DIY Metcal 13.56 MHz RF Supply

[parasole]

Continue design study 

mamalala design could be improved by simply adding pre-amp stage (as in Sergey design) which will solve the driver overload issue and will raise output power.
Another aspect is the output transistor blow-up at the power-off time, which is a well-known phenomenon in high power RF amplifiers, that is the energy accumulated in the output filter which applies to the closed transistor and is well higher than its drain breakout voltage. In sergey design a TVS is used which is OK at this relatively low frequency, but a better solution is simple to use a transistor with a higher breakout drain voltage.
One common technics used in industry is to taper drive oscillator signal down before switching it off, which means implementing soft power off in firmware. Here same might be done by driving down the output stage power supply.

Otherwise, two designs are equal, probable a little bit more tweak might be done by optimizing low pass filter design...   

[rfmerrill]

mamalala design could be improved by simply adding pre-amp stage (as in Sergey design) which will solve the driver overload issue and will raise output power.

I don't know about that. Sergey's design uses a power transistor (STP19NF20) with 3x current rating and 1/3 Rds(on), but also 3x gate charge, and that is what necessitates the preamp stage. The IRFP540 design you mention uses a gate driving transformer because the IRFP540 has even higher gate charge than that (150 nC vs 8 for the IRF540 and 24 for the STP19NF20).

150 nC * 13.56MHz = 2 ampere, which is on the order of magnitude of the output of the station!

Another aspect is the output transistor blow-up at the power-off time, which is a well-known phenomenon in high power RF amplifiers, that is the energy accumulated in the output filter which applies to the closed transistor and is well higher than its drain breakout voltage. In sergey design a TVS is used which is OK at this relatively low frequency, but a better solution is simple to use a transistor with a higher breakout drain voltage.

Sergey mentions that the Metcal design does use such a transistor, but he used the TVS to allow for easier part selection.

One common technics used in industry is to taper drive oscillator signal down before switching it off, which means implementing soft power off in firmware. Here same might be done by driving down the output stage power supply.

The bigger concern is not when the transistor turns off, but when the load is disconnected, as we don't know when that is going to happen.

[parasole]

I don't know about that. Sergey's design uses a power transistor (STP19NF20) with 3x current rating and 1/3 Rds(on), but also 3x gate charge, and that is what necessitates the preamp stage.

Using 3x more current rating transistor sure would require more input power as usually gate input capacitance will increase, but my point was about possible improvement based on what I read about problems with that design.

Sergey mentions that the Metcal design does use such a transistor, but he used the TVS to allow for easier part selection.

yes, as I mentioned, using TVS for this frequency is ok, and it is not a linear amp that would have totally different requirements to its output. But in mamalala design used transistor is quite at the low end with its 100V drain voltage...

The bigger concern is not when the transistor turns off, but when the output is disconnected, as we don't know when that is going to happen.

Right, in this case, the whole power would be reflected, but for this, we have a feedback loop controlling DC supply... however I admit it might not be that quick, and combined with low-grade transistor the result is as is...

[rfmerrill]

Respecting one of the PDFs you linked:  Going by their gate voltage graph they have something that looks vaguely like a ~12V amplitude sinusoid on the gate. With the gate capacitance of 2.6nF, that means the current waveform in and out of the gate is 1.6A RMS, so that's roughly 20 VA of reactive power just to drive the transistor. Doable of course but has to be kept in mind.

[rfmerrill]

yes, as I mentioned, using TVS for this frequency is ok, and it is not a linear amp that would have totally different requirements to its output. But in mamalala design used transistor is quite at the low end with its 100V drain voltage...

I'm not sure what voltage the transistor in the mamalala design sees as it's not connected the same way as in Sergey's design. Sergey's design is basically class C. Mamalala's design might also be class C (could be A?) but it's transformer-coupled, so I don't know if reflected power hits the drain of the transistor the same way. (I'm an RF noob)

[parasole]

Respecting one of the PDFs you linked:  Going by their gate voltage graph they have something that looks vaguely like a ~12V amplitude sinusoid on the gate. With the gate capacitance of 2.6nF, that means the current waveform in and out of the gate is 1.6A RMS, so that's roughly 20 VA of reactive power just to drive the transistor. Doable of course but has to be kept in mind.

That is one more point for pre-amp use, final stage gain is not that much and usually require substantial power to drive them...

I'm not sure what voltage the transistor in the mamalala design sees as it's not connected the same way as in Sergey's design. Sergey's design is basically class C. Mamalala's design might also be class C (could be A?) but it's transformer-coupled, so I don't know if reflected power hits the drain of the transistor the same way. (I'm an RF noob)

it also looks to be class C as no bias network, and it is 1:1 transformer, then final amplitude depends on how well the output network is matched...

[quadtech]

Ocela over at vrtp.ru successfully transplanted the RF feedback from Sergey's design into the
Mamalala board (can see via google translate) -

https://vrtp.ru/index.php?showtopic=30618&st=300

Mamalala reviewed some of his design choices and scope for improvement here, including converting
from a single driver to a push pull design -
https://www.eevblog.com/forum/projects/diy-metcal-13-56-mhz-rf-supply/msg1467251/#msg1467251

[parasole]

Ocela over at vrtp.ru successfully transplanted the RF feedback from Sergey's design into the
Mamalala board (can see via google translate) -

https://vrtp.ru/index.php?showtopic=30618&st=300

Mamalala reviewed some of his design choices and scope for improvement here, including converting
from a single driver to a push pull design -
https://www.eevblog.com/forum/projects/diy-metcal-13-56-mhz-rf-supply/msg1467251/#msg1467251

Implementing RF feedback is a must, good to see it is already done.
Regarding push-pull design, that is the easiest way, and I found one good "paper" 
Just use your transistor of choice and downsize the output transformer, just pay attention as it is transmission line type thus wound with coax cable...

[quadtech]

This one below looks good - for the output filter, the 20m band values in the first link are quite close -

http://pa-11019.blogspot.com/2016/11/diy-kits-70w-ssb-linear-hf-power.html
http://wa5bdu.blogspot.com/2019/03/70-w-ebay-mosfet-hf-amplifier.html

[parasole]

This one below looks good - for the output filter, the 20m band values in the first link are quite close -

http://pa-11019.blogspot.com/2016/11/diy-kits-70w-ssb-linear-hf-power.html
http://wa5bdu.blogspot.com/2019/03/70-w-ebay-mosfet-hf-amplifier.html

Yes, you may use the same ferrites and overall amp, it is wideband and will work happily on 13,56 Mhz.
One aspect to consider, they are linear and that affects efficiency, in solder station use we don't care about linearity hence bias network should be excluded which will transform it in class D amp. 

[t0m]

If you want to try hacking a buck converter module to add feedback, maybe start with this. Based around LM25116, less than $10. https://www.aliexpress.com/item/32845237362.html

[parasole]

If you want to try hacking a buck converter module to add feedback, maybe start with this. Based around LM25116, less than $10. https://www.aliexpress.com/item/32845237362.html

I am considering using a buck-boost converter and being less dependent on the power supply side. Anyway, right now my focus is on RF power amp, now study some papers so to understand what could be done to remain on the cheap side and with better efficiency and more power...

Regarding LM25116, it looks to be no different from any other converters as feedback pin is available, do I miss something over here?

[t0m]

Regarding LM25116, it looks to be no different from any other converters as feedback pin is available, do I miss something over here?

The chip I think would be suitable, I'm just not sure if the feedback pin is easy to access on the module that I linked to.

[parasole]

The chip I think would be suitable, I'm just not sure if the feedback pin is easy to access on the module that I linked to.

Doesn't look to be a problem, there is a picture from the bottom side and it could be seen that chip is fully accessible, and it should be relatively easy to solder a wire to feedback network resistors.

[quadtech]

The Mamalala RF board provides VR and VF outputs for the Reverse and Forward power.
An LM3914 based SWR display can be used if someone wants to run standalone (no CPU board)

http://m1kta-qrp.blogspot.com/2009/11/lm3914-and-swr-and-protection.html

https://www.qsl.net/lw7daa/LM3914LED.pdf

[Serdg]

Can anyone give me a link for a soldering iron connector?
 Or his model. For example on ebay.

[rfmerrill]

Can anyone give me a link for a soldering iron connector?
 Or his model. For example on ebay.

If you mean the connector that the metcal handpiece connects to, it's just a normal type F coax connector.

I have tried a few different ones and I think this one works the best: https://www.digikey.com/en/products/detail/cinch-connectivity-solutions-vitelec/VF320/4430389

In general I would look for bulkhead connectors that are designed for the nut to go on the outside (as otherwise you have to disconnect it to remove it from the housing).

You also probably want one that has good length on it so it's easier to be sure the handpiece connector is fully seated. Note: It took me a while to learn how to best connect the handpiece connector, as it's kind of weird: You want to unscrew the collar on the connector all the way, then push it down over the jack as far as it will go, then spin the collar down and tighten it by turning the collar clockwise and the *cable* counterclockwise. You are not screwing onto the jack.

At first I tried connectors that crimp onto coax directly, but in my project there is not a lot of space for the cable to come in from the side of the housing, so I switched to using the above kind of connector and having the coax meet it at an angle.

Note that these soldering iron designs appear to all be 50 ohm *except* for the F connector, and most F connectors designed to go on a cable will go on 75 ohm cable!

[Serdg]

Thanks! Good connector.
The fixing nut in the body is probably more correct. Maybe I'm wrong.

[quadtech]

You also probably want one that has good length on it so it's easier to be sure the handpiece connector is fully seated. Note: It took me a while to learn how to best connect the handpiece connector, as it's kind of weird: You want to unscrew the collar on the connector all the way, then push it down over the jack as far as it will go, then spin the collar down and tighten it by turning the collar clockwise and the *cable* counterclockwise. You are not screwing onto the jack.

These are the instructions which are bundled with a Metcal handpiece -

[richard.cs]

Another aspect is the output transistor blow-up at the power-off time, which is a well-known phenomenon in high power RF amplifiers

The blow-up at switch off can be fixed with a trivial circuit mod that disables the device if the input voltage is low. I added very crude circuit with Zener-based thresholding to mine which acted on the tip detect line to force a disable. Other more elegant ways are possible.

Yes, going to more power requires either a change of topology, probably to push-pull, or a larger transistor which then needs more gate drive. If I were to play with it I'd probably start by paralleling a second gate driver IC as the minimum-effort solution (the gate driver after all is a kind of preamp), but ultimately resonant matching of the gate drive would be better as the gate is really a reactive load and driving it with a resistive gate driver wastes a fair amount of power.

Ocela over at vrtp.ru successfully transplanted the RF feedback from Sergey's design into the
Mamalala board (can see via google translate) -

That looks interesting. 

[Serdg]

These are the instructions which are bundled with a Metcal handpiece -

    Great swivel connector! Very similar to the connector from an oscilloscope.
But where is the analogue of this connector sold?

[rfmerrill]

These are the instructions which are bundled with a Metcal handpiece -

Ah, right, it comes with instructions 

The main thing I wanted to add (which is not in those instructions) is if you try to get it tighter by *just* turning the collar, you may end up loosening the jack (especially if you're using whatever F connector rather than the specific one Metcal chose). Turning the cable end the opposite way or holding it still makes sure your torque is applied in the right place.

[rfmerrill]

I have one assembled and two blank boards, plus a little protoboard I made that generates the 12V and 5V, has an UVLO circuit and a bicolor LED that shows rf on/off.

quadtech hit me up for them first but they didn't answer a ping email--if anyone wants this stuff I can ship it for just a little over the cost of shipping.

[Splader]

Hello, I am rather new to all of this, I was hoping someone could point me in the right direction to actually build this. Im not sure on which design to go with, it looks to be that Sergey's is better but I don't know.
I need something that can push 80W and run off 120V mains as I have a Hakko FX-1001 that I plan to use with this. I'm not to great at designing my own circuits so it would be nice if someone could share Gerber's from they're build. Any information is highly appreciated.

[eevbstedt]

On the downside, the range of tips that are available is very small.

It's very hard to find side by side comparison of the size of the handpieces. This makes me suspect that the 470K handpieces are significantly thicker, as the cartridges certainly are (going by the diameter of the cartridge compared to the tip).

I have and use a Metcal SP200 (470kHz unit) if anyone would like any info or details on it.