My refrigerant leak detector has one flaw – from the factory it uses C-size alkaline cells. Considering I don’t normally stock these (nothing else I have uses them!), and they’re big & heavy, I decided to do a modification to run the unit from lithium cells instead. This would allow charging via USB.
Here’s the factory moulding, with extensive strengthening ribs around the battery compartment. Some of this will have to be removed to get the new cells into place. Cue the scalpel! The battery contacts will also have to be removed from the casing.
After removing some of the plastic, the new cells are a snug fit inside the battery space, with plastic ribs still in place to stop them falling out of the end of the case.
A small hole drilled in the back of the unit allows access to the USB port for charging, through this TP4056 3.7v charge module.
A bit of rewiring, and the original battery connector is refitted to the board.
A while back I found myself in the need of an adjustable RF attenuator capable of high-GHz operation. As luck would have it I had an old Spectrum analyser on the shelf at work, which we had retired quite some time ago.
Spectrum analysers being quite capable test instruments, I knew that the input attenuation would be done with a standalone module that we could recover for reuse without too much trouble.
The attenuator module
Here’s the module itself, with the factory drive PCB removed from the bottom, showing the solenoids that operate the RF switches. There are test wires attached to them here to work out which solenoid switches which attenuation stage. In the case of this module, there are switches for the following:
Input select switch
For me this means I have up to -75dB attenuation in 5dB steps, with optional switchable A-B input & either AC or DC coupling.
Drive is easy, requiring a pulse on the solenoid coil to switch over, the polarity depending on which way the switch is going.
Building a Control Board
Now I’ve identified that the module was reusable, it was time to spin up a board to integrate all the features we needed:
Onboard battery power
Indication of current attenuation level
The partially populated board is shown at right, with an Arduino microcontroller for main control, 18650 battery socket on the right, and control buttons in the centre. The OLED display module for showing the current attenuation level & battery voltage level is missing at the moment, but it’s clear where this goes.
As there weren’t enough GPIO pins for everything on the Arduino, a Microchip MC23017 16-Bit I/O expander, which is controlled via an I²C bus. This is convenient since I’m already using I²C for the onboard display.
Driving the Solenoids
A closer view of the board shows the trip of dual H-Bridge drivers on the board, which will soon be hidden underneath the attenuator block. These are LB1836M parts from ON Semiconductor. Each chip drives a pair of solenoids.
The bottom of the board has all the power control circuitry, which are modularised for ease of production. There’s a Lithium charge & protection module for the 18650 onboard cell, along with a boost converter to give the ~9v rail required to operate the attenuator solenoids. While they would switch at 5v, the results were not reliable.
A bit more time later, some suitable firmware has been written for the Arduino, and the attenuator block is fitted onto the PCB. The onboard OLED nicely shows the current attenuation level, battery level & which input is selected.
Since I inherited an old HP 8753C Network Analyser from work, I figured updating a few things to relatively modern standards would be good. The factory CRT, being 28 years old, is definitely getting a little tired, not to mention being slow to warm up. I read over on the EEVBlog forums about a DIY modification to integrate an LCD display into place instead. There was also the option of a ready-made kit for these instruments which would integrate an LCD, but the cost at over £300 was very prohibitive!
The CRT display unit is a self-contained Sony unit, taking RGBHV signalling from the graphics control card of the analyser. Power is 65v DC which will definitely come in handy for powering the new LCD & control gear, after some conversion.
Doing a quick test with some wiring stuck into the video connector from the graphics controller, proved that I could get a decent video signal out of the unit! The only signals used here are RGB, along with the vertical & horizontal sync.
The video is converted to VGA by way of a GBS-8200 arcade machine video conversion board, which will take many different video formats & spit out standard VGA signalling. The power supply to the left is a standard 100-240v to 12v PSU, which is happy to run at 6t5v DC input voltage, albeit with a ~5 second delay on output startup when power is applied. This is due to the massive 6.6MΩ resistance of the startup resistor chain, which I did reduce by 50% to 3.3MΩ with no effect. Since it does start OK even with the delay, I think I’ll not tinker with it any further. I doubt I could pull the full rated power from it with such a low input voltage, but all included, this mod draws less than 600mA at 12v.
A custom 20-pin IDC cable was made up to connect to the analyser’s graphics board, and this was then broken out into the required RGB & sync signals. Quite a few of the grounds are unused, I’ve not yet noticed any issues with EMC or instability.
There is a quad-XOR gate deadbugged to the PCB, which is taking the separate sync signals & combining them into a composite sync. The conversion board does have separate sync inputs, but for some reason doesn’t sync when they’re applied separately. This gate IC is powered from the 3.3v rail of the converter board, with the power lines tacked across one of the decoupling caps for the DRAM IC.
The donor 8.4″ LCD came from eBay in the form of a POS auxiliary display. I pulled the panel from the plastic casing, along with the control boards, and attached them all to the back. This LCD also had a sheet of toughened glass attached to the front, no doubt to protect against the Great Unwashed while in use! This was also removed.
A cut piece of plexiglas allows the boards to be mounted in the cavernous space the CRT once occupied, with some brass standoffs. 12v power & VGA are routed down to the LCD on the front of the analyser.
The LCD itself is tacked in place with cyanoacrylate glue to the securing clips for the glass front panel, which is more than enough to hold things in place. The input board which just has the VGA connector & power connector is glued edge-on to the metal back panel of the LCD, and is under little strain so this joint should survive OK.