New Feature in ALICE Adds Input Divider Frequency Compensation

In this blog installment we are going to talk about a problematic limitation of the ADALM1000. To keep production costs of the board low, certain trade-offs were made. One was to forego programmable input gain ranges that use resistor dividers and perhaps adjustable frequency compensation capacitors. This limited the usable input voltage range to 0 to +5V. Many users complained about this restriction when measuring circuits powered by supply voltages other than (read larger) the built in supplies.

Back in this earlier Blog I introduced the equivalent circuit for the ADALM1000 analog inputs (in Hi-Z mode) and offered some ideas for input voltage dividers and buffers.

This limitation on the allowable voltages that can be measured can of course be expanded through the use of an external voltage divider. The input capacitance, CINT, of the analog inputs in the high Z mode is approximately 390 pF (for the rev D design and slightly higher for the rev F design). This relatively large capacitance along with relatively high resistance dividers can significantly lower the frequency response. In figure 1 we again revisit the input structure of the M1k and connecting an external resistive voltage divider R1 and R2,3. The contents of the blue box represent the input of the M1k in Hi-Z mode. To introduce an optional DC offset for measuring negative voltages resistor R2 is included and could be connected to either the fixed 2.5V or 5V supplies on the M1k. The CINT and effective resistance of the divider network form a low pass pole in the frequency response. To give you a rough idea let's use 400 pF for CINT and 1 MΩ for the resistor divider. That would result in a low pass response with a 3 dB roll-off starting at around 400 Hz,

A capacitor would generally be needed across the input resistor R1 to frequency compensate the divider. Such a hardware solution generally requires the capacitor (or alternatively the divider resistors) to be adjustable.

Figure 1, External voltage divider options.

Wouldn't it be nice to not have to use a compensation capacitor, adjustable or otherwise? The ALICE Desktop could always adjust for any DC gain or offset when using an external divider. A new digital (software) frequency compensation feature has been added to the ALICE 1.2 Desktop software package (down load the latest version from GitHub).

The software frequency compensation for each channel consists of a cascade of two adjustable first order high pass filters. The time constant and the gain of each stage can be adjusted. Normal first order high pass filters do not pass DC so a DC gain of 1 path is added to the overall second order high pass software compensation filter. This structure is often called a shelving filter because of the shape of its frequency response.

In figure 2 we show the new controls for the input compensation. To turn on and off the compensation for Channels A and B check boxes are added under the Curves drop down menu. Turning on compensation applies to both the Scope and Spectrum tools (time and frequency measurements). The filter time constant and gain settings can be set using new entry slots in the Settings Controls screen. The DC gain and offset adjust controls are unchanged.

Figure 2, New software controls

The following examples use resistor values from the ADALP2000 Analog Parts Kit and the intention is to keep the input resistance greater than 1 MΩ. No external compensation capacitor was used. A 500 Hz square wave from the Channel A AWG output is used to observe the step response of the example resistor dividers and adjust the compensation filter settings.

As a simple first example we can just use the 1 MΩ R1 resistor and not include the other resistors from figure 1. This gives us a total input resistance of 2 MΩ.

Figure 3, Settings for just 1.0 MΩ R1

As we can see the DC gain setting is slightly more than 2 which is to be expected based on the internal 1 MΩ resistor and external 1 MΩ R1 resistor forming a 2:1 voltage divider. There is a small DC offset due to the leakage current from the ESD protection diodes on the M1k inputs and the parallel combination of RINT and R1.

The input gain factor of 2 (2.17 to be exact) increases the allowable measurement range from 0 to +5 V to about 0 to +10 V. Enough to work with circuits powered from a 9 V battery.

The stage 1 filter Time Constant is adjusted to correct for the majority of the frequency roll-off and the stage 2 filter Time Constant and Gain are tweaked to take out the remaining higher frequency (2nd order) roll-off. A number of TC and Gain combinations are potentially possible and there may be more than one "right answer".

The following screen shot in figure 4 shows the before and after response to a square wave input from AWG Channel A with and without compensation.

Figure 4, Single 1 MΩ R1 with (orange), without (dark orange) compensation

A factor of 2X might not be enough of an increase in the maximum voltage to be measured. We might also like to measure negative voltages. For a second example we use two 470 KΩ resistors for R2 and R3 along with the 1 MΩ R1. R2 is connected to the fixed +5V supply to introduce some positive offset.

Figure 5, Settings for R1 = 1.0 MΩ, R2,3 = 470KΩ

As we can see the DC gain setting is slightly more than 6 based on the internal 1 MΩ resistor in parallel with the equivalent parallel combination of the two 470 KΩ  R2,3 resistors (235 KΩ) and the external 1 MΩ R1 resistor forming a voltage divider of about 6:1. The input range is now slightly more than 30 V p-p. More than enough for dual supply circuits powered from 2 9 V batteries.

The Screen shot in figure 6 shows the step response for this divider configuration with and without compensation.

Figure 6, R1 = 1.0 MΩ, R2,3 = 470KΩ with (orange), without (dark orange) compensation

For a third example with an even bigger input voltage range we can use a 200 KΩ resistor for R2 and a 470 KΩ resistor R3 along with the 1 MΩ R1.

Figure 7, Settings for R1 = 1.0 MΩ, R2= 200 KΩ, R3 = 470 KΩ

As we can see the DC gain setting is slightly more than 9 now which means that the input range is now slightly more than 45 V p-p. The offset nearly centers the range around ground (approx. +/- 20 V). Now large enough for circuits powered by +/- 15 V supplies.

The Screen shot in figure 8 shows the step response for this divider configuration with and without compensation.

Figure 8, R1 = 1.0 MΩ, R2 = 200 KΩ, R3 = 470 KΩ with (orange), without (dark orange) compensation

Finally it might be nice if we could use a common 10X (passive) scope probe. To connect the probe to the Channel B input of the M1k just a BNC connector with short leads terminated in male pins is used. The input end of the probe is connected to the Channel A output as shown in the photo. It is difficult to inject a DC offset when using the probe so the input voltage range will be just positive voltages up to 10X the 0-5 V default range of the M1k or 0 to +50 V.

Figure 9, scope probe connected to M1k

Figure 10, Settings for 10X scope probe

The step response of the 10X probe without compensation is very poor. With compensation the step response lines up with the output of Channel A. The Screen shot in figure 11 shows the step response for 10X scope probe configuration with and without compensation.

Figure 11, 10X scope probe with (orange), without (dark orange) compensation

Now with the new software frequency compensation feature in ALICE 1.2 and a couple of resistors you can measure just about any range of voltages you need. This Blog post will serve until this new information can be added to the ALICE Desktop Users Guide.

This new feature is also of potential value in Signal and Systems classes to teach different filter configurations and compare in hardware (analog) vs software (digital).

As always I welcome comments and suggestions from the user community out there on other useful and fun ways to use the ADALM1000 and the ALICE software tools.