Last updated : 16/Jan/2006
[Back]


Wilkinson Type Peak Hold ADC


We used a Wilkinson type peak hold ADC (WPH-ADC) at E570 extension experiment. WPH-ADC was expected to have no odd-even effect and small differential non-linearity, not like successive approximation type peak hold ADC (SPH-ADC). But we found WPH-ADC had some hardware problems and larger odd-even effect (from bottom 2 bits). Here I summarize WPH-ADC analysis by comparing SPH-ADC.


Figure 1
Figure 1: WPH-ADC vs SPH-ADC. The events are selected by “ sdd1, run 418-538 and multiplicity = 1. ”

Figure 2
Figure 2: The close-up of Figure 1. A negative correlation can be seen in the bottom 2 bits of WPH-ADC.

Figure 3
Figure 3: The correlation of WPH-ADC flipped the bottom 2 bits and SPH-ADC. There are still strange overflow and some dispersion correlations.

Figure 4
Figure 4: The close-up of Figure 3. The negative correlation is disappeared. But the line of the correlation has about 5 ch width. This is too huge.

Figure 5
Figure 5: The projection of Figure 3 on X-axis (SPH-ADC). Due to odd-even effect periodical increment errors are exist. They make the differential non-linearity (DNL) bad.

Figure 6
Figure 6: The projection of Figure 3 on Y-axis (WPH-ADC). Please compare the DNL of this and of SPH-ADC (Figure 5). There are larger odd-even effect and worse DNL than SPH-ADC. The close-up are Figure 7 and 8.

Figure 7
Figure 7: Ti Ka region of SPH-ADC. 1 bin = 1 ch.

Figure 8
Figure 8: Ti Ka region of WPH-ADC. 1 bin = 1 ch. Mainly the odd-even effects influences bottom 2 bits.

Figure 9
Figure 9: Ti Ka region of SPH-ADC. 1 bin = 4 ch. Still the odd-even effect exits.

Figure 10
Figure 10: Ti Ka region of WPH-ADC. 1 bin = 4 ch. Does the odd-even effect disappeared ?

Figure 11
Figure 11: A graphical cut of WPH-ADC overflow region. The projection of this region are Figure 12 and 13.

Figure 12
Figure 12: The projection of the graphical cut of Figure 11 on X-axis (SPH-ADC). This is a proper histogram for SPH-ADC.

Figure 13
Figure 13: The projection of the graphical cut of Figure 11 on Y-axis (WPH-ADC). Though proper SPH-ADC events are incremented, WPH-ADC events are overflowed. This is a hardware problem of our WPH-ADC.