spend some time modeling and building another VTTC. I will have to see if I can throw together a simulation for my own curiosity when I have time.
Title: Re: 833A Cathode Modulation
Post by: Weston on February 02, 2021, 09:05:14 PM
I had some free time last night and was curious, so I created a model with both cathode modulation and grid modulation in LTSpice, using the secondary model for my QCW tesla coil (so impedance are going to be off a bit). I could not find a 833A model online, but I found a 811A model, which should give similar results. 

The bias between the grid and cathode modulation is different, but both show the modulation. I suggest playing around with the simulations a bit. From these simulations it seems like you are going to need a grid leak network, it plays a role in stabilizing the oscillator. You probably want the RC time constant to be something like 50kHz or greater so it does not interfere with the audio modulation. Here is a screenshot showing the supply current for the two designs.



Cathode modulation is going to be more linear, because you can regulate the current going through the tube, which more directly relates to power. But grid modulation is going to be easier. You could just drive the end of the grid leak network with a suitably bypassed audio amplifier + matching transformer. In either of these cases, you are probably going to want a fuse or some other form of current limiting on the grid. If you are trying out various modulation schemes it may be easy to over-drive the grid and melt it. The grid is a lot more fragile than the plate.

Cathode modulation is going to need to dissipate a power equal to the cathode voltage * average plate current. This could possibly be 100+ watts for a 833A coil, depending on how you dial in your bias. Grid modulation only has to supply a power equal to the average grid current * the modulation voltage, grid current is a lot lower. For the two sims, the power supplied by the source for the grid modulation is only 2W while the power dissipated for the cathode modulation is 12W.

Zipped file of the simulations attached.

Title: Re: 833A Cathode Modulation
Post by: kyledellaquila on February 04, 2021, 07:35:06 AM
Oh thank you Weston for sharing these simulations!
I always get stymied up when I get to the SPICE directives. This helps alot.
I did come across this page (http://www.intactaudio.com/forum/viewtopic.php?t=523 (http://www.intactaudio.com/forum/viewtopic.php?t=523)) Which claims to have 833A models... but I am having trouble importing these.
(https://i.imgur.com/YhSp7UT.jpg)
I will keep fumbling around until I get what I want to see

Quote
You could just drive the end of the grid leak network with a suitably bypassed audio amplifier + matching transformer. In either of these cases, you are probably going to want a fuse or some other form of current limiting on the grid. If you are trying out various modulation schemes it may be easy to over-drive the grid and melt it. The grid is a lot more fragile than the plate.
If I modulate at the cathode with a constant current shunt method (under the MOSFET)... that would assure that I don't draw more than the plate could handle. What is a respectable & minimal way to minimize the cooking the grid? My plan was to trial and error different tickler coils.

Quote
In the schematic you showed in the earlier post the resistance is not enough. I assume you want to modulate the cathode current, and the resistor that you have as 1.5m Ohm is a shunt resistor. The feedback loop regulates the current flowing through the FET to be the input voltage / the resistance. I assume you want a max cathode current of ~ 0.5 amps or something. Scale the resistor depending on your expected audio input amplitude. The 1.5m Ohm resistor is way to low in value.
So if I had a 2.2vpp max input signal, I should use a 4.4ohm shunt resistor to assure 0.5Amps max thru the cathode right?

Quote
Cathode modulation is going to need to dissipate a power equal to the cathode voltage * average plate current. This could possibly be 100+ watts for a 833A coil, depending on how you dial in your bias. Grid modulation only has to supply a power equal to the average grid current * the modulation voltage, grid current is a lot lower. For the two sims, the power supplied by the source for the grid modulation is only 2W while the power dissipated for the cathode modulation is 12W.
How wrong would it be to have multiple MOSFETs situated in parallel? Distribute that cathode current heat across a few FETs.

What about this? A hybrid solution?
(https://i.imgur.com/9JeO2Tn.jpg)
Another idea – is to have the cathode circuit regulate the maximum current + modulate like a staccato circuit and the grid is modulated with a voltage controlled circuit to protect the grid. (I forgot to add a RC blocking circuit on the grid)
Title: Re: 833A Cathode Modulation
Post by: kyledellaquila on February 05, 2021, 04:15:07 AM
The last diagram was a late night spitball with regulating the amount of grid current in fear of damaging the delicate part of the 833a.
But if simple has been satisfactory in the past, why mess with something that works fine. (i.e. typical feedback coil and RC network)

I think what I want is something that can regulate the current to .5A and allow for great control over the amplitude as well...
I imagine that it would have to be the combination of something below (I want to say 2 and 4):
(https://i.imgur.com/SzNdheB.jpg)

This combination allows me to control the cut off of the input signal and control over the current of the cathode:
(https://i.imgur.com/bTUfPEo.jpg)
Title: Re: 833A Cathode Modulation
Post by: Weston on February 05, 2021, 09:13:19 PM
Adding an RF bypassed analog current meter should be enough to allow you to monitor DC grid current. If you end up adding a grid bias supply you can add some sort of current limiting to it.

The grid leak circuit introduces a RC time constant which may impact your modulation bandwidth, you will need to be careful about component selection for that. The grid leak network will also impact the startup oscillation. It will be different for the real system, but the simulation shows a number of RF cycles to reach steady state once the system is powered on. This will limit the maximum modulation frequency if you want to pulse the coil completely on and off.

What are you trying to accomplish? Are you trying to linearly control the power to the tesla coil or are you trying to pulse it on and off?  I am unsure what the waveforms you drew represent. Is the sine wave an audio waveform and the rectangular pulses representing each RF cycle?

Controlling the voltage on the cathode is the same as controlling the voltage on the grid, except it will dissipate more power if you regulate the cathode voltage instead of the grid voltage. Your schematic of the current regulation is correct, but you might need some additional feedback network to stabilize it.


If you just want to modulate the coil on and off from some set power level, you can adjust the power by changing the value of the grid leak resistor, that impacts the DC grid bias.  It might be best so start off with that plus something like the comparator schematic you show, that would be ~ equivalent to a DC capable staccato mode.
Title: Re: 833A Cathode Modulation
Post by: kyledellaquila on February 08, 2021, 06:37:38 AM
Quote
Adding an RF bypassed analog current meter should be enough to allow you to monitor DC grid current. If you end up adding a grid bias supply you can add some sort of current limiting to it.

The grid leak circuit introduces a RC time constant which may impact your modulation bandwidth, you will need to be careful about component selection for that. The grid leak network will also impact the startup oscillation. It will be different for the real system, but the simulation shows a number of RF cycles to reach steady state once the system is powered on. This will limit the maximum modulation frequency if you want to pulse the coil completely on and off.

How would a clamp on meter like this fair?
(https://highvoltageforum.net/proxy.php?request=http%3A%2F%2Fwww.w1tag.com%2Fpix%2FRFA2.jpg&hash=dd255301280b3f9c016af500a28b1cbb06e76fc2)

For current limiting – this page...:
https://www.instructables.com/GU-81-VTTC/ (https://www.instructables.com/GU-81-VTTC/)
...suggests that the resistor size can be approximated with:
R= V(Anode voltage)/(4*I(Anode current))
W= (V(Grid Voltage)/R)*V(Grid Voltage)
But these values are for a GU-81 tube.

Quote
What are you trying to accomplish? Are you trying to linearly control the power to the tesla coil or are you trying to pulse it on and off?  I am unsure what the waveforms you drew represent. Is the sine wave an audio waveform and the rectangular pulses representing each RF cycle?

What I would like to accomplish is linear control over the power of a Tesla coil. I have labored for quite some time for a reliable 3kv DC power supply and I really would like complete AM control over the power to get some genuine ramped envelope waveforms.

At the audio input, I want to (1)control the cuttoff of a sawtooth wave (duty cycle) & (2)control the amplitude of the remaining active region.
And of course... (3)Assure that the cathode current is limited to 0.5Amps.

Quote
Controlling the voltage on the cathode is the same as controlling the voltage on the grid, except it will dissipate more power if you regulate the cathode voltage instead of the grid voltage. Your schematic of the current regulation is correct, but you might need some additional feedback network to stabilize it.
I think for all around reliable operation, I will go with grid modulation.. (although this may deviate away from the topic of this post)...
What do you propose for additional feedback? I am unsure of what is missing.

Quote
If you just want to modulate the coil on and off from some set power level, you can adjust the power by changing the value of the grid leak resistor, that impacts the DC grid bias.  It might be best so start off with that plus something like the comparator schematic you show, that would be ~ equivalent to a DC capable staccato mode.
(https://i.imgur.com/AF8MY9a.jpg)
I provide this picture to better illustrate what I desire..
Title: Re: 833A Cathode Modulation
Post by: Twospoons on February 08, 2021, 09:20:12 PM
I think you've missed a trick.
If you want linear control, but low power dissipation  I'd have thought you'd want to modulate the grid bias voltage, but with feedback from the cathode current to improve linearity.
Or think of it as controlling the cathode current by controlling the grid bias.

Circuit I've shown is incomplete - you will need to add suitable gain and bandwidth limiting and filtering
Conceptually like this:
Title: Re: 833A Cathode Modulation
Post by: kyledellaquila on February 09, 2021, 07:24:53 AM
Thank you for the tip Twospoons!
Very interesting method.. I am studying your simplified diagram
(as I learn about all the ways one could use op amps)
(https://i.imgur.com/2IApnTw.jpg)
Didn't have enough time to create a compelling simulation of the logic here.. I will make more progress tomorrow
Title: Re: 833A Cathode Modulation
Post by: Twospoons on February 09, 2021, 09:18:09 PM
BTW, I didn't check the input polarity of the opamps carefully - you'll need to check that the grid bias moves the right way to compensate for cathode current changes.  And as mentioned before, you may need to band limit the feedback around the opamps.

Oh, and that lower opamp wasn't meant to operate as a comparator - I was just too lazy to draw in the feedback resistors to control the gain.
Title: Re: 833A Cathode Modulation
Post by: Vdekje on May 14, 2021, 08:58:15 PM
Any updates on this? Two of us in a discord are looking into ramping VTTCs as well.

I'm planning on ramping the screen, as done by Lukáš Koval here
I'm using a 8890, and I have a 20kv 48µF bus cap. Since the anode idle current is relatively high without screen voltage (140mA avg with half wave doubled 4.2kv, so much higher with, say, 10-15kv smoothed DC), I'd like to cathode modulate to only have the anode on during the screen ramp.

The other person is trying to get a ramp by sawtooth modulated PWM cathode modulation. He's doing it on a mini-vttc, PL500.
Title: Re: 833A Cathode Modulation
Post by: pc50kd on May 16, 2021, 12:21:23 PM
Interesting topic. I also played around with a cathode modulated VTTC a while ago.

I used a GU81M vacuum tube and interrupted the cathode with a IGBT.
For modulation I used a audio modulated PWM signal at 0.33 times the resonant frequency of the VTTC driving the IGBT.

I also made a post about it on the 4HV forum. https://4hv.org/e107_plugins/forum/forum_viewtopic.php?p=1&id=29247#post-29348 (https://4hv.org/e107_plugins/forum/forum_viewtopic.php?p=1&id=29247#post-29348)
Unfortunately the schematic posted there is unreadable. But if there is interest I could dive into my archives and see if I can recover it.

Title: Re: 833A Cathode Modulation
Post by: SalinsLV on May 21, 2021, 07:28:55 PM
I have also built VTTC that plays music. Video of the coil in action:
/>
SimplePortal 2.3.6 © 2008-2014, SimplePortal