analog::blocks – A New Modular Analog Test and Development Platform

[khs]

The goal was to create a reusable analog development platform instead of designing a new PCB for every project.

Inspired by Jim Williams' LT Application Note 70, the platform is based on a solid ground plane. Instead of soldering every component directly onto the board, the circuit is divided into removable analog blocks.

Each analog::block plugs into the baseboard through a standardized 24-pin, 600 mil DIP connector, making it easy to reuse modules in different projects.

The modular architecture also simplifies debugging, since every analog::block can be tested independently before it is integrated into a larger system.

The analog signals leave the modules at the corners and are routed as controlled-impedance 50 Ω transmission lines with grounded guard traces between adjacent analog signals.

The measured analog bandwidth of a single module is approximately 1 GHz, mainly limited by the inexpensive 0.1-inch connectors.

The interface provides six power rails (±5 V, ±12 V and two auxiliary rails), an I²C bus and up to eight independent SPI interfaces. The digital interface is galvanically isolated and can be connected directly to an USB interface board.

A patent-pending variable module size allows efficient use of the available board area while keeping all modules compatible with the same baseboard.

If required the internal power bus allows the complete system to be powered from USB using plug-in DC/DC or other converter modules, eliminating the need for an external bench power supply during development.

The platform was originally developed for a regulated -200 V power supply. Input is +12 V,
see pictures.

[khs]

Here is a simple ripple measurement of one of the DC/DC converter modules.

The converter generates the ±12 V supply rails for the analog::blocks platform.

For this measurement, the output is routed through the stripline module directly to the coaxial analog connectors.

The output was measured at two different points: once via the SMB connector and once directly at the power rail pin.

[khs]

To use modules from other platforms, there are currently adapters for mikroBUS and the Adafruit QT Py boards.

The mikroBUS adapter is compatible with most modules, but modules requiring the TX and RX lines are not supported.

[khs]

In the past, generating ns pulses based on Jim Williams' design was very interesting to me. But the estimated time required to build it was simply not worth the amount of fun I expected to get out of it.

Now I have a reason to try it — I can use it to check the advantage of the analog::blocks framework.

To generate the first pulses, there was not much work to do. The HV module I had previously built for my 200 V power supply was still available, and I already knew that the 50 Ohm Transmission Module works up to 1 GHz. That was more than enough for the first step.

I only had to change the jumper configuration of the HV power module to connect the negative output to GND and the positive output to the +PWR line. The voltage was set to about 90 V.

As with the previous 200 V power supply, the HV power module is powered from an external 12 V supply.

The 2N2369 pulser transistor is mounted directly on the 50 Ohm Transmission Module.

The 90 V from the +PWR line charges a 2 pF capacitor through a 1 MOhm resistor. The capacitor is connected to the collector of the 2N2369. The emitter is connected directly to the transmission line and terminated with two 100 Ohm resistors. The transmission line is connected to the SMA connector on the baseboard.

Because my 7104 real-time oscilloscope is limited to 1 GHz, I needed a pretrigger to use my 7T11/7S11/S4 14 GHz sampling system.

To protect the sampling head, two 10 dB attenuators were used to reduce the signal level.

The pretrigger circuit is a simple HCT14 delay circuit similar to the one described by Jay_Diddy_B. The trigger signal is connected to the pulser module through the baseboard, so no additional wiring is required.

The pretrigger module was built using a prototype board with all signals already connected. This kept the construction time to a minimum.

The trigger is generated through the transmission line using a 2 pF capacitor connected to the base of the 2N2369.

The two HCT14 ICs on the pretrigger module require +5 V, so a regulator module is used to generate +5 V from the +12 V power rail. The PCB for this module was already available as well, so populating it took very little time.

Because all connections between the modules are part of the baseboard by design, the total time required to get the pulser running was about one afternoon, including debugging.

The measured rise time of the pulser is about 235 ps. There is some ringing after the pulse, which may be caused by the low-cost connectors.

So, with the analog::blocks framework, the amount of fun generated for the effort was really impressive.

[Doctorandus_P]

Skip the patents and make it an open standard. Then I may consider of giving you some very good tips on how to improve this.

[temperance]

If I understand this, this is a board with IC like sockets wired in parallel with pre defined signals?

[Doctorandus_P]

I was thinking of saving the picture of the big PCB overview to my personal archive with "neat PCB ideas", but then I discovered the picture is so vague that  the slikscreen text is not readable. Is this by accident, or on purpose and only available under NDA?

Also: there are probably more people who are collecting "PCB Ideas". For such it's OK to have a company or product name in the photograph (it's even prefered), but a watermark stamped all over the picture area is atrocious. Those usually don't end up in my collection.

As a small sample of my personal PCB photographs collection, I'll post one picture here:

[khs]

Skip the patents and make it an open standard. Then I may consider of giving you some very good tips on how to improve this.

The pinout and an example module are already public in my first post. That's basically all you need to develop compatible modules.

There will be also a free license for module development. The patent protects the underlying architecture, not the interface.

[khs]

If I understand this, this is a board with IC like sockets wired in parallel with pre defined signals?

Yes and no. All power lines, the I²C bus and SPI (except the SPI CE lines) are common to all modules.

There are eight separate SPI CE lines, allowing more than one SPI device to be connected.

The main difference from a pure digital bus is at the corners of the DIL connectors. These pins carry analog signals and can connect to neighboring modules via 50 Ω transmission lines, if the appropriate jumpers are installed or soldered for HF applications.

The jumpers also allow the SMA/SMB connectors to be connected to different modules, providing some flexibility in how the analog signals are routed.

[khs]

I was thinking of saving the picture of the big PCB overview to my personal archive with "neat PCB ideas", but then I discovered the picture is so vague that  the slikscreen text is not readable. Is this by accident, or on purpose and only available under NDA?

No, the slikscreen text is mainly pinout.

Also: there are probably more people who are collecting "PCB Ideas". For such it's OK to have a company or product name in the photograph (it's even prefered), but a watermark stamped all over the picture area is atrocious. Those usually don't end up in my collection.

There is no watermark, but the picture could be better.