Fullscreen HTML5Med Res (??MB)HTML4

Vintage Electronics The Tektronix 2465B Oscilloscope and electrolytic capacitor ageing The 2465B is an analog oscilloscope but with digital supporting infrastructure. Because of this, it has many features of a digital ‘scope, but without any sampling or aliasing concerns. By Dr Hugo Holden I t has calibrated frequency, time and voltage cursors and a memory for the scope’s panel settings. There is also an on-screen digital display, but otherwise, it behaves like an analog oscilloscope. It was rated for a bandwidth of 400MHz, but testing with a levelled sinewave generator and 50W termination shows that it is flat to 400MHz and only 3dB down by about 600-650MHz. Its trigger circuits are so good that you can visualise and lock a 900MHz waveform. Of course, at that frequency, the amplitude calibration is meaningless. I used it to diagnose and repair UHF TV tuners. It can also be configured for four-channel use, which comes in very handy fault-finding logic circuits. The scope is a masterpiece of application-­specific ICs (ASIC). Tektronix called them ‘hybrids’. They optimised every stage and function with dedicated ASICs and other ICs they designed themselves. The main multi-layer board is nothing short of awe-inspiring (Photo 1). Tektronix also designed and manufactured the CRT, a highly complex process. It is a shame that no company in the world now manufactures or 92 Silicon Chip repairs CRTs. This CRT is an extravaganza of precision metallurgy, glasswork and phosphor coating, all created by complex industrial processes. This article focuses on the oscilloscope’s power supply unit (PSU). It is a mixed switching and linear power supply. For a scope made in the late 1980s, the question is: do all the electrolytic capacitors in the power supply need changing now that they are 35 years old? It is an interesting question, especially for a product where the designers sought in the first instance to use the highest-quality parts. The oldest 2465B I have was made in 1989. The last time I powered it on, about a year ago, it was 100% functional. This time, it was totally dead. I performed the usual initial checks and found that power was being applied, and none of the fuses were blown. What could have caused it to fail? A2A1 Board A3 Inverter Board Photo 2: the power supply boards inside the chassis. Australia's electronics magazine siliconchip.com.au Photo 1: this is what you’re greeted with when you first open the case of the 2465B. The PSU is buried inside the scope. Once the chassis is slid out of its outer shell, you are greeted with a top screening cover. With the cover removed, there is some access to the PSU. Two boards are sandwiched together with a series of long goldplated plug-pins that connect the two PCBs. The lower A3 inverter PCB largely processes the mains voltage (Photo 2). There are some line input filter circuit components on the upper A2A1 board, where the power on/off switch, NTC surge suppressor and X2 EMI filter capacitors are located. However, the A2A1 board primarily handles the low-voltage side of things. With the PSU unit mounted in the scope, the access to the A3 inverter board is very poor. An aluminium shield partially covers it too. It is not practical to gain access to most of the PCB’s components initially. Photo 3: the A2A1 board removed from the chassis. Its construction, particularly the components used, changed somewhat over time. siliconchip.com.au Australia's electronics magazine One solution is to remove the whole assembly, attach flying leads to various test points and re-fit the PSU to the scope. In many cases (not all), it is better, if possible, to diagnose the PSU with it connected to its standard loads in the scope. This is the case with most SMPS repairs, unless specific dummy loads are substituted. Another method is extension leads for the supply’s output connectors, if you have them on hand. Notice the green-jacket 100μF/25V electrolytic capacitors in Photos 2 & 3. These are high-quality 105°C Nichicon parts. There are three on the A3 board and five on the A2A1 board. A year or two later, Tektronix moved to 100μF/50V Nichicon types with a brown jacket. Clearly, they had a lot of confidence in these Japanese capacitors. The blue-jacket capacitors (which sometimes have a clear jacket) are American made 180μF/40V and 250μF/20V 105°C types. The two brown-jacket electrolytic capacitors on this board are 10μF/100V parts, which seldom if ever give any trouble. The smaller electrolytic capacitors with black jackets are April 2026  93 Fig.1: part of the 2465B’s power supply circuit. 94 Silicon Chip Australia's electronics magazine siliconchip.com.au Photo 4: note the two large 200V capacitors inside the black insulating box on the right. C1025 1μF/50V bipolar types, which appear very reliable. If they require replacement, I recommend using 1μF/63V MKT capacitors instead. The two small dark-blue-jacket capacitors are 47μF/25V Nichicon types that, in my scopes at least, are still OK. On the right-hand side, where the mains power is initially processed and rectified, is a pair of Rifa 0.068μF X2 capacitors; more on those later. The A3 inverter board shown in Photo 5 is a 1990 vintage specimen, when they had moved to 100μF/50V brown jacket Nichicon capacitors. This board also contains some Rifa Y capacitors. On the right in Photo 4, there two large blue radial capacitors in a black plastic carrier. These are the main filter capacitors on the bridge rectifier outputs. In every 2465B scope I have assessed, these 290μF/200V parts have been perfectly normal, with an ESR of about 0.04-0.06W, no electrical or electrolyte leakage and replacement was not required. Along with other high-voltage electrolytic capacitors in the 2465B scope, these appear, for reasons unknown, to have much better longevity than the lower-voltage-rated electrolytics. It likely relates to the lower ripple currents that higher voltage parts experience. In terms of capacitor failures in the 2465B, the surface-mount electrolytics, if present, fail first on the A5 computer board (due to electrolyte leakage), followed by a similar problem with the 100μF/25V green-jacket Nichicon parts. Some A5 boards were fitted with tantalum capacitors, and while they can fail short-circuit, they don’t usually leak corrosive liquid. The horizontally mounted radial capacitor under the plastic carrier on the right-hand side of Photo 4 is designated C1025 and relates to the power supply’s initial start-up function. This capacitor was stopping my oldest 1989 vintage 2465B scope from powering up (although the one in that scope had a green jacket, rather than blue). The 0.068μF X2 Rifa capacitors had failed on the A2A1 board on this scope in the past, evolving smoke, and had been replaced. This is a common problem because the plastic casings crack and they absorb moisture as they are a metallised paper type. They swell up, opening the cracks further until they become conductive and burn. The partial PSU circuit is shown in Fig.1. The area shaded in blue is what we are concerned about. The incoming mains voltage was normal and the two large filter capacitors charged up. However, the prer­ egulator buck converter based on Photos 5 & 6: a later inverter board, from around 1990 (left). A close-up of some of the troublesome capacitors (right). siliconchip.com.au Australia's electronics magazine April 2026  95 Photo 7: this one had clearly been leaking through its rubber bung. Fig.2: a litmus strip changes colour depending on the pH of the solution it’s soaked in. Comparing it to this chart gives the reading. Q1050 (an IRF820 Mosfet) and the TL494 driver IC was not running. I connected extension wires to the gate and drain of Q1050 and, using a Tektronix 222PS scope (with isolated inputs), I found that there were no gate drive pulses. This buck converter supplies the pre-regulated potential to the primary windings of the main inverter transformer, T1060. The power supply system is moderately elaborate in that the switching drive pulses to Q1050 are modulated in their duty cycle at power-up to give a soft start and avoid current surges. Like many mains-powered switchmode supplies, this circuitry needs a way to get started. In this case, current sourced from the main bridge rectifier flows via 270kW resistor R1020 and a start-up circuit to get the driver IC (U1030) running. Once oscillations are established, the power for the start-up circuit and U1030 is derived instead from pins 7 & 6 of the buck converter’s own transformer, T1020. However, when the power is initially applied, capacitor C1025 (100μF/25V) is charged toward the rectified mains voltage via R1020. The voltage at the base of Q1022 follows at a level determined by the voltage divider composed of R1022 (100kW), R1024 (47kW) and the load provided by IC U1030, which is likely significantly lower than 47kW. This forms about a 1/3 voltage divider. When the voltage across C1025 reaches about 21.5V, Q1022’s base gets to around 6.9V (21.5V ÷ 3). This overcomes the 6.2V zener voltage and Q1022’s base-emitter voltage, and Q1022 switches on, biasing on Q1021, and then both transistors then remain 96 Silicon Chip Photo 8: at a certain angle, a small amount of fluid could be seen under some components. saturated. This effectively places R1024 in parallel with R1022, which reinforces the initial base drive current to Q1022. One job of R1024 appears to be to add some hysteresis to the switch-on function of Q1022 and Q1021. The initial positive voltage supply to the pre-regulator IC U1030 is then established via CR1023. If the pre-regulator IC (U1030) starts and runs, capacitor C1025 is recharged via CR1022 and the buck transformer, and it stays at 13.2V However, after this start-up process, the pre-regulator IC draws current from capacitor C1025 and its terminal voltage drops. If the pre-regulator IC and buck converter circuit didn’t run, for any reason, the voltage across CR1025 diode drops to about 8V. This causes Q1022 and Q1021 to switch off. Under a fault condition, this start cycle repeats. In other words, the start circuit becomes a relaxation oscillator in the event of a failure. My initial tests showed that this was not happening either; there was no activity of any kind in the power-up circuit. The likely culprit was the 100μF/25V electrolytic capacitor, C1025 (see Photo 7). A quick check showed its ESR was a little high compared to a new part. Initially, I had not noticed a couple of telltale signs on the PCB in the area of the start-up circuit. However, while manipulating the PCB at a certain angle to the light, there appeared to be a fluid meniscus under several components below C1025, R1025, R1024 and R1023, along with CR1023, VR1020, Q1021 and Q1022 (see Photo 8). In essence, the whole area shaded mauve in Fig.1 had become a Australia's electronics magazine conductive blanket from leaked electrolyte from C1025. To get Q1022 into conduction, its base voltage has to initially exceed around 6.9V. A leakage with a resistance no higher than 25kW across the 47kW resistor would prevent that. It was either that, or the leaked electrolyte was shunting current from the base of Q1022 to ground. In addition to leakage, the electrolyte had caused component lead corrosion. Leaking electrolyte from the base of capacitor C1025 was easy to see after it was removed for inspection. Despite this, the capacitor measured normally, at close to 100μF on my capacitance meter. The rubber bung in the base was softened, swollen and electrically conductive. I tore the corner off a piece of A4 paper to soak up the fluid under the components. It was yellow and a quick test with my meter probes indicated it was quite electrically conductive. The resistance measured in the order of 100kW across a small section of the soaked paper. Inside the capacitor, the electrical leakage effect of the electrolyte is greatly reduced by the fact that one of the foils is covered in aluminium oxide, which is an insulator. The other 100μF/25V green Nichicon capacitors I had removed, on testing with 20V applied via a 560kW resistor for 15 minutes, had a leakage of only 1.5μA, corresponding to a leakage resistance of about 13.3MW. Further investigations I tested the pH of electrolyte from inside another of the green 100μF/25V Nichicon capacitors and it had a pH siliconchip.com.au very close to 6-7. This is similar to other new capacitors I have tested; I see 7-8 with some brands, so there is variation in electrolyte formulations. I then tested the paper soaked in the leaked electrolyte. It was quite alkaline, with a pH around 9 (see Fig.2). I also put a sample of the A4 paper I had used in another bag and it was neutral (pH = 7). Not only is an alkaline solution corrosive, it is much more electrically conductive than a neutral solution, explaining the relatively low resistance I measured. I presume this is due to the electrolyte sitting on the PCB for a while, in contact with lead, tin (solder) and copper (leads, PCB tracks). This is not unexpected because, when metals are dissolved by weak acids, the result is an alkaline solution. When an acid and a metal react, the metal gives electrons to the H+ protons to form hydrogen gas. The oxidised metal (now positively charged) combines with the acid’s negatively charged anions to form a salt. Most soluble salts derived from weak acids form alkaline solutions. This is because the anions in the salt accept H+ protons from water. This leaves hydroxide ions (OH−) in the water. For example, a lead borate solution has a pH of 8.6 and a tin borate solution a similar value. There was little copper corrosion yet, in this case, but copper borate has a pH of about 9. These may seem like small differences from a neutral pH of 7 but remember that it’s a logarithmic scale; if you add or subtract one from the pH value, you are changing the ion concentration by a factor of 10! So a solution with a pH of 9 has 100 times as many OH− ions available as a neutral solution. Pure water (pH = 7) has the lowest electrical conductivity compared to alkaline (pH > 7) or acidic (pH < 7) solutions. As a solution becomes more acidic below pH = 7, it becomes more electrically conductive because of the higher number of aqueous H+ protons. Similarly, as it becomes more basic above pH = 7, there are more hydroxide OH− anions, again making it more conductive. Manufacturers of electrolyte solutions generally have tried to keep the pH of the electrolyte as close to neutral as possible, although most are a little acidic. Ageing effects inside the capacitors, especially where H+ has siliconchip.com.au reacted with the aluminium to evolve hydrogen gas, result in a shift toward a higher pH, so the electrolyte becomes more alkaline. Loss of hydrogen by way of gas evolution is obviously bad for the capacitor’s chemistry; a domed top is a sign of it. If the electrolyte leaks out of a capacitor, there are four main concerns: 1. The electrical effects of the electrolyte on the circuit. 2. Short- and long-term damage to components. 3. Short- and long-term damage to the PCB. 4. How to safely remove the electrolyte and avoid further failures. Electrical effects on circuits In this case, the circuitry involved was relatively ‘high resistance’ in that the resistor primarily dependent for raising the base voltage of the transistor Q1022 has a value of 100kW and the source resistance charging the capacitor is also high at 270kW. However, low-resistance circuitry, below say 10kW, could have electrolyte leaked all over it with possibly no apparent fault until the resulting corrosion becomes severe. In many ways, the fact this start-up circuit failed relatively early after the electrolyte leaked was a blessing, because significant corrosion damage was yet to occur. It must have been relatively recent leakage because most of it was still wet. Damage to components This must be considered in a cleanup operation. Corrosion can occur where tin-plated copper leads enter a resistor’s body, which is often made of ceramic with a metallised coating. As this continues, it expands, increasing its physical volume. This can result in the component failing at a later date. In extreme cases, the expansion can crack the entire resistor or component body. Electrolyte leaked from capacitors can also eat through the conductive films on surface-mount resistors, rendering them open-circuit. In my case, corrosion had already entered the ends of the resistor bodies. Although the 3kW, 1.2kW and 47kW resistors tested OK, I replaced them to be safe. The 100kW resistor also had one leg affected. I also replaced both diodes, the zener with a 1N4735A and the plain diode with a 1N4148 (see Photo 9). PCB damage Unfortunately, the PCB’s conformal coating (typically green) is not a total barrier to a contaminated electrolyte and its corrosive effects. The coating breaks down after a period of exposure to the electrolyte, and the copper under it beginning to corrode. With voltages applied to copper tracks, the copper corrosion is accelerated by electrolysis, and fine tracks can be eaten completely away. If you find tracks that are fully corroded through, likely the electrolyte leak occurred many months beforehand and the instrument remained powered after that for a considerable time. After the electrolyte has been in contact with solder for a while, the Mild conformal coating and track damage Photo 9: the leaked electrolyte had already corroded some tracks and component leads. Australia's electronics magazine April 2026  97 solder loses its shiny metallic surface and acquires a grey oxide like coating. The coating is a thermal insulator and can sometimes make the component difficult to desolder unless its surface is scraped down and fresh solder is added. PCB cleanup methods While a PCB can be cleaned with contact cleaner, this does not help the green conformal coating where the electrolyte has absorbed into the full thickness, with moisture and ionic species filling microscopic voids in the coating. If two adjacent tracks are disconnected from any components and the coating between them has previously been in contact with electrolyte, testing will show electrical leakage between the tracks. This is evident even after the surface of the coating has been thoroughly cleaned with contact cleaner. While the coating might look normal, it is no longer an electrical insulator. One remedy some have tried is to remove the coating by scraping it off or dissolving it with methylene chloride, but this ruins the appearance of the board. Also, methylene chloride is toxic and difficult to get in some localities, and restricted for public use. I prefer to use a leaching method to remove the electrolyte but it requires some patience. It involves letting a thin stream of warm-to-hot water run over the affected area of the board for at least a half an hour (ideally an hour). The retained ions migrate from the coating into the water and are washed away. If deionised water is available, it is superior to tap water for this. After that, standard contact cleaners (IPA etc) can be used to clean the water off the board. Ideally, the stream of water runs off the nearest corner of the board, with the board held a 45° angle. The whole board is not dunked in water. Although that can leach out ions, there are components than can absorb water and they will be very difficult to dry out. You could damage them that way. Some people have put PCBs in dishwashers to clean them, but it can damage parts, especially items such as trimcaps, some transformers, DIP switches, IC sockets etc. Thus, I never do it. In this case, because the plastic carrier was screwed to the PCB in the area being washed, I had to release the nuts from the carrier, to lift it away from the board surface a little, or water could have become trapped in that area around the stud’s threads. Component replacement I removed all five 100μF/25V green Nichicon capacitors from the A2A1 board and the three from the A3 board for inspection and testing. Some of the capacitors had visible electrolyte leakage. Others had conductive bungs, as revealed by a DVM on its ohms range. This is something not all technicians are aware of. If an electrolytic capacitor has leaked electrolyte in the past, it renders the surface of the rubber bung in the capacitor’s base electrically conductive, which is easily picked up with a DVM. The removed green-jacket 100μF/ 25V Nichicon capacitors all had higher ESR values than a range of new parts with similar ratings, and two of them had an ESR higher than the worst-case figure of 0.5W suggested by the ESR meter’s guidelines. The damage on the A2A1 board indicated that one capacitor had probably been leaking for longer than the one that caused the failure preventing the scope from powering up. There was damage to the board’s conformal coating, and the electrolyte had started to attack the copper traces. Also, where the electrolyte had dried out, there were white crystalline deposits. Failure or degradation of the rubber seal is one of the reasons why electrolytic capacitors leak. The leakage can also be encouraged by hydrogen gas evolution, pressurising the contents, and in many cases, doming the top of the capacitor. However, for these particular Nichicon capacitors, all their tops were perfectly flat. Thus I think they are failing due to drying out. In another instrument, a 1000μF capacitor dried out completely and failed. There was no evidence of any electrolyte leakage; it had lost nearly all capacitance and went to a very high ESR. I opened it up for inspection and found that it was as dry as parchment paper inside. As an experiment, I placed it in a container of deionised water for a few hours. It returned to a normal capacitance value and a normal ESR. It appears that the seals can partially fail to the extent that water vapour can escape in some cases, but not fluid. Comparison to another scope Photo 10: a close-up of some of the corroded tracks (circled in red). The solder mask helps, but it doesn’t stop the damage! I stripped down another PSU unit from a low-power-on-hours Tektronix scope for examination. This time, the green Nichicon capacitors had a date code of 8930. Their rubber bungs were in good order, without softening, and they were not electrically conductive. Their ESRs were a little above the normal range compared to new parts tested, but within the 0.5W guideline, and there was no significant electrical leakage. So they were probably OK. Likely, in the next five years or so, they will also leak and damage components and the PCB, so I elected to replace them anyway. So apart from the date of manufacture, the amount of running time is Australia's electronics magazine siliconchip.com.au 98 Silicon Chip Photo 11: the repaired inverter board, after I replaced all the troublesome capacitors with new ones from Nichicon. the other main factor that determines when the capacitor spills out its electrolyte. Indicators of a failed electrolytic capacitor include: 1. Visible electrolyte around the capacitor or corrosion of tracks and adjacent components. Loss of a metallic shine on nearby solder. 2. Damage to the conformal coating and tracks directly under the capacitor. 3. Visible fluid leakage on capacitor’s rubber bung. 4. The rubber bung has become conductive. 5. Softening or disintegration of the rubber bung. 6. ESR above the normal range for similar new parts. 7. If a capacitor of exactly same type has leaked elsewhere. 8. A very old device or a unit with long running hours Less reliable indicators are: 1. Measured capacitance outside of the normal range. 2. High electrical leakage. 3. Capacitor has a domed top. Returning to the 2465B I decided that the other capacitors on the A3 board should also be replaced. The manufacturer had attempted to ‘leak proof’ them by gluing resin over the rubber bungs. This appeared to have worked, except that in one case, some electrolyte had passed through the bung and around the sides of the leads as they exited through the section of resin. For that capacitor, again the ESR was a little on the high side compared to new parts. I removed the red-brown siliconchip.com.au resin from one of the blue capacitors to inspect the rubber bung and test its electrolyte. On the capacitance meter, the 250μF 20V part read 330μF, or abut 1.32 times its marked value. The 180μF capacitor also measured about 1.45 times its marked value. I opened one for pH testing and found it had an alkaline electrolyte, with a pH of 8. Interestingly, an increase in capacitance can be a marker of increased hydroxides in the capacitor. I performed an electrical leakage test on one of the 250μF/20V parts and found that its leakage current was low, at less than 2μA with 20V applied after 30 minutes, which is acceptable. However, a new part’s leakage current tested at 0.2μA, an order of magnitude lower. Rather than buying different values, I decided it would be reasonable to replace all of these with new 330μF/50V 125°C-rated Nichicon BT series capacitors, which have a rated ESR of 0.02W. These are similar to milspec parts. They can be recognised by their pale blue jackets. I replaced the original 100μF/25V parts with the 100μF/50V capacitors, as Tektronix did in their later model 2465B scopes. Photo 11 shows one board recapped with the new capacitors, including replacement ceramic Y-type capacitors. Should any electrolytic caps be left unchanged on the A2A1 or A3 boards? There are a few 10μF high-­ voltage electrolytic capacitors on these boards. The main filter caps in these scopes don’t appear to have any Australia's electronics magazine problems in the four scopes I own. For now, I have left these ones in place for observation. There are also some small electrolytic capacitors elsewhere in the PSU. They are elevated a little off the PCB on their leads and are easy to inspect and not prone to physical leaking or other failure modes, yet. To inspect these, apart from ESR testing, look closely at the solder on their tracks. The X2 & Y capacitor dilemma The 35+ year old Rifa capacitors should always be replaced because their outer plastic casings crack. They absorb moisture and swell up widening the cracks. The positive feedback continues until the X2 capacitors become electrically conductive, heat up and burn, evolving copious smoke and making a mess on the PCB. The internet is awash with stories about smoking Rifa X2 capacitors. When they were new, they were good performers. 30 years down the line, though, trouble can start. It may simply be that they were not designed for long service. So I don’t judge the Rifa parts too harshly. I previously replaced the two 68nF Rifa X2 capacitors in the mains voltage input area on the A2A1 board on all my scopes. It is better to move away from a metallised paper film product and use plastic film X2-rated parts. I fitted Wima MKP (polypropylene film) or other plastic film 100nF types instead. However, there are three other Rifa capacitors on the A3 board that now have surface cracking and swelling in April 2026  99 Photo 12: the rear of the Tektronix 2465B oscilloscope. connector. You can generally trust the X and Y capacitors inside that unit; being sealed in a metal enclosure, there is no risk of smoke or fire. In summary, for replacing the X2 capacitor, I prefer Wima X2-class film parts, and for the Y-class, capacitors I use Y2-class (labelled) ceramic types, which have similar proportions to 3-5kV rated ceramic capacitors. Battery-backed SRAM all of my 2465B scopes. Two are 2.2nF Y-class capacitors (C1020 & C1051). The usage in the 2465B is to bypass both the positive and negative outputs of the bridge rectifier to Earth. The customary use for Y-class capacitors is to bypass the incoming Active and Neutral AC lines to Earth; however, the application in the 2465B similarly relies on them not shorting out. Hence the use of Y-class capacitors. There is also a 10nF capacitor, C1052, that couples the negative side of the bridge rectifier output to an electrostatic screen behind the power switching Mosfets on the A3 board. I found visible horizontal cracks in the bodies of the two 2.2nF Y-class capacitors. The 10nF capacitor’s body was starting to swell up on one side, too. Generally, X2-class capacitors are designed for applications directly across Active & Neutral, while Y-class capacitors are designed to connect from Active or Neutral to Earth. Both types are often used to aid in the suppression of high-frequency interference either entering or exiting the instrument via the mains wires. Often, they are combined with inductors to improve the filtering. Before the Rifa-style metallised paper film ‘safety capacitors’ were invented, many manufacturers used waxed paper, oil-filled or ceramic types for Y-class capacitors. They got around the reliability problems and mitigated the risk of failure by using capacitors with a substantially higher voltage ratings than were required, and seldom had any troubles. Some products were encased in metal housings to mitigate the fire risk. 100 Silicon Chip The Y-class capacitor must be able to support sustained voltages over 1kV. Some manufacturers specify a 4kV DC rating for a Y-class capacitor to give a wider safety margin. This is because, on occasion, high-voltage transients can ride on the Active line. So capacitor failure can be made less likely by increasing the insulation withstand voltage. Tektronix also added some gas-­discharge voltage arrestors in the mains power input circuitry. They act as a negative resistance and a voltage clamp once they activate. In any event, the X2- and Y-class capacitors in the 2465B’s power supply should be replaced, and they need to be suitably rated X and Y parts for the task. Ceramic capacitors generally don’t burn much, except for their outer coating; they are a minimal fuel source compared to a plastic part. I prefer them for this reason. Y-class ceramic capacitors usually have a flame-proof coating and are designed to fail open-circuit. X2 capacitors frequently fail short-circuit, which is why they burn up. Fortunately, in the 2465B, the mains input is protected by fusing prior to the Y- and X2-class capacitors. Tektronix were also clever with the X2 capacitors, in that not only were they placed after the fuse, but they added small low-value resistors in series with them. If the capacitor shorts, the high current vaporises the resistor if the fuse does not blow immediately. That happened in one of my scopes when the X2 capacitor went low-­ resistance. Tektronix relied on a Japanese-made metal-cased commercial line power filter as part of the panel-mount IEC Australia's electronics magazine The PSU’s electrolytic capacitors determine the speed that most of the voltage rails collapse when the scope is switched off. The 2465B uses a Dallas DS1225 battery-backed non-­volatile SRAM with an internal lithium battery to store the scope’s calibration data and control settings. The DS1225 incorporates either the DS1210 or DS1218 control IC. When the 5V power rail drops below a specific level, this chip disables the SRAM and prevents any writes that could corrupt its contents. It works extremely well; I have been unable to corrupt the SRAM’s contents even by switching it on and off rapidly. I previously replaced the DS1225 with Ramtron FM16W08 FRAM because the DS1225’s battery was flat. This worked very well, and many people did this later with very little trouble. However, I noticed that power cycling could occasionally alter the FRAM contents. Fortunately, it did not affect the calibration constants, as those addresses are not active at the time of power cycling, but did affect the last panel control settings. In one case, I was able to ameliorate it with a 330W resistor from the WE line to +5V. Additional information I have written many other articles about repairing different sections of the 2465B oscilloscope. A list of them can be found below: • siliconchip.au/link/ac7b • siliconchip.au/link/ac7c • siliconchip.au/link/ac7d • siliconchip.au/link/ac7e • siliconchip.au/link/ac7f • siliconchip.au/link/ac7g SC siliconchip.com.au