AWG2021 troubleshooting - REPAIRED

[gianrisa]

Hi everyone,

I’m currently working on troubleshooting an AWG2021 that’s exhibiting a peculiar issue with its waveform output - a persistent jump of 223 mV — suggesting a possible issue with the DAC.

To get to the bottom of this, I need to take a closer look at the AWG2021's internal Memory to DAC 12-bit bus. ( There is a connector present on the A21 board that connects to the option 4 )

I believe that the Option 4 board, which is the ECL to TTL converter inside the AWG2021, could give me the insight I need.
However, I don’t have this board at hand.

I’m reaching out to the community to see if anyone can share a clear, detailed photo of the Option 4 ECL to TTL converter board from an AWG2021.
This would be incredibly helpful for my diagnostics and repair process.

If you have the board and are willing to help, please post a photo.

Thanks in advance for any help you can provide!

[gianrisa]

It has been an interesting repair for whom is interested here below the logic process I've followed.

At the beginning, I had only the service manual, which included a high-level block diagram.
The instrument wasn't built for low-level repairs, but rather for high-level repairs, meaning circuit board-level replacement.

Starting from the block diagram, I identified the signal path and noticed that from the output to the source, there were a few
possible blocks that could have originated the fault. It's important to remember that the fault was a jump in the waveform at the negative offset.

The first block in the signal path is the power amplifier and attenuator, the second is the DAC and the calibrator, the third block is the waveform
memory and the control circuit, and the fourth one is the CPU controller. I followed the signal through the amplifier attenuator to the DAC output
and discovered that the signal was already distorted there. Obviously, the analog stage was working correctly.

The culprit, then, came from the digital path. After analyzing the circuit and researching online, I discovered that the DAC is a 12-bit, full-custom unit,
 very similar to a Maxim DAC but with a specific pinout. Continuing my research, I found that Tektronix also uses this DAC in other instruments.
Fortunately, for those other instruments, schematics were available, and I was able to determine the chip pinout.

Starting from the pinout, I was able to identify the digital inputs and locate the voltage reference. I noticed that the AWG2021 has an option
that exposes the digital output of the DAC to the user. Following that circuit, the outputs are protected by ECL drivers, which could have offered
the possibility of connecting a Logic Analyzer to capture the digital input to the DAC. With this setup, I could have plotted, with the help of a small Python piece of code,
the bitstream captured from the logic analyzer to verify whether the data coming into the DAC were correct or not.

Observing the AWG2021's output waveform, it was noticeable that the output frequency was correct, meaning that the PLL was working correctly.
Testing arbitrary waveforms in the lower part of the offset revealed the jump in amplitude in the waveform.

Address corruption in the memory scan was to be excluded, otherwise, it would have represented multiple faults in the waveform at other offsets as well.

The logic analyzer test and the bitstream capture and reproduction confirmed that the DAC had a fault. This type of digital-to-analog converter has
R-2R logic switches and resistor banks, and one of them might have been shorted, resulting in a voltage jump during the conversion.

Replacing the DAC solved the problem.