MPS Tx
- 44 Turns (22 each half)
- ~33 mm long each half
- 18 strands of 28 AWG Litz wire
- ~60 mΩ, 8 μH @ 23.8 kHz
The purpose of the transmit system, as the name implies, is to transmit the excitation waveform which in turn magnetizes the particles in such a way that generates the signal of interest. Typically, this is a sinusoidal wave ranging from 1kHz upwards of 70kHz (although 15-35kHz is a more standard range.) Exceptions include Pulsed-excitation MPI (1) but for the purposes of this project, we have limited the scope to sinusoidal excitation.
The excitation (also called the “Tx” or “Drive” coil) is used to excite the SPIONs with an AC magnetic field and thereby enabling the examination of their behavior. In this case, we chose 23.8kHz (1MHz/42) to drive at, to be similar to the drive frequency of the imaging system. The reason 25kHz was not used is that there are many commercial devices which switch at 25kHz, or a harmonic of that so environmental noise may be higher at those harmonics.
The inner diameter was chosen as small as practical to achieve the highest possible field inside. To allow for the sample and sample holder to freely fit within the Rx coil, the ID of the first layer was designed at 20 mm. The homogeneity was a consideration when looking at the pairwise design of the Tx and Rx coils. The drive coil design was not explicitly optimized, rather to improve homogeneity it is clear that turns should be removed from the center. This is because in a simple short solenoid the field is highest at the isocenter. On the other hand, a Helmholtz pair design would be very homogeneous, but not practical due to spatial constraints and efficiency. Using FEMM 4.2, the design was iterated until a reasonable design was found. In the process the efficiency goal was to ensure at least 10 mT, and homogeneity remained within a few percent in the center 5 mm or so. Future work could improve this, but the gains will likely be marginal on both fronts.
The coil former is borosilicate glass (the 3D print material are just spacers that are glued to the glass tube), and was chosen for its thinness, and low coefficient of thermal expansion. As the coil heats up, if the coil former expands it will result in signal drift.
Waveform generation is done with an analog output (AO) channel on the NI-DAQ (DAQ=Data acquisition board/console) more on the DAQ selection. Ideally it would be a differential output, though depending on the DAQ which is used, it may not be an option due to limited AO channels such as with the 6211, and in that case signal generation is done in a 'single-ended' configuration.
We use the TDA3255 amplifier from Texas Instruments. Specifically, the amplifier chip is in a commercial amplifier that was purchased (AIYIMA A07) but substantially modified. The modifications included
- converting it to Parallel bridge tied load (PBTL) mode. This necessitated lifting some of the IC's leads and re-wiring them with jumper wires to convert the amplifier mode, and is only necessary to distribute the current it is outputting into the second on-board channel. With this change, there was no noticeable output performance.
- Removing the on-board low-pass filter, and instead using a band-pass design. This included removing all four output inductors (10uH each) and largest film capacitors (rectangular, 1uF each) and then rewired it as indicated in the filter design section of this page. Note that there is a series capacitor to the ground connection. This is essential to prevent the DC bias of the output from going into ground directly, and without this ground there is a severe reduction in noise performance.
- BNC input connector was added
See [The page for filter design](https://os-mpi.github.io/MIT-MGH_Drive-Filter/).
No signal, especially high-power waveforms, will be pure sinusoids. Besides inherent noise, there is harmonic distortion on the order of 0.1%. Besides the distortions and noise, the other goal of the filtering circuit is to transform the coil's impedance to a value that the amplifier can transfer maximum power into. If the load impedance appears to be too small, the amplifier will be current limited, and it will be voltage limited if the impedance is too high-- the goal is to find the point (or range) where it is equally current and voltage limited.
The filter is designed to be a maximally sharp band-pass filter with an additional stage that transforms the impedance. This topology restricts the use to a single narrow frequency range, but it performs very well. Filters that allow for a wide bandwidth will inevitably have less attenuation at the harmonics, be less efficient, and/or more complex.
We transformer-couple the filter to reduce the common-mode noise and match to the amplifier, and for more details see the page for filter design linked above. In short, by keeping the flux in the core low, it will likely not add any significant distortion.
The topology also allows for inherent stability: without feedback it drifts less than 0.1% over 500 acquisitions
The feedback utilized above is not optimized and causes a slight increase in noise, but it is essentially negligible.
It is probably useful for the reader to see our old designs to motivate some of the decisions. The first observation is that there are two layers with a plastic spacer. This is not an ideal design because the outer layer is far less efficient (because of its proximity to the tube) but also because it essentially traps the heat in the first layer. More turns also means higher voltages which will couple capacitively to the Rx coil (not ideal). The solenoid is unbroken (turns are the entire length), so the turns in the center don't contribute substantially to field.
Old Tx coil:
* (1) Pulsed Excitation in Magnetic Particle Imaging, Z. Wei Tay et al. 2019