HDAWG-PC Real-time Precompensation

HDAWG-PC Real-time Precompensation ensures that the signal at the device under test matches the one designed on the HDAWG Arbitrary Waveform Generator. Through the principle of inverse filtering, this feature minimizes the effect of imperfections in the signal path. At its heart is a widely configurable digital filter applied in real time before the generated waveform is converted into an analog signal. 

HDAWG-PC Key Features

  • Real-time waveform processing for in-situ tunability
  • High-pass compensation
  • Overshoot/undershoot compensation with 8 exponential filters
  • Bounce compensation for reflections and standing waves
  • Programmable FIR filter with convolution length 30 ns
  • Precompensation Simulator
  • Latency calculation
  • Filter reset by AWG sequence instruction

HDAWG-PC Upgrade and Compatibility

  • Option upgradeable in the field
  • Compatible with all other HDAWG options

HDAWG-PC Applications

  • Quantum computing
  • Flux bias pulses for superconducting qubits
  • Gate voltage pulses for spin qubits
  • Electron paramagnetic resonance (EPR)
  • Nuclear magnetic resonance (NMR)
  • Radar & Lidar
  • Semiconductor testing

HDAWG-PC Functional Diagram

HDAWG-PC Description

Typical use cases for the HDAWG-PC are experiments where the sample is located in a cryogenic environment. For these experiments, the wiring from the AWG to the sample, even if meticulously designed, introduces several types of signal distortion. A well-configured precompensation filter gives the system excellent performance, even in the presence of:

  • signal droop caused by high-pass filters in bias tees and DC blocks,
  • overshoots caused by spurious capacitances in chip bonds and planar circuit designs,
  • undershoots caused by spurious inductances,
  • reflections and standing waves due to impedance mismatches,
  • amplifier ringing.

Working in real time, the HDAWG-PC generates corrections even for long signals on the scale of seconds without consuming waveform memory. Real-time precompensation is the only method that tracks history effects over multiple pulses in a dynamically generated pattern, such as used for quantum error correction with fast feedback.

All filter parameters can be optimized based on a step response measurement of the distorted signal at the device under test. The LabOne Precompensation Simulator enables a direct comparison between the simulated AWG output, the theoretical signal shape that this filter would compensate, and the measured signal at the device as uploaded by the user. The user can treat the linear filters independently and eliminate signal imperfections sequentially, starting with the largest one in a well-structured manner. For each filter, the Simulator is used to match the measured signal with the simulation, and the filter parameters found in this way are transferred to the hardware with one click. If a subsequent measurement indicates that fine-tuning is necessary, this is accomplished rapidly by in-situ tuning while observing the measured signal evolve towards a clean, rectangular step response. This results in an optimized filter that is applicable to all signal shapes. The user can proceed to work in the LabOne AWG Sequencer and be confident that the signal at the device matches the designed waveform exactly.

High-Pass Compensation

The high-pass compensation is an infinite impulse response (IIR) filter with a finely tunable time constant integrating the signal over time and thus inverting the effect of AC coupling. To avoid overflow when the input signal has a non-zero average, the high-pass compensation can be reset from the AWG Sequencer between pulse trains. As the precompensated signal grows linearly over time and is ultimately limited by the AWG output voltage, there is a trade-off between the achievable step height of a square pulse and the maximum pulse width.

Exponential Compensation

The exponential compensation adds multiple exponentially decaying terms to a step in the signal. With up to 8 stages available, the user is able to correct for multiple spurious inductances and capacitances in the circuit. Exponential compensation works best for overshoots and undershoots smaller than about 10% of the step height. In this case, a sum of exponential terms is an accurate generic model for such defects.

Bounce Compensation

The bounce correction adds a delayed copy of the input signal to itself and effectively compensates back-and-forth reflections causing interference fringes in the frequency domain or delayed steps in the time domain. The available delay range covers standing wave dimensions up to about 10 m in an RG58 cable.

Finite Impulse Response (FIR) Compensation

For the shortest time constants, down to 1 nanosecond, a fully configurable finite impulse response (FIR) filter allows the user to tweak step edges in the signal to achieve minimum settling times to the sub-percent level. Amplifiers with bandwidths matching that of the HDAWG may show ringing on the nanosecond time scale and are typical use cases. The FIR filter also serves as an extension of the bounce correction in case there are multiple standing waves present in the setup. The FIR filter offers great flexibility and performance at the cost of a more advanced optimization workflow compared to the other filters. It is configured using the LabOne APIs. A tuning routine typically consists of applying FIR models available in LabVIEW, MATLAB, or Python to the measured step response and running a numerical optimization to minimize the deviation from the rectangular target signal.

HDAWG-PC Specification

Number of precompensation units 1 per AWG channel (4 or 8 total depending on HDAWG model)
Compensation filters per unit 1x high-pass, 8x exponential, 1x bounce, 1x FIR
High-pass compensation filter type IIR, configurable time constant
High-pass compensation time constant range 100 ns to 1 ms
Exponential compensation filter type IIR, configurable time constant and amplitude
Exponential compensation time constant range 15 ns to 1 ms (available range depends on amplitude setting)
Bounce compensation filter type FIR, configurable delay and amplitude
Bounce compensation delay range 0 to 100 ns
Bounce compensation delay resolution 1 sample period (417 ps at 2.4 GSa/s)
Bounce compensation amplitude range -1 to +1 (dimensionless)
FIR filter type Causal FIR filter, 72 coefficients
(corresponding to 30 ns at 2.4 GSa/s),
first 8 coefficients freely configurable,
remaining 64 coefficients pairwise equal
FIR filter coefficients range -4 to +4 (dimensionless)
FIR filter coefficients resolution 18 bit
Precompensation Simulator display options Forward filter, inverse filter, uncompensated
step response, measurement data
from CSV file upload

call +41 44 515 04 10
or leave us a message.

To help with your request and to comply with data protection legislation we will need to confirm that you agree for us to collect and use your personal data:

See here for details of the data that we hold on our customers and what we do with it. If you have any questions please contact privacy@zhinst.com