DIY Metcal 13.56 MHz RF Supply

[mamalala]

Note: There have been quite some updates, new PCB layouts, firmware, etc. So please read the whole thread (or start from the last post if you only want to grab the latest files). Thanks!

---------------------

Hello everyone,

i'm in the process of designing a DIY RF-Supply for use with the Metcal handpieces/cartridges. So far i have an initial prototype schematic and layout done. The supply consists of two boards, each 90 x 90 mm in size. One is the main power supply/controller, the other is the RF output stage.

Finished soldering the prototype yesterday, and did some tests today. Unfortunately i did not yet have a 13.56 MHz oscillator (it's on the way to me, however), so i ran the tests with a 11.0something MHz osc instead. So far, everything works as expected.

Attached are some images of the test-setup of the prototype. One is when the RG gen is turned off, the next is the heatup phase, the last is when it reached temperature. However, these are just very preliminary tests to check if the overall concept/circuit works. The circuit still has a lot of unnecessary trimpots that i used only to evaluate some parameters and values. On the controller-board there is a small aread just with solder pads, on which i soldered another trimpot. That was used during the very first testing to "simulate" the feedback from the RF gen and check if the DC/DC converter there can be adjusted by that as intended. For the real test run that was disconnected, using the real control signal instead.

No schematics/layout for the public yet, since this all is just the first step. As soon as i could check with the 13.56 MHz osc,  decided on some final part values and types, and finally wrote some usable firmware for it, i will upload the stuff here, in the hope that it will be useful for some folks

The LCD shows the actual output voltage of the RF stage's DC/DC converter in the upper left, the VSWR in the upper right (divide by 10 to get a real VSWR, i.e. 242 means a VSWR of 2.42). The bottom left shows the raw readout of the forward power meassured at the SWR bridge, the bottom right is the reflected power.

Greetings,

Chris

[mikeselectricstuff]

Cool! I've always wanted a small lightweight Metcal supply for on-site use - SMPSU for light weight and fan cooled to minimise heatsink. Maybe even a battery option...
Do you really need the MCU?

[mamalala]

Cool! I've always wanted a small lightweight Metcal supply for on-site use - SMPSU for light weight and fan cooled to minimise heatsink. Maybe even a battery option...
Do you really need the MCU?

Well, even the original Metcal RF supplies have a SMPS, at least for the RF part, as well as linear regulators for the remaining stuff. But then, they also have much more circuitry in general. I attempt to go a bit more minimalistic, if possible. Right now i have three buck converters, one for 5V (OpAmp, µC, oscillator and LCD), one for 12V (for the ISL55110 FET driver) and one variable for the RF generator itself.

Instead of the "discrete" oscillator circuit plus driver i go for a small SMD oscillator plus the ISL as driver. Output FET is a IRF510. The power supplies are a bit oversized right now, all are fed from the same input. I could do that a bit better, input to the 12V buck, from there to the 5V buck.

The µC is not really needed, the feedback-loop works on it's own. However, the µC is used to give a fancy display, as well as powering up/down the RF stage. I always hated it to switch of the Metcal supply to change tips....

What is missing is the detection whether a tip is inserted into the handpiece or not using some analogue circuitry. Right now the µC is supposed to detect a fault condition and shut down the RF stage then. Maybe i'll add some analogue fanciness, so that one can chose to leave out the µC completely but still have the RF final somewhat protected.

Not sure if a fan is needed. Right now i have a small heatsink (60 mm x 40 mm, 20 mm height). As you can see in the pictures it's even under the board, so no real convection takes place. While it does get hot, it's still acceptable. I have seen audio amps run hotter... The ISL chip might need some small heatsink glued on top of it, it gets hot as well. But so far no failures. For some reason Intersil decided to not put a thermal pad there... Only the QFN variant seems to have one, but i would prefer to avoid using QFN's.

My idea was to use a standard 100 mm x 100 mm heatsink as a base. The output stage board would then mounted directly onto that without any spacers. The whole bottom side of that one is a solid groundplane, no signals there. The controller board would then sit right on top of that, connection is made through a simple male/female header combination. The flat-ribbon cable in the image is just for testing, to keep the boards apart so that i can access all the signals for measurement purposes. So the whole thing would probably become something like a 10 x 10 x 10 cm block.

But then, this all are just the first steps and thoughts. Any input is welcome, of course. I hope to have some initial schema and layout ready to upload here in one or two weeks. Right now thermal recovery is not as good as the original. But i mostly blame that on the 11 MHz instead of the 13.56. After all, that's about 20% off the target frequency. Plus, the feedback loop needs a bit more tweaking, as you can see the supply voltage for the RF stage goes way too low. But that's mainly a matter of finding some proper resistor values. Which i will do once i get the 13.56 MHz oscillators.

Greetings,

Chris

[mikeselectricstuff]

Well, even the original Metcal RF supplies have a SMPS, at least for the RF part, as well as linear regulators for the remaining stuff.

The STSS and MX500 use a huge, heavy mains transformer - only the  MX5000 uses a switchmode mains supply, but it's still in a heavy cast case.

The µC is not really needed, the feedback-loop works on it's own. However, the µC is used to give a fancy display, as well as powering up/down the RF stage. I always hated it to switch of the Metcal supply to change tips....

The display on the MX5000 is nice, but in practice could be reduced to a LED that shows when power is below a certain threshold (i.e. heated up). Maybe a bicolour that did a gradual change of coluur over the power range.
 
Auto-reset on error is certainly useful.

What is missing is the detection whether a tip is inserted into the handpiece or not using some analogue circuitry. Right now the µC is supposed to detect a fault condition and shut down the RF stage then. Maybe i'll add some analogue fanciness, so that one can chose to leave out the µC completely but still have the RF final somewhat protected.

The Metcals put some DC through the heater and use this to detect open-circuits, but an RF fault detection cct could probably do this as well as protection against shorts etc.

But then, this all are just the first steps and thoughts. Any input is welcome, of course. I hope to have some initial schema and layout ready to upload here in one or two weeks. Right now thermal recovery is not as good as the original. But i mostly blame that on the 11 MHz instead of the 13.56. After all, that's about 20% off the target frequency.

I don't think the frequency is that critical as far as the heater is concerned- AIUI the choice of 13.56 is just to keep it in the ISM band.

As regards using a fan, this was purely to minimise size & weight, for cooling both the PSU and RF stage.

I had another thought for a more 'extreme' approach to a lightweight solution : generate the 13.56MHz direct from a rectified, unsmoothed  mains  supply, and use an RF transformer to do the mains isolation. 
 

[mamalala]

The STSS and MX500 use a huge, heavy mains transformer - only the  MX5000 uses a switchmode mains supply, but it's still in a heavy cast case.

Ah, sorry, my bad. I was talking about the stuff in the circuitry itself to generate local supplies and the supply for the RF stage. My design is open to whatever one wants to use. While i have a rectifier and caps on the controller board, so that it can accept AC voltage, one could directly feed a DC in there instead. But since i have a bunch of 24V/2A transformers here from old, broken soldering stations, i also placed the rectifier and stuff there.

In any case, no need for a dual-output transformer like in the STSS for example. Always wondered why the schema on the net shows a 2x19V transformer, when the converter output for the RF stage goes to only 21V max. anyways....

The display on the MX5000 is nice, but in practice could be reduced to a LED that shows when power is below a certain threshold (i.e. heated up). Maybe a bicolour that did a gradual change of coluur over the power range.
 
Auto-reset on error is certainly useful.

Agreed, generally it's pretty useless. but i like that bling-factor somewhat. Plus it helps me a great deal with the developing stage. After all, this is the very first circuit and layout i did on this project.

However, i'm pretty sure that for now i keep at least the µC. Makes a lot of things much easier for me. Instead of putting a lot of analogue circuitry on the board, i can do all that with the µC instead.

The Metcals put some DC through the heater and use this to detect open-circuits, but an RF fault detection cct could probably do this as well as protection against shorts etc.

Yea, i have seen that in the schematic that floats around on the net. Some transistor to generate the bias, plus some comparator circuitry to process that. As written above, for now i prefer to get the detection done with the µC.

In the end, there is no real need to have extra circuitry for that, i think, since the µC should be able to handle all that. As far as the RF final is concerned, all that matters is to protect it against too much reflected power due to mismatch, and against too high a supply voltage from the feedback loop. The former detected by the µC, the latter simply a matter of having a low enough input voltage to begin with. My idea was to use 24DC or thereabouts for supply. That way the output of the buck converter for the RF can't get too high at all.

From the FET the RF is AC coupled into the filters. So any random short on the output would just mean a mismatch. No "real" short is happening there, as far as the RF stages supply is concerned.

I don't think the frequency is that critical as far as the heater is concerned- AIUI the choice of 13.56 is just to keep it in the ISM band.

Yes and no. True, it could be done with basically any frequency. However, the filters of the RF stage, plus the inductance and capacitance of the tip/handpiece are somewhat tuned. After all, the tip _is_ like an antenna to the RF output. It just happens that due to the Curie effect it will go into a mismatch once the temp. is reached. But then, i will see how it behaves once i use the right frequency.

As regards using a fan, this was purely to minimise size & weight, for cooling both the PSU and RF stage.

True. but since i plan to make all the stuff available, everyone can modify it to whatever he/she wants

I had another thought for a more 'extreme' approach to a lightweight solution : generate the 13.56MHz direct from a rectified, unsmoothed  mains  supply, and use an RF transformer to do the mains isolation.

Uh, not sure that this would be a good idea. For one, generating such high frequencies directly at these voltage levels is no simple feat. At least not if it is supposed to be efficient. The next problem is regulation. Once the tip reaches temperature, you will have a mismatch. That means that a most of the RF is reflected back into the supply. The purpose of the buck converter on that supply is to reduce the voltage the more power gets reflected. Otherwise you end up with several hundreds of volts reflected back.

Greetings,

Chris

[mikeselectricstuff]

My idea was to use 24DC or thereabouts for supply. That way the output of the buck converter for the RF can't get too high at all.

Looking at the MX500 schematic and description, although it feeds about 50V into the buck converter, the description indicates that the output voltage only goes to about 21V - I wonder why they used such a high input voltage, unless maybe they were initially unsureabout how much power they'd need and just didn't bother changing the design.

    I don't think the frequency is that critical as far as the heater is concerned- AIUI the choice of 13.56 is just to keep it in the ISM band.

Yes and no. True, it could be done with basically any frequency. However, the filters of the RF stage, plus the inductance and capacitance of the tip/handpiece are somewhat tuned. After all, the tip _is_ like an antenna to the RF output. It just happens that due to the Curie effect it will go into a mismatch once the temp. is reached. But then, i will see how it behaves once i use the right frequency.

If the element is absorbing a lot of power I'd expect the Q is going to be pretty low, so it shouldn't be too sharply resonant. 

    I had another thought for a more 'extreme' approach to a lightweight solution : generate the 13.56MHz direct from a rectified, unsmoothed  mains  supply, and use an RF transformer to do the mains isolation.

Uh, not sure that this would be a good idea. For one, generating such high frequencies directly at these voltage levels is no simple feat. At least not if it is supposed to be efficient. The next problem is regulation. Once the tip reaches temperature, you will have a mismatch. That means that a most of the RF is reflected back into the supply. The purpose of the buck converter on that supply is to reduce the voltage the more power gets reflected. Otherwise you end up with several hundreds of volts reflected back.

Yeah I doubt it's very practical... it was just a crazy idea!
Having said that, the fact that you don't need DC for heating applications, and the AC reservoir cap is one of the biggest parts of a SMPS means there might be some scope to do something interesting, even if it's just a non-isolated, buck converter followed by RF isolation, or an unsmoothed DC supply Or even just phase-angle control to use just the leading and trailing parts of the mains waveform.
I don't know enough about RF power stuff to  know how practical an RF transformer at this power level with mains isolation would be.

Another approach might be to use a SMPS that has a wide voltage adjustment range, and use that to do the control, although that only saves the second buck reg which isn't a big deal in size/weight terms.

Would a fixed RF output and PWM control be another option? Could get a bit noisy I suspect!

[mamalala]

If the element is absorbing a lot of power I'd expect the Q is going to be pretty low, so it shouldn't be too sharply resonant. 

Well, i'm no RF expert either (or even analogue stuff in general, my main focus is digital electronics). Simulating the RF stage in LT-Spice, however, shows that the power level at the output varies depending on the inductance of the load at a fixed frequency, or depending on the frequency and a fixed load inductance. So at least there it seems to have some effect. But as said, i will find out soon enough.

Yeah I doubt it's very practical... it was just a crazy idea!
Having said that, the fact that you don't need DC for heating applications, and the AC reservoir cap is one of the biggest parts of a SMPS means there might be some scope to do something interesting, even if it's just a non-isolated, buck converter followed by RF isolation, or an unsmoothed DC supply Or even just phase-angle control to use just the leading and trailing parts of the mains waveform.
I don't know enough about RF power stuff to  know how practical an RF transformer at this power level with mains isolation would be.

Putting the issue of regulation and isolation aside, the biggest problem is with unwanted RF emission. To use RF as a general means for heating, even on regular irons, you would have to modify the whole thing. Shielded coax cable, and grounding the metal sleeve. Otherwise you make a lot of HAM folks unhappy

Another approach might be to use a SMPS that has a wide voltage adjustment range, and use that to do the control, although that only saves the second buck reg which isn't a big deal in size/weight terms.

Would a fixed RF output and PWM control be another option? Could get a bit noisy I suspect!

Here the problem is indeed with RF noise. Due to the non-sine waveform you will produce a massive amount of harmonics that get radiated as well. It will result in a rather nice spread-spectrum RF noise emitter

Greetings,

Chris

[mamalala]

Just some updates.

The frequency is indeed important. I did some more measurements today (well, at least as far as my half-broken HP54201D got me). The signal for the feedback is picked up before the last series coil in the RF path (400nH). After that it goes through the stripline-style SWR bridge to the output.

If i connect a purely resistive 50 ohms dummy load there, i get roundabout the same peak readings on both ends of that coil. If i plug in the handpiece, the peak voltage drops by over half _after_ the coil, while still at full level before it. Then i used a 10 MHz osc, even lower in frequency, and the voltage after the coil dropped even more. Using a 12.8 MHz resulted in a much higher voltage there.

So i'm pretty confident now that things will look much better once i got the 13.56 MHz oscillators.

Then i did some more adjustments/part-value changes. The RF stage buck converter now idles at roundabout 16-17 volts and peaks at around 22-23 volts during heatup. So that got much better now as well.

Also changed a cap at the RF detector section. Had a 10nF there, which resulted in a rather slow response (after all, it got charged through a 160 resistor....). Now placed a 56pF there, and the response is really fast.

Alright, that's it for now, now i'm waiting for those stupid oscillators

Greetings,

Chris

[mamalala]

Small update....

I have reworked the layout of the RF stage board. Now the OpAmp used in the feedback loop is on that boards as well (was on the controller-board before). Also designed some simple circuitry for tip/cartridge detection. Build that detection on a perfboard, and it works just fine.

The new layout means that the RF stage PCB can basically be used on it's own now. All it requires are the supply voltages and enable signals for the osc and buck converter. The output of the tip detection circuit is open-drain through a small N-FET (2N7002). This means that it would be sufficient to pull-up the enable signals through 1k to the +5V rail, and also connect them to the tip detection signal. That way it will automatically shut of the buck converter and oscillator when a tip is removed, and switches it back on once a tip is re-inserted.

Also added two 3.2mm holes next to the FET driver chip. This is to install a small heatsink onto it.

Will do a new circuitry/layout for the controller board as well. The one i'm doing will be using a PIC microcontroller, LCD, rotary encoder, etc. It also contains the buck regulators to generate the +12V and +5V. The bridge rectifier is now optional, the board can be fed either AC (with bridge rectifier installed) or DC, through separate pads. It will be fairly easy to have different controller-boards there. Minimum requirement is basically just the main supply, +12V and +5V supplies. Everything else is completely optional.

Attached is a screenshot of the current layout of the RF stage board.

Greetings,

Chris

[mamalala]

After some testing it turns out that the RF transformer toroid seems to be way too small. It gets rather hot, meaning a lot of core losses. So i will change that part once i got a fresh set of toroids. Same applies for the first 1µ filter coild after the xformer, albeit not hat badly.

Wondering what material they used in the original circuit. I mean, even there the toroids do have a small diameter, and i used the same.

It's not that it would prevent proper functionality, just that it means a lot of losses that can be avoided. Thinking of battery operation here ....

Greetings,

Chris

[cwalex]

Awesome project!  I'll be watching this closely

[poorchava]

Could you please share the schematic of power stage? RF has always been a total voo-doo for me.

I always assumed that RF-guys (plus EMC engineers) can do their work only after they sacrifice a goat/virgin/infant/whatever during full moon Probably some illegal substance use is also involved [no offense, rly, I'm impressed with your design ]

[mikeselectricstuff]

Could you please share the schematic of power stage? RF has always been a total voo-doo for me.

I always assumed that RF-guys (plus EMC engineers) can do their work only after they sacrifice a goat/virgin/infant/whatever during full moon Probably some illegal substance use is also involved [no offense, rly, I'm impressed with your design ]

Here's the schematic of Metcal's MX500 (pdf)
And a description (rtf)

[amyk]

Wondering what material they used in the original circuit. I mean, even there the toroids do have a small diameter, and i used the same.

What colour are they?

http://www.micrometals.com/materialchart.html

[mamalala]

Wondering what material they used in the original circuit. I mean, even there the toroids do have a small diameter, and i used the same.

What colour are they?

http://www.micrometals.com/materialchart.html

Yellow, depending on the light with a greenish tint. Roundabout 0.5" outer diameter. No matter where i look, i can't find any toroids with which i can arrive at the 9µH per winding (about 11 turns) (Scratch that, the ones in my unit have 7.5 turns). And all the ones i can find have slightly less height as well.

In any case, pumping 30+ watts of HF energy through such a little core is silly, i think. Maybe someone can check on a real Metcal if the xformer there gets hot as well. It's not that it interferes with functionality. It works well, that's not the issue. I just would prefer to avoid such losses there.

Greetings,

Chris

[mamalala]

Could you please share the schematic of power stage? RF has always been a total voo-doo for me.

I always assumed that RF-guys (plus EMC engineers) can do their work only after they sacrifice a goat/virgin/infant/whatever during full moon Probably some illegal substance use is also involved [no offense, rly, I'm impressed with your design ]

Sure, attached is the current schematic as zipped PDF file. Keep in mind that this is preliminary, once i got all the kinks sorted out i will release the Eagle files for the whole project.

Yeah, RF is a bit tricky sometimes. What looks like a short for DC can be perfectly fine for RF

Greetings,

Chris

[KD0CAC John]

Couple of things , I have a SST unit and have not found any schematics or manual , I did check on Metcals site , but nothing as old as my unit .
On the torrides , sounds like you may not know the mix ?
I am a ham radio operator and have come across a lot of discussion of torrides , and the mix is related to frequency .
a couple of my bookmarks ,
http://www.kitsandparts.com/faq.php

https://www.amidoncorp.com/

http://www.cwsbytemark.com/

http://www.cliftonlaboratories.com/estimating_q_of_ferrite_cores.htm   

[mamalala]

Couple of things , I have a SST unit and have not found any schematics or manual , I did check on Metcals site , but nothing as old as my unit .
On the torrides , sounds like you may not know the mix ?
I am a ham radio operator and have come across a lot of discussion of torrides , and the mix is related to frequency .
a couple of my bookmarks ,
http://www.kitsandparts.com/faq.php

https://www.amidoncorp.com/

http://www.cwsbytemark.com/

http://www.cliftonlaboratories.com/estimating_q_of_ferrite_cores.htm

Thanks for the links. Yes, i have a STSS unit as well. The circuit is basically the same as the one in the MX500 circuit diagram. A very few things are different, for example the MX500 has an output section to hook up a power output meter, which the STSS does not have. Also, there is no auto-sleep thingy in the STSS. But the main RF circuit and stuff is identical.

About the toroids, yes, i know that the material is important and that not every random core can be used for every frequency. But short of desoldering the actual xformer and coils from a real unit and measuring them, there is little i can do. Plus, even if i would unsolder them i simply lack the proper tools to analyze these parts.

The two toroids for the filter are yellow core with a green side. It's 12.7mm o.d., 6.8mm i.d. and 5.7mm height, 16.5 turns of 0.65mm magnet wire. The toroids for the transformers (for driver and final) are yellow-green in color, 13mm o.d., 7.5mm i.d. and 6.5mm height, 7.5 turns each winding.

But in any case, a 0.5" diameter toroid sounds awfully small compared to the power it has to handle. Also, i can only get Amidon iron powder cores here in Germany, as well as ferrite cores from Amidon and a few other manufacturers. I have to live with that and simply go through trial & error to see what works best here. I use the -2 and -6 materials for the Amidon iron powder cores, which also suit the frequency we have here. As for ferrite, on the xformer the one which seemingly runs a little less hot is the -61, also from Amidon.

Already ordered iron poweder cores in larger sizes, should arrive here on Monday.

Greetings,

Chris

[mamalala]

Just checked in a toroid calculator. No matter what, using a T50 sized core is too small to handle that amount of power at that frequency. The -6 is for 2-50 MHz, and the -2 for 1-30 MHz. The flux density is just too big at the resulting voltage that i get at the RF transformer, resulting in quite a temperature rise of the core. Changing the size to T80 already brings it down into the acceptable range. Well, at least as far as the calculations go...

Also just calculated the output power, it nicely dumps over 30 watts into a 50 ohms dummy (i have around 115 volts Vpp at the dummy), so that part is OK.

Greetings,

Chris

[mikeselectricstuff]

I've uploaded some hi-res PCB pics of the MX5000 here
This has higher output power, although ISTR the spec mentions a temperature alongside the power so maybe it scales back if it gets too hot.
I can't easily power it open to measure inductor temps.
 BTW PIC is unprotected if anyone feels like a hack.
I have no idea what they were thinking putting an RTC in this.. ridiculous!

Here's a pic of an old (slightly hacked) US STSS-002. If you want to borrow the inductors to test/compare you're welcome - not sure if I have suitable kit to measure here.
Main output device is a TO-3 IRF130 on the back.
Also a thermal image running - the hottest coil is only hitting about 35 deg.C

[nukie]

Nice parts there look at those expensive silver micas, they are very stable, military like them very much. Love it!

[grenert]

Here's a pic of an old (slightly hacked) US STSS-002.

What did you do to the older STSS-002?  Think it is possible to boost the power to MX-5000 levels?

[mikeselectricstuff]

Here's a pic of an old (slightly hacked) US STSS-002.

What did you do to the older STSS-002?  Think it is possible to boost the power to MX-5000 levels?

I was looking at repackaging it with a SMPS to make a smaller PSU thin enough to fit a laptop bag. I had problems with the RF upsetting the SMPSU and never got round to finishing it.
Doubt it's practical to increase power much. increased power on 5000 doesn't make a huge difference except maybe on big tips like long blades

[mikeselectricstuff]

Also just calculated the output power, it nicely dumps over 30 watts into a 50 ohms dummy (i have around 115 volts Vpp at the dummy), so that part is OK.

Might be interesting to try measuring the actual impedance of a Metcal tip to optimise matching & make an accurate dummy load for testing.

[KD0CAC John]

I have an antenna analyzer , but would have to make up an adapter to the meter .
But again , I would assume [ we know what that spells ] that Metcal would make the system resonant on the one frequency it uses ?