I recently had the pleasure of working on a 1934 Silvertone 1708A which was brought to me for repair locally. This was great – having a radio repaired can be a big decision, so I’m happy to show off my workspace and chat for a few minutes and go over the radio briefly in person. This particular radio itself is very interesting, too. Sears, owner of the Silvertone brand, liked to re-use model numbers. I discovered 2 completely different radios, one with two slight variations, both sharing the same model number so it also involved a bit of detective work.
The Silvertone 1708A is an 8-tube radio with a dedicated oscillator and two IF stages for additional selectivity, and a tube line-up that showed it was still in a bit of a transition period: 6A7 78 78 37 37 37 42 83V. In most radios even just a year or two later, the 37s would likely have been replaced by 76s in a high-end radio like this one. The 83V is a bit unusual, too. It’s functionally not much different from an 80, and in fact upon a close inspection, it even had an 80 in place when it came to me.
The more knobs the better, and with five, this is near the top of the line. Power, volume, tone, tuning and AM/Shortwave. I went through some intake checks and found 4 tubes were bad, and that transformer looks especially nasty and tested an open winding as well. Underneath was otherwise in decent condition.
It showed evidence of being worked on a few times, and one of the filter caps was put in across a failed capacitor (as was common, but still very bad, practice back then) but no major issues. The speaker was fine too:
Testing showed the other components to be good, so off to replacing parts. I tested the resistors; within tolerance were left alone but others were replaced:
A 2W flex resistor broke along the way. These are incredibly fragile and break if you look at them wrong; they can be replaced with a standard resistor.
With most of the parts erplaced and ready to go, I replaced the bulb and managed a first power-up using a bench clipped replacement transformer.
The lights are on but nobody’s home – and despite good voltages coming off the unloaded transformer, and a normal current draw, there’s only about 20V B+ available. Closer inspection and testing of the bias circuit revealed the resistor in the B+ was cracked and reading very high, around 500K, when it should have been 350 Ohms. I replaced it with a very close substitute with some extra capacity.
She powered right up after that, and while I was poking around, I discovered the original transformer appeared open because of a break just a little ways back; I was able to re-solder the connection to the rectifier and all was well. In my opinion this was one of the nicest radios I’ve worked on – there was plenty of room to work and attention was paid to make sure everything was wired neat from the factory. (Contrast with the Simplex Model P Dual Band from the same year.)
I also added a line input; a simple resistive stereo to mono converter into the high side of the volume control. This way, you can use the radio’s volume control for the input source volume too.
It was time for an RF and IF alignment using my vintage signal generator and digital storage oscilloscope.
The generator puts off a messy waveform, but it comes out as a nice sine on the radio side. Tube AM circuits are pretty forgiving.
While I was working on the electronics, the radio’s owner spent some time reconditioning the cabinet and it came out incredible.
This radio is going to play beautifully for many years to come and will look great in anyone’s living room – especially with the upgrade of adding a stereo line input, it’s also future-proof.
Andrew of Andrew’s Telephony/IT Blog came up with a fascinating project where he built a 3-tap kegerator out of a 7 cu. ft. GE chest freezer, an Arduino, and a variety of flow rate sensors to display exactly how much beer is left in any keg at a given time. It’s a very in-depth project, and it looks like it’s turned out both awesome and functional.
Go check it out!
I’m looking at launching some easy to use kits of the most commonly needed parts, and maybe some helpful accessories and good instructions for the Bose 901 Series 1 and Series 2. No need to order the parts individually, just grab a kit and spend a few hours soldering and you’ll know you have exactly what you need.
Would anyone be interested in that? What other easy repair kits should I put together?
Let me know! Leave a comment or send me an e-mail.
Marty KN0CK sent me some details to publish about his great looking miniature HF upconverter board for the RTLSDR, the HF Alchemy DVB-T Active HF Upconverter. It’s an incredibly miniature SMT board with an SA612 mixer and SMT oscillator, and with some very careful soldering the entire board fits inside the housing and draws its power from the tuner’s USB port. The design upconverts at 120MHz, which is well out of the FM band to reduce the possibility of interference from strong local stations. A 40 MHz low-pass filter on the input further reduces interference. Marty reports it works GREAT!
I also had the opportunity to test an identical dongle and it was very easy to use. It requires a PAL adapter, but most of these dongles need an adapter and this one wasn’t difficult to locate; the integrated form factor is excellent. It’s very sensitive, a bit more-so than my other tuner module even, and the integrated form factor is perfect. It would be very easy to purchase an active USB extension cable and locate this integrated SDR in a shielded enclosure at your antenna’s feed point for even lower losses and versatility.
Update 5/1/2013: KN0CK is releasing a Rev B! I have a teaser.
Thanks for sending this in, Marty!
I started working on a Jamo MPA-101 amplifier back in August for a friend and after some early work it sat for a while. He’s re-doing his audio system at home so I spent a few hours to finish troubleshooting while waiting on parts for every other project on my bench right now.
The MPA-101 is a nice compact desk amplifier for a stereo speaker system or a subwoofer. It’s 50W/channel into 4 Ohms or 100 into 8 Ohms bridged mono and has a very quiet cooling fan which is almost totally silent and doesn’t even come on all the time. Great understated styling, too. They’re still in production and you can even buy one on Amazon for about $200. This one was $20 at a thrift store, if I remember the story.
Jamo is a part of the Klipsch group, and these amps are pretty well regarded. They’re daisy-chainable with cascading inputs, so several of these would make a nice independent amplifier system when paired with a digital speaker controller or similar.
I e-mailed Klipsch and they sent me the schematic to help with the repair process. You can download a copy here.
The amp wasn’t coming out of protect or when it was, it was incredibly distorted and with basically no volume control, only loud crashing. It looked like the power supply had suffered a failure at one point, with the resistor being discolored. Some of the capacitors looked pretty suspect so I shotgun’d it and replaced all the capacitors on this board with new ones. The power supply board was solder jumpered to the main amplifier boards at an edge connector, which was a pretty annoying connection method.
Jim KJ7QT wrote me a note talking about his experience with a similar problem on this model:
I pulled the boards out of the amplifier, and carefully examined them with a 10x magnifing glass – a 220MF electrolytic capacitor (labeled C39 on the schematic) showed signs of leakage at its base, and less than 3 Ohms resistance across its plates in circuit – which should have been around 1K Ohm based on the value of resistor R88.Capacitor C39 is part of a sensing circuit that takes 32VAC from the main transformer, rectifies it to a 12VDC reference voltage, which is compared by the amplifier’s protection circuitry. I’m assuming that this circuit is intended to sense an overload on the transformer caused by a short-circuit and shut down the amp – so when the capacitor failed, the voltage dropped, and the amp was shut down.We also replaced resistors R78 (2.2K, 2W) and R85 (2K, 2W) with 5W parts, upgraded R90 (39 Ohm, 1W) to a 2W part, and re-flowed the solder joints on all of the main power transistors, as the back side joints were quite dry, and one had been visibly arcing under load.
I did also get my new Rigol oscilloscope, which really let me see what the amplifier was doing at each internal stage.
I used Audacity to generate a 600 Hz test tone about 25 minutes long and saved it as an MP3, then played it back from the laptop. The garbage waveform it produced and the laptop rendered makes me want to move my HP 200CD precision audio oscillator up the repair queue, it needs its power supply reconditioned and to be calibrated. One probe was attached across a dummy load at the output terminals. The other probe I held on to and used it to probe the amplifier stages from the back forward. The idea was to compare the waveforms and see where the distortion was being generated in the circuit.
It’s almost 600 Hz.
I started probing the input ICs on the preamp stage.
It was handy having the entire schematic visible at the same time, more or less, working right under it.
Output trace with the volume turned about half-way up. Terrible distortion.
Even worse when the amp was being adjusted
The signals phase better when both are connected together. I assume it’s something to do with the triggering; I’m still learning how to use the new oscilloscope since even this functionality was just not possible on my old EICO 460. This new scope has around 60 years worth of improvements built in.
Here I am probing one of the driver transistors on the amplifier board. The distortion has cleared up a bit it seems.
And at an earlier stage. At this point all I’d really done was clean some connections, cycle the volume knob completely a few times, and reseat connectors but it looked like the amplifier stage gain was working properly. I decided to switch to some music.
I’m not entirely sure how to make the scope snapshots a consistent size. The software isn’t the most intuitive. The communication protocol has been pretty well hacked, though, someone might write a replacement UI for the scope. I switched to probing the volume control, since the distortion only came back when it was moved.
It looked like the volume control might have a broken track internally. It worked fine when not being touched, and must have been worked into making a better connection by moving through its travel but was still very badly distorted and didn’t seem to be getting any better at the low end. I was feeling confident enough to attach an actual speaker to it at this point.
I could hear the distortion, but it sounds much better than it did before.
I ordered a http://www.mouser.com/Search/ProductDetail.aspx?R=RV122F-20-15F-A50Kvirtualkey14860000virtualkey313-1240F-50K from Mouser for about $4, and set to replacing.
I removed the control from the mini-board it was mounted on. Here you can see I split the control to see the carbon tracks under it.
There was one minor hold-up where the new control has a different footprint than the old one. Not the end of the world: each control has 3 wires, so I used a section of Ethernet cable and removed the extra pair. It’s about 4 inches long. These new controls unfortunately had the reverse pinout of the previous ones so I had to remove and re-solder the outside connections for each one to make the control work in the correct direction.
I mounted the potentiometer board to the LED board using a common screw and hole. That’s convenient!
I reassembled everything and set to test waveforms with speakers hooked up and my oscilloscope. Yellow is left channel, Blue is right channel. 600Hz synthetic wave software-generated MP3 tone:
Alternative Endurance streaming station:
Looks perfect to me. It didn’t sound like there was any excessive hum or buzz in the dead time. The original volume control had an additional grounding lug which the replacement doesn’t have. I’m betting this isn’t a significant issue, but if it is, I can reconnect it fairly easily.
Looks and sounds great. These large 6800uF 50V snap-in capacitors fit within about 1mm of the footprint, it can be kind of hard to find good-fitting parts out of all the possibilities out there.
This was a really fun project where I got to use a bit more in-depth troubleshooting techniques, and the end result sounds as good as you’d expect from something by the Klipsch group. I’m excited to hear it out powering a set of Bose 901 Series 1 vintage speakers.
Reader David Forsman, WA7JHZ, read the round-up of RTLSDR upconverter choices and sent me a photo and build schematic of one he designed which was featured in the January 2013 issue of QST, the ARRL‘s monthly magazine.
This circuit up-converts frequencies between 2.3 MHz and 43 MHz by 125 MHz (127.3 MHz to 168.0 MHz) for driving the Realtek RTL2832 quadrature decoder DVB-T device with Elonics E4000 tuner chip with USB 2.0 output. It also incorporates an input bandpass filter (BPF), diode limiter, RF attenuator, and amplifier.
The filter’s response curve:
David reported he’s built two of these up-converters with good sensitivity on 20 and 15 meters. I think I have all the parts except the crystal and coils to build this on hand, so this might be a good excuse to sting my antenna up again and try it out. Thanks David for sending this in!
While browsing Hack-a-Day, I ran into an awesome hack for the original Pong game. One enterprising hobbyist built his own Pong from scratch following the schematics. It’s a big job but it’s manageable since it’s all discrete logic. I’ve just been re-watching That ’70s Show and in one episode Red and Kelso modify the Pong to make smaller paddles. I was shocked to learn this is both feasible, and even somewhat believable for them to have done in their garage with a screwdriver and a soldering iron.
I recently came up with 3 mostly-stripped HP ProBook 1415s laptops that were being thrown out. They’re decent little machines – WXGA (720p) screens, dual-core AMD processors with decent onboard graphics and HDMI output. I have compatible laptop memory and a hard drive from the Dell I found a while ago, so I’ll probably try rebuilding one of them.
The other two, I think I’m going to harvest the LCD panels. Until recently, it’s generally not been cost-effective to reuse laptop screens because the controller boards were integrated into the laptop and aftermarket ones weren’t widely available. That’s changed recently with inexpensive external LVDS controllers like these sold on eBay which make it possible to reuse a laptop LCD as a regular monitor.
I’m probably going with the inexpensive VGA option which comes with an inverter to power the backlight. I don’t have any application in mind just yet, but I’ll think of something.
I’m trying to catalog all the different HF upconverters on the market for the RTLSDR. The RTLSDR stack, a combination of radio software such as GNUradio or SDR# with a Realtek RTL2383 + Elonics E4000-based DVB-T tuner using a modified driver, is getting to be very popular among amateur radio enthusiasts because it’s cheap and highly versatile to allow you to receive on a very wide frequency range. Unfortunately, the chip has some limitations – that wide range only goes from about 50-2200MHz with most in the 64-1700MHz range. That’s well above the HF bands where many ham radio operators and shortwave or AM stations are found.
Fortunately, the marketplace has taken care of that limitation, and quite a few choices for an HF converter / up-converter are now available which use a mixer and crystal oscillator to add around 100MHz to the incoming signals, shifting them into the tuner’s frequency range. There are several choices out there, including pre-built boards, partially built kits, and plans depending on your skill level and interest. An assembled board or a partially assembled kit will probably set you back about $50-100; if you’re savvy you can probably make it for $10-20 in parts – if you don’t mind winding your own coils. Your mileage may vary.
So, getting to it, these are your choices arranged with built boards near the top, partially built boards and kits near the middle, and plans and project logs near the end.
If you have any others I’ve missed, send me an e-mail!
1. There’s a new Ham It Up v1.2 HF upconverter for software defined radio produced by Opendous which has a large amount of documentation including layouts, and can be purchased mostly-assembled for only about $50. v1.2 makes some small improvements, including a 125MHz crystal instead of a 100MHz crystal to ensure there’s no interference from the U.S. FM Broadcast Band. It also features an input switch and an optional hardware noise source is an interesting, if possibly seldom used, feature. Ham Radio Science has a rather extensive review of the original revision and were pretty happy with it.
2. CT1FFU v5 HF converter is a brand new iteration of the long-standing design which was originally one of the first released. This model features an improved smaller PCB size, improved filtering and am improved ring mixer, and phantom power to eliminate a power cable when attached to certain compatible RTLSDR receivers. The LO is still 65.520MHz or 106.250MHz which might interfere with an FM Broadcast band in some countries, such as the U.S., but there is some filtering incorporated to eliminate that. The price is reduced to 55 Euro shipped worldwide with tracking.
3. Marty KN0CK has developed an incredible upconverter which fits inside the tuner dongle’s enclosure using all 0402 SMD parts. It’s a complete stack, including the RTLSDR tuner and hand-assembled precision converter with a SA612-based mixer and Mini-Circuits MAR-8+ amplifier and 120MHz oscillator. Very powerful, great at picking out weak signals, and requires no external hardware or cabling everywhere – just hook up an antenna and go. Available for $100 US plus shipping (U.S. and International shipping available), this is a addition to your ham shack. Buy it over at Easy-Kits. Bare boards are also available for incorporating into your own project.
4. I’ve been using CT1FFU’s v3.1 dongle. German retailer Wimo sells completed kits, which have been in stock even when CT1FFU’s own kits have sold out. A no frills HF up-converter, mine came assembled except for the SMA connectors. It’s one of the early generation upconverters but still offers solid performance. V3.1 uses a 106.250MHz IF.
Wimo also sells the FunCube dongle, if you don’t already have an SDR.
5. JA7TDO has produced the Soft66RTL, including an RTL2383u+R820T and an HF converter with a 50MHz local oscillator frequency in what looks to be a nice, 3D-printed case capable of receiving up to 30MHz, or 50MHz and above bypassing the converter.
6. JaniLab has started selling a derivative of the CT1FFU v2.0 on eBay for a bit lower of a price than some of the others. The older revisions work well, although with some more leakage than subsequent designs, and don’t have quite as fancy filtering or switching features.
7. Janilab also sells another DBM mixer-based converter, with an antenna switch to bypass the conversion.
9. If you’re looking for a more rugged commercial solution, High Sierra Microwave has an upconverter (FCD-1-55-UC) with a 133MHz IF frequency and BNC terminals with an integral amplifier. I’m a fan of converters whose LO frequency shifts the entire HF range above the FM broadcast band in general and the shielded enclosure will definitely cut down on noise. Looks like it’s suitable for mounting outside at your antenna’s feed point, and it also looks like you’re going to pay for those features. If anyone owns one and wants to share their experience with it, or if High Sierra Microwave wants to send me one to evaluate (*wink wink*) I’d
If you’re more ambitious, you can roll your own up-converter from parts. FAR Circuits appears to have manufactured PCBs for sale for many QST projects, including one upconverter. I’ve only recently discovered this site and there seems to be a lot of good stuff.
10. W9RAN developed a RANverter kit which was featured in the January 2013 issue of QST magazine, using a 125MHz local oscillator. It’s gotten a lot of great buzz on the Internet and offers good performance and even a little bit of conversion gain through the mixer. Unfortunately, he appears to no longer be offering it for sale. I did manage to get a set of boards and will be offering the W9RAN precision converter through Easy-Kits.com before the end of the year!
11. If you’re comfortable speaking Dutch, or just with Google Translate, you can buy the Kit RF Converter for RTL SDR Sticks DC – 65 MHz. Unlike most other models, this one HF up-converter takes a BNC 50 Ohm antenna input and has an SMA 50 Ohm output with a 100 MHz oscillator frequency and built-in protection. This one also looks like a great starter kit with through-hole components and large coils and looks easy to build. You can also purchase the completed assembled kit in an enclosure, which also includes a power cable and SMA-MCX adapter cable. Looks interesting.
12. Kalle over at DGK Electronics has a great looking compact 100MHz HF converter designed to fit inside of a pre-made RF shielding box. It uses the ADE-1 mixer and an ASEM oscillator. It has some of the most complex filters and great filtering on the incoming power line, it probably performs very well. He describes the filters on his page, and there’s also a full schematic available. There’s a photo of a pile of boards, and he says there’s still some available, one might be left! DGK Electronics
13. David Forsman, WA7JHZ, sent me a photo and plans of his 125MHz HF up-converter with a diode limiter, attenuator, and amplifier all in one from plans featured in Jan ’13 QST magazine. Click through there to the article for a schematic and explanation for more details and a full schematic. Thanks, David!
14. Matt Dawson GW0VNR has a very interesting HF converter using more discrete parts than some of the other ones I’ve seen. It uses hand-wound transformers, an actual discrete diode mixer, and a Saronix oscillator running at 106.25 MHz. It uses a total of 23 parts and looks like it would be pretty easy to build. He doesn’t have any photos of the completed board, but does have a full schematic, overlay, transfer mask and Gerber files for the PCB. I’m pretty sure I have all the parts to build this one in my box as well. It looks interesting and simple. Check it out.
15. Radio amateur Paulino Sato has posted schematics and specifications for using the TA7358AP FM Front-End as an HF up-converter you can build yourself, using small coils wound on your own forms. It’s a bit of a commitment, but only has about 40 parts. The instructions are in PDF format. Download from me directly or the original is available on DropBox. The PDF contains PCB masks and silk screen layouts and a full schematic.
16. Over in the UK at the George Smart Wiki, we see homebrew plans by M1GEO using hand-wound coils with an SBL-1 mixer and 100MHz crystal oscillator. It has around 20 parts to assemble. These coils look like they could be hand wound on a dowel coil form, and the crystal and mixer are very large parts, so you could probably build this on perfboard without any trouble.
17. Bryce Salmi KB1LQC built a very rugged-looking clone of George Smart’s above with some modfiications dead bug style.
18. Romanian amateur Alexandru YO2LDK built a simple HF upconverter circuit using an NE602. This has an amplifier, limiter, regulated supply and 100MHz frequency like several of the ones pictured, but the circuit itself looks quite different. The amplifier stage is ahead of the limiter, which looks like this one is offering a constant gain versus the adjustable gain some of the others have offered. It looks like this one has more tunable components, which means a little more work to dial it in. I didn’t see any photos of the completed product.
19. Nick G0CWA built an interesting switchable upconverter design, complete with instructions and board layout PDFs.
20. Japanese amateur JA2GQP build a rather minimalist upconverter with only a 19 components offering a +50MHz frequency shift. He’s included a schematic and PCB mask for you to build your own very easily!
There are plenty of options to build or buy for getting HF signals into the VHF range for your RTLSDR. With these choices, there are plenty of options for getting HF signals into the VHF ranges for use with the RTLSDR. It’s not difficult to modify these plans for even higher fidelity and accuracy, such as by increasing filtering on the power lines, building a shielded enclosure, improved antenna systems, and more.
I’ve been using my v3.1 Converter from Wimo for a few months, and have been very happy with its performance so far – there’s little noise and leakage, and I use it to pull in shortwave broadcasts from all over the world including the BBC, China, Cuba, Japan, Russia and more from my home in Seattle with only an 80″ long wire and a string of adapters – I highly recommend that model, or any of the models in this family, for great performance. I’m looking forward to trying out some of these other ones.
If you have a design you’d like to see featured here, let me know!
Edit 12/1/2013: Added JA2GQP’s minimalist SDR upconverter.
Edit 10/29/2012: NooElec offers “Ham It Up v1.0″ upconverter which looks to be based on a different design, and seems very high quality. Check it out!
Edit 2/2/2013: Better info about the Opendous Upconverter, KB1LQC’s DIY Upconverter, and WA7JHZ’s upconverter.
Edit 2/3/2013:Added 9A4QV HF Upconverter SDR UP-100, G0CWA 2012 upconverter, YO2LDK upconverter.
Edit 2/13/2013: Added Vandijken Elektronica upconverter and W9RAN RANVerter 2.0.
Edit 2/19/2013: Added KN0CK SMD HF Upconverter
Edit 4/13/2013: Added JaniLab converters, High Sierra Microwave converter.
Edit 4/25/2013: Informed 9A4QV Out of Stock – Thanks Adam!
Edit 5/14/2013: Ham It Up v1.0 > v1.2, Now Ships with 125MHz Crystal
Edit 8/13/2013: CT1FFU v5 replaces CT1FFU v3.1, and some copy-editing!
Edit 9/2/2013: Updated to reflect availability of some items.
Edit 10/21/2013: Added Soft66RTL
My friend recently purchased a 1991 Mazda Miata from its original owner, in immaculate condition and with incredibly low miles. Except for one issue: the Air Bag light on the instrument cluster was indicating a trouble code. The number of blinks of the light indicates the fault and the light was flashing 10 times, meaning the System Down Fuse had opened. This fault keeps the air bag system from working resulting in reduced safety in the event of a crash, so it’s important to take care of. Used modules can run around $100, with new computer modules starting over $200.
As these cars are getting to be over 20 years old, these module issues have been known for a while and the cause identified: faulty electrolytic capacitors cause the thermal fuse to blow, disabling the system. Capacitors are at the root of pretty much every electrical problem, it seems. We decided to try repairing the module after some research that showed it’s a common problem with a fairly straightforward fix.
This rest of this article demonstrates modifications to your car’s occupant restraint system that are not approved by the manufacturer and if executed improperly could very likely result in your serious injury or death from the air bag failing to deploy, or deploying unexpectedly.
The repair involves sensitive components which can be damaged by even slightly improper handling, furthering the risk of an unexpected failure.
This information is provided only for experienced automotive and electronics technicians as an academic exercise, and KF7LZE is not liable for any consequences arising from following or failing to follow these instruction.
The module is a little blue package that lives up near the steering wheel on the 1990-1993 Miatas. This part wasn’t used in the entire range of the first generation’s production due to revisions that happened along the way. It was also used in similar years of the Ford Taurus, the Mazda RX-7, and there may also be other Ford and Mazda cars using the same air bag module which have similar faults.
We unmounted the board from its housing, then got down to business by removing the bad capacitors:
If one capacitor of a set is bad, it’s very likely the rest of them will be soon. This board uses 5 x 100uF 35V electrolytic capacitors and 3 x 10uF 35V capacitors, all rated at 105°C, along with an assortment of other components that aren’t subject to failure the same way. The System Down Fuse is the long, red-tipped object parallel with the connector on the right side of the photo. A quick continuity check revealed yes, it was in fact open.
From the black spots around the bottom of C7 (center-left in the photo, lower right of R42) you can clearly see the electrolytic fluid had leaked from the bottom of the cans and etched the board a little, but the damage wasn’t that bad. Rubbing alcohol took off some of the residue, but it’s more cosmetic than operational damage and since this board lives inside of a plastic housing itself located inside the steering column nobody is ever going to see it. Replacing the capacitors was very straightforward: de-solder pads, pull old caps, insert leads, and re-solder pads. My Hakko de-soldering tool makes this job very easy, but with any tool it’s important not to overheat the joint or the traces could de-laminate from the board and that usually means the part is destroyed.
Here’s the thermal fuse on its housing. The fuse is the center component, and wrapped around it is a flexible trace completing a circuit between the two center pins. I’m not sure what purpose it serves, but it makes a complete circuit so it’s important to save it. It might be a current sensing winding around the fuse to send a signal when the inflators are triggered, but that’s just speculation based on its placement. Be very careful – on this one, the foil contact pads on the flexible trace split from the mounting points and it was very frustrating trying to get them back together during reassembly.
As you can see, the foil came off the terminal when it was removed. Not good, but not the end of the world either – it’s fixable.
Here’s another view of the board, showing the replaced capacitors mounted up.
Repairing the foil trace was a delicate process. The original connection was a very small contact area, and when it snapped off it removed a bit of foil. There was a thin layer of resin-like insulation over the remaining portions of the foil wrap that needed to be scraped off to expose the bare metal underneath. We first tinned the metal contacts on the fuse mount body, then heated the foil contact pad from the back while applying pressure to force the foil pad into the metal terminal. Once the solder starts to flow, remove the heat but continue to apply pressure – rolling the iron back so it wasn’t applying heat, but could apply pressure as the joint cooled. This took quite a few tries – looking at it wrong the first few times caused it to break off, taking a little more foil off each time. It finally made a good, solid connection and we wrapped it back around the new thermal fuse.
There was a lot of controversy about the thermal fuse replacement on various Miata forums while we did pre-op research. Most commenters who have attempted this repair in the past have been held up on a lack of information about the part number for the thermal component. I can only assume the flexible trace wrapped around it completely destroyed the part numbers on the fuse for most other modules – this particular car has been garaged its entire life, so maybe it was luck. For a safety-critical part such as this one, it’s important it have the right ratings or it could fail to allow the air bag to ignite in a crash or cause an electrical fire after a crash. Several people suggest to replace it with a 1/4W 10 Ohm fusable resistor, a standard metal-film resistor, or a standard fuse instead. I don’t recommend this shortcut.
…replacing it with a standard fuse, metal film resistor, or whatever would circumvent its primary function. Thus, the bottom line is that replacing it with anything but an identical item would risk air bag deployment at the improper time…
After unrolling the film from the old thermal fuse, there were still some very faint, but readable, part numbers listed – very surprising. It’s a Motorola part, which goes well with the Motorola controller chip onboard. Part number 4283A.
The bag the old fuse is resting on gives away the next step a bit. When looked up in the NTE Cross Reference Search, 4283A brings up a modern part number replacement that happened to be stocked at my favorite local electronics shop, Vetco. It’s an NTE8139: 141°C, 15A thermal cut-off. That’s a definite part number! And it’s available in modern production. No need to worry about replacing it with a different part and changing the operation of a safety-critical circuit when an identical component will do!
This bears repeating: the air bag system down fuse in a first-generation Mazda Miata is an NTE8139.
The new fuse gets installed and soldered into place, wrapped again in the original flexible trace that covered it before. It’s very important to remember this is a heat-sensitive device, and you’re soldering to it. We only applied heat for about 10 seconds max at a time, and it was just barely enough to get it to take the solder. The first 10 seconds were a cold joint with the solder holding it in place, the second 10 seconds reflowed the cold joint to be a proper joint to ensure the fuse wouldn’t open up while it was being installed. If you overheat the thermal fuse, you’ll destroy it.
Finally, clip the extra-long leads down to size, reinstall the board in the housing, the housing in the car, and fire it up: The light came on briefly at the start like it should, then blinked off. NO CODES! The air bag system passed all self-diagnostics. We think it’s ready to protect him in the unlikely event of a collision, but there’s of course no way of testing that short of crashing the car. The controller thinks the system is fine, and the parts were replaced with identical new replacements, but full system functionality testing is impossible.
The total cost of the project was about $8 worth of parts, and a couple hours standing over a soldering iron.
This is a fairly straightforward rework job, but you must take special care to not overheat the thermal fuse when installing the replacement or it will fail. I’d recommend using an a solder clip or other heat-sink between the terminal and the fuse body, and it really helped to have two pairs of hands working on this to hold the fuse in place while the other person soldered. If you don’t have a friend who can hold it in place, definitely use a soldering assist device.
And remember, use this information at your own risk. You should not attempt this repair yourself, and KF7LZE is not responsible for the consequences of failing to follow these warnings.