I have now completed the design and construction of both PCBs that make up a high speed sampling front end plugin for the ADALM1000. The programmable DDS sampling clock generator board was discussed here.
A new smaller all SMD board based on the AD783 sampling front end was designed and constructed. The whole thing plugged into an ALM1000 is shown in figure 1. The AD9833 based DDS clock generator board is plugged into the digital connector on the ALM1000. The dual channel sampler board is plugged into the analog connector and the output connector of the clock board. Both boards are powered from the fixed +5 V supply of the ALM1000 including the ADM660 -5 V DC-DC converter that produces the negative supply for the two AD783 SHAs and the AD822 dual op-amp.
Figure 1 ALM1000 with high speed sampler plugin
Also shown in the picture is a SparkFun MiniGen DDS function generator board attached through a wire swapping adapter to map the MiniGen pins to the clock board pass through connector.
Versions of the ALICE oscilloscope and spectrum analyzer Python programs have been made that can set the sampling clock frequency and adjust the display based on the frequency shift factor. The proper sampling frequency is calculated based on the target minimum frequency of the input waveform(s) and the desired shift or multiplication factor as follows:
Fsample = Fmin - (Fmin/MulX)
The Fsample frequency is made smaller to use the positive frequency upper side band of the down converted signal. If the sampling frequency were offset higher than the target frequency then time would go backward on the display. The minimum target frequency has to be the smallest common integer factor of all the input frequencies to be displayed. For example to display 100 KHz and 150 KHz signals, Fmin would have to be no higher than 50 KHz.
In figure 2 the sampling front end is connected to the two AWG outputs of an Analog Discovery board. The AWG outputs were set to generate +/- 3.5 V amplitudes with channel 1 (CA) producing a sine wave at 200 KHz and channel 2 (CB) producing a square wave at 400 KHz. The shift factor is set to 1000 and the minimum frequency is set to 200 KHz. With a shift factor of 1000 the 1 mSes/Div time scale becomes 1 uSec/Div and the equivalent sample rate is 1000 X the 100 KSPS of the ALM1000 or 100 MSPS.
Figure 2 AD783 SHA measuring Analog Discovery - time domain plot.
The measured apparent rise / fall time of the square wave is 7 V in about 100 nSec or 70 V/uSec. This is in line with the output performance of the Discovery AWG.
In figure 3 we see a screen shot of the spectrum analyzer program. Again we are measuring the AWG outputs of the Discovery. Channel 1 (CA) is a producing a 500 KHz sine wave and channel 2 (CB) is a producing a 1 MHz sine wave at around -9 dBV. The spectrum is plotted from 0 Hz to 2 MHz. The shift factor is set to 500 and the minimum input frequency is set to 500 KHz. The effective FFT resolution bandwidth is 3 KHz ( 6 Hz * 500 ).
Figure 3 AD783 SHA measuring Analog Discovery - spectrum plot
The spectrum analyzer program was further modified to include programming the SparkFun AD9837 DDS based MiniGen board. Figure 4 is a screen shot showing the output spectrum of the MiniGen producing a 300 KHz sine wave. Channel A is displaying the raw output of the board and channel B is displaying the signal after a single pole RC low pass filter with a -3 dB frequency near 300 KHz. The second and third harmonics are much worse than the AWG output of the Discovery. The other large non-harmonic spurs are likely due to the switching frequency of the ADM660 -5 V supply DC-DC converter mixing with the input signals and leaking through.
Figure 4 AD783 SHA measuring MiniGen - spectrum plot
In conclusion we can use a high input bandwidth sampler ( AD783 SHA ) to increase the bandwidth and effective sample rate of the ALM1000 to very nearly the equivalent of the Analog Discovery. Board and schematic files for the DDS clock and sampler are attached here.
As always I welcome comments and suggestions from the user community out there.