Today I put the Part Ninja to good use, it came in handy when I was troubleshooting an old Philips PM5132 function generator. While I was working on a project I wanted to take a break and switched the signal gen off, later that day I came back and the damn thing failed on me. I switched it on and the only thing I got from it were -22 volts at the output. Luckily I had disconnected the circuit under test or it would have gotten fried as well.
Next I opened the device. The negative voltage on the output indicated that either the output amplifier or the positive power supply had failed.
The construction is a marvel of european engineering of the late seventies/early eighties. Everything is made of discrete semiconductors, there are just a few oldskool opamps (LM308, uA741), no integrated function generator rubbish like a XR2206 or the likes :)
First I tested the power supply, and my first guess was right: The positive power rail was at 0 and not 22 volts, as it was supposed to be. They used a uA78GCU1C linear regulator, which I never heard of before. Googling turned up a datasheet, it's basically an adjustable 78xx linear regulator with a fourth pin in a crazy package (like a DIP-8 with four pins forming a cooling tab).
As I didn't have one of these long-time obsolete devices I cut some solder links connecting the power supply to rest of the circuitry and sustituted it with my symmetrical lab supply.
I turned on the power - nothing. Bummer.
Now I knew, I had to dig deeper, because the dying power supply took something else with it. Luckily I found a service manual on the internet.
With the schematics at hand I tried to locate the fault by dividing the system into its building blocks. I checked the main oscillator, which is the heart of the circuit, and it wasn't running. I followed the old principle of "divide et impera": I tried to find parts of the circuit which are not needed for operating the oscillator but in case of a fault might feed back into it erroneously and stop it from working. The TTL output was pretty closely coupled to the oscillator, one failed transistor could have stopped the thing from working, so I unsoldered a resistor to separate the circuits, but to no avail. So I turned to the input section, and disconnected the wobbling circuitry. Then I checked the function of a controlled current source that is converting the voltage from the frequency dial potmeter to a charging current for the timing capacitors. This is where I found the first discrepancies, the output current was too low and was independent of the frequency setting. So I narrowed the fault down to one opamp and one transistor. I unsoldered them both and put the transistor to the test:
I was too lazy to use the multimeter and took the Part Ninja instead, which revealed that one of the pn junctions in this transistor had gone open-circuit, degrading the transistor to a mere diode . . .
I did not bother to test the opamp, my guess is, that it failed first and put a overvoltage to the base of the transistor. As I did not have a 2N3964 transistor I substituted a BC560B, which is not a very close fit, but good enough. I wanted to replace the (most probably fried) LM308 with a TL081, but I only had 082s and 084s at hand, so I used a LF411 instead.
BTW, the signal gen is a very intelligent design. The slow opamps operate only at DC or very low frequencies, everything else is done by discrete circuitry. The opamps only fulfill support functions.
I'm happy to tell you that the PM5132 is working fine again. I have not yet replaced the faulty voltage regulator, because I'm going to replace both the positive and negative VREG with LM317 respective LM337 and that will need a separate daughter board due to incompatible pinouts.
It was a nice exercise of analog circuit troubleshooting and it has shown me the value of the Part Ninja. It has helped me saving this nice piece of equipment from ending up in the landfill.
I just wanted to show off my new Part Ninja, built on perfboard:
...testing a BS107 low power MOSFET.
I used a DIP40 package variant of the '4550, it's completely hidden below the display module. Everything is held together by M3 bolts.
The connections were made with enamelled copper wire AWG 28. The lacquer is burned off by tinning it with a very hot soldering iron (400 ° C). The fumes are pretty aggressive, I guess a fume extractor won't help much, you'll need a gas mask :)
I've left yome space for future additions.
BTW, I've made some component changes to fit what I had laying around: I used a LM385-2.5 shunt reference with a 4.7 k series (current limiting) resistor instead of the original Maxim part and a LP2950 voltage regulator. Jan
There is one more standard function gen IC, its the XR2206 by Exar. The ICl8038 and the XR2206 have been around since the late seventies. The waveforms they produce are of acceptable quality whereas the 2206 is better than the ICL8038. You'll find them in many of the cheaper function generators. As with any universal function generator the sine waveform has a too high THD for audio measurements. But leaving that aside they are good enough.
The (now again obsolete) MAX8038 is an beefed up ICL8038 in terms of max frequency range and sine quality.
Building a good function generator with the same specs isn't that easy. If you come across an instrument built around these parts, it does not have to be of inferior quality.
The whole thing is built from very basic integrated circuits, at least I have not seen a function generator IC or sine shaping network IC. Would be nice to study the schematics. Even the frequency counter is built of 4000 series logic. Nice instrument - you will never run out of spare parts :)
In my opinion it's not a good idea to cram every possible device tester into a single device. The easier, the better (in terms of portability, size, weight, product cost, serviceability, component availability). It may be nice to identify/test/measure most active and passive discretes, but to include a full-featured logic IC tester would just bloat it.
How many crappy square wave generators and 3-channel logic analyzers :) do you need? Buy or build a Bus Pirate for the job.
Regarding the USB connection: What's the point of having a one-size-fits-all instrument when you have to carry a computer around to use it? But without a PC or similar the user interface will be a hard thing to get right.
Regarding I2C: Common I2C DACs have a 400 KHz maximum clock frequency. With an 8 bit DAC, you'd only get a meager 50 KHz update rate (even less when adding the start and stop bits to the calculation).
The IC tester would make for a nice separate project.
More than ten years ago, Elektor magazine published an article describing how to build a 74xx and 40xx logic series identifier/tester. The device was build around a 80C535 or -537 MCU in conjunction with two Z80PIO chips and a lot of other stuff. This instrument was able to identify unlabled logic ICs - you just had to put the DUT in it's socket and push start. Sadly the software was closed-source.
Back then, times were easier, because most of the logic families could be powered from a 5 V supply. For modern logic families it may be safe to power them all from a 3.3 V supply, but what happens to old TTL, Schottky (S series) or 74F ICs at undervoltage? I expect them to work at low speeds, but I'm not sure.
Because of this it may become harder today to accurately identify an IC without knowing any preconditions. Maybe we should build an "IC verificator", which tests the DUT against the according truth table at a previously chosen supply voltage.
While populating the voltage clamping diodes a possible design improvement regarding component availability came to my mind. I'd suggest replacing the SOT 363 diodes by something more common - in a SOT-23 case for example. There are lots of common cathode double diodes (BAS70, BAV70, BAV74, BAT54).
I just wanted to share my Picqueno building experiences with you.
I've put the the power supply and FT232 part of the circuit together first and made a quick power up check using my laptop (USB should be protected against overcurrent, so I put it to the test :] )
I've omitted the 5 V regulator and the external power plug, mainly because I couldn't find any 1117-5.0 regulators in my improvised workshop where I do not have my parts inventory at hand.
Just in case you're wondering: the Q-tip in the picture wasn't used for digging in my ears. I use them to apply flux. I once bought a few one liter bottles of flux cheaply on ebay. This stuff is at least 20 years old but is still working well.
After successfully uploading the "blink" sketch nothing visible happened. The red LED I used for LED4 was soldered the wrong way 'round. It had anode markings while the other LEDs had their cathode marked.
After swapping the LED the board did what it was supposed to do.
In my opinion, this is a perfect application for an embedded Linux system. With Linux as a basis one could avoid to reinvent the wheel regarding all the network interfacing and filesystem stuff, maybe we could even use mplayer for audio decoding.
I've played around with an AVR32 board, the NGW100. The board itself is cheap and runs Linux. The AP7000 has plenty of power for audio decoding, the synchronous serial comm controllers can be used for I2S output, it even has a complete LCD controller (I've used it with a 320*240 TFT). One drawback is the missing USB host controller.
