Side projects and fun

CALIBRATION PARTLY DONE Calibration involves adjusting trimmer potentiometers to set various behaviors, as well as some testing of the limiters and fault detectors.  The adjustments completed were: 10V reference voltage set triangle wave generator frequency set step generator gain to 1V per step high voltage range to 200V max mid voltage range to 70V max low voltage range to 35V max CALIBRATION PARTLY DONE Calibration involves adjusting trimmer potentiometers to set various behaviors, as well as some testing of the limiters and fault detectors.  The adjustments completed were: 10V reference voltage set triangle wave generator frequency set step generator gain to 1V per step The adjustments with issues were: high voltage range to 200V max (erratic adjustment) mid voltage range to 70V max (erratic adjustment) low voltage range to 35V max (erratic adjustment) Verifications were done for: current limiting indicator operation triangle output in sweep and DC modes I am not...

FOUND THE ISSUE WITH THE OFFSET OPERATION After tracing the circuit on the back side which was readily accessible, I found a short across capacitor C52, but removal of that capacitor did not relieve the short. The schematic shows a resistor in parallel but mounted on the front side of the board which meant disassembly of the sandwich to gain access. I lifted the resistor and still did not clear the short. At that point, I referred to the additional PCB I have to look for potential shorts since this signal just runs to the output of the TL071 op-amp that produces the negative or positive bias.  I spotted a via for our signal that was just to the side of an unrelated capacitor and the solder on that component was shorting to the via. Usually vias that are possible to short are part of the same net, but in this case it was our op-amp output net instead. I removed the short, reassembled the sandwich, put all the cables on the sandwich and retested.  Via and adjacent capacitor The ...

CONTINUING CHECKOUT I set up the 1K resistor and scope probes in order to verify that the offset circuits work in both polarities and for both N and P type tested devices now that I have repaired the bad solder joint I discovered.  Generating base voltage steps Injecting offset to step voltage I also worried about a jumper that I had put on a header because its purpose was not documented anywhere and there were no diagrams to show which jumpers were to be inserted.  The assembly instructions stated that a jumper goes on the pulse polarity header on the left two of three pins and that a jumper goes on the P1en header. However, there is also a P2en header and no mention at all of this. I had put on a jumper but after looking at the schematics, it appears there are two ways to produce pulses for the triangle generator, one uses a second schmitt trigger (P1 route) and the other bypasses the second schmitt trigger (P2 route). In reading the detailed blogs about the design process t...

PRINTER PAPER FOR DECAL USED TO ATTEMPT LABELING I found a number of vendors selling printer paper that becomes a decal I could apply to the project box. Sadly I bought one named Koala No Need Spray; the poor English should been a clue. I made the labels and brought them over to soak in water then apply. I discovered that they don't really stick well at all. Worse, they are white background which is not what I want to apply to a black plastic project box. A white decal is just a label, one I could have whipped out on a label maker. I expected a clear plastic decal that would hold colored type while the rest of the decal was barely visible. 

CUT TRACES AND ADDED SHORT JUMPERS Since my PCB had the wrong connections for the op-amp, I did some adjustments to reroute the wiring. An op-amp has negative feedback, a connection from the output back to the negative input terminal, while the source signal is applied to the plus input terminal. The op amp follows the voltage of its plus input to produce a stronger output which is then fed to the analog to digital converter of a microcontroller.  My error had the input attached to the minus input and the output hooked to the plus input. This drives the op amp to its positive rail, 3.3V rather than varying based on the input signal. I cut the connection between output and + input as well as the connection of the source to the - input. Using blobs of solder across adjacent pins, I connected - input and output together. A short bit of bare wire was tacked onto the + input pin and connected it to the source input.  I tested the generator and it functioned as expected, taking inpu...

REPLACEMENT CHIPS PUT IN CORRECT SPOTS BUT INCORRECT OPERATION I was not getting reasonable results when selecting functions such as x squared so I began to troubleshoot using the schematic. It was at this point that I realized that I had made an error in the PCB. I swapped the + and - inputs of the Op Amp, which drives the amp to its rail instead of matching the input voltage with higher drive.  Even though this is a surface mount chip on the PCB, there is a way I can hack the board to correct the connections. Essentially I cut two traces per op amp, use a solder blob to bridge two adjacent pins, and tack a bodge wire to reroute a signal. The chip has two op amps in it, thus I repeat this on each side of the chip. 

FOUND THE CAUSE OF THE STEP GENERATOR PROBLEM A binary  counter applies voltages through a ladder of resistors to produce up to seven even steps in voltage as the 1-bit, 2-bit and 4-bit outputs of the counter turn on in the sequence 000 001 010 011 100 101 110 and 111. The patterns I saw had a step when the count would be 001, then nothing for 010, then the same single step for 011, a larger step for 100, etc. This strongly suggested that the 2-bit line was not getting to the resistor ladder.  Examining the chip on the board carefully, I found the pin for 2-bit had a blob of solder but it did not extend down to the pad on the PCB. Thus, an open circuit. I reflowed the connections to the chip and put everything back together. The stairstep output was perfect - varied from 0 to 7 steps as I turned the rotary 'steps' switch. The problem was resolved.  CONTINUING CHECKOUT Next up, the checkout had me install a 1K resistor across base and emitter connections and do some verifi...

REMAINING CHIP - LM555 TIMER - ARRIVED AND PUT ONTO THE BOARD I soldered that chip onto the board and then swung the two boards together. The various potentiometers and switches had to fit through the openings with the rear nuts setting the front panel at a consistent distance from the rear board.  ADDITIONAL NUTS ARRIVED TO COMPLETE ASSEMBLY Some of the controls I received did not come with the mounting hardware. The nuts used included various fine and very fine thread patterns and two sizes. I put on all the nuts and then installed the knobs.  BEGINNING THE CHECKOUT AND CALIBRATION PROCESS The first step is to verify the proper voltages are present, which is where I discovered my earlier error with the voltage polarity swap. I did discover a zener diode that was soldered on backwards and corrected it. This time, they all checked out properly. +5, -15, +15, -24, +24, +10 and 7.5 are the supply voltages, plus the device under test voltages from the 35, 70 and 200V ranges were ...

THE ANALOG THING (THAT) ANALOG COMPUTER IS ANOTHER TOY OF MINE Anabrid has produced a relatively low cost modern analog computer which they call The Analog Thing or THAT for short. I own one although I won't spend much time with it until I have the current restorations completed.  https://the-analog-thing.org/  The computer uses voltages from -10V to +10V to represent the analog of some variable. Its output is displayed on an oscilloscope or plotter.  DESIGN POSTED FOR A FUNCTION GENERATOR WHICH I WHIPPED UP An analog computing enthusiast designed a simple function generator which expands the range of problems that can be accomplished on a THAT. The functions it provides are: x^y x^2 x^3 sqrt(x) x^(1/3) sign(x) x/(10y) sqrt(x^2 + y^2) x^(5y) e^(x-1) e^(2x-2) 1+ln(x) 1+0.5ln(x) (atan2(y/x))/pi sin(x*pi) cos(x*pi) I don't have the labels on the box yet. The left side has the X input on top, Y input on the bottom and ground in the middle. The knob selects one of 16 functions...

INTEGRATED CIRCUITS AND DIODES REPLACED, ONE TO GO My shipment from Digikey arrived early and since I had lots of frustration working on the computer typewriter, I was able to relax by replacing everything I had onto the front board. The only part that was not in stock at Digikey was a surface mount timer (LM555) which is coming early next week from Mouser.  POTENTIOMETER ON FRONT PANEL HAD LEADS TORN OFF One weakness of the design for this curve tracer is that one of the potentiometers on the front panel, for sweep offset, is not soldered onto a PCB but instead mounted on the front panel and connected with discrete wires. However, to fit the space available the part chosen and listed in the bill of materials is a PCB part, not panel mount.  Thus the part has small fragile leads which tore off as I opened and closed the sandwich of front panel and front board in order to test, debug and replace parts. The broke off right at the base of the part.  REPAIR OF POT AND REPLACE...

EBAY PURCHASE OF MILLIVAC MV-864A DIGITAL MULTIMETER I found a DMM on eBay with the ability to measure very low voltages, currents and resistances, but at a nice low price. It arrived today and I tried it out.  The specifications are quite respectable. Microvolts, nanoamps and tenths of an ohm is pretty decent sensitivity. 

CABLING NEEDED BETWEEN BOARDS The four PCBs for this curve tracer are connected via a number of cables and wires. A few connectors such as the electrical power input use terminal blocks and there are three shielded cables running to the BNC connectors on the rear, but most use 2.54mm spaced connectors.  I wired up all the two, three and four wire cables to the connectors, allowing enough room to attach and remove the connectors while servicing the curve tracer. The connectors are keyed so that they fit only in the proper orientation.  CLEANED UP PADS FOR PARTS THAT ARE BEING REPLACED I had removed all the ICs and diodes that I suspected after the reversed power incident, but there were still blobs of solder on the pads. To prepare for the new parts to be soldered easily, I used some solder wick to remove the excess solder and get everything clean and ready for installation. 

ALL COMPONENTS INSTALLED AND TEMPORARY CABLING INSTALLED Once I had all the parts installed in place and after the main board had checked out properly, it was time to connect the front panel/front board sandwich to everything else so that I could finish the testing. I have male plugs on order to build the various cables for a permanent interconnection but used a number of breadboarding style jumpers in the meantime to be able to start on the testing.  BEGIN THE VERIFICATION AND CALIBRATION PROCESS Early on in the testing I discovered that the -15V 'step' power supply was missing but +15V was good. As I was tracing down the issue I discovered that one of my temporary connections between the main and front boards was not wired properly. I had reversed the + and - step supply connections to the front board.  When I fixed that issue, the +15V supply was now at zero while -15V looked fine. The fault is on the front board/front panel sandwich and I realized that I had blown out some...

AFTER THE THOROUGH CHECK OF EVERY RESISTOR, MISTAKE WAS LIMITED I found no errors at all on the front panel or front board. The main board problem was restricted to eight resistors which were intended to be 100K ohms but were installed with 1K ohm parts. This was likely an error I made when sorting the parts and assigning each bag to a list of resistor locations; I must have misread the bag. Alternatively it might be a sorting error at the distributor where what I received were different value resistors than listed on the bag.  I removed the incorrect resistors and soldered in the correct 100K ohm parts, then repeated the testing of the triangle wave generator, which performed as expected now. The rest of the procedure alternates between testing of sections and installation of the final two transformers which generate the higher voltages to drive the device under test - from 35V to 200V in three ranges.  MODIFIED TRANSFORMER TO ALLOW RELAY SWITCHING OF VOLTAGES One of the tran...

TRIANGLE WAVE RESULTS DON'T MATCH EXPECTATIONS The board is generating perfect 5V rectangular (square) waves and nice looking triangle (ramp) waves, but the adjustments are not working as they should. The frequency trimmer should be able to set the frequency of the triangle waves at 160Hz but the lowest it will operate is about 250Hz. The waves are substantially above ground and the 0V adjustment that should move the bottom above or below the ground line is instead widening or narrowing the wave shape.  INVESTIGATION REVEALED A FEW MISPLACED RESISTORS I found a couple of resistors in the triangle wave generator portion of the board had incorrect values. Rather than change only these and risk discovering other issues in further testing, I decided to double check the values of all resistors placed on the board. I generated a spreadsheet with the values from the schematics, from the bill of materials, and from my visual inspection of the board. This will flag all problem placements so...

EVERYTHING BUT POWER TRANSISTORS INSTALLED ON MAIN BOARD The assembly and checkout instructions for the main board ask me to defer installing the three power transistors and two transformers until after I complete a number of tests on the board, so these five parts are not installed. Everything else is soldered in place and I believed I wasready to start testing.  BACK PANEL COMPLETE AND FRONT PANEL WAITING ON PARTS I need a connector that was on back order and out of stock at all distributors, but I found a compatible one from another maker that I ordered today.  FRONT BOARD COMPONENT INSTALLATION PROCEEDING I installed all but one resistor and all but two surface mount capacitors, then added on some of the through hole parts and connectors. This board mounts the control switches and potentiometers that will emerge through the front panel which is sandwiched right in front.  CHECKOUT OF MAIN BOARD This involved a number of measurements to ensure there are no shorts on ke...

MOST PARTS ARE HERE Four shipments are still outstanding, assuming I don't have to buy any additional parts. It should be only a few more days until the last arrives and I doubt I will be done with the soldering and assembly before then since it is a filler task when taking a break from my main restoration tasks.  WORKAROUND FOR UNOBTAINIUM COMPONENTS Several parts are completely sold out at all the parts distributors with forecast availability out into the summer. For the negative voltage regulators (-5 and -15V) the surface mount (DPAK) versions are the issue, while the more usual through hole parts are more attainable. One of the two regulators was not even available in through hole from distributors but I found a more trustworthy US based eBay seller who offered chips pulled from recycled systems, so I bought from them.  Since the order of the three leads is the same between the two form factors and I have a large ground plane zone where the surface mount part would be sol...

RECEIVED 10 OF 15 SHIPMENTS OF PARTS FROM SUPPLIERS In several cases, the distributor Digikey broke up my order to ship the bulk immediately and then any items that were temporarily unavailable in a slightly later package. One of my Mouser orders is also in the mail. At this point, I know of two additional shipments that should arrive in the next day or two. There is a package from China, bought from Aliexpress, which is festering in Customs in China since it is the midst of Chinese New Years over there - eight days so far for two connectors. That is likely the last package that will arrive; currently projecting Feb 9 to Feb 16th as the date range but it all depends on customs in two countries plus the third-tier delivery service that actually comes to my door with some shipments from China.  A single part has not yet been dispatched by Digikey, thus the date of arrival is unknown, so it is a contender for the last package award. Finally I am building a 'catchup' order with Dig...

There were 346 resistors and capacitors required to build the project, although my purchases were somewhat higher since often it was cheaper to buy 10 of a part rather than the actual count such as 3 or 4 because the per unit price declined sharply.  Semiconductors, which comprised ICs, transistors, diodes and a few LEDs, totaled 92. Most of them, like the other components were surface mount to increase density on the boards.  This design required 29 switches and potentiometers plus another 108 parts from other categories such as connectors, transformers etc. In addition, there are the four PCBs on which almost all the parts are mounted - a front panel, a rear panel, a main board and a board sandwiched behind the front panel. Because of minimum buys at board factories, I received five of each meaning a total of 20 boards. 

SEEKING A CURVE TRACER FOR SEMICONDUCTORS Recently I put together a curve tracer for vacuum tubes based on a design from xxxx yet almost no projects I work on use tubes any more. I am a big believer in the ability of a curve tracer to flag issues with components that simpler testers fail to find. However, the good instruments, even bought used and restored, are very expensive, typically in the thousands to tens of thousands of dollars.  Much as with the tube tracer I built, I wondered if someone had designed and shared a suitably good tracer that I could build myself, saving money while enjoying the act of construction as a side benefit. I located a design, the VBA Curve Tracer Designer's blog about the device , which seemed to be sophisticated and accurate enough to handle all the devices I expected to work with. This includes junction and MOS transistors, plus diodes of all types.  THE COSTS KEPT CLIMBING AS I BOUGHT THE COMPONENTS I sent the four PCBs out to be fabricated a...