Lab Power Supply Design Part 5

[amspire]

The thing that concerns me the most about this design is the lack of an isolated RS232/USB.

Power supplies need floating outputs, and if you couple that with a USB cable internally connected to the power supply negative supply rail that is then plugged into a grounded PC, you have a nasty problem.

The thing I will be looking to do with some Arduino projects I am planning is to add an Arduino Ethernet shield to the instrument's board.  Driving the LAN shield is extremely simple on the Arduino thanks to the smart WIZNet W5100 chip on the shield, and the Arduino Ethernet library functions.

Shields like this http://solderpad.com/solderpad/arduino-ethernet-shield/ are available on ebay for about something like $24 including postage and I think it gives you a number of things:

1. Isolation to 1500VAC and 60pF total capacitance through the transformers.
2. If the instrument is on the local LAN, then any other computer on the local LAN can talk to it. You do not need to have a computer within a USB cable length.
3. You can talk to the instrument from Smartphones. This could be done using Telnet clients, or I know Android phones have several scripting apps available in languages like Python and Lua.
4. Includes a microSD slot, so an internal SD memory for data logging and instruction sequencing could be added.
5. Interfacing with a LAN port is simpler then with serial ports. Port number and IP  address can be set in the instrument, and there is no baud rate issues.
6. No drivers needed to talk to the instrument, so it will work with anything.

Using the SD card to store HTML and images, it is even possible to run a web server from the Arduino, so the instrument can have an Ajax/javascript smart web interface:

http://www.webweavertech.com/ovidiu/weblog/archives/000484.html

At the price, it is not probably not worth adding the parts to the main board - just buy the shield.

The cost is it uses the 3 ports on the ICSP connector, plus output 10 for Ethernet Select, and pin 4 for SD select (if you choose to use it). So that is 4 ports without a SD card and five ports with a SD card.

Obviously more ports then Dave has available in the current design.

Richard.

[McMonster]

The thing that concerns me the most about this design is the lack of an isolated RS232/USB.

There's a mention at the end of the blog that there will going to be a separate board so it won't be a problem to do isolated RS232 or just any interfece you could imagine.

[amspire]

The thing that concerns me the most about this design is the lack of an isolated RS232/USB.

There's a mention at the end of the blog that there will going to be a separate board so it won't be a problem to do isolated RS232 or just any interfece you could imagine.

You are right, Dave did talk about the isolation at the end.

It is just that to add the Ethernet, the simplest way is to add it to the main micro, by freeing up pins with an additional multiplexer IC or something.

Otherwise to add Ethernet, you need a second micro, plus a protocol for the two micro's to talk to each other. Might need that to get an isolated USB too plus an additional isolated supply, and the point is that I believe the Ethernet is really a superior solution to USB with the virtual COM port, and it does not need an iadditional supply to get isolation.

Richard

[Rerouter]

issue with that is, there isnt really any way for him to free up pins, atleast not without significantly rethinking his design,

its open source hardware, if you want some arbitary function that daves not aiming for, then modify the schematic to fill your needs, and lay out the board,

as for the usb port on it, its just for use as a charger at this point, i am pretty certain that if you really wanted to use usb to watch and / or control the supply you would just need to add an isolated serial to usb daughter board to the thing,

as for short circuit, for this revision of schematic, (and this is only as a higher voltage modification 30.72V)
2 x http://search.digikey.com/us/en/products/MC33172N/497-7409-5-ND/1038900 op amps are pin compatible with u3 and u12
and a few gain resistor changes;

2K in parrellel with R29
2K in parrellel with R31 (for a gain of 15)

10K in parrellel with R35
82K to replace R34 (for a maximum input voltage reading of approx 35.6V in, yeah not a pretty number, still working)

and for the mosfet, like a 82K resistor tieing its gate to ground, to divide input voltage by 2 (about 18V max), so its safe from 0 to 30.72V out,

please note the modifications are only something i am pointing out for others looking to modify it to there end

[amspire]

issue with that is, there isnt really any way for him to free up pins, at least not without significantly rethinking his design,

I fully understand that, which is why I made the suggestion. There is always a way to free up pins - if you have to - and adding an Ethernet port would transform the design from "just a power supply" to something that could really evolve in the Open Source world to something that is more then a power supply.

Richard

[amspire]

Just had a quick look at Dave's RevA schematic, and freeing up extra ports is easy, if they are needed.

There is a lot of redundancy as Dave is using seperate clocks and chips selects and data in lines to the A/D, D/A and display. They could all share the one data line and perhaps share clocks. Also four ports are used for the four push button switches. It is possible to reduce that to one port using one of several methods.

In the case of the DAC, it looks like the chip select is also the shutdown. Is there any need to shutdown the chip - as opposed to just setting the outputs to zero?

Richard

[EEVblog]

AFAIK all Arduino compatible Ethernet variants use the Wiznet W5100 chip which is fine pitch SMD.
Hardly suitable for a DIY kit.

Dave.

[EEVblog]

Just had a quick look at Dave's RevA schematic, and freeing up extra ports is easy, if they are needed.

There is a lot of redundancy as Dave is using seperate clocks and chips selects and data in lines to the A/D, D/A and display. They could all share the one data line and perhaps share clocks. Also four ports are used for the four push button switches. It is possible to reduce that to one port using one of several methods.

Yes, I mentioned freeing up pins in the video.

In the case of the DAC, it looks like the chip select is also the shutdown. Is there any need to shutdown the chip - as opposed to just setting the outputs to zero?

No, the MCP4922 has separate /SHDN pin that is tied high.

Dave.

[amspire]

AFAIK all Arduino compatible Ethernet variants use the Wiznet W5100 chip which is fine pitch SMD.
Hardly suitable for a DIY kit.

Dave.

My suggestion was to use the Ethernet shield directly- plug it into the main board. They cost about $24 delivered from ebay.

For $24, you get the isolated Ethernet and the microSD card slot.

[amspire]

No, the MCP4922 has separate /SHDN pin that is tied high.

Dave.

My mistake - I was confusing the A/D and the D/A chips.

[Rerouter]

asmpire, just as a recap from his video, he needed to tie the adc and dac onto the same spi pins to be able to use a oscilator accurate enough for serial communication

[McMonster]

I think the main board should stay the way it is for use as a kit, in my opinion youre trying to overcomplicate it here. Better to leave more freedom with the separate comm board which is perfect for adding modifications. For example it would be much easier to prepare different options for this separate board, from simple isolated RS232 to complex ones including a separate micro for isolated USB and sockets ready to plug arduino ethernet board. This way everyone can choose, but as for the kit it's enough to include just the basic one, the rest should be simple enough for etching at home.

Anyway if someone wants special features he can modify everything himself.

[amspire]

Anyway if someone wants special features he can modify everything himself.

I totally agree that is one sensible approach, but the point I was making is if the Ethernet Shield plugin pins are standard, it opens the supply up to being a real community software development project. Dave is already adding the ICSP pins to the board, which the Ethernet Shield uses.

If someone does a one off mod, you will not get community development happening, and it would be a pity to see forking of the project.

This is a device that has digital programming of output voltage and current, plus independant digital monitoring of output voltage and current plus Daves microCurrent circuit for measuring currents down into the microamps accurately.

Add a strong open source project developing the Ethernet interface, perhaps a web server, and PC software, the ability to log gigabytes of data to an SD card and it becomes a very powerful device. This would make it something different to any other power supply in the market. At the moment, it is just another power supply.

And I am not suggesting anything complicated - just free up some pins, add the socket for the shield. How is that complicated?  The supply will work perfectly without the shield, and as the software improves, add the shield (looks like it is more like $20 delivered now) and you have this amazing device.

[EEVblog]

Whilst I agree with the shield sentiment, there are some practical limitations restricting this. The current case and heatsink design would not allow the shield to have access to the rear of the case for plugging in the ethernet cable. Sure, you could connect the cable internally and have it run out the back of the box, but that's not very pretty.
Plus, there is a twist to this project that I haven't mentioned yet that may answer a lot of people's questions...

Hmm, I just remembered, the case has a taller version. Whilst still not allowing direct access, it does open up possibilities. Let me ponder it...

Dave.

[firewalker]

Plus, there is a twist to this project that I haven't mentioned yet that may answer a lot of people's questions...

Dave.

We will have to wait for part six I guess. 

Alexander.

[RTC]

So, leave the current micro like it is, and just let the ethernet arduino talk to it via serial or i2c?
That would let you do what you want and not limit yourself to working with the few free pins that are available.

[PeterG]

From what i can make out, there are only 3 things i would like to see.

1. I would like one clear port brought out to an 8 pin header. (for a 4x4 matrix keypad for direct entry.)
2. I would like to see a push button to toggle the output on/off.
3. I would like to see the three led colors connected to the mcu to allow color selection via software.

Regards

[Anders]

That and a pin for temperature supervision of the LT30XX and another one for PWM output to run a fan.

[shebu18]

We will need more PIN's, there is a need for a bigger micro, this means no arduino compatibility(maybe mega).

[Bored@Work]

And I like to have 3D animation,  flux capacitor and a do-what-I-mean function.

Gee, people, get real. The design is almost done. It will do what it will do. If you need something else, well, bad luck, and what about doing your own?

[shebu18]

And I like to have 3D animation,  flux capacitor and a do-what-I-mean function.

Gee, people, get real. The design is almost done. It will do what it will do. If you need something else, well, bad luck, and what about doing your own?

i have the same opinion, every one who wants something different or add a function should make their design. There will be the little problem named CODE.

If there is a need for a fan Dave would had sad something about it. So no need for temperature supervision.

[PeterG]

A fan  controller can be done off-board easily enough. To be honest, given the target client base for this project, i dont believe making a port available to a 8 pin header is a major change. My keypad idea was for my own use. It makes the kit more flexible for end users.

Code is easily adapted for users own setups according to there mods if the mcu is Arduino compatible.

I believe this design change would make this a very attractive kit. Given its ability to be customized via the available port and being able to select he screen back color. The back color can be set to change to indicate different conditions for example.

This is just an idea.

Regards

[firewalker]

All those could be implemented as a user extra with a separate mcu connected to the i2c bus.

Alexander.

From what i can make out, there are only 3 things i would like to see.

1. I would like one clear port brought out to an 8 pin header. (for a 4x4 matrix keypad for direct entry.)
2. I would like to see a push button to toggle the output on/off.
3. I would like to see the three led colors connected to the mcu to allow color selection via software.

Regards

[Rerouter]

yep, totally agree, the whole point of us having access to the i2c and serial interfaces is so that we can tack on what ever we like to the thing and treat it like a platform design,

yes each person will hold small changes they want to implement, such as myself, but atleast i am approaching it objectivly by working out what is the smallest amount of change on the original side i can get away with to accomplish what i want,

as for the keypad, yes it could be run through a resistance matrix, being how 3 of the keypad buttons are already on an adc pin, but really all that would need to be done is software modification for it providing one can enable 1 adc port without conflict, giving 2 more adc pins and a gpio if its needed for modders, (who knows maybe put some thermocouples there and drive a fan with the last io)

[Psi]

Yeah, we dont want to load the main micro with a large number of extra non crtical features, it should be reserved for powersupply stuff like voltage/current limit control.

Otherwise it may start lagging critical featurers like code that changes the current limit under special circumstances.