DIY Modular Test Equipment Project

[void_error]

The User Interface schematic is almost finished but I thought I'd ask for a few more opinions before I call it finished.
A few component values still have to be calculated.

[Mastrofski]

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.

I think for one-off pieces of equipment, using odds and ends is fine, but for something going into production, or something designed to be recreated by others(such as this project), sourcing components begins to become an issue.

Love the idea though! I'll definitely be following along.

[timofonic]

Amazing project!! Are you going to release as Open Hardware and Open Software at GitHub?

There's already a lab power supply project. Maybe a colaboraron would be possible of some kind.


Thanks.

[void_error]

Amazing project!! Are you going to release as Open Hardware and Open Software at GitHub?

There's already a lab power supply project. Maybe a colaboraron would be possible of some kind.


Thanks.

You mean this?
Not sure yet if I'm going to make it open source, will decide after I build it.

I still have to make some changes, thinking of moving the USB micro of the UI to a separate board and have an off-board connector for UART. USB connectivity was something I don't plan to add before I have the UI board built and tested and I'll need some separate USB-UART modules anyway.

[void_error]

If you're all wondering why I didn't use a SPI connection directly to the LCD and instead went for I/O expanders it's because SPI doesn't allow reading the LCD data (see UC1701 datasheet) such as BUSY status.

Modifications to the Waveform Generator have been completed, it's now time to switch from I2C I/O expanders to SPI ones for the UI as the Waveform Generator already contains one.

[SaabFAN]

I think for one-off pieces of equipment, using odds and ends is fine, but for something going into production, or something designed to be recreated by others(such as this project), sourcing components begins to become an issue.

Didn't realize that this one is intended as a (maybe commercial) product. In that case, only Components that at least "Active" or "Recommended for new Designs" should be used.
For standard-parts like diodes or transistors, there's normally a pin-compatible replacement, but ICs can become a pain real quick.

Btw. how are you planning to implement the Modularity?
- Mainframe-Style: One device with a backplane and Plug-In Modules
- Modular Construction: Complete devices that look different on the inside, but always have the same front-panel and case

[bktemp]

If you're all wondering why I didn't use a SPI connection directly to the LCD and instead went for I/O expanders it's because SPI doesn't allow reading the LCD data (see UC1701 datasheet) such as BUSY status.

Do you need to read the data back?
I generally avoid connecting GLCDs using I2C, because it is slow. Especially in your case, because you need to send many I2C bytes for writing a single byte to the LCD (setup data byte, toggle WRITE. It looks like you need 3 I2C bytes for each write to the ouputs, so thats 9 bytes for each data byte).
The same applies for the rotary encoder. Depending on the encoder, you may need to sample the levels 100-1000 time per second otherwise it misses steps or gets the direction wrong.

[void_error]

I think for one-off pieces of equipment, using odds and ends is fine, but for something going into production, or something designed to be recreated by others(such as this project), sourcing components begins to become an issue.

Luckily for me there's a local distributor about 1km (actually same neighborhood) from where I live and they can get pretty much any part I want from the large distributors like Digikey and Mouser at a slightly higher price but without the $50+ shipping that I would have to pay if I got the parts straight from the overseas distributors.

I usually try to avoid sourcing parts from outside Europe, most of the stuff I order from TME. I think the Waveform Generator has all the parts available there, I'll have to double-check to be sure.

Didn't realize that this one is intended as a (maybe commercial) product. In that case, only Components that at least "Active" or "Recommended for new Designs" should be used.
For standard-parts like diodes or transistors, there's normally a pin-compatible replacement, but ICs can become a pain real quick.

Btw. how are you planning to implement the Modularity?
- Mainframe-Style: One device with a backplane and Plug-In Modules
- Modular Construction: Complete devices that look different on the inside, but always have the same front-panel and case

It's going to a combination of both Mainframe-Style and Modular Construction.

The case won't be the same but the front panel is always going to be the User Interface board.

For Waveform Generator I intend to make it a plug-in board with the BNC connectors on the front panel, whule the Aux Power Supply modules being separate boards or even stackable, that depends on how the PCB layouts turn out.

For the Lab Power Supply there are going to be several modules, with the digital isolator board plugging into the back of the User Interface board just like the Waveform Generator. I'm going to stick to 0.1 inch pitch connectors so I maintain some sort of compatibility with breadboards/arduino boards/etc.

If you're all wondering why I didn't use a SPI connection directly to the LCD and instead went for I/O expanders it's because SPI doesn't allow reading the LCD data (see UC1701 datasheet) such as BUSY status.

Do you need to read the data back?
I generally avoid connecting GLCDs using I2C, because it is slow. Especially in your case, because you need to send many I2C bytes for writing a single byte to the LCD (setup data byte, toggle WRITE. It looks like you need 3 I2C bytes for each write to the ouputs, so thats 9 bytes for each data byte).
The same applies for the rotary encoder. Depending on the encoder, you may need to sample the levels 100-1000 time per second otherwise it misses steps or gets the direction wrong.

I could get away without reading the data back but I might as well have the option to do it for example just updating a certain portion of the screen instead of the whole screen.

I did suspect I2C would be quite slow so I replaced the I2C I/O expanders with SPI ones.

The MCP23S08 expanders used have an interrupt output pin with programmable polarity, which is triggered depending on how the input pin interrupt registers are set. In other words, all the micro has to do when the expander's INT pin changes state is transfer the data. With SPI  that would be really fast, up to 10Mbits/s, depending on the UI micro's clock. With the USB-UART micro moved off-board the UI micro can run off an 8MHz oscillator (I had to use 16MHz when the clock was shared because the USB micro had x3 or x4 PLL to achieve the 48MHz clock required for USB operation), add x4 PLL and SPI can go up to 8MHz (Fosc/4).

The UI code will be heavily based on interrupts, taking advantage of all the core independent peripherals the PIC16F18857 has.

[prasimix]

Didn't realize that this one is intended as a (maybe commercial) product. In that case, only Components that at least "Active" or "Recommended for new Designs" should be used.
For standard-parts like diodes or transistors, there's normally a pin-compatible replacement, but ICs can become a pain real quick.

Btw. how are you planning to implement the Modularity?
- Mainframe-Style: One device with a backplane and Plug-In Modules
- Modular Construction: Complete devices that look different on the inside, but always have the same front-panel and case

It's going to a combination of both Mainframe-Style and Modular Construction.

The case won't be the same but the front panel is always going to be the User Interface board.

For Waveform Generator I intend to make it a plug-in board with the BNC connectors on the front panel, whule the Aux Power Supply modules being separate boards or even stackable, that depends on how the PCB layouts turn out.

For the Lab Power Supply there are going to be several modules, with the digital isolator board plugging into the back of the User Interface board just like the Waveform Generator. I'm going to stick to 0.1 inch pitch connectors so I maintain some sort of compatibility with breadboards/arduino boards/etc.

Glad to see that you decided to define multi-modular solution with Mainframe-style possibility. That is ambitious project and hopefully more people will be attracted soon and decide to join you. If I "dream" about something then it's interoperability of what I made with other people designs. I don't have a problem to adopt everything that is already created to share the same "DIY-instrumentation-bus" and works under same software framework and successfully communicate with commercial instruments.

Like timofonic I vote for GitHub/Bitbucket openness in one moment because it could make a whole thing more interesting in the long run (and hopefully you are already aware of how looooooong could takes to make something in correct/usable way ).

From my side I'm looking forward to see if your design could become a potential "target" for firmware that we are developing. Not sure for now, but if currently developed code could be ported to PIC that can save literally hundreds of hours for you.

One more thing because I been there: it's a LCD display or better to say its 128 x 64 (and monochrome) resolution. TFT LCD becomes really dirty cheap and they are not of equally dirty quality! Even if you hate "touch screen" you don't need to rely completely on it: there is no reason why encoders shouldn't be on your front panel together with touch-screen.

Finally something that I should mention on the beginning: I'm suggest to not take mechanical aspects too lightly! That can really generate a lots of frustration: because you realized that something is not accessible, cannot be mounted, have EMC or heating issue, etc. Of course today's 3D modeling, thermal modeling, etc. could help you to avoid such mistakes that will push you into another revision, but remember that you are still alone and even with 96 working hours day that will consume all your time (again I'm talking about something that will be usable for you and eventually many others). So, mind the mechanics! Define boundaries within everything will happen first.

 

[JPortici]

+1 on the tft screen. nowadays they are much cheaper than GLCDs.
One thing i wanted to try on a new project is going old school: encode the framebuffer in 4 bit per pixel and use a palette. a 320x240 screen would then need "only" 37.5 kB of ram for the framebuffer (and 16 colours are more than enough for my application.. and yours)

[void_error]

Glad to see that you decided to define multi-modular solution with Mainframe-style possibility. That is ambitious project and hopefully more people will be attracted soon and decide to join you. If I "dream" about something then it's interoperability of what I made with other people designs. I don't have a problem to adopt everything that is already created to share the same "DIY-instrumentation-bus" and works under same software framework and successfully communicate with commercial instruments.

Like timofonic I vote for GitHub/Bitbucket openness in one moment because it could make a whole thing more interesting in the long run (and hopefully you are already aware of how looooooong could takes to make something in correct/usable way ).

From my side I'm looking forward to see if your design could become a potential "target" for firmware that we are developing. Not sure for now, but if currently developed code could be ported to PIC that can save literally hundreds of hours for you.

There's still a long way until I start coding. Planning on ordering a few ERC12864-655 displays this month, it'll take another month for them to arrive so that's when work starts. They're about $10 a piece.
My general habit is to write modular code and reusing already written code, which I have a lot of, at least for peripherals, so only the display is something I might run into difficulties with.

One more thing because I been there: it's a LCD display or better to say its 128 x 64 (and monochrome) resolution. TFT LCD becomes really dirty cheap and they are not of equally dirty quality! Even if you hate "touch screen" you don't need to rely completely on it: there is no reason why encoders shouldn't be on your front panel together with touch-screen.

LCD description updated.

I don't hate touchscreens, I just like knobs and buttons more. I'm not using a touch screen because it would add more complexity to the firmware and it'll take me more time, and I've never worked with touch screens. I did look into using TFT color LCDs but I decided to go the simple way, again, because of the limited amount of time I have for this project, I also have a full-time job.

Finally something that I should mention on the beginning: I'm suggest to not take mechanical aspects too lightly! That can really generate a lots of frustration: because you realized that something is not accessible, cannot be mounted, have EMC or heating issue, etc. Of course today's 3D modeling, thermal modeling, etc. could help you to avoid such mistakes that will push you into another revision, but remember that you are still alone and even with 96 working hours day that will consume all your time (again I'm talking about something that will be usable for you and eventually many others). So, mind the mechanics! Define boundaries within everything will happen first.

 

For the User interface at least, I found all the 3D models that weren't already in the DipTrace library, except for the display.

Starting next week I'm back to work at my full-time job, had this week off, so don't expect much progress on this project. I may have 3-4 hours of free time every weekday which I could use and a bit more during the weekends.

[void_error]

Having changed the I/O expanders going to the UI dispay from I2C to SPI I have another decision to make.

The MCP23S08 is addressable, up to 4 devices on the same bus using the same chip select. The User Interface uses two which means I have 2 addresses unused, so I can hook up 2 more chips on the same bus. Another board that uses an I/O expander is the Waveform Generator so it could share the CS pin with the ones on the UI if the address is set different from the other two.

I'm inclined towards sharing the chip select pin as it frees up an I/O pin on the UI micro. What do you think?

[void_error]

Made a few more changes to the UI. Replaced the I2C I/O expanders with SPI ones (MCP23S08), both share the Chip Select pin and SPI lines. For the Chip Select that goes to the off-board SPI connector there's a separate pin.

I actually used up ALL the pins on the micro except for MCLR/Vpp (input-only and high voltage programming).
Pull-downs (33k) are ok with the programming pins, the two XOR gate inputs' capacitances are very unlikely affect programming. With the buzzer being driven by the ICSPDAT pin signal during programming I'm prepared to hear some weird noises.

PCB layout is progressing well, about two to three hours a day this week, all except one of the traces are laid out, they only need some tweaking.

Before I finish laying out the PCB I thought I'd ask for opinions/advice, just in case I missed something or made a stupid mistake on the schematic below. Feel free to ask any questions.

[void_error]

The UI board layout is almost compete, some small tweaks and via stitching are still required.
Found the 3D models for almost all the parts so here's a sneak peek of the board. It's going to be the front panel for almost everything I build from now on.

The dual row connectors on top will either be regular pin headers for connecting to piggy-back boards or shrouded headers for using IDC ribbon cables.
The power connectors on the side can be either straight or angled.

[void_error]

Moved the mounting holes and the 6-pin UART connector on the UI board so I can make the Waveform Generator PCB narrower.

Speaking of the Waveform Generator, the schematic is almost complete, I only have to sort a few details out, spent most of the evening looking through datasheets and I still have a few questions unanswered, like the following:

If I use MLCCs for decoupling on the AD9835, do I still have to use a 10uF || 100nF pair or can I get away with just one cap? Same question for the LMH6321 output buffer.

While looking for MLCC characteristics I ran into this nice selection tool...

An unfinished Waveform Generator schematic can be found below. As always, suggestions are welcome.

[void_error]

Realized the digital output driving solution is dumb just because I can get away with using MOSFETs. Oh, and I'm at the 500 pin limit of my DipTrace Lite license, that's another reason using 74AHC125s is dumb, and yes, I was too lazy to come up with something else. I guess I just can't get away from DipTrace because I've been using it for a long time and I didn't need any tutorials to figure it out. Eagle would be another option, but the free version has a 80x100mm board limit, useless. KiCAD is a pain in the ass to use, installed it for the third time and then uninstalled it after 10 minutes, Circuit Maker has the annoying thing of having to be always online... anyway, rant over.

Mounting holes might cause me some problems, mainly because everything has to be stackable. The UI board is quite crowded, as you can see in one of the previous posts, but the 80(H)x100(W)mm size is pretty much what I'd like to stick with, or if it's not possible, something with nice round numbers. The main problem with the size comes from the fact that the Waveform Generator board has to be at least 70mm wide because of the locations of the mounting holes on the UI board and has to extend about 20mm below it because of the BNC connectors. So far it's 100(H)x80(W)mm.

On top of that there are going to be 2 Aux Power Supply boards, one with the Analog supply rails, the other one with the Digital supply rails. They'll have to be supported mechanically by the Waveform Generator board, that means 4 extra holes. No room for 4 mounting holes on the UI board but I won't need them there anyway.

Another plug-in board would be the digital isolation module for the Lab Power Supply, that will require only +5V which will be provided by the same board that powers the digital part of the Waveform Generator. I'll have to keep that in mind then I design the Aux Power Supply modules.

An alternative to the boards directly plugging into each other is to use angled pin headers and a pieces of stripboard (soldered or with connectrors) linking them together. Not very elegant but it would do the job. In any case spacers will still be required to keep everything from falling apart.

Finally, I've decided to make this Open Source, and I'll probably do some (rather amateur-ish) YouTube videos if I have enough time after I assemble and test the first boards. GitHub stuff coming after that. Stay tuned for more.

Oh, one more thing, does anyone know any free 3D modelling program that accepts .step or .wrl models and it's to use so I can check if everything fits together perfectly?

[altaic]

Finally, I've decided to make this Open Source, and I'll probably do some (rather amateur-ish) YouTube videos if I have enough time after I assemble and test the first boards. GitHub stuff coming after that. Stay tuned for more.

Oh, one more thing, does anyone know any free 3D modelling program that accepts .step or .wrl models and it's to use so I can check if everything fits together perfectly?

Very cool, thanks!

On the open source CAD front, I've used FreeCAD and BRL-CAD, both of which have matured quite a bit in the last few years. The easiest option in this case, though, is probably Autodesk Fusion 360, which is "free for students, enthusiasts, hobbyists, and startups."

[void_error]

FreeCAD seems to do the job, if I export a .wrl model of the board. Easy enough to use for basic stuff like adding boards on top of each other which is all I need to do.

With that said, having the BNC connectors of the Waveform Generator sticking out right below the UI board seems to be a bad idea so I'll move them to the left side. That implies rotating the Waveform Generator board 90 degrees and relocating a few connectors.

The result will look like this: (from left to right) BNC connectors -> Display -> Encoders. Buttons right under the display as seen in one of the previous posts.

I'll start working on it after my badly needed 12 hours of sleep...

[void_error]

Had a few 3D models for the pin header corresponding sockets used to interconnect the boards and started looking around. TE Connectivity seemed to have all the 3D models I needed.

Unfortunately their servers give me this after a few attempts.
"(error 21) [ERR_NO_RESOURCE_AVAILABLE] CoreApi(allocate) A predetermined resource limit has been exceeded. Further connection attempts will be refused until sufficient resources are available. Your patience is appreciated. Please try again later."

[void_error]

Found the 3D models elsewhere but had to register to be able to download.
Anyway, problem solved which means I can post a nice 3D preview of two of the nearly completed boards, the UI and the Waveform Generator.

[void_error]

I was doing thermal calculations for the different components which are going to heat up, like the linear voltage regulators for the +/-15V rails on the Aux Analog Power Supply module and the sine/triangle output driver of the Waveform Generator, the LMH6321.

Since the LMH6321 comes in a surface mount package there's a limit of how much power it can dissipate with only the PCB as a heatsink and also trying to stay away from the SMT heatsinks.

The datasheet says 300mA maximum output current, that's 15V into a 50 ohm load, assuming a dead short on the output. The output of the buffer doesn't swing to the supply rails so it's not really going to have to output more than 250mA. Worst case is outputting half the supply voltage - 7.5V - and that's 1.125W.

If the using the PCB as a heatsink doesn't do the trick I've got another solution which is already implemented in hardware. I can set the output current limit depending on the set output voltage, all in the micro on the UI board.

[void_error]

Connectors, connectors, connectors... and some more connectors.

This is when the 3D preview comes in really handy. I'm trying to make the boards as versatile as possible which means they must accept both straight and angled connectors. For the whole modular assembly of boards straight 0.1 inch headers will do the job just fine, no drama here. Angled PCB headers are not a problem either, unfortunately they're not keyed and you can plug stuff into them the wrong way around and we all know what happens afterwards.

Messing around with connector placement over the past two days trying to make the boards accept different types of 0.1 inch pitch connectors I found that the best placement for the connectors is at the top and bottom, with none going off to the sides.

This allows using both angled and straight pin headers as well as the keyed KK 2.54 connectors from Molex pictured below. Just making sure I don't plug the power connectors in backwards...

[void_error]

Looks like the first layout pass for the UI board wasn't good enough and it also made the Waveform Generator board layout not to easy so it was back to almost square one. I wasn't expecting everything to be perfect first try anyway. Not even second try but I'm on the right track.

Being a modular design implies the use of a lot of mounting holes and a lot of boards that have to stack up.
Since the UI is actually the front panel it'll most likely be the only board which has connections on only one side (back side, the front side is the front panel). Most of the other boards will have connectors to both sides.

I managed to stuff a lot of components in a small space for cost reasons but I realized that for more flexibility I'd need more mounting holes, 8 just wouldn't cut it.
This would allow for extra combinations between modules stacking up in more ways, like this:

[UI] : [Waveform Generator] : [Aux Digital Supply][Aux Analog supply]

or

[UI] : [Aux Digital Supply][USB - UART]

or

[UI] : [Digital Isolators] : [Aux Digital Supply]

For the bigger modules like the DC Electronic Load or the Lab Power Supply the data connections (SPI/I²C) will be made using ribbon cables. I'm trying to keep the off-board wiring to a minimum.

[void_error]

Took a small but needed break from this project and it seems to have paid off.

The UI board seems to be sorted out, only a few more minor tweaks required... maybe(*)... wait, did I say that before?

Anyway, I moved on to another board that has to fit on the back of the UI board and that's the USB-UART module.
It's nothing fancy, just a PIC16F1454 with a few connectors and some optocouples so I thought why don't I make it more versatile, able to support some add-on modules such as RS-232 or RS-485 outputs.

Back to the Waveform Generator board again, I'll have another go at it this weekend probably. The tricky part is getting everything correct, both mechanically and electrically. In more detail, that boils down to BNC connector and other component placement relative to the LMH6321 output buffer while keeping the analog and digital signal paths clear of each other and at the same time have enough room to fit two power supply boards on top of that which will have to share a few components in some configurations (don't want to go into more details, it's a bit hard to explain).

(*) If you look below you might notice U5 (EEPROM) can be moved and J5 (Fan Connector) taking its place. Also, I color-coded a few traces as follows:

• YELLOW:   +12V
• ORANGE: +3.3V
• PINK:       UART
• GREEN:    I²C
• BLUE:       SPI

Here's the UI board as viewed from the back, the display and controls along with the buzzer are on the other side:



And in 3D, with an example of the possible connector configurations:

[pmc]

I just registered, and read this topic. So here goes the n00bish first post...

A couple of things:

1. You were looking for an alternative to DipTrace and free 3D modeling.
For schematics & layout, check out EasyEDA. From the little I've played with it, it seems to work. And it's free.
For 3D, check out OnShape -- a relative newcomer offering no-apologies parametric CAD for free. They claim to import step. Free accounts limit *private* documents. I like it, for what that's worth.


2. Maybe I'm just not understanding what you're doing. It sounds like a UI front panel, some stackable modules and a framework for making arbitrary other stackable modules. Then I imagine that means a UI panel can manage a mixed stack behind it. Is that the idea, or did I get lost?
If I'm not completely lost, then here's a thought: (assume easy answers to fab questions and) arrange modules to launch their real-world connectors/terminals from one side (both?) and angled ~45deg forward. The result would be stacks with a UI panel on the "front" face, and behind that along the "side" of a stack, all the external connectors facing front-ish.
(You're placing connectors at top and bottom not sides - but maybe the constraints that lead that way apply more to stacking connections than to external connections?)
Or more simply just go straight out the side - at cost of requiring more directly perpendicular access to the side of a stack.
Something kinda like this, but with the knobs on the front and connections down the side: http://w140.com/tekwiki/wiki/File:Tek_211_front.JPG