
Repairing a dead Tektronix TDS 340A Oscilloscope and making some modifications.
I have been in need of a general purpose two channel oscilloscope since selling my Tektronix 2430A several years ago.
I came across this TDS340A on Ebay with a starting bid of $25.00 USD. Description stated “For Parts and/or not working and is sold strictly in “AS-IS” condition”, it also stated in bold print “Does NOT Power ON”. The shipping was reasonable, and I was the only bidder.
The Tektronix TDS 340A is a dual channel 100 MHz Digital Real-Time™ Oscilloscope with a 8 bit 500 MS/s sample rate. The specifications state that it samples at five times it’s bandwidth. The unit I received also included Option 14, the communications option, which includes GPIB and RS-232‑C interfaces, 9‑pin VGA monitor output, and a Centronics-type parallel printer port. The display utilizes a 6–5/8″ diagonal green monochrome CRT monitor.
I noticed in the sellers photos that it was missing it’s floppy drive, which was not a problem for me as I didn’t plan to be using that function of the scope.
The unit arrived in great shape and was well packed. It had many years of dirt, dust, and grime, along with the usual scrapes, stickers, and sticker residue included. It was surprisingly clean on the inside with a very light coating of fine dust.
The service manual is downloadable from Tektronix, and provides basic troubleshooting information, circuit descriptions, and parts lists. The troubleshooting section almost immediately sent me to the power supply section. There it had me check for 1 to 1.5V on pin 13 of J1 the main power connector. I checked it and was reading 0 volts.
This is the low voltage signal from the standby power section that passes thru the oscilloscope front panel ON/STBY switch to turn on the main power supply section.
The next step said to replace the Low Voltage Power Supply “LVPS”. It was at this point that the Tektronix manual stopped being useful. It included schematics for the rest of the scope, but nothing for the power supply.
The power supply is easy to remove by disconnecting 4 cables and two Torx screws. Once out of the unit I disconnected the three Faston connectors to the power input filter, and 7 Torx screws attaching the circuit board to the metal frame. Upon carefully removing the board I immediately discharged all the high-voltage capacitors before proceeding.

I didn’t see anything suspicious on the front side of the board, but while discharging capacitors I noticed an overheated area on the back side which turned out to be the location of a ceramic power resistor. I performed some basic checks to verify that the DC high voltage was being generated in the rectification section, and everything was as it should be. Next after discharging the capacitors again, I checked the resistance of the power resistor I had noted earlier. It read greater than 20 Meg ohms.
After de-soldering the resistor I noticed that it was supposed to be a 3.9 K ohm resistor rated at 3 watts. A quick check of my higher wattage resistors yielded nothing close to 3.9K ohms, so I decided to get creative.

I built up a 2.5 watt 3.9K resistor using 6 each 33K, and 4 each 56K resistors in parallel. After installing the temporary resistor, reassembling the power supply, and placing it back in the unit, I now had 1.3 volts on pin 13 of J1.


After pressing the front panel power button the fan immediately spun up, and the front panel LED indicators lit-up. Shortly after that I noticed the CRT display with several patterns that changed during the startup self-test. After a few more seconds of relay clicking I was presented with a Tektronix logo and “Power-On self test PASSED”.
It was time to place an order from Mouser Electronics for a proper resistor. I ended up with an Ohmite TWM7J3K9E metal oxide 3.9K ohm 7W 5% Radial resistor.
After the resistor arrived and was replaced I also replaced a couple of electrolytic capacitors, one of which was very close to the failed resistor so may have been subject to higher than normal heat for a extended period of time. After a quick check of several other capacitors and the big bulk rectification capacitor I decided to replace them as well.


Even after a thorough cleaning of the original 120x25mm NMB fan, it was still very noisy, so I decided to replace it with a Noctura NF-F12 PWM 4‑Pin fan I had on hand. I made a small adaptor to go from the locking two pin connector on the power supply to the 4‑pin connector for the fan. Wiring the fan yellow conductor to +12 volts, and the fan black conductor to 0 volts/ground. It made a significant improvement in sound level, but I still would not say it is quiet though.


One issue that I am not quite ready to tackle just yet is the Dallas Semiconductor battery backed SRAM and real-time clock module. It is used to store calibration constants along with date and time. The device contains a small battery that maintains the stored data inside the module, and has a limited lifetime before the non-replaceable battery fails. The module is obsolete and no longer available, but there are several workarounds for it.


As I mentioned earlier the unit was missing it’s floppy drive leaving a large opening in the front panel, it was also missing it’s carrying handle leaving more openings in the side panel. So I decided to 3D print some plugs for those openings.


I 3D printed these on my Prusa Mini printer using PLA filament. I also included a small flat panel to restrict airflow between case sections. The blue colored plug for the handle opening was printed using Printed Solid — Jessie PLA Pure Cyan filament.
Without the floppy drive the full diagnostic test generates a Cpu section FAILED error stating a diagCpu_floppyDrive problem occurred, which I assume means it couldn’t communicate or access the floppy since it was missing.
After a 45 minute warmup I ran the Signal Path Compensation SPC utility from the Cal menu which ran with no issues.
It looks like I can add a working scope to my bench again.


I came across a interesting website while researching Tektronix power supply issues, by ToughDev on using the RS-232 port to retrieve raw signal data. They also included a PC interface utility using a .NET application that allows the user to retrieve the frequency measurements, waveform data as well as taking a screenshot of the oscilloscope.


The next project for this scope is to work on the Dallas RAM/RTC module hopefully before it completely fails.
STL file for 3D print TektronixHandlePlug.stl
STL file for 3D print FloppyPlug3.stl
I purchased two of these on EBay. One works fine but the other had a failed DS1644. I got the upgrade module from an EBay seller and installed it. It works much better than the one with the standard DS1644 so I’ll be upgrading the working one as well. I had a difficult time removing the DS1644 about 3–4 pins wouldn’t desolder so I eventually removed it with a Dremel by sections.
@Chris : Can I ask what part you used for the replacement of the DS1644 ? And while I’m at it : where did you source the new part from ?
I’ve had a look around for similar chips and ebay had listings for a few bucks and mouser had them for ~30$ so I’m wondering if the ebay listings just sell fake parts or not …
Hi Albertino and Chris,
The one I have been looking at on Ebay is this one:
https://www.ebay.com/itm/303821132831?hash=item46bd26941f:g:RJQAAOSwif9f4gjO
I was able to remove my DS1644 module with no issues using a Hakko FR-301 de-soldering gun.
Before purchasing the Ebay unit, I am going to try out a new Maxim DS1230AB instead, which doesn’t have the RTC.
I will update once the new part is programmed and installed.
Greg (Barbouri)
I’m about to try replacing the DS1644 on mine as well. If you have an tips on how to read from the DS1644, they’d be much appreciated! Mine still has the calibration constants and I’m really hoping I don’t lose them. I did pick up a TL866II Plus programmer to try to read it once I get around to de-soldering it.
Hi Andrew,
I read my DS1644 using a TL866II Plus programmer and used the DS1230AB device type, which should be identical except for the RTC.
The address range was 0000 to 7FFF, and the last 7 bytes of the read data had the hour, minutes, seconds, day, date, month, and year.
One unverified tip I’ve seen suggests letting the DS1644 cool down to room temperature after de-soldering, before reading the data.
Good luck,
Greg (Barbouri)
Thanks Greg—It was a success!