DIY Modular Test Equipment Project

[void_error]

Oh, hi there!

If you scroll down you might realize I've been consistently biting off slightly more than I can chew.

What started off with me trying to design and build myself a proper bench power supply has grown into a small collection of bits of test equipment being designed as functional modules sharing a common communication bus which is yet to be named.

There is no set date for completion of anything as I'm doing this in my spare time and don't expect regular updates. Since I'm stubborn and I've put a lot of time and effort into this I'm not going to abandon it and I've learned a lot so far.

Enjoy!

Almost forgot... UPDATES below.

UPDATES

2018.02.25

First draft of the UI is now on GitHub. Check the draft branch of the repo.

2018.02.26

First draft of the Waveform Generator + Frequency Counter module is up on GitHub. Repo here.

2018.02.27

First draft of the Regulator for the Digital part is also on GitHub. Link to the repo.

2018.02.28

First draft of the Regulator for the Analog part is also on GitHub. Here's the link to the repo. Damn, too many firsts

2018.03.04

Finished the PCB layout for the UI and yes, it's rev.0 again. Some MCU pins were moved around. It probably won't be the final version as some tweaking is still required.

2018.11.25

Updated everything. Switched to STM32F101 MCUs.

2019.12.27

Waveform Generator is now almost finished.

[void_error]

User Interface

This is the control panel shared by all the different modules. I tried to keep it as basic as possible, as well as easy to use. Nothing too fancy here.

User accessible features:

• 128x64 2.8 inch monochrome LCD display - ERC12864-655
• 4 square push buttons, backlit, function displayed on LCD - Omron B3W-9
• 2 pushable quadrature encoding switches
  
• Buzzer, in case you want those annoying beeps, can be turned off
• USB connectivity via a USB-UART module, initially as USB-serial device, planning to include SCPI commands
• Bootloader for relatively easy firmware updates

Specs:

• DC input: 5V (3.3V onboard LDO)
• Optional fan control/monitoring pins
• Microcontroller generated synchronization signal, used to synchronize DC-DC converters (SYNC)
• Microcontroller clock output (BUS_CLKR)
• Off-board connector (2x10 pin IDC) for communication with other modules via 1-wire, I2C, UART

Hardware:

• STM32F101VC
• EEPROM for saving calibration data and presets
• more details coming soon^TM

USB-UART module

• No idea anymore, could be anything
  
• SiLabs digital isolators on the UART side

[void_error]

These two are going to share the same PCB

Waveform Generator

Features/Specs:

• Sine wave / Triangle wave output - 0.1Hz - 10MHz for sine, 0.1Hz - 250kHz for triangle
• Adjustable amplitude (up to 10V_PP @ 50ohm) and DC offset (-5V...+5V)
• Square wave (uses DDS chip) / PWM output (from UI MCU or local MCU) adjustable output level, frequency values for the PWM are limited by the capabilities of the User Interface MCU or local MCU
• Output driver with adjustable current limit for Sine / Triangle

Hardware:

• AD9834 DDS waveform generator chip
• AD603 Variable Gain Amplifier
• LMH6321 output driver (sine/triangle only), adjustable output current limit
• REF2041 dual output voltage reference
• [https://www.analog.com/media/en/technical-documentation/data-sheets/ad5761_5721.pdf]AD5721[/url] 12-bit DACs for setting offset & amplitude

Frequency Counter

Features/Specs:

• 100MHz maximum input frequency
• Adjustable input trigger level
• Selectable divide by 10  or by 1prescaler
• External 10MHz clock reference input

Hardware:

• STM32F101RC
• LMV7219 high speed comparator
• MCP41050 digital potentiometers
• glue logic

[void_error]

Modular DC Electronic Load[OUTDATED]

Features/Specs:

• 200W maximum continuous power dissipation per module depending on the cooling used, calculations were done using one cheap desktop CPU heatsink for each two MOSFETs
• Two ranges: 50V maximum @ 4 A or 12.5V @ 16A - not a final spec, subject to change
• Fan Control via the Aux Digital Power Board
• Overtemperature protection
• Overvoltage protection
• Overcurrent protection
• Optional on-board microcontroller for the "smart" load configuration
• Two modules can be connected together to make a 400W load - subject to change (it's quite tricky to implement)

Hardware:

• DAC: AD5662
• ADCs: AD7683
• LM56 temperature sensors in thermal contact with each MOSFET
• LM4132-4.1 4.096V reference for DAC and the ADCs
• 2 IRFP460 N-channel MOSFETs as current sinks
• A bunch of op amps (see schematic)
• PIC16F1619 optional microcontroller for the "smart" load configuration, will use the PID loop of the Math Accelerator module for Constant Voltage/Resistance/Power functions
• Dirt cheap AMD CPU cooler

[void_error]

Bench Power Supply[OUTDATED]

Features/Specs:

• Configurable for different output voltages, output currents and number of channels
• Two current ranges
• Overtemperature protection
• Overvoltage protection
• Overcurrent protection

Hardware:

• DACs: AD5662
• ADCs: AD7683
• LM56 temperature sensors in thermal contact with each post-regulator BJTs
• LM4132-4.1 4.096V reference for DACs, ADCs and OTP circuit
• A bunch of op amps
• It can be built with or without digital control

[void_error]

Aux Power Supplies

Provide the required supply voltages for the User Interface / Waveform Generator / DC Electronic Load Modules

Voltages required by each module:

• User Interface: +12V or +7.5V
• Waveform Generator:
  
  • Digital: +7.5V
  • Analog: +5V, -5V, +15V, -15V
  • The digital supply rails use switching regulators, while the analog supply rails use linear regulators.
• DC Electronic Load Module: +12V, +3.3V provided by the User Interface via the IDC ribbon cable
• Lab Power Supply: +3.3V provided by the User Interface via the IDC ribbon cable, powering the digital isolators

Aux Analog Power Board:

• +/-15V& +/-5V (maximum current limited by the maximum allowable power dissipation)
• uses 78xx/79xx and 78Lxx/79Lxx linear regulators

Aux Digital Power Board:

• +7.5V or +12V @ 2A
• uses an LT3680 step-down DC-DC switching regulator

[void_error]

[reserved for future use]

[BFX]

I have similar idea for such 15 years But never was time to start 
Looking forward to progress 

[void_error]

You would spend thousands of hours of work to complete this project. Is it really worth it?

Totally worth it. It's not even that complicated and most things are almost complete as far as the simulations and schematics go.
I'm doing this to learn a few more things, I could have bought all the test equipment.

You can buy a very nice used HP 33120A arbitrary function generator for ~300€.

Aux power supply function can be fullfilled using an off the shelf multi-output industrial power supply. You can find plenty of them in surplus sales for a bargain price. Example : http://www.ebay.com/itm/291682783473

Where's the fun in buying?

[SaabFAN]

You would spend thousands of hours of work to complete this project. Is it really worth it?

You can buy a very nice used HP 33120A arbitrary function generator for ~300€.

Aux power supply function can be fullfilled using an off the shelf multi-output industrial power supply. You can find plenty of them in surplus sales for a bargain price. Example : http://www.ebay.com/itm/291682783473

Nothing hard about building an AWG. In the simplest form, it's just a N-Bit counter, an SRAM and a R2R-Ladder as a DAC with an amplifier behind it.
Can be easily assembled on a euro-card, or put inside a CPLD (the cheap chinese Signal-Generators are proof that the hardware is simple and easy to make).
What's a little bit more difficult is the software, but nothing one can't handle by smashing together some already made pieces of software - The hardest part is probably the software that runs on the PC to create and upload the waveform.

The other modules are excellent beginner-projects and provide good learning-experience.

[German_EE]

You would spend thousands of hours of work to complete this project. Is it really worth it?

Hell yeah! Why? Because it's FUN.

Rightnow I have a home made DC Power Supply, DC Dummy Load, 10 MHz Frequency Standard and RF Power Meter. On the ham bench there's also a home made DC Supply, home made Antenna Tuner and a home made transceiver that is a continuous work in progress. Only the scope, multimeter and VNA were made in a factory.

[TiN]

Subbed.

I'd suggest 14-bit DAC for DC load. That will allow to set current in 1mA steps with some wiggle for calibration. Sometimes it's nice to have.
Or at least two ranges (low current, high current) like in some production DC loads.

[void_error]

Subbed.

I'd suggest 14-bit DAC for DC load. That will allow to set current in 1mA steps with some wiggle for calibration. Sometimes it's nice to have.
Or at least two ranges (low current, high current) like in some production DC loads.

Most likely going for the higher resolution DAC. 12-bit should do, that's 4096 steps and a suitable candidate is the DAC121C085. Same pinout as the DAC081C085, just more bits. Same 12-bit DAC will be used for the Lab Power Supply.

There's also a 10-bit variant, the DAC101C085.

[spudboy488]

User Interface

Hardware:

• PIC16F18857 8-bit microcontroller, relatively new part but extremely versatile, this is the brain of the User Interface
• PIC16LF1455 8-bit USB microcontroller, used as an USB-UART interface
• MCP7940N Real time clock/calendar with battery backup, for optional data logging, hmm, I might even program the UI as an alarm clock

Look at the MCP2200 if all you're using the PIC16LF1455 for is a UART. IC plus 12 MHz resonator is all you need. No programming of the part required.[/list]

[void_error]

Look at the MCP2200 if all you're using the PIC16LF1455 for is a UART. IC plus 12 MHz resonator is all you need. No programming of the part required.

I intend to keep the micro for USB-UART, offers more versatility which might come in handy later. Maybe I drag a friend of mine who's a programmer, to write the PC software to control things, he's pretty good with C#. That would mean programming the micro as a USB HID device, not too complicated since it's going to be based on one of Microchip USB examples / demos, porting the code from the 16(L)F1459 to the 1455 isn't too difficult as I've already ported USB stuff from the 16F45K50 (used in the microchip demo code) to the 25K50 and added my own stuff on top of it with o problems.

I probably haven't mentioned it before or at least not too clearly, but the User Interface board is also designed to be used as development board. For example, since it has both SPI and I2C it can be used to test code for different peripherals with the LCD providing debug data... and then there's the alarm clock . I'll actually program one for that sole purpose, just for fun, it already has buttons, a buzzer , RTCC and a display.

[JPortici]

Why use two pics and not directly a usb capable one?

Keep posting. I think i am in a rather simillar situation (on the verge to design an electronic load because i'll soon need one.. and i have to keep my mind occupied these days)

[void_error]

Why use two pics and not directly a usb capable one?

Time and money.
I did look into some 16-bit PICs with USB, they're only slightly more expensive. However, what I looked at didn't have as many peripherals and/or relocatable pins which would have made both firmware and PCB layout a bit more difficult.
There's also the 18F series with USB, but they don't have a PPS module or as many peripherals (which work independently of the core by the way). Initially I wanted to use those, having previously used them.

The USB capable micro I mentioned is an optional extra and can be left out.
Another reason is the "smart" version of the DC Electronic Load which will use the same 16F18857 micro as the User Interface, also sharing some bits of code like the fan control and I2C code (not too much but it's less work so it takes less time).

[JPortici]

Yes. Those pic24 are more like glorified pic18.. I never liked to use those.
and the dspic with usb is much, much, much more expensive.

I also gave another quick look at the catalog and usb pic16 are only up to 20 pins. Too small, i agree with you. A 28 pin smart pic16 with usb would have been the best.

[void_error]

Ran into a small problem:
The User Interface uses 3.3V logic while the Waveform Generator uses 5V. Logic level shifting is required.

There's a level shifting circuit using a small N-channel MOSFET which is quite popular and it'll do the job for I2C although I think it adds a singificant amount of capacitive loading. I don't plan to go anywhere above the 400kHz I2C fast mode so I think signal integrity won't suffer too much but correct me if I'm wrong.

SPI is going to run at the highest speed supported by the slowest device on the bus - the MCP3201 - which is 1.6MHz, so the simple circuit used for I2C won't do the job.

The Waveform Generator already uses a level shifting buffer for the square wave / PWM output, a 74LVC1T45 so I could go that route.

Since I have both SPI and I2C I'm using an I2C I/O expander for the SPI chip select on the Waveform Generator board so the SPI connector only has SCK, SDI and SDO.

There's going to be a +5V rail available on the Aux Digital Supply, sharing the same connector that goes to the User Interface board so I could do the level shifting there.

What's the best choice for level shifting on the SPI bus? I could post an unfinished schematic if it helps more.

[hans]

Yes. Those pic24 are more like glorified pic18.. I never liked to use those.
and the dspic with usb is much, much, much more expensive.

I also gave another quick look at the catalog and usb pic16 are only up to 20 pins. Too small, i agree with you. A 28 pin smart pic16 with usb would have been the best.

PIC18 and PIC16 architectures are very similar. And I don't mean that in a positive way. The architecture is clearly not optimized for todays complex programs written in C and accessing 4K of RAM (that PIC in particular has 64 banks!). It will probably run and do it's job, but given it's limits.

PIC24 is completely different architecture. The instruction set is tailored to C which results in much faster code, it has a vectorized interrupts, proper software stack, etc.
dsPIC shares the same PIC24 basic CPU core, but has DSP instructions you specifically have to use. They are nice for higher throughput systems, but not a necessity for general purpose stuff.

Personally the only reason I would choose a PIC16 is if I can save 1-2 chips using an (unique) onboard peripherals (like the 24-bit measurement timer). However if that is not the case, I would always start searching at PIC24 or ARM.
I can understand bringing out the USB UART chip for a different reason; so you can isolate the UART to PC (this is test equipment after all). This is possible with USB but likely more complex/more expensive.

However for a 28-pin PIC24 my go-to chip is the PIC24FJ64GB002. Sure it has a few less I/O and PPS pins than the PIC16F18857, but if that's a problem I would opt for the PIC24FJ64GB004.

[void_error]

Is it really necessary to isolate the UART? I know it can be done with 2 optocouplers for less than $1... H11L1 comes to mind.

The Waveform Generator will be mains earth referenced, the DC Load modules will be isolated from the UI, same thing with the Lab Power Supply.

Actually... changing* from a 16LF1455 to a 16F1455 and powering it off the USB +5V and adding two optos with some resistors would do the job.
*so no 3.3V regulator would be required

I did use some (actually quite a lot) of H11L1s in a design, if resistors are chosen carefully you can easily reach 230.4k BAUD with a small but almost symmetrical delay between input and output rising and falling edges.

PIC18 and PIC16 architectures are very similar. And I don't mean that in a positive way. The architecture is clearly not optimized for todays complex programs written in C and accessing 4K of RAM (that PIC in particular has 64 banks!). It will probably run and do it's job, but given it's limits.

Yup, they're almost the same, the only thing I noticed on the PIC18s that PIC16s don't seem to have is the 8-bit x 8-bit hardware mutiplier. Most 18F also support higher clock speeds.

PIC24 is completely different architecture. The instruction set is tailored to C which results in much faster code, it has a vectorized interrupts, proper software stack, etc.
dsPIC shares the same PIC24 basic CPU core, but has DSP instructions you specifically have to use. They are nice for higher throughput systems, but not a necessity for general purpose stuff.

Personally the only reason I would choose a PIC16 is if I can save 1-2 chips using an (unique) onboard peripherals (like the 24-bit measurement timer).

At some point I thought about using a PIC24 for the UI but after looking confused at the datasheet and thinking about how much code would be required to make it work since it doesn't have the onboard peripherals the 16F18857 has I gave up on the idea.

Two relatively uncommon peripherals the drove my decision to use the 16F18857 are the Signal Measurement Timer, makes reading the fan RPM a piece of cake, just spits out the period between pulses, and the Numerically Controlled Oscillator which will be the Waveform Generator's square wave source. Haven't completely understood how to use it so I've got some reading to do.

[JPortici]

Older PIC24 are very close to pic18 in both the core and the peripherals.. the pic24hv family for example. The FJ is a much different beast, nearer to dspics but with no dsp core. still, I have no real use for most pic24 (if i need 16 bit arithmetics i also need the dsp... and dspics have nicer peripherals.)

Void Error has a point: the power of the modern PIC16 lies in its peripherals. with these new chips i mostly just have to move bytes around, call tables and the software is done.
(And they have had linear memory access, along bank switching, for a very very long time.)

[SaabFAN]

Subbed.

I'd suggest 14-bit DAC for DC load. That will allow to set current in 1mA steps with some wiggle for calibration. Sometimes it's nice to have.
Or at least two ranges (low current, high current) like in some production DC loads.

Just a little note here: Using old components bought on ebay, or from Grandpas part-box is often really worth it.
I have an almost finished (Haven't built a case yet) DC-Load flying around somewhere that uses a Ferranti 8-Bit DAC, a 4066 to switch between 2 current-ranges and an LM324 as the controlling Op-Amp on a Euro-Card. It's quite a mess because of the parallel data-bus, but you can debug the circuit with a Multimeter instead of a digital scope with Serial Bus-Decoding.
Also it looks kinda cool having some ceramic DIP-Package with a golden cap on it
And you can build it on Vero-Board in an afternoon instead of waiting for 3 weeks until the boards arrive.

[TiN]

You can get away with a NFET and pullup resistor on SPI if speeds are slow or use more expensive Silabs digital isolators or AD ADUMs if you wish.

As of ebay parts. Its often a good way to get fancy parts, but also a non-zero chance to get remarked fakes. I got fake AD7534 DAC and even fake LM339s one time.

[void_error]

You can get away with a NFET and pullup resistor on SPI if speeds are slow or use more expensive Silabs digital isolators or AD ADUMs if you wish.

I'm going to choose the safe route and go with some proper buffers for SPI.
The USB micro is going to be electrically isolated.
The User Interface will be mains earth referenced, same thing applies to the Waveform Generator. They're going to share the same transformer.
The DC Electronic Load modules will be isolated from the User Interface via ADuM1251 I2C isolators and probably powered off a different transformer or winding, haven't decided yet.
Right now I'm focusing on finalizing the schematics for the UI and Waveform Generator, I might post them this weekend.
One of the goals is to have as few different parts across all of the boards as possible, at least for the glue logic.