NIKHEF, Amsterdam, 10-08-99, JDS. Pre-amplifier-shaper at 500 µA. Introduction: Normally it will take some time after a pulse appears in a detector system to decide its part of an event or not. The information of the pulse is stored often in an analog memory cell. The system to fill these cells is normally clocked. For this design, we want to use a clock of 40 MHz or 25 nsec period time. The consequence for the pre-amplifier-shaper due to this clock frequency is a settling time of 25 nsec. In the set-up with a bias current of 200 µA this settling time was to fast. The only way to get that fast was by increasing the bias current. In the below described set-up the bias current of the pre-amplifier is 500 µA. The schema. In figure 1 the schema of the pre-amplifier is drawn. Figure 1: The schema of the Pre-Amplifier. The schema is the same as in earlier designs, the difference is found in the sizes of the components. Because there were two parameters of interest, noise as low as possible and settling time 25 nsec, the outcome is a little different as before. At the end of the simulations the outcome was: Pre-amplifier-shaper at 500 µA. 1 NIKHEF, Amsterdam, 10-08-99, JDS. Parameter Value Bias current M0 Bias current M16 (cascode) Bias current M28 (level shifter) Size M0 (width) Size M0 (length) Size M16 (width) Size M16 (length) 500 µA 10 µA 1 µA 15 x 100 µM 0.6 µM 20 µM 0.6 µM The shaper. In figure 2 the schema of the shaper is drawn. Figure 2: The schema of the shaper. Also in this schema the changes are found in the sizes of the components. Pre-amplifier-shaper at 500 µA. 2 NIKHEF, Amsterdam, 10-08-99, JDS. Parameter Value Bias current M33 Bias current M37 (cascode) Bias current M34 (level shifter) Size M33 (width) Size M33 (length) Size M37 (width) Size M37 (length) 10 µA 1 µA 0.5 µA 2 x 50 µM 0.6 µM 10 µM 0.6 µM Pre-amplifier-shaper at 500 µA. 3 NIKHEF, Amsterdam, 10-08-99, JDS. The simulations. The Transient Response. Again this design is simulated. This time I started to simulate the whole schema at once, because both parts influence each other. The shaper is the load of the pre-amplifier. By going around through the schema I have tried to optimise the setting of the schema. The results of the simulations are: Figure 3: The transient response of the design. In figure 3 the transient response of the pre-amplifier-shaper is drawn. The signal is caused by an input pulse of one MIP (500 nA during 3.84 nsec) on the moment 10 nsec. The result of this pulse on the pre-amplifier is drawn in figure 4. The settling time of this signal is 30 nsec. The settling time on the output of the shaper is 15 nsec. This is due to the time constant made by the capacitor C6 and the resistor M38. These two components form a high pass filter. The shaper there for has some gain only for a small frequency band around 18 MHz. During this simulation is the detector capacitor kept 0. The influence of this capacitor will be shown later. Pre-amplifier-shaper at 500 µA. 4 NIKHEF, Amsterdam, 10-08-99, JDS. Figure 4: The transient response of the pre-amplifier. The AC response of the circuit. In figure 5 the AC response of both the pre-amplifier and the total schema are drawn. Well visual in this graphic is the fact that the shaper boosts the gain around the frequency of 5 MHz, where the gain of the pre-amplifier is already falling. Output pre-amplifier Output shaper Figure 5: The AC response of the Pre-amplifier-shaper. Pre-amplifier-shaper at 500 µA. 5 NIKHEF, Amsterdam, 10-08-99, JDS. To boots the gain of the circuit at 5 MHz the AC response of the shaper must be: Figure 6: The AC response of the shaper. The highest gain for the shaper is reached at almost 20 MHz. The difference in bandwidth between the pre-amplifier and the shaper is found in the sizes of the components. In the pre-amplifier the size of the input FET must be big. The reason for this is partly noise and partly a high gm. A wider component has a lower internal resistance and therefor a lower noise contribution. The gm of the FET must be as big as reasonably possible. The input signal is very small. This results in a low voltage at the gate of the FET. To get from this signal a reasonably drain current response the gm must be big. The noise response. In figure 7 the noise response of the circuit is drawn. The graphic in figure 7 is not very informative, because the dimension of the y-axe is totally different from the input signal. With other words you can't say the noise is x % of the input signal. To be able to do this a calculation method is developed with on the end the formula: v ⋅V ⋅ C ENC = rms s s q ⋅ Vop The result of this formula is the number of electrons at the input the noise of the circuit represents, which can be compared to the input signal. Pre-amplifier-shaper at 500 µA. 6 NIKHEF, Amsterdam, 10-08-99, JDS. Figure 7: The noise response. The results of the calculation with the formula for ENC are given in table 1 for 5 value of the detector capacitor. Table 1 C a p a c i t o Er l (ep c Ft r) o n sE ln e oc i t sr eo n sE ln e oc i t sr eo n s f r o m d e t pe ce tr o pr F c a p a c ito r 0 2 3 2 5 4 1 8 1 8 6 3 7 .2 1 0 6 2 0 3 8 8 3 8 .8 1 5 8 2 5 5 9 3 3 9 .5 3 3 2 0 1 0 3 0 7 9 8 3 9 .9 The size of the input signal is 12000 electrons. With a detector capacitor of 20 pF the noise is 1030 electrons, 9 % of the input signal. In a graph the contents of table looks like a line (see figure 8). Pre-amplifier-shaper at 500 µA. 7 n o i s e NIKHEF, Amsterdam, 10-08-99, JDS. electrons noise 1200 Electrons noise 1000 800 600 400 200 0 0 5 10 15 Detector capacitor (pF) Figure 8: The noise verses the detector capacitor. Conclusion. The gain of the pre-amplifier shaper is high enough, at least 100 mV signal from an input of one MIP. The settling time of 15nsec up and 30 nsec down is fast enough. The noise with no detector capacitor (232 electrons) is within the limits, just the noise contribution due to connecting a detector capacitor is to high (38 electrons per pF). Pre-amplifier-shaper at 500 µA. 8 20