The completed instrument.
This blog will cover the construction of a standalone version of the DIY Tektronix PS503A Dual Tracking Power Supply.
The explanation, prototyping and design details are in another Blog that can be found here:
https://www.paulvdiyblogs.net/2025/12/diy-ps503a-rebuild-prototype.html
This is a picture of the original instrument that occupied a slot in a TM5XX mainframe:
The details of this original instrument can be found here : https://w140.com/tekwiki/wiki/PS503A
Design process
I use KiCad for my design, schematic capture, part and footprint selection and layout processes, and since I switched to KiCad, I never looked back. It's such a joy to use (most of the time (;-)).
PCBWay created a KiCad plugin that can be invoked directly from the layout screen, and produces all the required files for their manufacturing process. There are actually two plugins, one that saves the files in your local KiCad workspace and another plugin sends the data to their website and opens it with the information loaded and ready to go with the ordering process. Simple!
The first version of the PCB that I designed resulted in a working system, but needed some changes and additions and was the basis for the Version 2 that was send for production at my sponsor PCBWAY.
Here is a 3D view of the V2 PCB in KiCad.
After I got the boards back from manufacturing, they looked as good as always when I inspected them. I always check for solder mask issues and if holes are drilled in the exact middle of the pads. This is where the quality of the process typically breaks down, and it makes applying solder paste and reflow soldering of tiny parts so much easier.
Especially the black solder mask color they use in combination with the sharp white silkscreen creates very nice front panels for my instruments.
I really like working with the folks from PCBWay, they are very friendly and responsive. They also provide a complete manufacturing service by which they obtain the parts and populate the boards for you. This is where the rubber hits the road, for me as well, because when I'm not precise in my part selection, footprint or availability, they contact me to resolve the issue. They do a great job in user satisfaction, and I have never had a complaint from another Maker for the PCB quality or the manufacturing option. They do their best to resolve issues with me, so their customer get's the finished product without getting involved in the part procurement and manufacturing process.
Populating the V2 PCB
Because I had a working unit, th parts were transferred from the V1 board to the V2 board, and the additional parts added.
I then used my Lab PS to supply set to + and - 32V volts with 100mA max and connected the leads to the sine wave connections of the two bridge rectifiers. When I turned the Lab supplies on, there was an initial current limit because of the reservoir capacitors filling up, but after that, there was no indication of a short on the rails. I verified with a DMM that all rail voltages were present.
The next step was to connect the mains transformer and give it another go, turning it on, and quickly measuring the key vitals. All rails and the reference voltage where there and no smoke.
Next step was to connect the power transistors and the front panel with all the connectors to get a fully functioning supply. When turning it on, there was no smoke and all controls seemed to work as they should.
Initial calibration
I used my DMM and calibrated both output voltages to 22.00V at the maximum level.
Voltage rail measurements
Without a load, the +33V rail was +32.51V, the +27V rail was +26.32V, the +6V4 rail was +6.43 and the reference was +10.000V.
The negative rails : -33V was -32.48V, the -27V was -26.28V and the -6V4 was -6.48V. All very close and good.
Testing the current limits
With the current limit setting at 300mA, I used my Dynamic Load to see where the current limit kicked in, and that was 330mA for both supplies. At the 600mA setting, the current limit kicked in at 650mA for both supplies, with the intended little headroom that I created.
Stress testing the voltage rails
Applying the full 600mA load with my DL, I measured the rails again.
+33V became +23.67V and the +27V became +22.88V.
-33V became -23.48V and the -27V became -22.70V.
This is according to the original design although I would have liked a little bit more on the +/-27V rails feeding the Opamps. This is the result of selecting a transformer with 24VAC outputs, instead of the 25VAC that the TM5XX mainframes provide.
The heat sink stayed cool, I felt no warmth at all.
Calibration of the output voltages
Use a DMM and calibrate both output voltages to 21.00V at the maximum level to have a little head-room with the maximum current as a load.
Verify the lowest voltage output
The lowest voltage you can adjust to should be positive, and not negative, otherwise it could trip the tracking voltage protection. The minimum voltage should be around 100mV. If it deviates significantly, you need to change the value of R58 for the positive supply, or R34 on the negative supply. These values can be different and are also determined by the voltage setting potmeter tolerance.
Verifying the lowest current limit setting
With no load, verify that when you turn the current limit setting all the way CCW, to the minimum current, the red LED could come on to show the limiting, but does not need to be. Use a Dynamic Load or a power resistor to draw about 10, 20 or 30 mA and verify that the Current limit LED does come on at this level. You can change the value of R50 for the positive supply or R34 for the negative supply to either raise or lower the trigger point where the current limiting starts.
Verify the half way point of the current limit
Set an output of 10V and turn the current limit fully CW to maximum current. Set the Current limit to 300mA. Using a Dynamic Load, or a power resistor to create a current of 150mA. Turn the current limit potmeter slowly CCW until the red LED turns on, signaling the trigger point. This should be about half way of the potmeter travel.
Verifying the fluctuations of the output voltage
With no load, and with the output voltage set for 20.0076V, I used my 6.5 digit DMM in the trend mode to look at the fluctuations over a longer period. They were between 20.0080 and 20.0072V, a mere 800uV difference.
Verifying the voltage reference
Below is the reference voltage after turning the power on. As you can see it will take some time for it to fully stabilize, but also note the voltage level. It only drops 800uV during the whole warm-up process.
Output stability with maximum load
Under a 600mA load, and an output voltage set to 20.0076V, the output voltage dropped 11.3mV and the fluctuations over a longer period were now between 19.9967 and 19.9961V, a mere 600uV.
The screen shot below was taken with an output voltage of 10V and the full 600mA load. The output voltage only varied about 20uV.
Testing the over-voltage protection.
The easy way is to use a Lab supply connected to the positive (or negative) outputs. Set the Lab supply to 20V with a maximum current of 50mA to prevent issues. Set the PS output to 20V and switch on the Lab supply. Slowly turn that voltage up until the output of the PS as shown on the display will go down. The trigger point will be at about 26V. Note that the output voltage does not go all the way to 0V, this is due to the Lab supply back-feeding. When you turn that output off, the PS will go down to about 100mV or lower.
To reset the PS supply, you need to turn the main power off and wait a few seconds until all rails have gone to zero and all LED's are off. Turning the unit back on will reset everything. If the PS does not come back to 20V, turn it off again and wait a little longer.
Now do the same with the negative supply, but note the polarity please.
Testing the tracking over-voltage protection
The easy way is to use the same method as above, but in this case, use only a 5V output setting and apply a voltage just below 5V from the Lab supply. Turn it on and slowly increase the voltage. The tracking over-voltage protection will kick-in at 5.1V or close to it. This is due to the fact that you raise both the output voltage but you are also back-feeding the injected voltage into the voltage regulation. They both go up which is why this verification only works reliably below 7V, above it this verification will not work.
To completely verify the tripping voltage above the set voltage for the whole output range you need to open the enclosure and get your solder iron out. You need to remove R57 for the positive supply or R61 for the negative supply and apply the Lab supply voltage through a 1K series resistor to the resistor pad that is connected to the first diode (D1 for the Pos or D2 for the Neg). I suggest you use a tiny wire, like a wire-wrap wire, to solder to the SMD pad, and solder the other end of the wire to a 1K THT resistor to avoid stress on the PCB pad.
With the tree diodes in series, in addition to the voltage drop over the 1K series resistor, you will need to apply about 1.7V above the set voltage to trigger the over-voltage protection.
The 1.7V delta is the intended and normal voltage I selected for the tracking over-voltage protection. If you want to lower the trigger point, bridge one of the tree diodes in series with a piece of wire or use a 0R 1206 SMD resistor instead.
As an example for this extended test, set the output to 20.00V, and apply the Lab supply voltage to the 1K resistor. You will need to apply about 21.7V to trigger the tracking over-voltage protection. To reset the supply, you need to turn the main power off and wait a few seconds until all rails have gone to zero and all LED's are off. Turning the unit back on will reset everything. If not, you were impatient, so wait a little longer.
Note
Normally, there is serious fault somewhere to trigger either over-voltage protection, so it's wise to first remove your DUT and trouble shoot the supply before you use it again with a DUT.
Verifying the "dual tracking" operation
The easiest to see if it works correctly and how it works, set the positive supply to +18V and the negative supply to -9V. Turn the dual tracking potmeter fully CW and activate it. The voltages should stay the same. Now adjust the positive supply from +18V to +9V (-50%) and verify that the negative supply is now at -4.5V (-50%).
Set both supplies to +/- 20V and the dual tracking fully CW and activated. Using the dual tracking potmeter, reduce the voltage of say the positive supply to +15V and verify that the negative supply is also at -15V or within a few mV. Turn the supply further down to +10V and verify the negative supply. Do that for +5V and +1V as well.
Turning the PS on and off through the mains switch
Turning the power on with the mains switch and the output on. (not recommended)
Turning the power off with the mains switch and the output on (also not recommended)
Turning the outputs on and off (recommended)
Turning the output on
Turning the output off
Thermal tests
I used my IR-probe to look for hot spots. There are none that concern me at all.
An appropriate stress test is to use a lower output voltage, like 3V3 and then use the 600mA load. In this case the output transistor needed the heatsink, but did not get above 54 degrees C.
The square in the lower left is the small transformer for the display. In the upper right below the tracking voltage protection switch is the positive diode bridge.
Passed with flying colors...
Output voltage noise
The noise level is only about twice that of my DSO, without any hum artifacts. The ambient noise in my office is too high to really show anything that comes from the PS. In other words, it's clean.
Final assembly
During this whole process, I had the unit open, with the transformer next to it.
After checking everything, it was time to mount the transformer to the top shell of the enclosure and close it up for some more testing. The transformer is literally turned upside down and located in the space just south of the two large capacitors, and away from the cooling slits in the back of the top cover.
Final words...
So far, I'm very happy with the results. It looks like I will have a very nice additional power supply for the bench, and that's where it is located now.
Github project repository
