DIY Metcal 13.56 MHz RF Supply

[richard.cs]

Progress update - it nearly works. I've seen lots of RF transiently when messing about.

The op-amp appears to be pulling the output right down because it thinks there's RF from the detector when there's not (much). It's non-inverting input (pin 3) is at about 2.1V, as is the slider and the top of R45. There are no volts across R26 (160k) though so it's clearly not coming from the detector, something somewhere must be leaking about 0.4 mA into here to hold it up against the 5k of R45. Naturally I've looked for solder splashes and poked around with the meter, no real success. Winding R45 to the midpoint doesn't change anything narrowing the source of voltage to the op-amp side of the pot, and given the impedance to ground has fallen and the voltage hasn't changed it suggests a short to some relatively low-Z 2.1V node.

Early on I did see about 10V peak of RF into the load, now I see more like 2V. It was a slow drift downwards and may simply have been something changing as things warmed up given it's not designed to regulate that low. Op-amp output is around 2.7V, it was a tiny bit lower. Volts from the switcher are now about 600 mV, was about 2V. The op-amp should be able to swing almost to the rails so the fact it's not saturating is interesting, it feels almost like a short to another op-amp pin only I can't find one.


One last thing, now that I'm examining the op-amp part of the circuit in detail, I don't see any connection between the input of IC7G$2, a node labelled VREG, and the output of the buck converter which is labelled V+. Have I missed something or is this an error?

[richard.cs]

Another update - Pin 11 on IC7 was floating so the whole op-amp was without a ground connection. I now get 7.4V on L2 and about 25V peak RF out whatever I do with R45. Not sure if it's another bug in my construction or if I've killed the op-amp.

When I power up with R45 wound all the way clockwise I draw 1.75A @ 30V, I get 55V peak of RF into the load and 17.4V on L2. I can adjust R45 a little but I'm limited by my PSU going into I-limit long before I get to 21 Volts (there's a lot of ripple, I set a limit of 2A and it limits in an odd intermittent way when the meter indicates 1.8A). I can find more 30V power easily enough, but drawing nearly 2A with R45 all the way clockwise is quite different from the "few hundred milliamps" in the build guide. I notice however that there's a note in the schematic saying 80k next to the 160k resistor R26, and elsewhere in the build guide it says 120k.

If 80k were fitted to R26 then half the RF voltage would be needed to get into regulation, and a quarter the input power giving perhaps 400-500 mA input. What have other people fitted here? I guess it sets both the minimum power that can be regulated and the loop gain.

Edit: I've used Zayance's numbering,R45 is R19 in the original pack, R26 is R18.

Edit2: With a beefier PSU I can get to 21V on L2. At this point I have 62V peak RF into 50 Ohms (just shy of 40 Watts) and about 380 mV on VFB. I can hear a hissing sound but I'm not sure if it's from the buck converter or the RF transformer, maybe the control loop is unstable and oscillating at audio frequency? Thoughts?

[zayance]

Hi there,

Sorry again for the Silk numbering issue, that was  a major bummer that passed my quick QC, since it took a while before ordering the batch, i went too fast at the end and s..it happened.
Anyway, back then i had asked mamalala if there were some things i could change (revision) since i was ordering a batch, and he basically just replied, "don't need to bother really this one works as is..."  and i can only trust him.
Now of course any revision could make things better, i was not willing to take the deep, and wanted this one just to be doing what it was meant to do haha.

Cool that you've started building this thing, can't find the time on my side.
concerning your problems, i'm sure Mamalala will help, he's been very nice to me when asking for infos..., even if the silk bummer might confuse a little   
All i can say is that the Boards were just copy pasted and silk moved a bit for helping visibility, all the rest was going from original file.

[richard.cs]

A couple of photos of waveforms attached. The RF output looks fairly nice, assuming 60 V peak into 50 Ohms is right when adjusted for 21 V on L2.

The voltage on VFB is interesting though, it seems to be a series of pulses at about 170 kHz, although it wobbles around a lot and I can't get it to sync any better than that. Maybe mine is being odd or maybe everyone's does that? In any case my digital multimeter thinks it looks like 380 mV and the DC on L2 looks fine with just a bit of the 13 MHz on it. Apologies for the poor photos but I think they show what I'm seeing well enough. I suspect the wobble is the much lower frequency stuff that I can hear from somewhere, presumably one of the inductors, as a white-noise like hiss. Sometimes it goes quiet, I think at lower output voltages but it's not entirely predictable and I don't have any scope traces from when it's silent.



Zayance - the silkscreen thing is unfortunate but I'm very grateful to you for taking the time to make and distribute some boards, don't worry about it

[hardware_freak]

If anybody has any PCB left I am really interested in this project. On the other side, if somebody thinks about ordering a batch, I am in with 2-3 PCBs. I live in Germany so if any parts from reichelt are needed... I could order them. I also have an farnell account if this helps.

[extide]

I am definitely planning on building up one of these with both boards. It may be a couple months out though. We might be able to work out something where I can send you boards and you can order and send me parts, and I can send money if the parts are more expensive than the boards. I will probably get them made locally here in the US at www.oshpark.com -- the purple PCB place.

[mamalala]

Whilst you could use the origonal gerbers I would suggest a few of changes:
1) Add in pads for the extra 56 pF capacitor.
2) Put some thermal relief in. I've found a lot of areas where something that carries no real current (d.c. or RF) like a logic pulldown is unsolderable with a small iron because one end is hard-down onto a large area of ground plane.
3) Consider swapping the LCD pinout to the standard one.
4) A little bit of BOM rationalisation would be nice. For example the Aux supply capacitor for the switcher (can't remember the number now) is currently the only 1206 10nF in the design and could have been the same as the 10nF 0805's scattered everywhere.
5) Depending on your skill with surface mount soldering the pads might benefit from tweaking slightly - they're perfect for reflow but a little tight for hand work.

Good points. About the 10nF in 1206, i used that to squeeze the incomming supply track under it.

Greetings,

Chris

[mamalala]

Edit2: With a beefier PSU I can get to 21V on L2. At this point I have 62V peak RF into 50 Ohms (just shy of 40 Watts) and about 380 mV on VFB. I can hear a hissing sound but I'm not sure if it's from the buck converter or the RF transformer, maybe the control loop is unstable and oscillating at audio frequency? Thoughts?

Yes, that looks about right. I have the hissing as well, during heatup. Funny enough, the original STSS as well as the MX500 supplies i have also give out some slight noise while heating up. But then, i haven't looked into the overall stability of the control loop, so there may well be some room for improvement there. However, be aware that there is the output cap's and a whole filter section bewteen the DC output of the buck converter, and the place where the RF signal is picked up for the feedback. Without making the whole response way slower, i would guss that some noise/"wobbling" is normal and expected.

That loop can't get too slow, or too much RF would go into a suddenly mismatched load, likely to blow the output FET. And in the other direction, it may negatively impact performance if there is a sudden change in thermal load on the tip, and the supply needs to crank up the output power.

But i can only repeat myself again: I'm not an RF expert art all, this was basically my first real RF circuit, so i'm certain that there are quite some places to improve the thing. At least it is something that basically works, so people can use it, while those who have an actual clue about the RF stuff (which i just don't really have) can improve the thing

Greetings,

Chris

[richard.cs]

Thanks for the replies, it sounds like my board is working normally then

I built up the power supply part of the control board this morning, it'll be nice to get away from having a big stack of bench supplies for the three rails and also rule out any interaction between the switcher and my bench psu.

[richard.cs]

It works!

Heated a sttc-147 tip in a few seconds, dropped back to about 500mA (at 30 V) in shortly after the solder melted and I could see it respond to cooling the tip by adding solder or wiping it on a sponge.  This is with the complete RF board and a part-assembled control board providing the 5 V and 12 V supplies.

Just need to finish the control board now and do the mechanics.

[richard.cs]

A quick question - how is the encoder connected?

The software refers to the four inputs as A, B, K and S, the schematic as A, B, C, D, and as far as I can tell from the documentation the intention is to have an encoder with a built-in push button and use long and short pushes. To my mind that gives two inputs for the rotation, one for the button, one spare input and GND as the return for all four, is that correct? And if so which is which?

I have a few spare encoders but none with buttons and it sounds like a nice way of doing it. Any suggested part numbers?

Regards
Richard


Edit: A repeat of my question from #395, I think it got lost amongst the rest of the post. Are the nodes VREG and V+ on the RF board connected? Should they be?

[richard.cs]

Another update: I unmounted the boards from the heatsink and re-assembled in a large diecast box. The MOSFET sits in a hole in the box directly onto the heatsink with only a mica washer in between, elsewhere the box is sandwiched between the RF board and the heatsink with liberal amounts of thermal paste because the surfaces aren't entirely flat and there is kapton tape over the non-ground tracks on the underside.

I powered it up to test it after that and it still works though I noticed some odd behaviour. I am uncertain if it is new, I was cooling the tip much more than in previous tests because I wanted to check the output MOSFET didn't get too hot. The supply seems to work normally except that sometimes I can cool the tip heavily and the output power observed on the LCD does not seem to increase as would be expected, likewise the tip seems to cool rather too much. I hope I've not broken anything but I imagine the RF board was flexed a little as it was screwed down and I did give it a brush with some alcohol after getting thermal paste everywhere. The other difference is that this is now powered from a transformer rather than a bench supply. I'm going to try taking some measurements of the buck converter output while doing different things later but it seems almost as though the change in temperature isn't "noticed" but I don't really see how that could happen.

I have identified what I hope is a nice boost converter module that I intend to fit between the existing PCBs and the box wall above the transformer. That would give me the option of 10-30V d.c. input which might be useful on occasion.

[mamalala]

A quick question - how is the encoder connected?

The software refers to the four inputs as A, B, K and S, the schematic as A, B, C, D, and as far as I can tell from the documentation the intention is to have an encoder with a built-in push button and use long and short pushes. To my mind that gives two inputs for the rotation, one for the button, one spare input and GND as the return for all four, is that correct? And if so which is which?

I have a few spare encoders but none with buttons and it sounds like a nice way of doing it. Any suggested part numbers?

Regards
Richard

TA and TB are for the encoder itself (RB1 and RB2 of the PIC), TC is the pushbutton (RB4 on the pic), the fourth one is unused.

Don't really have a part number at hand, i usually bus a bunch of encoders from a surplus dealer on eBay, like, 10 ALPS encoders w/ button function for 5 euros or so. Baiscally any simple mechanical encoder should work.

Edit: A repeat of my question from #395, I think it got lost amongst the rest of the post. Are the nodes VREG and V+ on the RF board connected? Should they be?

They are connected through the inductor L2. VREG goes into the inductor, V+ comes out (and powers the FET). Forgto to put a netname label in the schematic to make it more visible, sorry.

Greetings,

Chris

[mamalala]

I powered it up to test it after that and it still works though I noticed some odd behaviour. I am uncertain if it is new, I was cooling the tip much more than in previous tests because I wanted to check the output MOSFET didn't get too hot. The supply seems to work normally except that sometimes I can cool the tip heavily and the output power observed on the LCD does not seem to increase as would be expected, likewise the tip seems to cool rather too much.

Someone else (Widlarizer on IRC) also noted that behaviour on a few occasions, but it seems to be gone now. He also found out that the control loop is unstable, who knows, maybe that could be the issue. What he did was to a a 1k+22n series combination in parallel to R23 (close to the OpAmp) on the RF board. Dunno if that is what solved the issue for him.

Another possible source maybe the microcontroller. It uses one pin to manipulate the voltage that the buck converter creates. High-Z = nomral operation, low = boost power, high = reduce power for auto-sleep. It may be that there is some  glitch in the firmware that pulls that pin high. On the RF board it is the signal VLIM on pin 3 of the header. If the problem persists, you can try to remove R11 (close to the buck converter and the SMD pot R32 there). This would basically disable that function and make it work in normal mode always. If the problem is then gone, i may have a bug in the firmware. If not, the problem must be elsewhere.

I don't have my DIY station here right now, so i can't check, sorry.

Greetings,

Chris

[mamalala]

Also, a more general note regarding the power supply. Seems that there is an issue that, on rare occasions, may damage the FET or the buck  converter (or both). Originally, the  unit  was designed to be turned on/off through the encoder (simply pushing it once shortly to  toggle betwen on/off). Once it is turned off that way (or the unit has shut itself off after the selected sleep time), it is safe  to "pull the plug", i.e. remove  the main supplly into the  unit.

However, if the external power is  removed while the  unit is turned  on,  for example by simply unplugging it, that rare condition may show up and damage the FET or buck converter. So be warned. Will have to see what i can do about it once i get my unit back.

Greetings,

Chris

[richard.cs]

It sounds like I've got a few things to try. Tip impedance measurements I made a while back also suggest that the tips are the best match to 50 ohms at just a little below normal temperature and not quite as good when cold, that might not help either.

Regarding the failure mode on switch-off; perhaps it just needs an undervoltage lock-out on the input, that would be pretty easy to implement, probably tied into the same enable line as the tip detect. I'll admit I have just turned mine off a few times while it was heating and I haven't broken it yet (unless that's somehow the cause of what I've observed). I have spare mosfets but not buck converters.

[richard.cs]

A better way of doing an undervoltage lock out might be to use the enable pin on the 12V buck regulator on the control board. Pull it down with 10k in parallel with an LED or low voltage zener (clamping the enable pin voltage safely under all conditions). Pull it up with a 22V zener and a 1k series resistor to the dc input. This way neither the microcontroller nor the RF board get enabled below about 23V.

[Zero999]

I remember someone in this thread worrying about the impedance mismatch causing problems. I think it's unlikely to be that bad, given the wavelength at 13.56MHz is 22.1m the lead would have to be quite long to cause a problem.

[richard.cs]

It's finished! Or very nearly anyway.

Since my last post I have modified the board with an undervoltage lockout circuit, mounted a boost converter module in the case for low-voltage d.c. input (10-30 V), and made up a small reverse polarity / overvoltage protection circuit for the new d.c. input, photos attached. Still to do is some input filtering on the DC lines, debugging the occasional odd behaviour seen previously and some general tidying up.

Edit: added schematics.

[girolamo]

For now only two folks interested, and 8 PCB's left.

Anyone else?

Hi
Can i obtain a pair of this pcb?How can i proceed for this?
Thanks

[girolamo]

Hello everyone!
I have read with pleasure all posts and I decided to build such a soldering iron.I believe that the performance of the technology applied in this project ,make this soldering iron ideal for a service gsm ,where I work now. But now I'd like a colleague on the site to help me with the programming PIC18F2620.In this respect I would like last file hex required for programming. Thank you in advance

[ha7az]

Hello!

Interesting project.

What is the final pcb set and ARM hex?

[ha7az]

The topic is dead?

[girolamo]

 Is there anyone who can give me a help?It's about the value of R18 and R19 in limitid circuit power and boost control power on power suply board  ,original schematics.?
Thank you

[richard.cs]

I don't have the schematics in front of me now but I might be able to help.  What's your question specifically? Just what values they should be?