DIY Metcal 13.56 MHz RF Supply

[mamalala]

Hi all,

just a quick update. Attached is the current firmware used for the PIC 18F2620 microcontroller. It is an MPLAB-X project, and it uses Microchips C18 compiler. In the src/Docu folder you will find a small, rough docu of the main points if the user interface.

Not much in terms of source code docu yet, some cleanup needs to be done, etc. But at least it is the first usable version that works.

Some main features:

- Nifty status display
- Extensible menu handling/structure
- All power manager options are editable
- Settings can be saved/loaded to/from internal EEPROM
- One main system setting and 8 power-manager settings saveable
- Multi-Language support, lang selectable during runtime
- Uses only one rotary encoder with pushbutton-function
- Verbose status output over RS232
- Alternatively extensive realtime parameter logging over RS232

The logging can be used to characterize the tips used, so that one can do proper power-manager settings. The format used also allows to just dump the log into an ASCII file and have a plot done with GnuPlot (this is how i did the plots previously shown).

So far it works rather well. It allows to enter a sleep mode with reduced power output when idling for a while. It has configurable thresholds that will automatically wake up the unit when a load change on the tip is detected. It can also go into auto-powerdown after a given time of sleeping.

Greetings,

Chris

[mamalala]

On a slightly related issue:

Does someone here have a handpiece for me for an affordable price? Right now i have only one, that i use for my regular work as well as for testing/developing this unit. Would be great to have a secnd one just for testing/development, so that i can avoid plugging them back and forth all the time, as well as to lower the risk of killing it due to some mistake/mishap. The old MX-RM3E would be fine, but any other that can take the STTC cartridges would be OK as well.

If so, send me a PM or let me know here in the thread.

Thanks and greetings,

Chris

[zoltan]

Hi,

My RF boards arrived this Friday. If anyone need a board send me a pm. Btw, I'm open for parts-for-boards exchange ;-)

The board is a slightly modified version v0.3.
Changes:

- IC1 modified to accept LM22670, added R35 0R resistor on pin 5 of IC1. Leave it unpopulated for LM22670 or solder 0R resistor if using LM22676.
- R19,R25 changed to SMD footprint, same as R32
- SV1 changed to thru-hole
- Added extra vias on many solder-pads to strengthen them, eg. XFMR1, L3-L5, output pads.
- SV1 pin description on back side

At the moment I've finished one board assembly, but had to stop due I don't have a proper dummy load yet (It' on the way from Greece). The SMPS part is alive and working properly, the signal generation is also working (I think), but the output level looks way lower than needed (see image) and that overheats the mosfet.

By now I managed to kill one LM22670 by accidentally shorting the output pin with multimeter probe and cooked one gate driver (forget to put thermal paste below it).


RF board

EDIT: The scope readings are false - the signal is over the range of the scope - sorry guys.


Driver output, 3.59 Vpp


Drain voltage, 6.63 Vpp @ 22V


Z

[mamalala]

Uh, the driver should not get hot at all, at least not if you use the IRF510 as output FET.

What the heck are all these holes in the pads for the coils, RF-Output and one side of L6 (near the output). Who placed them there and why? And please tell me that they are not through-plated to the other side, maybe even to the GND plane at the back?

What toroid cores did you use? There has been a problem with them in the past, meaning that what seems to be an equivalent to the Amidon core did not work out well. Also, did you wind XFMR1 properly?

And in any case, how did you use the RF board if you have no dummy load? With nothing connected at all? That sure will cause havoc if the tip-detection is somehow bypassed. Or did you use a tip? Not really usablee, since it changes characteristics when it heats up (although there is a way to do the adjustments with a tip instead of a dummy).

Greetings,

Chris

Edit: Driver output of 3.59V is dead wrong, so is the sine-wave there. It should be square-ish with about 12V p-p.

Edit 2: I also see that you have changed R19 to a single-turn pot. Good luck getting a stable and precise adjustment with that one!

[Paulinho19]

Sine-wave Driver output.... What happened there, that does not look good. Also most of your changes make it almost impossible to mount the board flat on a head sink. For example "SV1 changed to thru-hole".
If you want a set of boards (Chris's RF-Gen Board + Chris's Controller Board or my Arduino Controller board) just send me a PM, I still have a few, 10 Euros plus shipping.

Greetings,

Paulo

[zoltan]

Hi,

First I have to say, our friend Nyquist was making some fun with me (and the Chinese to). I'm using a cheap USB scope that's good for maybe 4MHz (never used it over 1-2MHz before) sqare wave and it's "just rounded the edges". Plus the probe is on the cheap side. Sorry guys.

The holes are vias I put there to give it a bit more strength, they are not connected to anything but the solder pad on top layer. Please don't tell me this have any effect on the RF circuitry.

The cores are from kitsandparts, exact same as you ant Paulo used. The windings are as in your manual, carefully put together and checked for short. If the cores are not good for a given F, they would most probably overheat - is this correct?

Yes, I tried with the tip only. Hopefully next week my dummy load will arrive so I can start again - or not, if the vias are affecting the circuit. Until then ill try to source a better scope.

Paulo, I have some extruded profiles that make it perfect fit that way. If you mount the board on the heatsink, you have 1.5mm to compensate by bending legs upward.
With my profiles I need to bent them to, but 1.5mm down. That gives me 3mm between board and heatsink, what is more than enough for pins. The reason I've changed it to th is that I have lot of them in every shape except SMD.

Chris, the driver is overheating if no thermal paste is put between driver and the board even with no fet connected. With little thermal paste everything is under control.

[mamalala]

Hi,

First I have to say, our friend Nyquist was making some fun with me (and the Chinese to). I'm using a cheap USB scope that's good for maybe 4MHz (never used it over 1-2MHz before) sqare wave and it's "just rounded the edges". Plus the probe is on the cheap side. Sorry guys.

Hmm, OK. Not Nyquist, but bandwidth limitation. I would suggest that you use a "real" scop instead, with at least 20MHz bandwidth. Who knows if the voltage levels it shows are correct, when it already messes up the waveform so badly.

In any case, the output of the driver shoud reach roundabout 12V p-p. What you have there is much too low (if that reading is correct), meaning that the FET can not turn fully on.

The holes are vias I put there to give it a bit more strength, they are not connected to anything but the solder pad on top layer. Please don't tell me this have any effect on the RF circuitry.

Alright. As long as they don't touch anything at the bottom that should be no problem.

The cores are from kitsandparts, exact same as you ant Paulo used. The windings are as in your manual, carefully put together and checked for short. If the cores are not good for a given F, they would most probably overheat - is this correct?

Well, that's the thing. The ones that Paulo got first turned out to not work that well, while it was claimed that they are equivalent. Once i sent them a set of cores from the exact types i use here, it worked for him as well.

The cores are good for a rather wide frequency range. However, small differencies in material composition can have big effects. RF is a nasty thing when it comes to such things...

Also note that later i updated the docu, so the number of turns for each core changed. If you used the very first one i wrote, it will be slightly off.

[quote author=zoltan link=topic=12578.msg240931#msg240931 Chris, the driver is overheating if no thermal paste is put between driver and the board even with no fet connected. With little thermal paste everything is under control.
[/quote]

That does not sound right at all. Without load (i.e. no FET connected) it should not get hot, not even really warm, no matter if there is thermal paste or not. That, in combination with the awkwardly low output level you get from the driver, tells me that there is something wrong. Either with the supply voltages or with the solderwork on the circuit itself. Maybe soldered in a wrong (too low) value for R10, the pulldown connected to the FET's gate? Is the supply voltage to the driver a clean 12V DC?

You need to check and verify the RF board step by step. First make sure that all supply voltages are OK and that the +5V DC and +12V DC are clean. Then check (with a suitable scope!) that the output of the oscillator is good. If it is, next check the output of the driver (roundabout same waveform as input, just with 12V p-p).

Unless you know what you are doing, using a tip to setup the circuit is a bad idea. If it drives the output stage too hard (improperly adjusted R19), it will almost instantly kill the FET when the SWR goes up (tip is hot).

You should also limit the output voltage from the dc/dc converter to a lower value using R32 first during testing. That will help to protect the FET  bit in case something goes wrong.

Can you upload an image of the bottom of the PCB you have?

Greetings,

Chris

[zoltan]

Hi Chris,

Hmm, OK. Not Nyquist, but bandwidth limitation. I would suggest that you use a "real" scop instead, with at least 20MHz bandwidth. Who knows if the voltage levels it shows are correct, when it already messes up the waveform so badly.

I'm using a Hantek DSO-2090 40MHz, 100MS/s. Last night I've made made a test - I've set up a 50MHz ARM microcontroller clock output to one of it's pins. Started with 50MHz all I got was a nice 50MHz sine wave. By decreasing a clock little by little the usable shape showed up at 4MHz. It's time to buy a new scope I think 

In any case, the output of the driver shoud reach roundabout 12V p-p. What you have there is much too low (if that reading is correct), meaning that the FET can not turn fully on.

By now I can say that the reading IS incorrect for sure. I'll try to get access to something better this week.

Alright. As long as they don't touch anything at the bottom that should be no problem.

They are not touching anything. Maybe the via stubs add up a little bit of capacitance, but not much.

Well, that's the thing. The ones that Paulo got first turned out to not work that well, while it was claimed that they are equivalent. Once i sent them a set of cores from the exact types i use here, it worked for him as well.

The cores are good for a rather wide frequency range. However, small differencies in material composition can have big effects. RF is a nasty thing when it comes to such things...

I have this from kitsandparts:
T50-6 powdered iron toroid
T80-6 powdered iron toroid
FT82-61 ferrite toroid

While searching for cores I've found many good reviews from HAM people regarding kitsandparts. Are they correct then or not? How can I test them?

Also note that later i updated the docu, so the number of turns for each core changed. If you used the very first one i wrote, it will be slightly off.

Used the docu from post #68.

That does not sound right at all. Without load (i.e. no FET connected) it should not get hot, not even really warm, no matter if there is thermal paste or not. That, in combination with the awkwardly low output level you get from the driver, tells me that there is something wrong. Either with the supply voltages or with the solderwork on the circuit itself. Maybe soldered in a wrong (too low) value for R10, the pulldown connected to the FET's gate? Is the supply voltage to the driver a clean 12V DC?

Tonight I'll put together a new board and see how it behaves. Maybe there is some residue left from soldering - any kind of mess can behave badly in RF world. When I say'd no load I meant without R8 and R9 populated - R10 should have no effect at all.
The supply voltage looks OK, I've checked with scope and it is not showing any variations in voltage level. Even with BW problems, the scope should show if there is significant level change, but it's clean line.

Unless you know what you are doing, using a tip to setup the circuit is a bad idea. If it drives the output stage too hard (improperly adjusted R19), it will almost instantly kill the FET when the SWR goes up (tip is hot).

It does not heat the tip, only the FET    But it clearly does something RF, because my hot-air station switch on when RF power is enabled. At least I can use it as remote control for hot-air.
I will not do any testing until my dummy load arrives. I'm getting short on parts.
 

You should also limit the output voltage from the dc/dc converter to a lower value using R32 first during testing. That will help to protect the FET  bit in case something goes wrong.

What level you suggest? 12V or more?

Can you upload an image of the bottom of the PCB you have?

Tonight. I can send you the gerbers if you wish, but I will not upload them to public until the board is working.

Thanks for help,

Z

[mamalala]

Ho zoltan,

the cores i use are from here:

FT82-61:
http://www.reichelt.de/index.html?ACTION=3;ARTICLE=7937;SEARCH=FT%2082-61

T80-6:
http://www.reichelt.de/index.html?ACTION=3;ARTICLE=20024;SEARCH=T%2080-6

T50-6:
http://www.reichelt.de/index.html?ACTION=3;ARTICLE=20012;SEARCH=T%2050-6

I have no way of knowing if the kitsandparts cores are the same or not. The name would suggest that they are....

Regarding the docu, in post #82 i already made some corrections, and in later posts are even more changes. So it's a good idea to check them all. When i have some time i will upload an updated doc.

With the 2k2/18k combination for the feedback resistors onthe dc/dc converter, it should output a little less than 12 volts. You can simply remove R4 (that disconnects the OpAmp from the feedback loop) and make sure that the limit/boost pin is left open (or remove R11).

Please double check the driver chip. As said, it should not get hot at all without any load attached. It's easy to short two pins with solder, as well as not properly soldering the GND pin (due to thermal mass attached through the PCB's copper). Also double-check the orientation.

The output pins are the two outmost ones. Between them are the +12V and GND pins. Accidently shorting two pins there will very likely show the symptoms you see.

Greetings,

Chris

[mamalala]

Attached is a ZIP with the latest Eagle files for the RF stage. R23/R24 is now 22k/200R. Added some text for the toroids/windings in the schema as well.

Greetings,

Chris

[mamalala]

BTW,

i'm in the process of assembling two more units. That means that soon i have again a unit to play with.

So, if anyone wants to send me a set of torroids to try out, i can then test them against the "original" ones, one by one, and check what the difference is so as to add some extra hints how to wind/use the alternative ones.

Anyone interrested just send me a PM here in the forum, and i will give you my contact details.

Might be useful to know what is going on with them.

Greetings,

Chris

[zoltan]

Hi Chris,

Attached is the bottom side of the board, as you asked.

I will send you sample cores. Just give me your mailing address in pm.

[mamalala]

Hi Chris,

Attached is the bottom side of the board, as you asked.

I will send you sample cores. Just give me your mailing address in pm.

Looks OK then, nothing suspicious.

That said, you should really check out what is wrong with the driver there. The p-p voltage is way too low, indicating a short somewhere (or a broken driver chip). Instead of just visually checking, you might also want to actually meassure for resistance/conductivity there. While rare, it can always happen that something went wrong during PCB fabrication. I had the occasinal PCB's where a hair-thin short was present between two adjacent traces/signals.

In any case, for reference i attach a hi-res image of  one of the latest RF boards (sans the toroids) i'm assembling. Part values should be quite readable there. Keep in mind that i mostly use 1% resistors, so most of them have 4 digits on them.

Greetings,

Chris

[zoltan]

Hi,

I had no time to build another board these days but I managed to find some data to compare.

Here are the A_L values for different cores and vendors:

Vendor: reichelt.de (data source: www.amidoncorp.com)
FT 82-61: 75 (turn number not defined, may be 10)
T 80-6: 45 µH/100 turns
T 50-6: 40 µH/100 turns

Vendor: kitsandparts.com (data source: toroids.info)
FT 82-61: 79 (Actual measured AL using 10 turns #28 wire - comment from toroids.info)
T 80-6: 4.5 (It is not defined, but I believe the number is for 10 turns, that is 45/100)
T 50-6: 4 (It is not defined, but I believe the number is for 10 turns, that is 40/100)

Chris, what are the values of your inductors? Cannot find them in schematic and the values from older files are not accurate (L3=1u, L4=1u, L5=400n) based on Paulo's measurement (0.6, 07 and 0.3uH).

Edit:
Tonight I rebuilt the signal stage on a new board, and the signal is much better - 4.69 Vpp after oscillator output and 11.6 Vpp on driver output. It still heats up without thermal paste, but it's not  critical. Re-checked every component before soldering and inspected every solder with microscope. Paranoia kickin'   
Z

[mamalala]

Hi zoltan,

sorry, but my crap LC meter simply does not register the inductors. The values i had in the schematic were derieved from the original circuit. I then tweaked the inductors and number of capacitors in the filter section to give a decently high output voltage on a dummy load. But i really can't say what the actual value of them is, since i am unable to meassure them, sorry.

Greetings,

Chris

[richard.cs]

Hi All,

I've been planning to build a 13 Meg Metcal power supply for some time now and I've been following this thread with interest. I'm not intending to make a clone of either the circuit posted here or of the Metcal power supply, simply something that works well and fits with my own style of building stuff.

I've a few questions that people who are building this might be able to answer:

1) This PSU and the Metcal one both regulate based on Peak voltage at a certain point in the filter, It seems to me it should also work regulating for constant RF current or perhaps for constant reflected power. Does anyone see a problem with either of those?

2) Power requirements - I've seen 100V Peak mentioned during heating, so 100 Watts for a few seconds assuming the tip looks vaguely 50 Ohms below its Curie temperature. The SWR vs time graphs are very nice but would it be possible to plot some Power vs time for typical usage, or perhaps post the raw data so I could have a go at estimating it.

3) What does the tip really look like in impedance? My understanding is that when below the Curie temperature it will look like a very lossy inductor, but I don't have a feel for the real and reactive impedances (or do the tips have a small capacitor to resonate out the L?). Above the Curie temperature the ferrite "disappears" and you should just see the copper core as a shorted turn and therefore the tip presumably approximates a short. Some odd behaviour must then happen in between over a very narrow temperature range. It'd be nice to be able to build an accurate dummy load for testing and setup.

I have the ability to measure the tip impedance, but I haven't yet got around to buying a handset. The only ones I have around here are for the 450 kHz system - if anyone wants one of those measured give me a shout and I'll put it on the network analyser, I *think* we've got one that goes that low.

My current thinking is to full wave rectify the mains supply to get a 350V dc bus and run a small non-isolated switcher off of that to power the control circuitry. I plan to generate the RF directly from the high voltage dc and provide the output isolation and much of the matching via a suitable RF transformer. I'm currently toying with the idea of using a pair of line output valves for the high voltage RF source - designed for switching, suitable for the peak voltages expected, tolerant of overloads on timescales of ten seconds or so (tip heatup time), generally bombproof. What's holding me back at the moment is not really knowing what the load is like, and being a little unsure of the best parameter to regulate for.

I don't intend to include the LCD, etc, although I'll probably put a bit more status information on it than the standard Metcal source. I'm currently thinking a few LEDs for "ready, standby, fault" and an analogue meter showing forward minus reverse power.

[mamalala]

Hi Richard,

1) This PSU and the Metcal one both regulate based on Peak voltage at a certain point in the filter, It seems to me it should also work regulating for constant RF current or perhaps for constant reflected power. Does anyone see a problem with either of those?

According to the original patent (which is now expired), it is indeed a regulation for a constant RF current. The original patent:

http://www.google.com/patents/US4626767

In my circuit i use the same point in the RF path for the feedback loop, however, i end up at the dc/dc converter at a different point compared to the schematics of the MX-500P you can find here:

http://www.eserviceinfo.com/downloadsm/33981/Metcal_MX-500P-11.html

In that circuit they use a zener + trimmer, on the input side of the regulator, which goes to the feedback pin together with the signal picked up before L7 (which in my circuit would be L5, the last coil before the output). In my circuit i feed that signal through an op-amp and series resistor into the feedback pin of the regulator.

2) Power requirements - I've seen 100V Peak mentioned during heating, so 100 Watts for a few seconds assuming the tip looks vaguely 50 Ohms below its Curie temperature. The SWR vs time graphs are very nice but would it be possible to plot some Power vs time for typical usage, or perhaps post the raw data so I could have a go at estimating it.

I think you got the numbers a bit wrong. At 100V pp into a 50 ohms load, it should be around 25 watts. In reality i get a bit higher even, up to 120V pp, but that also depends on the maximum voltage that the dc/dc converter can supply, of course, and thus can be adjusted.

3) What does the tip really look like in impedance? [...snip...] (or do the tips have a small capacitor to resonate out the L?).

Good question, and i simply don't know what the real tip cartridge characteristics are. I dont have the equipment to test that.

However, you can take a look at the relevant patents to get an idea of their workings. For the tip cartridges themselves, specifically the heater method, check these:

http://www.google.com/patents/US4745264
http://www.google.com/patents/US4769519
http://www.google.com/patents/US4877944

As for the handpiece, the patent is

http://www.google.com/patents/US4839501

According to that patent there seems to be a capacitor in series to the tip cartridge.

In any case, continuing to pump lots of power into the tip even after the temp is reached will cause it to heat up more, up to a point the cartridge starts glowing where the coil assembly is inside. So it really is important to regulate the RF power into the cartridge once it is heated up and the mismatch happens.

My current thinking is to full wave rectify the mains supply to get a 350V dc bus and run a small non-isolated switcher off of that to power the control circuitry. I plan to generate the RF directly from the high voltage dc and provide the output isolation and much of the matching via a suitable RF transformer. I'm currently toying with the idea of using a pair of line output valves for the high voltage RF source - designed for switching, suitable for the peak voltages expected, tolerant of overloads on timescales of ten seconds or so (tip heatup time), generally bombproof. What's holding me back at the moment is not really knowing what the load is like, and being a little unsure of the best parameter to regulate for.

While it would probably look nice with tubes, i'm not sure that it is really practical. Tubes need heaters, which by themselves consume quite some power. You might end up using more energy there compared to the actual soldering cartridge. Also keep in mind that other countries have different line voltages. If you intend that others may build your circuit as well, i think that directly using the line voltage is a rather risky thing to do. But then, this is of course completely up to you

About usage data, attached is a zip file with the raw data log i made a while ago, using a STTC-126 tip. The .gpd contains the data, one sample every 100 miliseconds, the .plt is the gnuplot commands i used to create an actual plot. Vsup is the voltage out of the dc/dc converter, in volts. Vfwd and Vref are more or less raw ADC readings. SWR is what i calculated as SWR out of the Vf/Vr, and DELTA is just what i internally use in the firmware to detect load changes, it is a combination of various parameters.

The file start with a normal heatup with a cold tip. It starts out with a VSWR of 1.00, then slowly rises. When it reaches temp it makes a dip to almost 1.00 again and then quickly rises, which is also the point where the regulation kicks is and reduces the supply voltage. The next event is when the unit goes to sleep mode with reduced power output. it then wakes up with boosted output for a while, then switching back to normal power output. Then follow a few quick solder joints, a pause, another set of quick solder joints, then again sleep, wakeup, and that's it. Hope it is helpful for you.

Greetings,

Chris

[richard.cs]

Chris,

Thanks for the reply, the patents look very useful. I'll need to think about the voltage to current conversion, it's not really clear to me yet how that works. I've not yet decided between that approach and regulating the current directly - probably with a current transformer based measurement.

I think you got the numbers a bit wrong...

Oops, somewhere along the way I'd read that as peak rather than p-p. That makes things a bit easier.

Good question, and i simply don't know what the real tip cartridge characteristics are. I dont have the equipment to test that.

I have the equipment to measure it (I can do it at work), cold characteristics are easy anyway. Hot measurements might end up being a bit crude. Looks like the first thing I should be doing is buying the handpiece and tip  .

While it would probably look nice with tubes, i'm not sure that it is really practical. [snip] Also keep in mind that other countries have different line voltages.

I'm well aware of the disadvantages with a tube based output stage but I do like the idea of using the RF transformer for isolation. Given there are only really two line voltages in common use worldwide I would imagine a switchable doubler similar to older computer PSUs would be fine, smaller variations would then be ironed out by the feedback loop. Of course a modern design with a power factor correction switcher at its input inherently produces a stable dc bus from a wide range of supply voltages.

I haven't really settled on a design yet and may well end up with something closer to a standard Metcal supply. Or given the lower output power than I thought I wanted a single PL504 across the rectified mains and live with the 9 Watts of heater. Or maybe a reasonably priced MOSFET exists that can handle transients of a kilovolt and can switch at 14 MHz, I'll admit I haven't looked yet.

Thanks for the data files, I'll see if I can make sense of them later.

Regards
Richard

[mamalala]

Chris,

Thanks for the reply, the patents look very useful. I'll need to think about the voltage to current conversion, it's not really clear to me yet how that works. I've not yet decided between that approach and regulating the current directly - probably with a current transformer based measurement.

From what i gather, that last coil (L7 in the original, L5 in my circuit) is the actual CV-CC converter. At least in my circuit, once the regulation kicks in, the voltage at that node is kept constant, which should result in a constant current through the load.

What is different compared to the original is just that i use an opamp and different way of building the feedback loop. In the original, in case of some failure, the dc/dc converter can generate a rather high voltage. In my circuit i still have the regular voltage divider there (R2=18k, R3=2k2) which causes the converter to drop to roundabout 12 volts if no external feedback is coming in. Since the opamp output goes through a series resistor (fixed R4 and variable R32), i can limit the maximum output voltage it can generate (unless boost mode is activated), which is about 21.5 volts.

The opamp is configured so that it references to 2.5 volts instead of GND. That means that any input below that will have 0 volts as output, thus causing the dc/dcv converter to generate the maximum voltage as adjusted with R32. Only if the output of the peak detector goes above 2.5 volts, something comes out of the opamp.

Greetings,

Chris

Edit: as to the Vpp readings i gave, please be aware that those are just what my old scope tells me. But it is half broken, and i have a really cheap, shitty probe. So those values might be off somewhat...

[mamalala]

Just in case:

http://dangerousprototypes.com/2013/05/28/diy-metcal-13-56-mhz-rf-supply/

A few things... For a few days i'm thinking where to go next with this project. There are several, not mutually exclusive, options:

- Single board design that has everything on one PCB
- A version that uses a boos or SEPIC converter, so it could work, for example, with a 12V supply
- One that has "no frills", that is, basically just the RF board designed to work on its own from a single 26V minimum supply

What do you folks think would be usable? In case of the boost/SEPIC option, i would rather prefer the SEPIC variant: This would allow to use a wide range of input voltages, below, equal and above to what the RF stage would need. However, i'm still looking for a suitable converter chip. Since the buck version already eats up to 3 amperes at 30V input, i think that a suitable boost/SEPIC would have to be able to at least handle 5 or 6 amps.

A "no frills" version would result in something like the old RFG3/STSS styles of supplies, that is, no sleep/boost mode stuff, no auto-off, etc. Only the tip-detection would still be there (and of course, still would not require a power-cycling as with these old original units).

What do you folks think about the placement of the output FET? Right now the whole thing is made so that it has to sit flat on a heatsink. Of course it can be made so that a heatsink has to be mounted vertically to the board.

Let me know what you guys think/want/prefer, and i see forward to it that the next step will include that as much as possible

Greetings,

Chris

[Paulinho19]

Hi Chris,

for me it's the single board (10 x 15 cm) SEPIC, bases for example on this chip:

http://www.linear.com/product/LTC3805-5

The FET position is good, but the IRF510 is too close to his limits and it’s easy to kill them. I think a bigger (TO-247 for example for better heat transfer) and better (200V 10A) FET would be great, but I don't know if there exists one. Overkill is always appropriated.

My unit is 99% done, next week I will post some pics, videos and scopes.

Paulo

[richard.cs]

My metcal handpiece just arrived and I was somewhat surprised by the connector. I *think* it mates with a normal F-type but it's not threaded, it grips with some kind of collet. Is this normal?

If it does fit the F to N type adaptor I have then I should have some impedance mesurements by the end of today.

[mamalala]

My metcal handpiece just arrived and I was somewhat surprised by the connector. I *think* it mates with a normal F-type but it's not threaded, it grips with some kind of collet. Is this normal?

If it does fit the F to N type adaptor I have then I should have some impedance mesurements by the end of today.

Yes, it fits on a standard F connector. Untighten the collet, plug it in, then fasten with the collet.

Greetings,

Chris

[richard.cs]

First measurements are in. At 13.56 MHz the handpiece with tip (STTC-147) looks like 42.3 + 13 j Ohms. So a 16 dB return loss / SWR of 1.4 in a 50 ohm system. Not too bad a match I suppose. I'll try and get some hot measurements done later and if I can maybe some of the tip alone. The handset alone looks like 280 pF.

Bear in mind that I haven't taken out the handpiece cable so the reactive components are what is measured looking into the F connector, not what actually exists at the tip end.

[mikeselectricstuff]

Just in case:

http://dangerousprototypes.com/2013/05/28/diy-metcal-13-56-mhz-rf-supply/

A few things... For a few days i'm thinking where to go next with this project. There are several, not mutually exclusive, options:

- Single board design that has everything on one PCB
- A version that uses a boos or SEPIC converter, so it could work, for example, with a 12V supply
- One that has "no frills", that is, basically just the RF board designed to work on its own from a single 26V minimum supply

What do you folks think would be usable? In case of the boost/SEPIC option, i would rather prefer the SEPIC variant: This would allow to use a wide range of input voltages, below, equal and above to what the RF stage would need. However, i'm still looking for a suitable converter chip. Since the buck version already eats up to 3 amperes at 30V input, i think that a suitable boost/SEPIC would have to be able to at least handle 5 or 6 amps.

A "no frills" version would result in something like the old RFG3/STSS styles of supplies, that is, no sleep/boost mode stuff, no auto-off, etc. Only the tip-detection would still be there (and of course, still would not require a power-cycling as with these old original units).

What do you folks think about the placement of the output FET? Right now the whole thing is made so that it has to sit flat on a heatsink. Of course it can be made so that a heatsink has to be mounted vertically to the board.

Let me know what you guys think/want/prefer, and i see forward to it that the next step will include that as much as possible

Greetings,

Chris

My thinking was to integrate the PSU, user interface, oscillator etc. using a MCU, promarily to minimise the part count - Microchip have  recently brought out a PIC with built-in building-blocks for switchmode PSUs - I've not looked at it hard enough to see how useful it would be for this.
It would also be nice to have something that fits an off-the-shelf shielded/heatsink case - maybe something like this: 
http://www.evatron.com/index-pag-products-cid-13-sid-91-l-2.html
using PCB as front & rear panels.