EEVblog Electronics Community Forum
Electronics => Beginners => Topic started by: stcoso on November 22, 2016, 01:35:14 pm
-
Hi guys, it's my first post here and i go ahead excusing me for my english ;D.
So, I need a generator capable of give out 100W of RF power (something around 5/10Mhz sine wave 500Vpp).
It's going to be used for a little project that i have (RF sputtering) just for the time to show that works and obtain the money to buy a proper designed, commercial generator.. so the cheaper the better ;D
At the moment, i'm using this : http://www.sciencezero.org/vasil/romtcb/groyer.html. (http://www.sciencezero.org/vasil/romtcb/groyer.html.) Instead of 100Ohm i'm using 10Ohm gate resistance. Instead of Ultra fast diodes (even with 25ns diode i get a lot ringing on the gates and an awful output) i'm using 400V schottky rectifier. With this setup i can get to 2 Mhz easily. I tried both IRF840 and IRFP250. Altough IRF840 has lower Qg and lower rise/fall time i get better results with the irfp250.
I've also tried with this circuit http://users.skynet.be/BillsPage/TeslaPage.html (http://users.skynet.be/BillsPage/TeslaPage.html) that works up to 4 Mhz (with a triangular wave shape as output) but tends to produce a lot of "magic smoke" ;D
For both the circuit i tried using ferrite core and air choke (varing the tank capacitor to adjust the frequency)
I know that there are many design suitable for the purpose but i'm really looking for something cheap without caring too much about efficiency and output waveform.
-
Define the load because your figures are potentially quite a lot off...
500 volts peak to peak into a 50 Ohm load (the 'usual' impedance) is quite a lot more than 100 watts for instancebut it depends on the characteristics of the load.
If it needs to be bulletproof then the cheapest/simplest route might be a valve amplifier.
The sputtering 'power supplies' I've seen tend to be 13.56MHz which is in the ISM band, designated for use in such processes.
Generating a hundred or more watts of RF on some random frequency might be an invitation for a visit from some regulatory authority or other so I'd probably try and stick to the ISM frequencies, you should probably also pay some attention to the quality of the waveform as well, matching an output which is horribly rich in harmonics and spurious signals outside the band of interest is likely to be rather difficult and proportionally related to the amount of magic smoke you release but I don't know enough about sputtering to know if the waveform has an effect on the process?
-
Hi There is a thread on here https://www.eevblog.com/forum/projects/diy-metcal-13-56-mhz-rf-supply/ (https://www.eevblog.com/forum/projects/diy-metcal-13-56-mhz-rf-supply/)
You will need to use 13.56MHz , that's the frequency allocated to induction heating etc.
Some plasma light bulbs use a 13.56MHz driver, you could re-purpose one (if you could find one!)
The METCAL range of soldering irons , and knock offs, use 13.56MHz at ~ 50W.
Googling 13.56MHz may yield some useful schematics.
You will need physically smaller coils than the tesla coil example, for the higher frequencies, and a shorter length of wire on the secondary.
The sciencezero link seems dead. ?? no here it is ?? http://www.sciencezero.org/vasil/romtcb/groyer.html (http://www.sciencezero.org/vasil/romtcb/groyer.html) , seems like you go here http://www.sciencezero.org/vasil/romtcb/ (http://www.sciencezero.org/vasil/romtcb/) then get the groyer.html???
Ok this type of oscillator is intended for 20kHz-100kHz induction heating, there will be too much gate capacitance to make it work at RF without neutralising it somehow.
The capacitors are critical too, need PP or some other low loss type.
-
you might get some inpiration by looking up "10m transmitter" and "17m transmitter" (used by hams) and change the fequency slightly
ok the light bulbs are electrodeless fluorescent lamp or induction lamps
https://en.wikipedia.org/wiki/Electrodeless_lamp (https://en.wikipedia.org/wiki/Electrodeless_lamp)
images:
https://www.google.com.au/search?q=electrodeless+fluorescent+lamp&tbm=isch&tbo=u&source=univ&sa=X&ved=0ahUKEwjX5qqoybzQAhXKlJQKHUzcAvQQsAQIJw&biw=1661&bih=1005&dpr=0.9 (https://www.google.com.au/search?q=electrodeless+fluorescent+lamp&tbm=isch&tbo=u&source=univ&sa=X&ved=0ahUKEwjX5qqoybzQAhXKlJQKHUzcAvQQsAQIJw&biw=1661&bih=1005&dpr=0.9)
-
The load is a plasma chamber. It's rather complex to define the properties of the load.. The generators used for sputtering does have that kind of figures ( i think they're all almost the same).
The commercial generators does actually work in the ISM band (13.56 or 27Mhz) but i'm trying to use standard components (no expensive RF mosfet/BJT ) to achieve the goal.. So after searching through the internet i learned that it's rather difficult get that power at the ISM band. For this experiment i'm using a grounded faraday cage to avoid any problem with autorities.
I want to emphasize that this generator (once it'll works ;D) will be used one or two times for just a couple hours and then forgotten. I just need it for the time to prove that the chamber that i built actually works and then get the money to buy a proper ,ISM frequency, auto matching, safety compliant, commercial generator
-
Ok this type of oscillator is intended for 20kHz-100kHz induction heating, there will be too much gate capacitance to make it work at RF without neutralising it somehow.
If you replace the diodes + pullup resistors with small mosfets, it works up to 10MHz:
Connect source from the small mosfets to gates of the larger mosfets and drain to the LC tank like the diodes in the original cirucuit. The gates of the small mosfets need to be connected to an auxilary voltage like 16V to make them work as voltage limiters (source follows the drain voltage as long as it is lower than about 13V).
The power for driving the gates is now taken from the LC tank, therefore you have no large losses in the gate pullup resistors. Because of the very low gate resistance the circuit can work at much higher frequencies.
I have done this up to a couple of MHz, but it gets difficult because of the parasitic inductance of the wiring.
Single ended class E circuits are much better suited for higher frequencies, but require a constant load impedance to work correctly.
-
Hi There is a thread on here https://www.eevblog.com/forum/projects/diy-metcal-13-56-mhz-rf-supply/ (https://www.eevblog.com/forum/projects/diy-metcal-13-56-mhz-rf-supply/)
You will need to use 13.56MHz , that's the frequency allocated to induction heating etc.
Some plasma light bulbs use a 13.56MHz driver, you could re-purpose one (if you could find one!)
The METCAL range of soldering irons , and knock offs, use 13.56MHz at ~ 50W.
Googling 13.56MHz may yield some useful schematics.
You will need physically smaller coils than the tesla coil example, for the higher frequencies, and a shorter length of wire on the secondary.
The sciencezero link seems dead. ?? no here it is ?? http://www.sciencezero.org/vasil/romtcb/groyer.html (http://www.sciencezero.org/vasil/romtcb/groyer.html) , seems like you go here http://www.sciencezero.org/vasil/romtcb/ (http://www.sciencezero.org/vasil/romtcb/) then get the groyer.html???
Ok this type of oscillator is intended for 20kHz-100kHz induction heating, there will be too much gate capacitance to make it work at RF without neutralising it somehow.
The capacitors are critical too, need PP or some other low loss type.
The thread on metcal RF gen it's one of the first that i've seen and evaluated for the purpose, searching through the internet.. I found some very interesting schematics (oscillator+ class E/F amplifier at most) and if needed i'll go down that way but i'm asking your help to spend the least possible (IF possible ;))
About the coil.. i used 40mm PVC pipe but i didn't really care about it's self resonant frequency (well.. i know it's higher) because the load have a capacitance almost certanly greater than the coil capacitance ( which i can't even estimate without a working generator)
-
The load is a plasma chamber. It's rather complex to define the properties of the load.. The generators used for sputtering does have that kind of figures ( i think they're all almost the same).
No, it won't be easy to match or define but it can be done, of course putting things in the chamber to sputter will also change the characteristics so while it's possible to design and calculate the chamber characteristics you need to accommodate a range.
The commercial generators does actually work in the ISM band (13.56 or 27Mhz) but i'm trying to use standard components (no expensive RF mosfet/BJT ) to achieve the goal.. So after searching through the internet i learned that it's rather difficult get that power at the ISM band. For this experiment i'm using a grounded faraday cage to avoid any problem with autorities.
Nope, I suspect you've been looking in the wrong places and not thinking of the wider picture as it's fairly trivial to get that sort of power, especially on a spot frequency that doesn't vary.
You could even do it with off the shelf MOSFETs designed for use in motor control circuits if you're really intent on cheap, they're pretty awful for any RF work that requires wide band operation but again, a single frequency removes many of the challenges.
There's a seller on eBay who occasionally lists parts salvaged from KW range RF generators for use in mass spectrometers, he often has all the ferrites and even occasionally boards, but a pair of RF MOSFETs which would be capable of in excess of 1KW generally list for around £48. The last set of ferrites I bought from him cost me £10 and he doubled up the order.
You'll need a source signal to amplify of course but again, that's pretty trivial at 13.56MHz.
I want to emphasize that this generator (once it'll works ;D) will be used one or two times for just a couple hours and then forgotten. I just need it for the time to prove that the chamber that i built actually works and then get the money to buy a proper ,ISM frequency, auto matching, safety compliant, commercial generator
Do it right and you might not need to buy the 'proper' generator.
-
Oh, you live in Italy....
The home of the cheap CB amplifier.
Look up Zetagi, Bremi, RM Electronics and doubtless many more.
There are many schematics on the net for their products, while they're mostly junk as amplifiers (no proper filtering, dreadful linearity, poor circuit design for transmitting and other reasons) they could give you a nice starting point and an idea of how simple this can be.
-
No, it won't be easy to match or define but it can be done, of course putting things in the chamber to sputter will also change the characteristics so while it's possible to design and calculate the chamber characteristics you need to accommodate a range.
The big problem here is that the targets (which covers one of the electrode) are dielectrics with different dielectric constant. So, basically my load is a capacitor with a series resitance.Infact any commercial generator have a proper output matching stage (well.. this is true for almost any RF gen/PA ;D ).
Do it right and you might not need to buy the 'proper' generator.
That would be great, but i don't have the experience needed to do that properly, especially regarding the safety part. I'm a chemistry student.. If i get a working sputtering chamber, even with heavly prototypal gen ;D there are a lot of chance to get some money for the generator. I'm doing this for passion (and i think i've learned something in the process) but i'm not into stealing EE job :D.
-
Hahah, fair enough, if you can get a grant or other funding then yes, go that route, if nothing else it's a learning experience.
-
Hi stcoso welcome to the forum. Strictly speaking your circuit is not a Royer oscillator, a lot of people post schematics of push pull oscillators and then for some reason call them Royer oscillators. The original Royer oscillator used BJTs and core saturation controlled the operating frequency usually a few tens of kHz, primary and secondary voltages are square wave. A variation of the Royer oscillator, with tuned primary winding, is still popular in small CCFL inverters and I've even seen one used in an early IBM PC power supply <100W. Many PC power supplies rated at less than 200W used a variation of the Royer known as the Jensen converter which used a separate base drive transformer fed from a winding on the main transformer. The smaller base drive transformer was designed to saturate before the main transformer and core losses could be substantially reduced.
-
Ok, but i just posted a link not giving any name xD ;D
-
Hi stcoso realize you just posted the link so it's not you that are at fault it's all the people who post schematics of push pull oscillators are call them Royer oscillators. One day I will get around to designing one for low power HV supply. I just remembered going for an interview at a company that made power supplies for plasma etching, I think that most of them worked at 13.56MHz and supply was "matched" to the load using a large tuning coil and a motor driven capacitor.
-
Strictly speaking your circuit is not a Royer oscillator, a lot of people post schematics of push pull oscillators and then for some reason call them Royer oscillators.
Wikipedia calls it a "Resonant Royer oscillator":
https://en.wikipedia.org/wiki/Royer_oscillator
The circuit is almost identical, but instead of a saturating transformer it uses a tuned circuit for controlling the operating frequency and achieving zero voltage switching. So it is basically an advanced version of the classical Royer oscillator.
-
Ok this type of oscillator is intended for 20kHz-100kHz induction heating, there will be too much gate capacitance to make it work at RF without neutralising it somehow.
If you replace the diodes + pullup resistors with small mosfets, it works up to 10MHz:
Connect source from the small mosfets to gates of the larger mosfets and drain to the LC tank like the diodes in the original cirucuit. The gates of the small mosfets need to be connected to an auxilary voltage like 16V to make them work as voltage limiters (source follows the drain voltage as long as it is lower than about 13V).
The power for driving the gates is now taken from the LC tank, therefore you have no large losses in the gate pullup resistors. Because of the very low gate resistance the circuit can work at much higher frequencies.
I have done this up to a couple of MHz, but it gets difficult because of the parasitic inductance of the wiring.
Single ended class E circuits are much better suited for higher frequencies, but require a constant load impedance to work correctly.
This is interesting.. but.. between the mosfet drain and ground there'll be something like pi*V (V= voltage at the center tap of the coil). Eliminating the diodes wouldn't be dangerous for the little mosfet? the gates of the little mosfet will be at "-something" ( where something>Vgs breakdown)?
-
If it needs to be bulletproof then the cheapest/simplest route might be a valve amplifier.
Agreed, if the load is really ugly you will blow up a lot of expensive transistors trying to get something going. A solid state PA that can tolerate that kind of load needs both overrated output devices and fast, reliable protection circuitry.
I would start with something like this: https://pa0fri.home.xs4all.nl/Lineairs/Frinear150/fri150eng.htm (https://pa0fri.home.xs4all.nl/Lineairs/Frinear150/fri150eng.htm) or it's big brother: https://pa0fri.home.xs4all.nl/Lineairs/Frinear400/frinear400eng.htm (https://pa0fri.home.xs4all.nl/Lineairs/Frinear400/frinear400eng.htm). Tip: suitable TV line output valves in order of increasing power dissipation (and cost/rarity) are PL500, PL504, PL509 and PL519. The EL versions are also fine but much more expensive.
-
Makes me shudder to think of all the old TV parts I chucked out...
-
Can anyone post a schematic on some resonant gate ZVS ?
-
It is not the typical resonant gate driver, but this circuit works well and does not need many parts.
(https://www.eevblog.com/forum/beginners/medium-power-(100w)-rf-generator/?action=dlattach;attach=272708;image)
The auxillary supply must be turned on before the main power is switched on, otherwise the oscillator does not start correctly (both mosfets will be turned on and will therefore short the power supply until something melts).
The wires between Q1/2 and C1 should be as short as possible otherwise the mosfets will be exposed to large voltage spikes at turn off.
The auxillary mosfets can be small mosfets, because they only need to switch the gate current for the main mosfets.
Here is a fixed frequency driver using a resonant gate driver for the power mosfet:
http://richieburnett.co.uk/hfsstc.html (http://richieburnett.co.uk/hfsstc.html)
It is a bit outdated (there are much better mosfets for high frequency operation today), but this type of circuit is good for generating a lot of rf power using cheap mosfets.
-
I found a lot of CB very cheap (40/50$) power amplifier on ebay (2/3/400W in the CB at 27Mhz). So thank to CJay and richard.cs to making me look in that direction :-+
I'll pick one of those in the next days :D
BUT
i'm curious so i'm still gonna try building one. When you say "better mosfet" do you have a mosfet family in your mind?