Soldering with new TOMOV 3D Digital Soldering Microscope

My employer was kind enough to gift me with some Award points at Amazon for 40 years of service.  So, I decided to spend it on a Digital Soldering Microscope, as one does.  The full description on Amazon is: "TOMLOV 3D Digital Soldering Microscope: Magnetic Dual Lens Micro Scope TM3K-3D Max For Adult, Coin Microscope with Flex Arm, 3K 10.1-Inch IPS Screen, HDMI/USB Output, Gooseneck Light, Repair Mat, 16GB".

 

 

 

After getting it set up in my shack, beside all the other clutter of pre-existing magnifier, solder stations, hot air, and solder sucker, it looked like this:

 

 I wanted to try a microsoldering repair to build some skill.  As it happens, I had a DFM-17 radiosonde kicking around that I had damaged.  I was putting on a programming header with a soldering iron (bad idea) and managed to rip out a few traces when I bumped it after a few solder joints had grown cold.

 

That header is tiny.  It's a 0.05-inch (1.27mm) pitch, 5-by-2 (10-pin) header.  For scale, that resistor above it is an 0402 size.  In the image above, you can see that I accidentally ripped off the upper and lower right hand pad (SWDIO and nRESET respectively).

To begin, I scraped a little conformal coating off each of the traces, and soldered a dab of solder on the exposed copper.  You can see that in the image above as well.  I then stripped a stranded copper wire, and took one strand to use as my substitute trace.

 

 

In the image above, you can see the piece of wire soldered into place.  Next, I intended to use some conformal coating to "glue" the wire in place, before forming the remaining length into a pad.  This was my first mistake.  As you can see, the (very fine) nozzle on the conformal coating was HUGE for this application.  To make matters worse, it's UV Cured, but apparently does not cure well if you do too much at one time.  After covering the wire, and hitting it with UV light, it basically didn't set up.  After a bunch of fighting with it, I ended up removing the attempt and starting over.

 

 

For the next attempt, I used a very fine pick to dab on conformal coating in fine layers, and hitting it with UV, and then adding more layers.   Above you see an "in progress" shot of layering the conformal coding with that really sharp pick as a "paint brush".  That worked fairly well.  After fully securing and covering the wires, I then scraped off the dried conformal coating over the pads, to expose the copper wire.

 

 

Here's the result after a bunch of fighting with it.  It was ready to add some solder, so I fluxed it and added a dab of solder to each.

 

As I did the repair, I did confirm that I had connectivity back to the SWDIO and nRESET pins on the MCU.  Note, in this picture, you can see yet ANOTHER pad missing on the left.  I forgot to get a decent image prior to my first (failed) attempt to install the header.  In the act of removing THAT failure, I accidentally ripped off that pad on the left.  Fortunately, that one was unconnected, so other than for stability of the header, it's not really needed.

 

S0, about soldering on the header.  The root cause of this problem was using an iron, rather than hot air, to install the header.  It got on crooked, and then I ripped traces while trying to heat and straighten it.   Hot air is a challenge with this part, as it has lots of plastic.  You can't just blow hot air on it.

The solution I settled on was to heat the board from beneath.  Aside from a battery holder, there are no components on the back side in that area.  So, I desoldered the plastic battery holder so it wouldn't melt.  I was then able to heat the board from both above and below.  Once I got the solder to melt, I held my hot air on the bottom while placing the part from above.  After a few adjustments, it was in place.

 

Note, the header still had a plastic retainer installed, holding the pins in position.  That was removed once I was satisfied with the placement of the header.

Once I reinstalled the battery holder and cleaned it up, I tested it by flashing firmware onto the DFM17 with my standard cabling.  It worked like a champ.

 

 

 Lessons learned:

• I should not attempt to install these headers with an iron. Hot air is definitely the way to go.  It takes just a moment to remove the plastic battery caddy that's closest, and the work is very simple after that by heating from the top, and then from below.
• I bought cheap conformal coating.  I need to get better stuff that doesn't have to be painted on in thin layers.  In the meantime, thin layers DO work OK.
• I had also bought some thin copper "trace repair" things that could theoretically be used for this kind of repair, vs just a strand of wire.  The kit of copper traces I bought were tragically small, even for THIS repair.  The pads were sized for 0402 components and were too small for this repair.  I'll keep them, but if I want to use pre-cut copper trace repair pads, I need to get bigger ones.
• Using tiny lint-free swabs to frequently clean the board was really handy.  It's tough to get small enough portions of the conformal coating, and even flux, into place.  At first, I had some leftover flux on the board, and that inhibited the conformal coating from sticking.  So, note to self, clean the board thoroughly before using the coating. 

 

                                                 

 

 

An (un)bonding experience - Predator 9000

 

I was shocked to discover that, by default, my Predator 9000 generator bonds the Neutral and Ground together... Well, not ACTUALLY shocked, but I could have been! Since Neutral and Ground are bonded in the main panel in the house, you're not supposed to bond them together anywhere else. Otherwise, you run the risk of current running over your ground wires, which is a nokie-dokie. I happened to bump into a comment in a youtube about it, and investigated my generator. Sure 'nuff, in the generator, you can clearly see all of the white (neutral) wires piled onto the same frame bolt as the ground (yellow & green) wire (circled in pink). 

 

 

 

 

The solution is simple. Immediately adjacent is a set of binding posts, and the nearest (circled in green) was unused. The binding posts are isolated, and not grounded to the frame of the generator. You simply move the neutral wires over to the binding post, isolating them from ground. Note, the Predator didn't have a nut on this binding post.  It needs a #10/32, for the record.  I had one kicking around, fortunately.  Voila. Done.

 

 

 

I don't intend to ever run this as a standalone generator, but if I did, I could pick up a Neutral/Ground bonding plug and slap it into one of the unused outlets on the front of the generator. Then it would be safe to use for standalone applications.

 

 

I've also got a Predator 4550 (that I got for $50, so why not?!?) and a Predator 2000.  I'm going to check on those.  I'll post another blog with results. 

LoRa Preamp shoot-out

 

 

Two-Way Lora Signal Booster Comparisons

As part of the Alpha Cubesat project at Cornell, we want to set up a "TinyGS" ground station to receive telemetry from the flight in December.  There has been a little back and forth about which pre-amp to use for the station.  I tried using a Uputronix 440 mhz amplifier, but didn't have great results with it.  In fact, I had better results with a cheap LNA I bought off eBay years ago.  We heard through the TinyGS community that this other preamp, available on Amazon, works really well for LoRa applications:

 

 

 

I picked one up and ran some tests on the bench.  The spec sheet says it wants a supply voltage of  3.6V-6.0V.  Since I had a USB Breakout board kicking around my parts bin, I used one to make a USB adapter for it.

 

 

 

That done, I hooked it up to my service monitor and ran some tests.  I generated a 433 Mhz signal, and ran it into my Spectrum Analyzer. 

 

 

 

 

 I adjusted the output until it was reading -50dbM.

 

 

 

Then, I inserted the "Two-way Lora Signal Booster" preamp.  With no power applied, no signal was passed.  Once I applied USB Power, I got a gain of about 11-12 db (-50db up to -38db)

 

 

 

Double-checking the spec sheet:   "Receiving gain: 11dB±2dB"

 

It seems to be dead-on what I would expect.

 

Comparison to Uputronix Preamp

I then simply swapped out the LoRa preamp for the Uputronix. 

 

 

 The result was about a 20dB gain (-50db up to -30db).

 

  

The spec sheet shows a bandwidth with a gain of about 19-20db from 427-452 mhz, so LoRa at 433 mhz would fit comfortably inside the bandpass filter, even at maximum (500khz) bandwidth.

 From the Uputronix data sheet.

 

So, the Uputronix, at 20db of gain, seems dead-on spec.

 

Generic eBay preamp

So, finally, I attached the generic "eBay special" preamp I purchased years ago.  It claims 30db of gain.  I powered it with 9 volts.

  

Sure 'nuff, the gain was about 28db (-50db to -22db).

  

Conclusions

The two-way LoRa signal booster performed the most poorly, with only about 11db of gain.  That said, it has a place in applications where you want gain on LoRa output, as it's a bi-directional amplifier.  However, for TinyGS, where we are receive-only, it appears to be the poorest performer.

The Uputronix preamp performed precisely on spec, delivering about 20db of pre-amplification.  That said, I have seen poor performance with it, compared to the generic eBay preamp.  I will note that the uputronix is designed to work with bias-T or USB input power.  When powered by USB power, it back-feeds DC up to the LoRa board.  According to Google, the Semtech SX1276 is sensitive to DC on the RF input, and it should be blocked.  It's quite possible that this is causing the deafness I've observed.

El-cheapo for the win.  The $12 eBay preamp board showed a gain of about 30db.  It is the best-performing on my home TinyGS station.   I'm honestly shocked, as I expected the 30db preamp, with no filtering, to be filling the LoRa board with noise.  Apparently, the LoRa board doesn't mind.

Future Experiments

After the initial tests, I did check to see if any of the preamps were back-feeding DC to the LoRa board.  Only the Uputronix was doing that.  The Googles report that the Semtech SX1276 must not have DC on the RF input.  So, it's quite possible that's why the Uputronix preamp doesn't work with the LoRa board.  There are little DC blocking doodads that I could pick up and try to put inline between the preamp and the LoRa board.  I may try that at some point to see if the Uputronix performance improves.

 

 

 

 

 

Hojo and the Case of the Deviated Spectrum

 

 

 I was working on another of our passel of Alinco DR-1200T's.   This one is our spare radio which I recapped, aligned (as best I could) and set aside in case our primary APRS radio failed. Well, a different radio failed, and we attempted to use this one in its place.  Unfortunately, it didn't work.

The presented problem was "Won't TX".  I brought it into the shack and ran it through a bevy of tests.  It was transmitting and was pretty much right on frequency.  I ran a tone through it, and it made audio.  I chalked it up to "testing failure" when we tried to use it, and asked our APRS guru to give it another go.

He did, and it still didn't work.  This time, he was able to give me a little more information.  While he attempted to transmit, he had a Baofeng on the APRS frequency, and was listening for the tone.  He noted that it WAS, indeed, transmitting a tone.  However, in his mind it seemed a bit short.  He sent me a short video clip.  When I heard it, It definitely sounded like the tones were WAY off.  

So, back to the bench it came.   I hooked it up to my service monitor and confirmed I could still get a tone through it.  I busted out my function generator, and had it transmit tones from 300hz on up to 5khz.  It sounded OK to me.  I was confused.

After noodling on it for a few days, I had an epiphany.  What if the deviation is WAY off?  Perhaps my service monitor is being very forgiving with a very wide, or narrow, deviation and I still hear the sound, but when it's transmitting, it's not working well with APRS receivers?

I commenced some additional testing.  I've got an old IFR 1200S service monitor, but I've never trusted the deviation meter.  It was time to try something new.  I researched using the "MIN/MAX" method on a spectrum analyzer.

Well, the results were pretty striking.  The radio's deviation was  10khz!   I didn't manage to get a screen shot, unfortunately.  It would appear that my service monitor's deviation meter is WAY off.

So, I decided to adjust the deviation without using my service monitor.  Here's what I had to do.  First, the alignment procedure calls for a 1khz tone with a P2P voltage of 25 millivolts for deviation alignment, and 4 millivolts for the mic alignment.  My service monitor doesn't go that low, and it doesn't really provide for reading the voltage.  So, I opted to use function generator.

  

 This crusty old Instek does the job.  The amplitude of the signal was still way too high.  So, I had to attenuate it even more than the function generator's internal attenuators could do.  I bodged together a line of attenuators to get the tone's voltage down where it was needed:

 That was sufficient to get the P2P voltage down to what I needed.  I tested it with my oscilloscope:

 

Now, confident that I was feeding the signal in according to the alignment procedure, I proceeded to do the MIN / MAX method to measure the deviation on my Spectrum Analyzer.  I pumped the radio's RF output into a tapped dummy load, and fed the tap into the analyzer.

After some trial and error, I was able to measure the deviation.  This is a screen shot after I finally got the trim pot adjusted such that the deviation was the required 4.9 khz, according to the service manual.

 

 

 Once it was properly aligned, I went ahead and plugged it into my IFR 1200S.  The deviation meter was reading 3khz, with all the same settings.  That would explain why my FM Deviation alignments always seemed off.  The IFR must have a problem in the deviation meter.   That also explains why an APRS tracker I built, years ago, never managed to work.  I was using the IFR to adjust the deviation.  Whoopsie.  I guess I'll have to use this method from now on.

So, I ended up fully aligning the deviation on this radio using MIN/MAX.  The proof will be in the pudding when we hook this up and try to beacon with it.  I suspect it'll be MUCH better, since the deviation is now less than 1/2 what it was.   While I was at it, I readjusted the deviation on the other of these radios I had on hand.  The deviation wasn't AS far off - 6.9 khz where it should have been 4.9, but i was definitely worth doing.

I'll hand both radios back off to our APRS guru, and see if they're working better now.  I'm hoping that he won't have to readjust his TNCs to compensate for the changes in the radios.   However, these are "in spec" now, so if the TNC needs adjusting, it's probably best to do it there.

 

I had a serious buzz

 So, I was testing out my new POTA Monster setup before I took it to the field.  

 

The audio reports I got were not great.  One friendly ham was even set up to record QSOs and played back my audio.  It was awful.  There was an extremely loud electrical buzz in the background.

I went through and put more clip-on ferrite chokes on things (I had picked up a bag of assorted  ones on Amazon.  Nothing much helped.  I also tried moving the head off the top of the box and turning off audio processing (PROC) on the radio.  No joy.  I also replaced the frayed cord I had on the mic with a different (better) one.  Nope.

Nearing my wits end, I decided to try swapping the mic out.  Yup, that was it.  I was talking to a POTA operator and he very kindly took the time to let me swap mics back and forth.  The problem went with the microphone.  Phew! 

So, I've swapped the hand mic.  I'll tear apart the other one and see if I see anything obvious, like a short to ground or something.  I've got spare microphone elements, so I could always just swap that if push comes to shove.

So, the POTA Monster is now tested and ready for the field.  I hope to get out with it tomorrow night.   I repaired a 20m hamstick I had laying around (corroded connection on the choke wire).  I also have a new one coming in the mail tomorrow, so one way or the other, I should be good to go.

POTA Monster

 So, I've gotten more interested in POTA activations, recently.  I had a lot of fun with my first activation in the Spring.  I bought myself a Xiegu G90 for my birthday, and put together a go bag and EFHW antenna.  It works OK, but honestly, at 20 watts, kind of lacks the "punching power" I wanted.  There is an "Erie Canal Bicentennial Celebration" POTA event coming up this week.  It runs for 8 days, and I'd love to be able to run out to a local park and activate a few times.  That said, my last two attempted activations were flops, as no one was hearing me.  Also, the portable setup with EFHW is nice, but it involves throwing a wire into a tree in a State Park.  Many parks don't care for that.  I've tried the Xiegu with a Buddistick and, frankly, it sucked.

Considering the problem, I recalled I have some hamsticks, and a Kenwood TS-480HX (200 watt) in a box.  I removed them from our previous truck when we got rid of it, and haven't used it since.  I decided to resurrect it.  Slapping a hamstick on the car roof, and using a powerful radio from a go-box should allow for quick activations and more satisfying results.  So, I decided to make a GO box for my TS-480.

 

 First, I busted out my hamsticks and checked them over.  Unfortunately, the 20m stick (the one I'd use the most) has some kind of failure and won't tune.  I ordered a replacement.

Looking at my TS-480 and taking some measurements of it, along with the tuner, I decided I needed a somewhat large box.  I decided I could make it work in an Apache 4800 case from Harbor Freight.  Note, that I always use separate battery boxes.  I prefer to be able to grab a box with a radio, and a box with battery in an a-la-carte fashion.  That seems more versatile, and frankly, a lot easier to move around, than the huge 19" rack-in-a-box kind that many people assemble.

I'd really >like< to be able to operate the radio without having the case open, so I purchased some bulkhead RJ45 adapters.  I learned, upon assembly, that I had forgotten that the radio head actually uses RJ12.  Fortunately, you can use RJ12 in an RJ45 jack, but I've since ordered the correct bulkhead adapter.

I had an SO-239 and Powerpole bulkhead adapter kicking around my parts boxes, so I threw it together over the weekend.  Pictured here is the complete go box with all the cabling necessary to plug it into a separate, powerpole, supply.

 

I stacked my 200W tuner on top of the radio.  I left a channel underneath everything for air to flow from a hole in the right side of the case through the radio, the fans, and out holes on the left side.  Ultimately, I'll 3D print some "mesh" to cover these holes, and perhaps even print up some "channels" so that nothing can block the flow if the foam gets compressed.

 

  It has Panel and Mic jacks on the front.

 

 The right side has an air hole, and jacks for the antenna and power.  You can see why I need some "channels" to keep wires and foam inside the box from blocking airflow.  The power wires are right in that area, and they keep migrating to right in front of the hole.

 

 

 

I cut and crimped some short patch cables to go from the radio to the internal ports on the bulkhead adapters, and a shorter patch cable for the radio head.   The cable that comes with the TS-480 is really long.  I also slapped some ferrites along the control cabling, since the stock cable came with one.  I put one near the radio head, and another inside the box, just before the internal cable plugs into the radio.

 

 
Testing completed, the last thing to do was to give it a name.  I give you the "POTA Monster".  200 watts of cloud-burning fury.

 

I'm looking forward to trying it out when my new 20m hamstick arrives. 

Hojo and the Case of the Silent DR-1200T

We seem to have a fleet of Alinco DR-1200T's kicking around.  The local APRS expert uses them, along with a TNC, for APRS nodes in the area.  I've repaired, recapped, and aligned several of them over the years.  A new one (to me) hit the bench today.

 

 

 

I popped on the service monitor and ran a few tests.  It generates no audio either to speaker or headphone jack.  However, I do see the S-meter doing sane things as I adjust my service monitor up and down, so it's receiving, but not generating audio.  TX tests seemed good.  I fed a tone into the Mic in and threw the PTT.  It generates a signal of the appropriate strength - about 20 watts or so.

So, something's amiss with the audio out.  My first guess is that there might be something wrong with the final audio amplifier.  That seemed like a reasonable place to begin.  I could check the AF-in and AF-out to see what I might see.

Consulting the block diagram:

 

It appears I'm looking for IC4.  I see an "AF Mute" signal there too, which might be suspicious.  I'll look into that.

Reviewing the (blurry) schematic:

 

It appears that IC4 is a "uPC 1241H".   With a few googles to double-check my assumptions and reading of the schematic, it would appear that 

• Pin: AF In
• Pin 2: Bypass cap?
• Pin 3: Unknown
• Pin 4: GND
• Pin 5: GND
• Pin 6: AF Out
• Pin 7: Unknown
• Pin 8: Vin

Looking at the board layout, I located the chip:

 

 

Busting open the case, I found it against the rear heat sink: 

 

 

Starting with the multimeter, I read 11.6 volts on VIN.  That's down a bit from the 13.6 I'm feeding it.  The current draw showing on my bench multimeter doesn't seem unusually high compared to another unit I had, nor is the chip particularly warm.

So, I grabbed an amplified speaker, and touched it to Pins 1 and 4 (AF In and Ground).  Sure enough, I hear static.  Turning the volume knob on the radio, static volume changed.  Turning squelch, static went away.  Seems reasonable.

I hooked my service monitor back to the antenna port and sure 'nuff, I'm hearing tone on AF In.  Volume and squelch continue to work as expected.  Switching my probes to AF Out, I get nada.

I looked at the remainder of the output chain, wondering of something might be pulling the AF Out down to ground or something.  I don't see much.  C90 seems to connect AF Out back to Pin 7.  Probably some kind of feedback thing.  C91 is along the way to the remainder of the output chain.

I pulled out my ESR meter and checked all of the caps in the neighborhood.  Though not highly scientific about it, all showed an ESR of about 1.5 ohms or less.  Nothing stuck out as a dead short or anything.  

I'm reasonably content that the amp chip is the problem.  If I replace the chip and all is well, I'll probably recap the entire radio, just on principles, but I'd rather not change too many variables at once.

 The chip is, of course, long out of manufacture.  There seem to be plenty of options on eBay for $20, shipped, or less.  

 

Two weeks later 

 The replacement amp chip arrived and I swapped it in.  

 

 There was an "improvement" when I tested.  Whereas, before swapping, I could hear static on the input pin when attached to an amplifiied speaker, the output pin was silent.  With the new chip installed, the output pin was definitely making louder static.  Progress!

I then plugged into my service monitor, and observed the 1khz tone on the input pin (1), but the output pin (6) remained loud static.   Referring to the schematic, the first capacitor after the output pin was C91.  

 

Though I had spot-checked the capacitors in this area with my ESR meter, perhaps I missed it somehow, and it was bad.  I pulled out C91 and immediately, I got very loud tone on the output pin.  Checking C91, sure 'nuff, it was reading open. 

I replaced C91 with a fresh cap, and tested the radio.  I was providing good, loud audio now.   Yay! I suspect that the cap failing probably lead to the failure of the amplifier chip.

Though I had loud audio, it was a bit "fluttery" sounding.  My plan has been to recap the radio after getting the audio to work, so now seemed like the time.  As I went through, I ended up finding 3 capacitors that had failed in a "leaky" kind of way.

 

 I cleaned up the board as I went through the recap process.  I also pulled the inductor (L11) whose footprint you see on the image above.  Goo had leaked under it, so I wanted to make sure it didn't corrode that inductor.

 

 

 

Interestingly, it was the three green Sanyo caps that leaked.  A few of the others looked a little swollen, but they didn't fail outright. 

I buttoned the radio back up and ran it through tests on the bench.  It was more on-frequency when transmitting than it had been prior to the recap.  It was close enough that I didn't fiddle with it.   The audio quality was decidedly better than before the recap.

Fixed.  It's ready to go back into service as one of our regional APRS Digi / iGates.