The Art (maybe a bit of Science too) of Troubleshooting!
Three Easy Steps To Troubleshooting!
What are you kidding? There simply are no three easy steps. So get over it! That indeed is the heart of the problem –we all want to slide into home, take the easy way out, do the minimum work, send out 100 emails in hopes of having someone else resolve your problem. It just doesn’t work that way. I repeat there are no three easy steps! No friends this is not Fake News!
That said there is however a process for troubleshooting a project and that is what I will share with you. A long time ago I came up with my process after observing how heathkit was successful in the kit market until they became a dinosaur and were overrun by the black boxes from the land of the rising sun. Heathkit’s success was build and test a portion of the circuit before moving on to the next element. In effect the portions completed actually became a part of the test system as you added new elements. How simple!
I like to scratch build my radios and avoid the many kits that are currently on the market. Here is the rub –current kits –you get a board and bag of parts. Some suggest installing all the resistors first and then the capacitors. Or the reverse, caps first. Next install the devices, hook up the wires and give it the smoke test –often it not only smokes but catches fire too! So now you have this semi-burnt mess or if it hasn’t burned it plain don’t work! How in the world do you methodically go about isolating the problem areas? There was/is no testing of what you built as you progress through the bags of parts. One notable exception was the ‘softrock” kits from KB9YIG. This was the heathkit model.
So here is my process:
Turning on the soldering iron is weeks if not months away. The first steps are to get smart about your project. I call this noodling! Do your homework and research the hell out of it. Be wary of the “Internet boasts” There is no way you can run an IRF510 with 12 volts on it and from a single device have a 500 Watt PEP linear stage. Ain’t gonna happen! Collect data on similar projects and set up a system that is easy to use so when you need the data it can be readily found. Think in terms of circuit chunks such as an audio amplifier block, a microphone amplifier block, a product detector/balanced modulator block, an IF block, a Mixer stage block, a receiver and transmitter RF amp stage, Band Pass and Low Pass filter blocks. Most of my transceivers are just that, a collection of blocks. An article I wrote for QRP Quarterly on a SSB transceiver was entitled JABOM (Just A Bunch Of Modules).
Using the modular approach serves the purpose of ease of trouble shooting as well as providing a pathway for circuit improvements. Want a Digi VFO instead of that drifty analog VFO –pull the analog module and install the Digi VFO. Want a better audio amp –Boom same drill! The module approach serves another purpose –simply flexibility in how to cram, stuff, and/or shoehorn in the boards into that neat metal tin that formerly housed cookies. I often make card board cutouts that are the same size as my module boards and using these cutouts can move them around the enclosure to determine a best fit –all the while mindful of minimizing circuit interaction and unwanted coupling of signals. I can’t think of one of my projects that is just a single circuit board. I have over 100 you tube videos of my projects and you can confirm this for yourself.
Backing up from the troubleshooting side is the decision you made early on to homebrew something and with that brings the other side in having suitable test equipment. Don’t begin a homebrew effort if you have no means to verify if it is working. Bill, N2CQR has a monthly podcast where I just ride along and have lots of fun talking about radios. Almost two years ago we started an effort to encourage homebrewing a simple 7 part 80M CW transmitter, known as the Michigan Might Mite. We even supplied 3.579 MHz crystals (mostly Bill) for the project. That seemed to be a hit with the listeners. One ham who competed his MMM emailed Bill and asked how to know if it was working –as all he had was a crystal set and couldn’t hear any CW note. Yes a crystal set –no meters, no O’ Scope not even a receiver that could tune to 80 Meters. This clearly illustrates the issue –what a downer to have built the MMM and not know it was working. Another ham put a Bitx20 on the air and while he could readily hear other stations –no contacts. He asked how would he know if his rig was working as he had no other receiver in his shack so he could listen for a signal. Bill suggested he look at his SWR meter and see if the meter moved in concert with the voice peaks (about 1/2 scale from tune). Yes, No SWR Meter in line and simply didn’t have one. So acquiring some test equipment is very much in order if you want to homebrew your own rig and is mandatory when you have to troubleshoot same. There are about 6 or 7 pieces that I would put on the list including an SWR meter [Note one of the reasons for low output or oscillating finals is too high an SWR but you would never know unless you knew how was the antenna (load).] Next would be a DVM (Digital Voltmeter) such as the ExTech series from Jameco Electronics and a frequency generator/counter such as the Feel Tech which can be had for around $50 on eBay. An LC meter (Marlin P Jones is selling one for about $40). To that I would add a DSO (Digital Storage Oscilloscope) such as Bill has the 100 MHz, Rigol about $400. For the ham with the Bitx20 who didn’t know if it worked — if he had DVM, he could have built an RF probe (three parts) where he could see if there was an RF voltage when the rig is fed into a 50 Ohm dummy load which is another item to the list. Dummy loads are easily homebrewed! Take twenty 1K 1/2 watt resistors in parallel –Boom 10 watt, 50 Ohm resistive load which is good for QRP levels. OK come up for air and swallow hard — that comes to around $700. The biggest item is the DSO but the others can be acquired over a few months of time and are far better than just a crystal set. Should not forget about finding an all band inexpensive shortwave radio (Sanjean) and that would give you a radio where you could at least find your signal. To that I would add an isolated low wattage soldering iron– that Radio Shack 80 watt iron helps heat the shack but not good on surface mount soldering or repair. While we are at it, throw out all of those $1.95 12VDC WalWarts –they are noisy and will make it seem like your project is not working! Get yourself or build a 12VDC, 2 Amp Linear power supply (one with a real transformer). Hand tools are still more items to the list. Trash those crappy Harbor Freight screwdrivers and pliers. I could go on an on — but why try to do grain surgery with nothing more than a rusty spoon is the message.
The build philosophy is the next part of my process. Bill, N2CQR and I have talked about this several times. Bill likes to start with what he considers the “tough nut to crack” and since he is partial ( I should say very partial) to analog VFO’s –he always starts there. My approach is to start at the back end and build the audio amplifier stage wherein I have a couple of standard designs. [I have committed these designs to circuit boards that are cut on my CNC mill. If I want more “poop” out of the audio I use the NE5534 driving the LM380. A less demanding application would be a 2N3904 driving an LM386-3.] I know these designs work AND I know how much “poop” they will deliver. So once built I can easily tell if they are working properly! The next element is the Product Detector/ Balanced Modulator. I favor the use of the SBL-1 or TUF-1 DBM for this application. Again I have templates in my CNC files. Once installed I can hook up to the audio amp and being careful to not exceed 1.414 Volts Peak to Peak (7dBM) inject that signal into the LO port –you can tell if it working — any LO signal (any frequency at this point) will give you that test. At one time I had a small single transistor test oscillator using a 2N3904 and a socket built from a SIM socket ran various crystals into the LO port. The Feel Tech generator will do the same BUT watch the 7dBM –you’ll smoke the DBM otherwise. Now with these two stages working build the IF amp stage (consisting of the IF amps and crystal filter) and when that is completed hook that up to the product detector. If you have that single transistor test oscillator use that to supply the BFO signal to the SBL-1 and the Feel Tech can supply the signal into the 1st IF. Caution, Caution, Caution!!!! Loosely couple into the IF stage and run the signal level way down like may 50 Millivolts. Loose coupling means using a small value capacitor like 10 PF. The signal generator can be used to sweep the filter about the center frequency. If you hooked a voltmeter up to the audio amp output and if you were using a homebrew crystal filter –this is now a test system! By sweeping the filter and noting the audio output readings you can plot the voltage versus the frequency and this will tell you much about your filter! Do you get the point? What you built becomes part of the test system and when you add a circuit it and it does not work –the problem is pretty much confined to what you just added.
Most problems with a circuit can be traced to some common roots, especially when you are dealing with a proven design and what is replicated simply does not work. These root tracings include: wrong component installed, bad component, improper wiring of the circuit, cold solder joints, wiring not connected. To digress I love receiving emails that start by saying “Your Circuit Doesn’t Work!”. One chap who generated such an email sent me a photo of the circuit and had marked voltage levels at every location on the module. At one transistor (a 2N3904) the collector had a “0 Volts” reading which quickly caught my eye. Hmmm 0 volts tells me there is a problem at the collector. Upon close examination of the photo –there was absolutely nothing connected to the collector –it was just sitting there soldered to a pad. Well here is the rub — the chap had all of the data in front of him –Clue 0 volts, Clue nothing connected, Clue why it doesn’t work. I politely responded that he hook up the circuit as designed and then it would work. Never heard back from him. But this illustrates the point of how the taking of simple data can often provide the answer. Another similar email/ photo form another ham reflected the need to improve soldering skills –yes it was a cold solder joint. Many of the components are surface mount and that opens the door to solder bridges shorting out pins or worse hot pins to ground. My process therefore before applying power is to reaffirm: right components right place, tracing the wiring to assure correctness and to observe polarities on diodes and electrolytic caps, checking the quality of solder joints especially grounded connections and to look (with a headband magnifier ) for any solder shorts! Needless to say –is the power connected with the proper polarity.
I have a stable of proven horses (circuit modules) so I well know their level of performance. I know what works and how well it works. Some argue I would never try anything new — no –it is simply I will try a new circuit in an existing well understood design. Then if that new circuit adds to the performance it will now become one of the horses in the stable. A friend is building a project where he is utilizing the Simpleceiver project that was featured on this blog as a part of is project. For the front end RF amp he was using the Dual Gate MOSFET (two J310’s in a cascode circuit) developed with the Simpleceiver but wanted to compact things. I suggested the use of a single 2N3904 as an RF amp (have used this over and over). Both provide about the same level of gain BUT the J310’s offer a way to apply AGC– but for his application not needed thus the 2N3904 choice. He reported back the 2N3904 was not as good as the J310’s –well 15 dB of gain is 15 dB. So that is a clue that there is something needing further attention with the 2N3904 build (not the design). I further suggested simulating both circuit in LT Spice and if you get the same answer on gain then it is how it was built.
Soldering skills must be learned! We do not weld our projects (that is a different hobby)! Many problems are encountered that are traced to lack of soldering genes. If the joint looks like it is welded there is a good chance the parts being “welded” have suffered heat damage. Too much solder can lead to poor connections –cold solder joints and often a large welded joint spills over and shorts out other parts. There is also the problem of how to “unsolder” the welded joint. Take time to learn how to flow solder on to joints and how to do it without overheating the parts being soldered. This is where a temperature controlled iron is vital. Grounded irons are important to prevent static discharge to very sensitive (CMOS) electronic parts. Like I said the soldering part is like the last thing you do!