DIY reflow oven

So I’ve been messing around with SMD component level repair and building for some time, but for more complex boards with things other than discrete components – a regular hot air station just doesn’t cut it. Also when working with 0402 or smaller, they just get blown away.

The solution to all of this, is to get a reflow oven. The bad part? Expensive. The solution? DIY!

I did some research and there are some things you need to look out for:

  • Ideally; you want quartz heaters – as these have low thermal mass, you’re able to switch them fast
  • The smaller, the better (well, in relevance to Wattage). Mine is 9L
  • Wattage: the more, the better. My 9L oven has 800W
  • Thermal Insulation. This is where the cheap ovens suck; they lose heat fast. This can be remedied with ceramic insulation mats. My oven had a double front window, so that already helps a bit.
  • A fine grill is preferred over a baking plate

I ended up buying a small oven on Amazon, a BlackPear BFO 09. Doesn’t look too bad and actually isn’t bad at all.

How do we turn this into a reflow controller – better yet – how do you reflow boards? The first thing to know, is that in order to do reflowing, you need to get two things:

  • Name brand solder: get good solder paste; that can be traced to a real manufacturer
  • The reflow profile for that solder paste

What’s a reflow profile look like?

You’ve got 5 phases in a reflow:

  • Ramp-up: generally to bring to oven and board to a certain stable point that isn’t too far away from the actual reflow temperature. This helps cope with “lag” due to thermal mass.
  • Preheat/soak: every component picks up humidity, even your PCB and solder paste. This phase makes sure everything is gassed out. It’s the longest phase in the reflow profile.
  • Ramp to peak: the flux will get activated, components will start to align themselves
  • Reflow: once you hit liquidus temperature – the components will further align themselves due to surface tension and get soldered to the PCB.
  • Cool-down: you want to cool down fast, but not too fast. We don’t want components failing due to thermal stress.

For home reflow, get leaded paste. Seriously. No you will not die from using this (just wash hands afterwards). No you will not inhale lead fumes (gassing temperature of lead is way higher). It helps keep temperature down, makes better joints, doesn’t grow whiskers,… Get lead!

Okay – back to the oven. So yes: it’s not a matter of just cranking the heat up and cooking it like a cake.

I found a nice looking reflow controller on Tindie (yep – no point reinventing the wheel and I’m also busy with work and family..). It’s a controller created by CraftyCrow. He’s based in the UK so that means no import duty for us EU citizens.

It works with a bluetooth app on your Android Smartphone, so you can have multiple profiles depending on the solder you use. To be honest, the app still has some quirks, BUT I’ve been in contact with CraftyCrow themselves and they reply pretty fast to any questions you have. Also, they’re working on an app update.

You’ll need a couple more things to finish the build: a Solid State SSR + heatsink (I use a cheap Fotek one, might be a clone, don’t really care), a K-Type thermocouple (an exposed tip is best – you know – the ones where you see the actual weld) and some bits and bobs like a housing and power supply.

The Fotek Zero-Cross SSR

The steps to take are the following:

  • Bypass the timer (I didn’t bypass the temperature control, as I keep it as failsafe as it’s maximum temperature is higher than the reflow temperature. Think of it as thermal runaway safeguard)
  • Put the SSR in series with the quartz heating elements
  • Put the thermocouple above the reflow area, i keep it “hovering” above the pcb area, as close to the PCB itself (or even touching it)
  • Hook up everything including the powersupply

In the end, you end up with something like this. You can barely see the controller hanging on the right side of the oven, mounted on the SSR heatsink (It doesn’t even get warm, so perfectly safe).

The oven tracks the reflow curve amazingly well, considering what I’ve payed for it. For the cool-down phase, I just crack the door open a bit. And it’s fine..

I love the smell of baked PCB’s in the morning

Works great, cost me in total about 80 EUR. Highly recommend building one. Much easier for building items in comparison to a hot air station. Don’t get me wrong: the hot air station is great for rework, but not to do initial assembly.

73’s de ON3AD


Sea 1612c Remote ATU – bias & mods

So I acquired a SEA 1612C remote ATU for cheap, but without an enclosure. I figured I’d buy my own enclosure anyway, so no big deal.

As I didn’t want to use any remote cable at all, I had a little challenge:

  • I need to give it power. Okay, that can be done via a bias-tee. The tuner isn’t made for it, but can be modified.
  • I need some way to be able to “tune-on-demand”. I settled on a small Arduino Nano, but will eventually do this via a ESP8266 (I can remote control it via WiFi then)

So what is a bias tee? It’s actually no more than a handful of components: a couple capacitors, some ferrite to make a choke, and some nuts, bolts, sockets.. and a fuse. You double up these components to make one for the other side as well.

So what’s it look like? Well, see below. C1 in my case is 2x 10nF – 3kV capacitors (in parallel – so 20nF in total). F1 is a Ferroxcube 4C65 ferrite, wound with about 30 turn of magnet wire. It makes a nice 3-30MHz choke when testing it with a Vector Network Analyzer. F2 is a 3A fuse.

So how does it work? Well – RF in to the left – the capacitor blocks any DC; the choke blocks any RF going back into the DC supply. Build the same thing twice, and you can run DC over the same coax. Easy, nice, elegant. Great!

Alright, that’s one solution. Now we need to be able to “tune on demand”. There’s a pin called “DTN” on the connector block of the SEA 1612c – “Demand Tune”. The thing is:

  • This only works when it already has a stored tuning
  • Keeping this permanently low (instead of the usual “high” impulse when triggering something) locks the microcontroller up, as it’s on an interrupt pin. So no easy fixes.

Simple solution: slap on an Arduino which pulls DTN down, only once, after being powered on 3 seconds. Just an NPN and an Arduino, collector to DTN; gate to the Arduino output pin and emitter to ground.

I piggybacked the Arduino on the 5V regulator on the board (7805 regulator) so in terms of power: no issue.

To force a re-tune, I press a normal-close momentary switch which breaks power to the tuner. Upon release power is fed back to the tuner. I press PTT which recalls the previous setting. After 3 seconds, the Arduino pulls the DTN line low, causing the tuner to clear the memory setting. I can then put some RF on the tuner and it’ll force re-tune for that frequency.

I found a nice enclosure for the tuner at, a German web store. The enclosure is IP65 rated, well priced, has a window, and comes with all locks and mounting hardware.

I used two aluminium “L”-profiles to securely place the enclosure in the ground. Here’s the end result feeding my 40m long multiband half-wave end-fed antenna.

Yaesu FT-450 – SDR Panadapter & SDR-Console

I was recently gifted a SDRPlay RSP1 (thanks again ON3HVP!) and have wanted to install an RF-tap for a long time in my FT-450.

Now, there’s a couple of ways you can do this:

  • Get an extra RX antenna (duh..)
  • Tap at the antenna port (but you’ll need to disconnect on TX)
  • Tap at the RX circuitry
  • Tap at the IF (in case you’re using an RTL-SDR which doesn’t handle HF too well..)

In our case, we’ll tap at the RX circuitry. It doesn’t see any TX power, so it’s all safe. There’s one caveat: you can’t just solder an extra wire and be done with it. If you connect an extra cable and receiver; you’ll load this stage down, causing your receiver to be less sensitive. Not something we want on HF!

Enter a great little board by G4HUP (who is sadly an SK at the time of writing…). The board Dave designed is essentially a high-impedance buffer. So it avoids loading down the RX section, while making sure it has adequate power to drive the input of the RSP1.

Seeing we want the full spectrum delivered to the RSP1 – I chose the “PAT V”-board: no filtering, straight from RX on the transceiver; to the RSP1.

Here’s how it’s all connected (click the image for a large version):

FT-450 Panadapter

FT-450 Panadapter

Basically, we’re tapping at the little BalUn next to the white relay; and we’re powering the buffer from a regulator providing 12V – but only at RX (it mutes the adapter on TX by not providing any voltage).

You might ask: “Why aren’t you using a thin coax?” – It’s a recommendation from G4HUP himself: it adds less capacitance which would in turn load down the RX and thus decreasing the signal.

OK – so we got the adapter installed, we have the RSP1 connected and OmniRig is working like it should. (See other post..)

Enter SDR-Console V3 with… External radio support (you can see where this is heading… HI). If you have OmniRig set up; you can enable external radio support in the SDR-Console options. You’ll have to restart the program once you do that.

You’ll notice an extra window in the DSP setting. Enable the “play” button and the “tracking” button. That’s it. Your PC (and thus RSP1) is tracking your radio, but also the other way around: If you see a signal in the waterfall: click it and you’re QRV on the other station’s frequency.

Even better, my logging program (Log4OM) uses OmniRig as well. So I could even click an entry on the cluster, Log4OM sends a CAT command; the transceiver tunes to the frequency; and so does SDR-Console V3.


sdr-console V3

SDR-Console V3

Yaesu FT-450 and Omnirig

Lately I’ve been having issues where my CAT commands from Omnirig would time-out or vice-versa. (Notably the dreaded “Rig is not responding”) I had been looking high and low why or who was to blame. Factory resetting didn’t change a thing, nor did changing the serial cable. The settings in Omnirig didn’t matter – nor did serial speed. Hmm, maybe RFI is to blame? – Nope.

Turns out – the issue is in the driver for the CP210x dongle! If you use the latest version (at the time of writing – v6.7.5) then you’ll experience these mishaps. (I’m betting other rigs using this chipset will be affected too!).

In short – you want to revert to a previous version. I got hold of v6.7.3 and the time-outs are 100% gone. Seeing it’s hard to find this version – here’s a link to it.

Edit 2018-apr-27: It’s better to select the FT-450D model, even if you’re using the FT-450 (original..). The “D”-profile will return the TX/RX status as well.