|
|
|
New CODD Pick-up Amplifier |
The head assembly is the part that collects the PU plate signals and combines them into sum (Σ) and horizontal and vertical displacement signals (X & Y). It also allows the application of test signals in order to calibrate the electronics chain. For details, see the schematic diagram.
 |
| Fig.1: Head assembly picture |
 |
| Fig.2: Head assembly test setup |
 |
| Fig.3: Head assembly frequency response T->Σ |
 |
| Fig.4: Idem, HA inserted into PU |
 |
| Fig.5: Wire to Σ frequency response |
 |
| Fig.6: Antenna calibration jig |
 |
| Fig.7: Σ channel S11 |
 |
| Fig.8: X channel S11 |
 |
| Fig.9: Y channel S11 |
Some measurements on the new head assembly (Fig.1) (Yves' prototype, October 2005). The 'gold' PU showed asymmetries between the electrodes of up to 4pF, as measured using the 'KARU 510' capacitance bridge. After some soul-searching, I tried to readjust the electrodes to balance using the capacitive plungers. One of the plungers is inaccessible, so it had to serve as the reference. While the symmetry is now within 0.2pF, that wasn't good enough to get consistent results. I decided to adjust the head assembly without inserting it into a PU.
The frequency response of the head assembly through the test input is smooth as long as it isn't inserted into a PU (Fig.3). However, when inserted into a PU electrode set, a resonance appears (Fig.4). I didn't verify, but I believe this to be the leakage inductance of the hybrid transformers resonating with the electrode capacitance. It isn't visible in Fig.3 because its frequency is beyond 200MHz. Fortunately, this resonance is only associated with signals applied through the T inputs, and when a signal is applied through the wire in a coaxial setup, (the Gillieron table), the plot is smooth (Fig.5).
The head assembly is tested and adjusted using the HP3577A Network Analyser (NA) (Fig.2), with HP35677A test set. Settings: CW, 10MHz, 15dBm, RBW 100Hz, average 64. Normalise with a short piece of coax between the ports. Then, using a 50/75Ω pad and a 75Ω cable connected to one, then the other, of the Test inputs of the head assembly, adjust X and Y outputs to -35.00dB. Terminate unused outputs! Measure the Σ output: It should be -31.23dB, so that X/Σ and Y/Σ are both -3.77dB. (Composed of -3.01dB inherent in the functionality of a hybrid, plus about 0.75dB of loss due to the sum hybrid.) With the head assembly inserted into a PU, the sum signal drops to just -35dB.
Now insert the head assembly into the PU. Use the antenna calibration jig (Fig.6). Use the HP3577A without the test set. Input A/R. Normalise using a splitter from Out to R and A. Use a symmetric splitter! Connect Out to the head assembly Test input using a 50/75Ω pad and a 75Ω cable. Use identical length 50Ω cables to connect Σ and either of X or Y to R and A respectively. A/R should be -3.77dB.
Disconnect Test input. Connect Out to antenna in positions +25 and -25mm and measure A/R:
| displacement | |X/S| | |Y/S| |
| -25mm | -14.31dB | -7.55dB |
| 25mm | -13.87dB | -7.69dB |
| displacement | |X/S| | |Y/S| |
| -25mm | 0.1925 | 0.4193 |
| 25mm | 0.2025 | 0.4126 |
| -------- | ---------- | ---------- |
| 50mm | 0.3950 | 0.8319 |
For a large PU, the same procedure yields the following results:
| displacement | |X/S| | |Y/S| |
| -50mm | -10.10dB | - |
| -25mm | -15.94dB | -7.62dB |
| 25mm | -17.00dB | -8.07dB |
| 50mm | -20.69dB | - |
| displacement | |X/S| | |Y/S| |
| -50mm | 0.3126 | - |
| -25mm | 0.1596 | 0.4159 |
| 25mm | 0.1413 | 0.3949 |
| 50mm | 0.2921 | - |
| -------- | ---------- | ---------- |
| 150mm | 0.9056 | 0.8108 |
| X | Y | |
| Normal | 82mm | 38.9mm |
| Large | 107mm | 40mm |
Note: There are two reasons why this may not be optimal:
This head assembly does not contain back termination like the old design. In order to provide some semblance of back termination at high frequencies, where the input impedance of the amplifiers is poorly controlled, it contains bridged-T filters with a corner frequency of 100MHz. Therefore, the impedance looking back into the head assembly's outputs progressively approaches 50Ω above that frequency. Here are some reflection coefficient measurements to demonstrate that (Fig.7,8,9). These plots have been measured using the HP8753D network analyzer.