Jump to content

Mr.RC-Cam

Admin
  • Content Count

    5,407
  • Joined

  • Last visited

Everything posted by Mr.RC-Cam

  1. Rather then sit idle waiting for the Arduino CAN-Bus shields to arrive, I shifted gears and got out some woodworking tools. The instrument cluster will be a bookcase display piece. It'll have a hollow oak base to hide the electronic parts. The instrument cluster will get a plywood "dashboard" that will either be wrapped in vinyl carbon fiber film or painted black. Just getting started. Oak base frame and plywood dashboard bezel. Oak base is assembled. 1/2" threaded rod risers with 3D printed brackets. Vintage '66 Mustang badge emblems will be installed on the base. DiY 3D printed speaker enclosure. There will be two speakers for the V8 motor audio. More details are expected soon. So drop by in a few days to learn more about the CAN2Cluster project.
  2. The clock struck 2019 last night. So I'm starting our brand new year with another project. And of course that means I'm going to write another workbench adventure story. Hang on to your horse, I think it's going to be a fun ride. Despite the title to this blog, I don't have a Ford Mustang. But I had a '66 Mustang when I was a teenager. She was a tired pony with 130K miles when I drove off the used car lot. Nice looking, ran well, and had a sweet 8-track player. Dependable too, but I occasionally had to get my hands greasy. Like the time the transmission blew on the way to work (quickly fixed with a $75 junkyard pull). Back then I enjoyed the troubleshooting challenges while working on everyone's cars. I rebuilt engines and did just about any auto repair that came up. In those days working on cars was easy; A determined guy with basic tools and a floor jack could fix anything. Being able to fix clunkers was an advantage to a cash-strapped kid. It allowed me to purchase a variety of used vehicles to play with. I even had some that were air cooled; That's VW Bug and Chevy Corvair territory. Besides a couple modified Beetles, I had a Greenbrier 6-door van and two Monzas (sedan & coupe). Then there's that time I built a "kit" car. Not really a kit, but essentially a pile of fiberglass body parts that were assembled by guesswork and ingenuity. Running gear was decided by the builder. Mine ended up on a shortened VW Squareback pan´╗┐ (Type 3 chassis) with a bored out 1500cc engine. The body's T-top styling looked good but the stock suspension made it an uncomfortable ride. See photo below´╗┐. Then the 1980's arrived and working on American cars became a hideous chore. Horrible mechanical reliability, bad paint, fragile plastic parts, and ungodly smog systems that ruined engine performance. At this point I was busy working in the electronics industry and my interests morphed from motors to microprocessors. And higher wages meant I could finally buy better cars. My last car project. Dropped a Accord motor into a '77 Civic. Easy swap. Fast forward a few decades. Now I'm an occasional weekend mechanic that drives a 20+ year old Ford SUV. Bought it new (special factory order) and over the years it has been a reliable friend. Other than DiY oil changes and some minor repairs, I really haven't needed to bust my knuckles. And I like that, because my love affair with grease ended long ago. Why would I tell you all this? Well, it's the introduction to the story on how a Mustang instrument cluster became the centerpiece to my latest electronic project. At the heart of the story is a grown up kid that liked tinkering with cars, lost his passion, then gets reunited by happenstance. So if you are interested in where this is going then hang around the electronic campfire while the story unfolds. The ending is not written yet, it's a real-time adventure.
  3. After two late night sessions there has been more progress. The ArbIDs to the Fuel and Oil gauges have been identified. Plus those that control the door status and tire pressure monitor. I should mention that these functions are not accessible from the HS CAN-Bus. Instead, they use the companion MS CAN-Bus for communication. Unfortunately none of the online posted Ford or Mazda CAN-Bus hacker information helped find them. It was a painful exercise of manual keyboard entry and patiently watching the cluster's reaction (if any). There's over 2000 possible ArbIDs and their data payloads have some bit field dependencies that enable/disable other functions. The proverbial Needle in a Haystack. The Fuel gauge was particularly fussy to identify since the control data has weird split scaling and a unexpected control field. To make things more cruel, it would secretly go dormant for 35 seconds on some written data values. I pulled out a lot of hair while deciphering the magic data that it wanted. But I won the battle and can now fill up the tank. The Oil Pressure gauge is interesting too. It's an idiot light in the disguise of a linear display. A single bit controls it, so it only has one active position. This makes sense since the engine's oil pressure sensor is just a brainless on/off switch. Here's the ArbIDs to the newly discovered MS CAN-Bus (MS-CAN) items: 0x3A5: Tire Pressure Monitor (TPM) 0x3B1: Door Status 0x400: Fuel Gauge 0x445: Oil Pressure idiot Gauge Without a real Ford Mustang's data to sniff, my technique was brute force. I used the Microchip CAN-Bus analyzer and manually entered experimental ArbIDs and 8 byte data payloads. The data fields would be populated with test bytes (0x00, 0xFF, 0xAA, or 0x55). Then I would watch the cluster for several seconds to see if anything changed. This went on for about 15 hours, with occasional bathroom breaks. Below is a screenshot of the Microchip tool's GUI. The four ArbIDs have been configured for 1/2 tank of gas, valid oil pressure, all doors closed, and good tire pressure. There's a handful of remaining ArbID's to find. Such as: Dashboard Backlight Intensity Turn Signals (Indicator) Handbrake (Indicator) Parking Brake On warning (message center) Fuel Level Low warning (message center) DTE Data ODO Data Error (message center) Brake Fluid Level Low warning (message center) Check Charging System (message center) I'm a bit worn out from the manual ArbID mining. Going forward I think it's best to wait for the Arduino CAN-Bus shield to arrive. I can use it with some custom code that will assist me with the search for the remaining ArbID's. But the board is coming from China and I don't expect to see it for a couple weeks. In the meantime I'll build a nice looking stand for the Mustang instrument cluster. Woodworking tools won't get me greasy, so I'm good.
  4. After several hours of experimental CAN-Bus data injection I have identified a few more ArbIDs. Besides rpm and speed, I can now actuate the temperature gauge. Several indicator lights are under my control too, as follows: Overdrive Off (orange), Overheat (red), Check Engine (orange), Charge System Fault (red), Powertrain Fault (orange), Cruise Control Enabled (green), Security Enabled (red). These items are handled by two HS CAN-Bus ArbIDs (0x201 & 0x420). As follows: ArbID 0x201, with 8 byte payload The packet takes the form: [RR, rr, 00, 00, SS, ss, 00, 00] Where RRrr is the tachometer rpm and SSss is the Speed mph. The following formulas are used: rpm = 0.25 * (RRrr) - 24 Speed (mph) = 0.0065 * (SSss) - 67 Byte 0 & 1 = Tachometer rpm. See formula above. Byte 2 & 3 = Unknown. Set to zero. Byte 4 & 5 = Speed. See formula above. Byte 6 & 7 = Unknown. Set to zero. ArbID 0x420, with 8 byte payload Byte 0, Temperature Gauge: 0x55 = LOWEST Temp, 0 line 0x7F = Middle Temp 0xA0 = High Temp (top mark) 0xA1 = Max Temp (red line) with Red warning symbol (Below Tach) Byte 1, Unknown. Byte 2, Unknown. Byte 3, Unknown. Byte 4, Indicators and Temp Gauge Override: Bit 0, Unknown. Bit 1, Unknown. Bit 2, 1 = Orange O/D OFF Indicator (Below Tachometer). Bit 3, 1 = Orange O/D OFF Indicator Blinking (Bit D2 must be zero). Bit 4, 1 = Force Max Temperature (red-line gauge), no warning indicator. Bit 5, 1 = Force Max Temperature (red-line gauge), no warning indicator. Bit 6, 1 = Orange Check Engine (Below Tach). Bit 7, 1 = Orange Check Engine Blinking (Bit D6 must be zero). Byte 5, Indicators: Bit 0, Unknown. Bit 1, Unknown. Bit 2, Unknown. Bit 3, 1= Red Charge System Fault Indicator (Below Tach) Bit 4, Unknown. Bit 5, Unknown. Bit 6, Unknown. Bit 7, 1= Orange Power Train Fault Indicator (Near Tach's minimum mark). Power cycle reset! Byte 6, Indicators: Bit 0, Unknown. Bit 1, Unknown. Bit 2, Unknown. Bit 3, 1= Green Cruise Control Indicator (below Temp Gauge) Bit 4, 1= Red Security Indicator, (Below Tach). Bit 5 must be zero. Bit 5, 1= Flashing Security Indicator (Below Tach). Bit 4 must be zero. Bit 6, Unknown. Bit 7, Unknown. Byte 7, Unknown. I haven't been able to find the magic bits for the fuel and oil pressure gauges. Or for an assortment of indicators such as the hand brake, turn signals, and fluid levels. To control them I suspect I'll need a second CAN-Bus interface connected to the cluster's MS CAN-bus port. But I haven't given up on finding a way to do it through the HS CAN-Bus.
  5. CAN-Bus is working and I can control the tachometer and speed gauges. It's been a good day. There's something satisfying about driving 80mph while sitting at my desk. And I have photos to prove I'm exceeding the office speed limit. 80 mph @ 4400 rpm. Oops, left the parking brake on. Despite some detours the project is moving along nicely. But there's a long ways to go. It's time to shift into forth gear and find out how to control everything else in the dashboard cluster. After I conquer all those mystery bits I can get to building what I have in mind. It's going to be a working bookcase display with race engine sound playback and a real key-start ignition. Parts are on order.
  6. The Mustang Instrument Cluster has arrived. It came from a wrecked car (eBay based recycler). There's some scuff marks but overall it looks good. Every part from a wrecking yard has a story. The cluster was tagged with the VIN and a online search provides a wealth of information. It was a 2009 Mustang 4.0L V6 coupe with 60K miles that had gotten sandwiched in a roadway skirmish. If by chance you were the owner of this pony when it took its last breath then please accept my condolences. And I hope everyone involved in the accident is OK. The 2009 Instrument cluster came from a pretty pony that has been put out to pasture. One of the reasons this particular cluster was chosen was because the seller's photos showed it included the harness connector plug (chopped off, with several inches of wire). I even email them before the purchase and they confirmed the connector would be included. I'm happy with the condition of the cluster I received, but the connector is missing. /* Guy shakes head in frustration. */ Without the connector plug I'm at a disadvantage. Besides the obvious reasons for wanting a factory made plug, the colored wires on it would have helped me determine the pin numbering (connector orientation). It's best to roll with the punches, so after a few minutes with an ohmmeter I have what I need. Connector orientation was determined by identifying the ground pins and matching them to a wiring list found online. Rear panel photo: Connector Pin numbering (plug orientation). Some push-on jumper wires and a variable power supply provides good news. The cluster powers up and the six gauges successfully perform their needle calibration dance. Its voltmeter gauge is centered with the supply voltage set to 12.0V, but the other gauges are zero (as expected). All the indicators light up and backlighting is working too. The warning buzzer chirps as well. The donor organ is alive. There's a pair of wires (Pins 6 & 7) that need a 3-button Message Center switch. The switch is expensive so it's DiY time. A couple hours of experimentation determines that the switch is a resistor ladder (voltage divider) for the cluster's analog input on pin 6. The resistor values are as follows: Info: 100 Ohms Setup: 1.5K Ohms Reset: 470 Ohms Default: 100K Ohms DiY Message Center Switch, 3D Printed Bezel Green = Info, Black = Setup, Red = Reset The DiY Message Center switch is wired to the cluster and working great. I have to admit I'm not thrilled by how the push-buttons feel when they are pressed. Maybe I'll swap them out with higher quality switches that have better action. But regardless, there's more to do before it can be installed on a bezel. The CAN-Bus is the next item to test. But rather than stuff more jumper wires on the naked connector I've decided to step back and solve the missing plug dilemma. Instead of complaining to the seller about the forgotten plug I decided to build my own. The cluster's connector is a 26-Pin (2 x 13) 2.54mm header. I have a female header that is 40-Pins and a close encounter with a sharp saw turns it into a suitable doppleganger. A test fit determines that the plug is difficult to install/remove in the cluster's deep opening. So I designed a 3D printed housing shell to make it an easier effort. The housing is keyed to prevent backwards plug installation. {A few hours later} All wires are soldered and safely potted in the 3D printed housing with hot melt adhesive. The DiY Cluster plug is complete. Instrument Cluster's Power/Signal plug is a 26-pin header modified to fit. See text. There's two cables terminated to the plug. One has twisted pairs for the CAN-Bus (HS CAN and MS CAN), plus some miscellaneous signals. The other has power and all the remaining signals. I won't use them all, but just in case they're all accessible. It's time to connect the Microchip CAN-Bus analyzer.
  7. The Microchip CAN-Bus Analyzer dongle arrived. It's always a pleasure to be visited by the big brown truck. I ordered the CAN-Bus tool directly from Microchip's official online store to ensure it was the latest release. Wishful thinking at best; In the box was a CD containing old V2.0 software (2011 release). So the CAN-Bus dongle hardware is probably aged like a fine wine too. I read that V2.0 wasn't compatible with Windows 10. So I downloaded the V2.3 (2016 release) upgrade from the Microchip web site and installed it. The app launched and I was able to setup the CAN-Bus tool. I currently don't have any Can-Bus devices to test out, but that doesn't stop me from investigating the tool's GUI. Wait a minute, the GUI's features look like someone half-hardheartedly phoned it in. It has basic functionality, but it's missing important features mentioned in the manual. For example, I can't save my tool configuration settings and there's no Trace Filter. And it's buggy too. For example, moving a trace window creates graphic ghosts that require a program exit to clear up. I found a possible explanation to these problems. The help screen reports that the PC Software is a BETA V2.2 released in 2014. That's not a good sign because the installed version is supposed to be V2.3 released in 2016. Dooh! /* Guy shakes head in frustration and mumbles a variety of curse words to himself. */ But there's another possible reason for the missing functions. The manual states when upgrading the PC software it is necessary to re-flash the tool's two PIC18F microprocessor chips using the provided hex files. The GUI's reported firmware version numbers match the two hex file names found in the downloaded software. It appears the dongle's chips are already the latest version. But I don't trust this Microchip tool, so out comes my MPLAB ICD3 ISP Debugger/programmer. With a 12V supply connected to the CAN-Bus tool I successfully flashed both PIC chips with the provided hex files. But the firmware flashing was a wasted effort. The GUI's reported firmware version information is the same as before and the missing features are still missing. Pardon me while I take a break and try to sort this out. There's closure to this mess. I found the Release Notes file in the upgrade distribution. It confirms I have the latest PC GUI software (V2.2). It also mentions that those missing features I wanted were dropped years ago in the V2.0 release. So some functionality is now vaporware and any bugs that appear are here to stay. I've concluded that this tool is an abandoned Microchip orphan. /* More mumbling & cursing. */ If this were a product review it would be generous to give Microchip's tool a one star rating. Hopefully my opinion gets upgraded after I've done some basic CAN-Bus experiments with the Mustang instrument cluster.
  8. After a long night of web mining I'm now confident I'll be able to control the Mustang's Tachometer (RPM) and Speedometer. But everything else is still a mystery. What I unearthed is courtesy of like-minded hackers that used reverse engineering tricks. That is to say, they connected CAN-Bus sniffers to their cars and drove around while logging the real-time data. Back at home, they reviewed the timestamped activity while looking for clues on which data packets involved the dashboard. The reverse engineering task is best described as finding a needle in a haystack. But choosing the right tools eliminates a lot of drudgery. I'd like to use the sniff method too, but I don't own a late model Ford with CAN-Bus. The car in my garage was built years before this networking protocol was a Ford staple. Fortunately I found the CAN-Bus ArbIDs to the Ford Tach / Speedometer. See this sourceforge wiki: https://sourceforge.net/p/ecu/wiki/canbus/ But the search for controlling Ford's indicator lights and gauges has hit a dead end. However, during the online investigations I noticed that Mazda partially shares some of Ford's CAN-Bus arbitration codes (these two car makers have been exchanging technology). I expanded my search and found OpenGarage's list of Mazda CAN-Bus ArbID descriptors and I plan to try them: http://opengarages.org/index.php/Mazda_CAN_ID And to make things a bit more interesting, the Mustang Instrument Cluster I selected has two separate CAN-Bus ports. The HS-CAN (High Speed CAN, 500kbps) port is for the engine and drivetrain data. It provides the RPM, speed, coolant temperature, and similar data. The MS-CAN (Medium Speed CAN, 125kbps) is for ancillary information, such as fuel level, oil pressure, seat belts, parking brake, doors, etc. To fully control the cluster I'll probably need to send data to both ports. But I'm hoping that MS-CAN data is allowed to tunnel into the HS-CAN bus so that only one Arduino CAN-Bus shield is needed. //I ordered a second shield in case my wishful thinking is cruelly trampled by reality.
  9. A quick search on eBay finds a nice looking 2002-2004 Mustang cluster. But this one is not a good choice for what I have in mind. It uses Ford's obsolete SCP (Standard Corporate Protocol) bus for communication. The fuel / oil pressure gauges are hard wired to their senders and many of the indicator lamps are also hard wired. And its two harness connectors have a couple dozen wires (Click for Pin-out). This means the Arduino will need a lot of digital I/O pins to control everything. All told, I'm not a fan of this choice. The next eBay prize that caught my attention was a 2012 Mustang cluster. This version uses CAN-Bus for communication with all indicator lamps and gauges appearing on its serial bus. But a key coded PATS Anti-theft module (not included) is directly connected to the cluster. I don't know if this missing module will disable some of the cluster's features. Keep looking. 2011-2014 Mustang Cluster Wiring Diagram But like Goldilocks, I came across a third pick that seems to be just right. It's a 2007-2009 series Mustang Cluster that uses CAN-Bus communication. But on this model year the PATS anti-theft module does not directly connect to the cluster. Maybe this is a good sign it will be less trouble to interface. 2005-2009 Mustang Cluster Wiring Diagram So a gut feeling tells me this is the one to hack. This 2007-2009 Mustang instrument cluster is on its way to me, thanks to eBay. Next I ordered the CAN-Bus shield for an Arduino MEGA2560 R3 that's been sitting on my workbench. See image below. I also ordered a Microchip APGDT002 CAN-Bus Analyzer Tool (see image below). This USB connected dongle should provide a convenient way to experiment with automobile CAN-Bus protocols. I could have bought a similar CAN-Bus tool from a eBay seller at a fraction of the cost. But I've lost confidence in the functionality of cheap China sourced tools. So I spent a bit more to avoid unexpected surprises. Keep in mind that I can't use a common OBD diagnostic scanner for the things I want to do. That tool is a beast with another purpose. That is to say, I'm not going to read OBD PIDs to troubleshoot a Check Engine light. Instead I need to get close and personal with the cluster's back panel CAN-Bus signals and read/write some mystery data. While all this hardware makes its way to me I'll continue my search for the instrument cluster's CAN-Bus Arbitration ID (ArbID) list and payload parameters. I understand that these Ford secrets are not officially released to the public. But I suspect I will find useful details with my friend Mr. Google.
  10. Every project deserves a name. This instrument cluster hack has been officially badged the CAN2Cluster Project. The name includes a reference to the CAN-Bus interface, something we will soon discuss. First on the to-do list is to research the communication signals used to control a modern vehicle's instrument cluster. Given the number of car brands out there I expect to find an assortment of proprietary protocols. This is where you interrupt me to explain that there's a OBD diagnostic port (big connector under the dash) that is interfaced to the Engine Control Unit (ECU). And it will be able to provide every little bit of data that my project would need using a communication protocol that is well documented. You end your lecture by saying that controlling the instrument cluster is going to be Easy peasy lemon squeezy. Hmm. I like what you're saying but I don't expect it will be that easy. My project wont have an ECU, so the OBD interface is non-existent. Beyond that, it's true that some automakers provide access through the OBD connector to directly control the instrument cluster and vehicle actuators. But this special bi-directional functionality is not designated by the OBD standards (and not all cars support it). So any protocol information gleaned from the OBD documents is not going to be helpful to my project. Typical OBD port. It's usually located under the dashboard on the driver side. OK, I just returned from researching deeper into this topic with the assistance of Mr. Google. As expected, simple it's not. I've learned that there are several automobile network protocols that are brand dependent. Some are for diagnostics and some are for inter-module communication. Some do both. I don't know if it's a complete list, but the protocol wiki will help paint a clearer picture: Click Me. I'll ignore the network/protocol choices for now. My immediate goal is to find a late model instrument cluster with gauges and warning indicators that are controlled by the fewest number of wires. A dream solution is a cluster that can control **everything** from a dedicated serial bus. //Somewhat confident that this probably exists, a typical thing a tech guy tells himself before reality strikes. A convenient place to start is to pick a instrument cluster for the project. All the YouTube hacker videos I watched only showed tachometer and speedometer operation using Mazda and VW dashboards. Those car brands don't interest me. I've decided that my Arduino interface will need to control a late model Ford Mustang dashboard. Perhaps a working instrument cluster, displayed on my office bookcase, will comfort me with memories of the good times I had driving my old pony. And since auto junkyards are full of dead Mustangs, getting a donor instrument cluster will be the least of my problems.
  11. Hang in there, I'm nearly finished with the backstory to my Mustang Cluster project. We'll get to the technical stuff in a moment. A couple weeks ago I noticed that my old SUV's instrument panel had a burned out light bulb; Half the speedometer was dark. In the old days I would reach under the dash and change the bad bulb. But easy access to anything in a car is a thing of the past. So I did the typical DiY guy thing and went to YouTube. I quickly learned that I have to pull out the instrument cluster. A few screws and some hidden fasteners. Nothing I can't handle. But then I got trapped in a YouTube rabbit hole. For hours I binge watched instrument cluster repair videos. Blown FETs, failed needle driving stepper motors, bad LED upgrades, and so much more. Fixing these problems is interesting to watch, at least to me. Then I tripped across some DiY hacker videos that show Arduino boards and magic software controlling automobile instrument clusters. Instantly I was hooked and felt the sudden urge to know more about modern dashboard systems. The video above shows an Arduino controlling a dashboard cluster. Too bad the project links are dead. Keep in mind that I grew up with mechanical cable driven speedometers, hot-wire ammeters, and brainless hardwired warning lights. Of course I've known that modern technology has infiltrated the dashboard, but until now I just didn't care. Intrigued, my adventure begins.
  12. Mr.RC-Cam

    3D Printed DJI Inspired Quadcopter

    The drone market has changed a lot in the last four years. Complex/expensive builds like this are no longer popular. The project discussed here uses the stock front bearing holder. I don't have an stl for a mod'd version. But if one exists, it is probably posted at RC Groups.
  13. I purchased some low cost plans ($7 USD) for a DiY 3D printable Quadcopter that was inspired by the DJI Inspire. It has a motorized arm transformation (retract) feature too. Plans are available here: http://www.rchobbysuk.co.uk/blogs/dji-inspired-3d-printable EDIT MAY-02-2016: The rchobbysuk site is now dead (404 error). But revised community designed plans are now available for free: http://www.rcgroups.com/forums/showthread.php?t=2399740 EDIT May-11-2015: Custom 3D printed parts I created are available at no charge (but you'll need the file set from rchobbysuk). My downloads start here: http://www.rc-cam.com/forum/index.php?/topic/4022-3d-printed-dji-inspired-quadcopter/?p=28111 A very lively discussion on user builds is found here: http://www.rcgroups.com/forums/showthread.php?t=2327044 Just so that you can see what I'm trying to duplicate, here's a CAD rendered image from the rchobbysuk web site: Here's the status of my printed parts (after 150+ hours of printing): EDIT / May-05-2015: It Flies! See post #32 for short video. Here's a beauty shot of the finished model.
  14. CAN-Bus hardware/software discussions are allowed in this forum area. Off-topic discussions not permitted. The Ford Mustang dashboard project has its own sub-forum. For details please go here: https://www.rc-cam.com/forum/index.php?/forum/30-ford-mustang-instrument-cluster-area/
  15. Mr.RC-Cam

    Tektronix 2245A Oscilloscope, SMPS repair & re-cap

    My scope has her official Kardashian moment; The photos of her naked chassis are going viral on the internet. Thanks to hackaday. https://hackaday.com/2018/12/08/the-guts-of-switched-mode-power-supplies-brought-to-you-by-oscilloscope-repair/
  16. I own three different o-scopes. My favorite daily rider is a Tektronix 2245A, circa 1990. I've had it about 20 years (purchased used) and have treated it like a little princess. Aesthetically it looks remarkably good and has maintained its measurement accuracy over the years. It has been a reliable friend indeed. But a few days ago I powered it up and she was dead. No lights, no screen trace, only a brief hiss sound at the moment of power-on. After a 911 panic moment I collected my thoughts and mentally prepared to troubleshoot the power supply. Here's an image of the Tektronix 2245A scope (photo courtesy of a random web search): I'll share all the details in this blog of the scope's repair journey. If fixing old test equipment is something that interests you then hang around the virtual campfire and listen to my story as it unfolds. Hopefully this ride-along workbench adventure has a happy ending.
  17. Mr.RC-Cam

    Tektronix 2245A Oscilloscope, SMPS repair & re-cap

    Here's a beauty shot showing the repaired TEK 2245A measuring the front panel's Probe Adjust test signal. This is the end of the campfire story telling. It was a bumpy E Ticket ride with a splash landing. But everyone survived. Hopefully all your repair journeys are successes too.
  18. Mr.RC-Cam

    Tektronix 2245A Oscilloscope, SMPS repair & re-cap

    During disassembly all the hardware was organized to help identify where it belongs. So re-assembly was fast and easy. It's too soon to install the outer case because the supply voltages need to be checked. All voltages are measured on the power supply's main connector that is accessible on the bottom of the scope. With a bit of trepidation, the scope was powered up. Good News: It's working! Bad News: It booted into the test menu splash screen. As feared, the dead battery has caused the loss of the stored calibration constants data. But all DC output voltages are within specification. This is a good time to adjust the 7.5VDC rail (POT access is on the side of the chassis) and recheck all voltages. Now it's time to address the nagging test menu screen. Fortunately loss of Calibration data is not as serious as I thought. The scope's built-in self calibration feature is easy to use and it stores new calibration constants. Self Cal Measurements Instructions (excerpt from page 1-6 in the Operators Manual): The calibration measurements function takes a couple minutes to run. There's some comforting status messages during the test too. So sit back, relax, and let it run. After the test I pressed the CH4 button to exit and return to normal operation. Some typical workbench measurements confirm everything is back to normal. The scope is repaired and it feels great to see her alive again.
  19. Mr.RC-Cam

    Tektronix 2245A Oscilloscope, SMPS repair & re-cap

    The repaired supply is back on the isolation transformer. The brief hiss noise wasn't present when power was first applied, only silence. But the voltmeter provides much joy; The supply is running. All DC output voltages are about 15% too high. But this supply's regulation is reliant on a proper load, so I'm not too concerned. As a sanity test a 15 ohm resistor was added to the +5V rail and the loaded output's voltage dropped from ~6.8V to 5.5V. Still too high, but the voltage will be further reduced when all the outputs are properly loaded. After confirming the supply is working I shut it down, did the capacitor discharge bleeding chore, and re-installed the original MC34060P Vreg IC. Another test and everything is still OK, so the original IC will stay in the circuit. Success, such a lovely thing. I'm on the home stretch.
  20. Mr.RC-Cam

    Tektronix 2245A Oscilloscope, SMPS repair & re-cap

    The two 39uF caps in the DC output filter needed minor lead bending. Some electronic-safe RTV was applied to hold them upright. The other output filter caps were an exact fit. The old 1000uF 50V pre-regulator filter cap (C2203) has a 3-lead pinout. So the much smaller replacement needs its two leads to be formed to fit the PCB. A 3D printed (ABS plastic) base mount helps support the new capacitor. The photo below shows the newly installed C2203 capacitor. The bent leads under the cap prevent it from resting flat on the PCB. The 3D printed base is a nice solution since it acts as a rigid standoff. Plus it eliminates the need for a big blob of adhesive to secure the elevated cap. The old 270uF 450V mains filter cap (C2202) has four leads. The replacement has two leads so it needs some rework to make it fit. A 3D printed base helps with the installation. The new C2202 is installed: BTW, here's the STL file for the two 3D printed plastic base mounts: TEK2245A_Cap_Base1.stl The remaining parts went in without any tricks. The rebuilt power supply is ready for testing.
  21. Mr.RC-Cam

    Tektronix 2245A Oscilloscope, SMPS repair & re-cap

    The replacement parts have arrived. It's always a pleasure to be visited by the big brown truck. As mentioned, this power supply repair includes a full electrolytic recapping. Keep in mind that I didn't find any bad caps. But this is a 30 year old scope and for long term reliability it seems like a good idea to replace the SMPS's aging capacitors. Plus it's the perfect time to do it. This is the graveyard of parts to be replaced: Here's the mouser.com parts list to re-cap the power supply (electrolytics only): 3 pcs 4.7uF 100V, 661-EKYA101E4R7ME11D 2 pcs 39uF 160V (150V), 647-UCY2C390MPD1TD 2 pcs 100uF 63V Non-Polar, 647-UEP1J101MHD 1 pcs 270uF 450V, 661-ELHS451VSN271MR3 4 pcs 470uF 25V, 667-EEU-FM1E471B 1 pcs 1000uF 50V, 667-EEU-FR1H102B 8 pcs 1000uF 25V (12V), 661-EKZH250EC310225S The shorted diode is a 1N4936, 863-1N4936G All of the replacement capacitors are smaller than the originals. Some have incompatible footprints (different lead spacing), so I'm prepared to get creative to make them fit. Here's a comparison of the size difference on some of the caps:
  22. Mr.RC-Cam

    Tektronix 2245A Oscilloscope, SMPS repair & re-cap

    Parts are on their way. While I wait for their arrival I visually inspected all the PC boards. And this good deed punished me with more bad news; The 3V lithium battery is dead. It measures 0.3V, so I'm afraid that the stored Calibration Constants data has been corrupted. I won't know for sure until the power supply is repaired. The battery is a Panasonic BR-2/3A that is mounted on the bottom of the Processor board. It looks like this: Before I removed the bad battery I temporarily bridged a 3V coin cell into the circuit on the top side of the Processor board. If the cal data is still intact then the coin cell will continue to retain the data until after the replacement battery is installed. Fortunately the industrial BR-2/3A battery is not hard to find. It is also a kissing cousin to the popular CR123A cell (non-rechargeable version), which I already have on hand. The CR123A battery is missing the through-hole mounting leads that are needed for PCB installation. I have a battery welder, so it's not a problem to replicate the leads on the new battery to mimic the BR-2/3A cell. Here's the welder: It is a DiY MOT (microwave oven transformer) monster that I use to build 18650 battery packs for ebikes and FPV equipment. It makes nice welds that are much better than those provided by the "attractively priced" Chinese machines. I duplicated the leads seen on both ends of the old BR-2/3A using 8mm wide nickel strip. Then welded them onto the new CR123A cell. The photo below shows the old and new cells. The replacement battery fits like the original.
  23. Mr.RC-Cam

    Tektronix 2245A Oscilloscope, SMPS repair & re-cap

    Shotgun troubleshooting begins. I decided to replace U2201, a MC34060P switchmode modulator IC. Long out of production, I was lucky enough to have a spare (date code 1986). Rather than direct soldering the replacement IC, it was installed using a Augat machine pin IC socket. Unfortunately the new IC did not fix the supply. BTW, if you need a replacement MC34060P then the NTE1753 is a direct substitute. It too is out of production but gives you a second chance in your hunt for this rare part. Using a DMM (Digital Multi Meter) to check components in-circuit is not a reliable way to confirm they are good. But I have good success if the parts are removed before measuring. It's a slow and tedious process, but remarkably effective in the right hands. All the output caps were removed from the board using a de-soldering pump tool. Their capacitance and ESR measurements confirmed everything was OK. Too bad, since I was hoping to find a bad one. But since this is a repair AND refurbish project, I'll be replacing all the power supply's electrolytic caps. With the forest of caps out of the way I had easy access to a long row of diodes. One lead on each diode was unsoldered, lifted, and the diode checked (via DMM's semiconductor test function). This provided some "good" news; CR2213 is shorted. Why would I say this is "good" news? Because the shorted diode is the smoking gun I was hoping to find. It is a failure that would trigger the pre-regulator's over-current protection. Replacement parts (diodes, caps) are on order and will be here in a few days. I'm optimistic that they will bring life to this dead scope. Come back soon to hear the rest of this campfire story.
  24. Mr.RC-Cam

    Tektronix 2245A Oscilloscope, SMPS repair & re-cap

    During my bench testing I discovered that the power supply briefly runs for a couple hundred milliseconds. Fortunately this is enough time to confirm the DC outputs have the correct voltages. But because the supply immediately shuts down the troubleshooting period is short lived. The brief hiss noise that I hear is due to an event called the "chirp mode." The faint noise originates from the magnetic component vibrations that occur when the supply starts up, fails, and shuts off. The chirp mode is designed to repeat every few seconds (continuous shutdown & restart), but my supply only does it one time whenever it is powered-up. I don't hear it again until I remove power, discharge the caps, and then try again. Time to study the schematic. There's a pre-regulator switching supply that bootstraps the output switcher. It is given about 100mS to begin running using the mains voltage that is stored on a 100uF cap (C2204). If the pre-regulator is successful in starting up, it generates an output voltage (43V @ 2A) that is routed back to the 100uF cap, which provides the ongoing voltage to power itself. This voltage is also supplied to the DC output's Inverter Power Switcher that does the heavy lifting (it generates the various low and high voltages that power all the scope's circuitry). Here's the pre-regulator circuit. So I learned that at power-up there's a split second situation where the power supply must run *before* it can run. A proverbial catch-22. I found it frustrating to troubleshoot given that only a fraction of a second is available to observe the signals before the supply shuts down. I was able to confirm that the pre-regulator's switching transformer (T2203) is working during the startup period. It is providing about 60VAC to the pre-regulation rectifier (CR2202). So that helpful information suggests to me that the shutdown problem is probably caused by a failure that is triggering the protection circuits. There's a current sensing circuit that will disable the pre-regulator if current draw is excessive. I used a Fluke peak reading multimeter and measured the current sense resistor's (R2201) voltage drop. It reported 0.072VDC max, which is an indication of safe currents. Spoiler Alert: I will soon discover this to be an incorrect/misleading observation. There's also a over-voltage crowbar that shuts down the supply if the pre-regulation voltage climbs too high. I disabled the protection (remove Q2206) and sadly the problem persisted. The final protection circuit is a thermal sensor switch (S2202) that is mounted to the pre-regulator's MOSFET driver (Q2201) heatsink. It shuts off the supply if the transistor gets too hot (Over-Temp protection). The ohmmeter reported the thermal protection switch is OK. For further confirmation its electrical contacts were temporarily bridged with a jumper wire, followed by a power on test. No improvement. So it seems I'm at the end of the rope. Time to use the shotgun approach. That is to say, check/measure all the components one-by-one to find the bad apples.
  25. Mr.RC-Cam

    Tektronix 2245A Oscilloscope, SMPS repair & re-cap

    With the power supply on the workbench it's time to start the troubleshooting. This always begins with a thorough visual inspection. I saw no bulging capacitors, no overheat discoloration, and found perfect solder joints. Other than a tiny bit of dust, the board looks like new. That's both good and bad news for me. Now for a gratuitous public service announcement: Don't apply power to the SMPS board without an isolation transformer. That's because traditional switching power supplies expose dangerous/deadly mains voltage. The isolation transformer is not the holy grail, but provides some basic protection from the earth referenced voltages. NEVER use any AC powered measuring equipment without the isolation transformer. Here's a photo of the 400W isolation transformer that I am using: The SMPS board has low voltages (5VDC-58VDC) mixed with very high voltages (130V, 3KV, 14KV) that can persist for several days after power is removed. So even with power disconnected, manually discharging the output capacitors before handling the board is a mandatory chore. While power is applied it's important to work slowly and methodically. I usually keep one hand in a pocket as a reminder to be safe and to help avoid fatal electrical shock / injury as I troubleshoot. Let me be clear, it's nerve-wracking to work on high voltage circuits. The currents can vaporize tools and kill you too.
×