Royer oscillator - any design tips?

Thread Starter

kingdano

Joined Apr 14, 2010
377
************************************WARNING**********************************************
THIS CIRCUIT PRESENTS LETHAL VOLTAGE LEVELS TO THE USER - DO NOT UNDER ANY CIRCUMSTANCES ATTEMPT TO RECREATE THIS CIRCUIT UNLESS YOU ARE A TRAINED PROFESSIONAL FAMILIAR WITH THE CONSTRUCTION AND TESTING OF ELECTRONICS. PLEASE BE CAREFUL
************************************WARNING**********************************************


This is my first post which is technical in nature - i will try to describe what needs to be done in detail - this is difficult as this (the system which i am driving) is company proprietary technology so revealing too much about it is not an option...

the "device" being driven in this particular sub-system, is pairs of electrodes - these electrodes are required to generate an electrostatic pressure field.

the proper driving signal for each electrode is a 500V RMS sinusoidal waveform - i believe the frequency is supposed to be 60 Hz.

there are custom wound transformers which i will be receiving that apparently convert a 5V input signal to the 500V RMS out.


edit: the design team is based over-seas (i am in the US) and there is a language barrier which keeps things a bit hazy in terms of technical communication

i do not know if i will be receiving the circuitry as well - or if i will have to design it.

now the design team told me that they used two of these oscillators in to drive the electrode pair (one per electrode) - but here is the part where i lose them...

the signals (in theory) need to be out-of-phase with each other by 180° (1/2 cycle) - how in the world can i ensure that this will happen if the two oscillators are not connected.

and if they DO have to be connected - how do i ensure that they stay 180° out of phase?

i understand that this is not a common circuit topology - in fact the only place i have seen it referenced was for CCFL applications.

in case i cannot use this oscillator - what other devices can i use to achieve a 500V RMS pair of sinusoidal signals?


hopefully someone can help me, this is the one thing i have been dreading tackling at work.


***note***

this is now the 9th hit on google for "royer oscillator"...
 
Last edited:

retched

Joined Dec 5, 2009
5,207
You can use a zero cross detector on one of the oscillators to ensure it is on the up-swing before triggering the negative swinging osc.

Or if they are pulling from the same AC source, they will be in time regardless. (At least the input side will)
 
I designed a Royer converter a couple of years back, as a precision temperature-controlled HT supply for an avalanche photodiode. Your guru here is Jim Williams over at Linear Technology, for it was he who dug out the old 1950s circuit and redeveloped it to design a high efficiency CCFL driver for an early Apple laptop. All the modern circuits I've seen are a variation of this.

I'll have to post again in a few hours when I get home, as my lunch break is ticking away, but do hang around. One salient point for now - the 50 Hz you talk of won't be the switcher clock, that will be running at 10s / 100s of kHz. What you need to do is modulate the output to get your 50Hz sine wave.

Lunch over... will send some links later.
 
Home now, but exhausted (hard week, long drive), but I'll see what I can remember before my brain fades completely.

The full story of Jim Williams' redevelopment of the Royer oscillator is told in his book (as editor) "The Art and Science of Analog Circuit Design" (EDN, ISBN 0-7506-7062-2) in the chapter "Tripping the Light Fantastic". As well as the circuits history and some early examples of Tektronix CRT drivers, he explains just about everything you need to know about designing Royer converters. If you're designing one, you need this book. There are some freely downloadable papers from the LT website too, I'll dig out some links at some point. Until then, search for "high voltage" and "CCFL" in the power products section, as I don't think "Royer" is mentioned in any titles.

OK kingdano, to run at 50 Hz your transformer will be a bit special, I'm guessing quite chunky and iron-cored rather than ferrite. The secondary output will have a sinusoid that's a bit distorted, so hopefully a few harmonics won't be an issue. If they are, an LC filter will clean them up a bit. Alternatively, R1 could be tweaked so the oscillator runs a tad sideways from peak efficiency, which will reduce the harmonics a bit more. The Royer oscillator frequency is unrelated to the frequency of the switcher circuit that drives L1. It will be free-running, so an accurate 50 Hz will require more complexity. C1 will be quite large too (relative to the 1 uF or so used in CCFL circuits, which run at 10s of kHz), and needs to be a low-loss type for decent efficiency. Maybe a motor-run polypropylene.

Royer transistors are a bit special, in particular having high gain and low Vce(sat), but they're not essential for a prototype working circuit, just for a decent efficiency and guaranteed startup every time. Sometimes low ambient temperatures can inhibit the oscillator startup; this can be helped by driving one transistor a bit harder than the other.

The phase of the feedback windings is crucial, but an easy way forward with the first prototype is to just guess, you've got a 50% chance of getting it right first time, and if you guess wrong then (in this instance) nothing bad will happen, and you know your next guess will have a better chance of success.

On the posted circuit there should also be a schottky diode connected to the open end of L1, anode to L1, cathode to +5 V (or whatever). Hanging off the anode/L1 junction is the output of the switcher circuit, which in conjunction with L1 and the diode forms a current source that controls the transformer drive. This switcher circuit runs at normal switcher frequencies - up to a couple of hundred kHz if you don't mind paying for a large L1 to get high conversion efficiency, maybe a couple of MHz to trade off efficiency against a smaller L1. I'm guessing the former in this case.

As for the 180 degree synchronisation between two oscillators, retched's fine idea has merit. Synchronisation will also be aided a little by a chaotic effect called mode locking, normally a nuisance in free running oscillators of similar frequency, as it causes them to pull into each other after a while.

If you're lucky though, someone has wound you a transformer with a dual secondary, and you can use a single drive circuit. The common centre tap will be grounded, and the two outputs will be in antiphase. The downside to this arrangement is that you can only tap a feedback signal from one of these secondaries, so the loads on each side will have to be reasonably balanced to get outputs that are reasonably matched in amplitude.

The usual high-voltage safety rules apply, and tinkering with such circuits should only be attempted by professionals (like kingdano) with the appropriate safety training and procedures, and even then never working alone. The output voltages here could easily be lethal, and even current-limited transformers can charge a capacitor to deadly energy levels. And if the control loop breaks, the voltage output will go as high as it can. Such voltage levels jump gaps and flash over normal insulation.
 
Last edited:
Just spotted something in the OP circuit - the architecture is not quite right. The transistor emitters should go to L1, not 0V. R1 and the primary centre tap should go to +5 V (or whatever). I'll dig out some example circuits tomorrow.
 

Thread Starter

kingdano

Joined Apr 14, 2010
377
i received a "black box" of electronics today from the design team in France - i glanced at it briefly before powering it up and testing it a bit initially.

i used a high-voltage differential probe since the 500VRMS would damage our scope.

i didnt really believe what the scope was telling me - as the scope would read out a 5V pp signal - but with the high voltage probe attached this 5V would have been something like 2500V...so something is up there...maybe i need to set the scope to read a 1x probe...

when i measured the VAC with a DMM i got more realistic Vi/Vo relationships...4VDC in = 500V out

the circuit came with a note describing how to operate it - apply DC voltage to the input and adjust until you have between 500 and 600 VRMS on the output.

another interesting note - what we thought was required to be 60 Hz is coming out of this circuit as a 50-70 kHz wave

i briefly looked at it as well...i will take pictures of all of this tomorrow...i am pretty anal about documenting what i did and why i did it.

what i did see was two small transformers - i assume the current required is very low because all the circuit is doing is loading two traces of wire essentially. and at such high voltage, i imagine the current will be very low (can anyone confirm this?) it also changes dramatically from DVM probe load, so i think i am correct.

so i assume that they are running off the same source, i am wondering if they are out of phase...i didnt get to measure it

if i have time i will draw up the circuit in capture too - would be an interesting reverse engineer for me anyways.

i
 

Thread Starter

kingdano

Joined Apr 14, 2010
377
Just spotted something in the OP circuit - the architecture is not quite right. The transistor emitters should go to L1, not 0V. R1 and the primary centre tap should go to +5 V (or whatever). I'll dig out some example circuits tomorrow.

i am not sure these guys really drew it up correctly, i have been skeptical since they showed that drawing to me.

that was the main reason for my concern -i really hope these guys know what they are doing and just half a**ed their example circuit to me knowing theyd be sending a working circuit...
 

Thread Starter

kingdano

Joined Apr 14, 2010
377
I just spent some time drawing up the layout in OrCAD.

I hope i got all the connections right, but its hard for me to look at spaghetti wire on the backside of a breadboard and make component connections etc.

Its too much R/L flip flopping and pin checking...at any rate...

I will post some pictures of the physical circuit as well - note that the transformers are probably not wound in the fashion the schematic symbol has them - but it was the only 10 pin symbol i could find.

i also did not see a "pin 1" marker on them so i just made up a pin order...i suppose we should assume the proper winding and pin matching is being used. i think the important thing to note is that the primary sides of the transformers are running in parallel. so pin 1 -> pin 1 down to pin 5-> pin 5.

i am concerned that there is no protection of any kind - input or output - no zener clamping the input so that it does not rise beyond the ab. max rating of the electrode (650 VRMS or approx 1800 Vpp if im not mistaken) - i think this would be about 5V on the input...not entirely sure yet.

i also have a scope capture of a 0.6VDC input producing 150Vpp/60VRMS out. the two sine waves are 180 out of phase - and looking at how the transformers are tapped this makes sense now - as there are two output pins - and they are tapping the opposing side of each transformer, and grounding the other.

if anyone can offer some insight into this i would be very interested to hear it.

i do not quite grasp how this circuit works - and i did see your suggested reading above - not sure if/when i will have time to read it though.
 

Attachments

beenthere

Joined Apr 20, 2004
15,819
This subject is causing some fluttery feelings. We normally discourage and prevent the discussion of circuits that present the user with lethal voltages. Please emphasize safety and good practice.
 

Thread Starter

kingdano

Joined Apr 14, 2010
377
images of the physical circuit are attached.

when looking at this circuit and the schematic - the orientations are the same.

ie) the top transistor is Q1, and the top transformer is T1.

the transistors used are motorola BD137 NPN transistors

the transformers are (i believe) coiltronics transformers designed for CCFL applications. there is no discernible part number marked on the package...'42CB05 F' is printed on the side - but this returns 0 hits on google.

the scope capture is from a 0.6VDC input being applied - note that the two sine waves are 180 degrees out of phase.


that is all the information i have at this point in time.
 

Attachments

Last edited:

Thread Starter

kingdano

Joined Apr 14, 2010
377
bertus

i hesitate to re-design the drive circuitry for a few reasons:

1. this is a high visibility, high priority project - i would like to not be the cause of major delays if possible.

2. i dont have a source for the transformers - they may be custom and i do not want to completely hose the project by ruining them.

my goal here is to ensure that this particular circuit is biased as safely as possible - and that the overall topology is correct and will not be unstable.

it seems that this circuit is at the very least missing an inductor on the input, and a zener to clamp the input to around 5V - since the max RMS the electrodes can see is 650.

can anyone confirm this?
 

bertus

Joined Apr 5, 2008
22,270
Hello,

Without the transformer characteristics it is hard to say what will be the result.

On the foto : what resistor did you use?
It is not the 330 ohms as in your schematic, but 82 ohms as far as I can tell from the colors.

Bertus
 
Firstly, my apologies for casting aspersions on the cromulence of kingdano's OP circuit. It's a Royer variation I didn't recognise, but can be found in this Royer-specific transistor application note:

Zetex AN14 "Transistor Considerations for LCD Backlighting".

Royer transistors are also available from other manufacturers.

Secondly, I can understand the moderators trepidation about high voltage circuits. Usual high-voltage safety warnings taken as given, the Royer circuit should have no appeal for the amateur constructor, and should not be attempted by same. It's not a suitable circuit for applications like rail guns, cattle prods or tazers, and I wouldn't post information on the internet that would help build such devices. If people want to build these things, I would rather it was done incompetently, and I'm not going to help. For instance, there was a thread recently on piezo transformers that appeared to be destined for a tazer, and the thread was quite rightly shut down. I know exactly how to do this, because I've done it (for a gas igniter for cookers and boilers), but this is information that I won't share as I regard it as dangerous and, frankly, just plain nasty.

Anyhoo, some promised links. The first one contains a lot of meat:

Linear Technology AN118 "High Voltage, Low Noise DC/DC Converters".

I feel slightly guilty about AN118, as when I was developing my last Royer circuit I was badgering the LT rep for info, and it seems that I was one of four customers at the time wanting more of the same, and this prompted Jim Williams to publish his work on the subject. The trouble is, he wrote it while he was on holiday on a paradise tropical island (as can be seen in the little cartoon on the last page), and I can't help but feel 25% responsible for that.

Some more on Royer circuits. If they're all a bit LT-oriented it's because that's where Jim Williams works, and he's the worlds leading expert in the area.

Linear Technology AN55 "Techniques for 92% Efficient LCD Illumination".

Linear Technology DN164 "High Power CCFL Backlight Inverter for Backlight LCD Displays".

Linear Technology DN133 "Low Input Voltage CCFL Power Supply".

A great paper on switched-mode power supplies in general ("A gentle guide for the trepidatious"):

Linear Technology AN25 "Switching Regulators for Poets".

Regarding kingdano's prototype; 50 - 70 kHz is about what I'd expect to see from a CCFL transformer with about 100 nF strapped across the primary. The oscillation frequency is a (rough) function of the primary inductance and the primary cap, which is I was thinking the custom transformer you were being supplied would be a lot larger than the one that actually turned up. I don't know how well a CCFL transformer would run at 50 Hz, but if you want to find out then you need a much bigger primary capacitor.

You're going to need some sort of feedback and control circuit to stabilise the output voltage or it will vary too wildly and won't pass parametric testing, yet alone be a viable product in the field. The feedback networks as shown in the CCFL circuits in the app notes above will work well. Time to ask the boss to buy you a high-voltage 'scope probe too, cheaper than a blown-up 'scope input stage, and they won't load the circuit as much as a lash-up resistive divider. The latter will do at a pinch, but they need to be very high value resistors, in which case an ordinary 10:1 probe will load the divider unacceptably to give DC errors and affect the transient signal integrity too.

Hope this helps.
 

Thread Starter

kingdano

Joined Apr 14, 2010
377
On the foto : what resistor did you use?
It is not the 330 ohms as in your schematic, but 82 ohms as far as I can tell from the colors.
The colors on the resistor are hard to decipher, even in person, but i think they are orange-orange-brown-gold - and when I measure the value (in circuit) i read 328Ω. The only resistive component in the circuit is that resistor - so i am fairly sure it is the right value.


Darren Holdstock said:
I don't know how well a CCFL transformer would run at 50 Hz, but if you want to find out then you need a much bigger primary capacitor.
I will ask the question to the design team in France, but after consulting with some of the other R&D guys over here, they do recall a mention of 60 Hz. I just cant imagine them not testing this before shipping it over though...so maybe they said 60 kHz? We will see.

Darren Holdstock said:
You're going to need some sort of feedback and control circuit to stabilise the output voltage or it will vary too wildly and won't pass parametric testing, yet alone be a viable product in the field. The feedback networks as shown in the CCFL circuits in the app notes above will work well.
I agree on feedback and control - however, note that this will NOT be productized - at least not by me. I work in the R&D group, so any "product-ization" would be handled by a more senior EE, and would be at least a year or two from now. That being said - i still want this to be safe for lab use - given that conductive fluids may be spraying near-by.

Darren Holdstock said:
Time to ask the boss to buy you a high-voltage 'scope probe too, cheaper than a blown-up 'scope input stage, and they won't load the circuit as much as a lash-up resistive divider. The latter will do at a pinch, but they need to be very high value resistors, in which case an ordinary 10:1 probe will load the divider unacceptably to give DC errors and affect the transient signal integrity too.
We do have (2) high voltage probes (Tektronix) - i have "borrowed" ;) both of them for now - to collect data with, and to test the circuit if i modify it.


Thanks much for the app note and book suggestions - a quick glance last week showed me that this circuit is missing an input inductor at the very least - i just really dont want to hack this breadboard apart for fear of not being able to re-assemble it. I will see what fixes i can do with minimally invasive surgery and go for those.

Thanks to all for the tips.



-Dan
 
Top