Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Sean Schult]

Hey guys,

I'm a long time lurker and have learned a ton at this forum (thanks!). Finally made an account because I have a question that nobody has asked before.

I've created small power distros before (120/240v 50A with 6 20A breakered outlet pairs, etc) and got a new idea for a project. I want to make a distro with built-in metering of voltage, wiring, and current. It would have a front panel display (probably an LCD) and possibly even WiFi so the power could be monitored from a PC. Why? Because I can. It doesn't solve any need that can't be solved by walking around with my trusty Fluke, or even a hardware store plug tester. But it would be cool, and convenient, and I want to do it. I have some experience with AC wiring, although I am not an electrician. I have extensive experience with embedded systems, so that part is no problem. My question comes in, how do I combine all of these things into a single unit while remaining code compliant? Is it even possible?

The low-voltage section would have:
- A microprocessor (Arduino, Parallex, TI, or similar -- I haven't decided yet)
- A display of some type, probably monochrome LCD unless I find a color one cheap
- A button panel
- A small transformer to power it from below

The high-voltage section would have:
- A 50A input, CS plug
- A digital voltmeter IC measuring across each input combination (H-H, H-N, H-N, H-G, H-G, N-G) which would double as a means of warning against a miswire
- A number of 20A breakers, probably 6 or 8, feeding regular 5-20 receptacles
- An inductive coil on each circuit, connecting to above, to measure current
- (Optional, this may be pushing it) A 20A relay on each line to allow remote open/close of each outlet. Perhaps it would even lock in the open state in the event a miswire is detected.

Maybe this is a bit absurd but it sounds fun to build. The programming and wiring all seem pretty simple, but the one thing that gets me is how exactly to run wire between the two sections. For instance, the output of a voltmeter IC to an input pin on the processor. Can you run HV and LV in the same box? I had planned to address this by building it in two sections, connected by a conduit, and only the necessary sensor wires would traverse it. This would minimize exposure, but by design it is impossible to eliminate. Does it even matter as long as it's all contained within a listed box? I know there's commercial solutions similar to this, but they're special purpose and very expensive. Still, that proves it's possible to do it right. Based on the scope of my project, what do I need to keep in mind to do it within code?

Thanks in advance!

[Scott Holtzman]

You need an 8 bit output for your embedded computer.  Some have TTL logic that will have to be buffered with a transistor to be able to sink enough current to pull a relay.  Don't forget the diode across the relay contacts. 

As far as digitizing the inductance coil.  You will need one that has 16bits of resolution.  The output will not be linear.  Don't get wrapped up in the math to do integrate the device response.  Plot the curve and use a table lookup.

See if you can find the 8bit io device and the simple circuit to integrate to a relay.  If you can't I will for you when in front of a real computer.

[Sean Schult]

Thanks Scott. Wiring the relay is no big deal, but it is very interesting to know that the inductive coil is not a linear response. I would not have discovered that until calibration. Why do you think a table lookup would be preferable? Is the curve really that complex (ie multiple knees), that a simple equation couldn't do it?

Anyway, I'm more looking ensure safety (top priority!) and code compliance (second priority!). Currently I intend to address safety by keeping as much of the LV in a separate section of the housing to minimize exposure, and by using insulated buttons on the front panel (think industrial PLC panels) so that in the event the LV section did become energized somehow, there is still a barrier to the operator. As for code compliance, I have no idea what I don't know, so hopefully I can discover anything necessary to know before the build starts

You need an 8 bit output for your embedded computer.  Some have TTL logic that will have to be buffered with a transistor to be able to sink enough current to pull a relay.  Don't forget the diode across the relay contacts. 

As far as digitizing the inductance coil.  You will need one that has 16bits of resolution.  The output will not be linear.  Don't get wrapped up in the math to do integrate the device response.  Plot the curve and use a table lookup.

See if you can find the 8bit io device and the simple circuit to integrate to a relay.  If you can't I will for you when in front of a real computer.

[Scott Holtzman]

I just find it easier to use the table, much simpler to calibrate.  Personal preference. 

You don't have to go overboard on the isolation.  You could go as far as opto isolators if you wanted to.  Certainly none of the computer controlled PDU's from APC do this, the HV and computer circuit boards are in the same chassis. 

Build quality is essential to make sure the parts stay put in transport.

Why would you need insulated buttons?   The cool side of me would say to use a touch screen or at least softkeys, however if other people are going to use it the more conventional illuminated switch to load (the switch would still tell the uP to turn the relay on).

This guy has a ton of cool info on his site:

http://www.c2o.pro.br/en/automation/x73.html

Take a look at 5.3.1 just what you want.

The project won't be complete without running a http server so you can manage the thing remotely!


Sent from my To Be Filled By O.E.M. using Tapatalk

[Lyle Williams]

Any safety critical item that cannot be visually inspected for correctness is very expensive to get approved.

And...  Are you sure you mean HV ?

[Jonathan Johnson]

I'd suggest working with a UL panel shop, at least for your prototype. They will have the experience to help you design a distro that will satisfy NEC and UL requirements. Once a solid design is in hand, file for any applicable patents then go for manufacturing and marketing if that strikes your fancy.

Panels built by UL panel shops generally are treated favorably by inspectors.

Most industrial electrical contractors have their own panel shops and many employ electrical engineers. Of course, you'll probably want to have your lawyer help you write up appropriate NDAs and non-competes if you'll be contributing any significant intellectual property to the project or if you plan on marketing it as a product.

[Steve M Smith]

I just find it easier to use the table, much simpler to calibrate.  Personal preference.

Same here.

About twenty five years ago, I had to repair something at work.  It had a strain gauge and an LED display.

The strain gauge was connected to an ADC and the ADC digital output was connected to a set of eproms.  The eproms had been programmed with a look up table to convert the strain gauge output directly to data to run the displays.  It didn't have normal binary to seven segment drivers, the eproms drove the segments directly via buffers - no microprocessor was involved.

It was a clever and simple design.

[Stephen Swaffer]

Low voltage (less than 50 volts) and "high" voltage (50-600 volts)  [the term "high" voltage means different thing to different people your local POCO considers under 600 volts "low voltage"]  commonly reside in the same cabinet in industrial controls.  Others on here know UL requirements better, but the bottom line is that any switch rated for 120 volts is insulated adequately and any industrial touch screen would provide adequate isolation outside the box.

This could all be done with an industrial PLC and touchscreen and you could have ethernet built in-but that likely will cost a bit more and  take  the fun out of the project.  You can get current transducers that output a voltage proportional to the current-that might simplify things in that you would just have whatever number of voltage inputs you need to do the A/D conversion-not a combo of current and voltage.   

[TJ (Tom) Cornish]

Hey guys,

I'm a long time lurker and have learned a ton at this forum (thanks!). Finally made an account because I have a question that nobody has asked before.

I've created small power distros before (120/240v 50A with 6 20A breakered outlet pairs, etc) and got a new idea for a project. I want to make a distro with built-in metering of voltage, wiring, and current. It would have a front panel display (probably an LCD) and possibly even WiFi so the power could be monitored from a PC. Why? Because I can. It doesn't solve any need that can't be solved by walking around with my trusty Fluke, or even a hardware store plug tester. But it would be cool, and convenient, and I want to do it. I have some experience with AC wiring, although I am not an electrician. I have extensive experience with embedded systems, so that part is no problem. My question comes in, how do I combine all of these things into a single unit while remaining code compliant? Is it even possible?

The low-voltage section would have:
- A microprocessor (Arduino, Parallex, TI, or similar -- I haven't decided yet)
- A display of some type, probably monochrome LCD unless I find a color one cheap
- A button panel
- A small transformer to power it from below

The high-voltage section would have:
- A 50A input, CS plug
- A digital voltmeter IC measuring across each input combination (H-H, H-N, H-N, H-G, H-G, N-G) which would double as a means of warning against a miswire
- A number of 20A breakers, probably 6 or 8, feeding regular 5-20 receptacles
- An inductive coil on each circuit, connecting to above, to measure current
- (Optional, this may be pushing it) A 20A relay on each line to allow remote open/close of each outlet. Perhaps it would even lock in the open state in the event a miswire is detected.

Maybe this is a bit absurd but it sounds fun to build. The programming and wiring all seem pretty simple, but the one thing that gets me is how exactly to run wire between the two sections. For instance, the output of a voltmeter IC to an input pin on the processor. Can you run HV and LV in the same box? I had planned to address this by building it in two sections, connected by a conduit, and only the necessary sensor wires would traverse it. This would minimize exposure, but by design it is impossible to eliminate. Does it even matter as long as it's all contained within a listed box? I know there's commercial solutions similar to this, but they're special purpose and very expensive. Still, that proves it's possible to do it right. Based on the scope of my project, what do I need to keep in mind to do it within code?

Thanks in advance!

Sean, "safe", "to code", and "listed" are three totally separate concepts.  How much you need to care about the latter two are dependent on what you want to do with your project.

Building something safe requires using components designed for your application - wiring and devices appropriate for the voltage and current in safe enclosures, etc.  The National Electric Code is a helpful starting point for this, however the NEC has no jurisdiction over products such as you're referring to; it is intended for buildings.

Listing agencies such as UL deal with the safety of devices, and the rules can be very different.  For example, the NEC does not allow any combination of low voltage and line voltage in the same enclosure, while nearly every device made intrinsically must have both high and low voltage together to function; however the listing agency has their own set of rules that dictate exactly how to safely isolate high and low voltages.

If you're making this in your basement for your own use and it will never be inspected by an electrical inspector, then an NEC-derived distro with common-sense isolation to low voltage is probably acceptable.  If you intend to use this in a professional capacity, the barrier is much higher, as the typical inspector du jour is likely to automatically flunk anything that looks homemade - particularly if it's more complicated than a few breakers and receptacles.

Getting a product listed costs tens of thousands of dollars even if you did it all correctly the first time, and more if you have to fail, modify, resubmit, fail, modify...  That starts making Motion Labs gear look pretty attractive.

[Stephen Swaffer]

Tj, for my own understanding, what would be the listing expectations for building an industrial control panel, using listed components following manufacturers instructions?  This is how I would tackle this project if brought to me-that is my background experience and comfort level and for me a one off would likely cost about the same if you consider labor cost-obviously Scott and the Sean have different backgrounds so a different approach makes sense for them-but would my approach help with listing requirements or make no difference?

[TJ (Tom) Cornish]

Tj, for my own understanding, what would be the listing expectations for building an industrial control panel, using listed components following manufacturers instructions?  This is how I would tackle this project if brought to me-that is my background experience and comfort level and for me a one off would likely cost about the same if you consider labor cost-obviously Scott and the Sean have different backgrounds so a different approach makes sense for them-but would my approach help with listing requirements or make no difference?

I am not an expert on this, but my understanding is that normally, industrial control panels are covered by the NEC/OSHA, and are generally a fixed installation, so if they are inspected, it's done at installation and considered for the known, intended use.  A listing agency listing such as UL is frequently, though not always, required for selling a product to a customer.

I have never submitted a product for listing, but I know that others on this forum have and can speak to the process.  The process is different and in some ways more stringent than the NEC, as they not only have construction guidelines that detail how to separate high and low voltages, but they do potentially destructive testing like sticking probes through enclosure holes as well as hipot tests to try to anticipate end users' behavior and see how the product handles it. 

An assembly of UL-listed components does not automatically mean that the assembly is UL-listed.

Having a licensed or experienced person do the design and/or construction will certainly make a material difference in the safety of the product, however it will have no bearing on the listing process.

For the purpose in question where some stages are inspected, and moreover the life-safety and liability consequences inherent in this kind of thing make me leery of going too far down the DIY route here.  Best case is it works perfectly, and you have some interesting monitoring and/or logging capabilities.  The medium case is that the inspector shuts you down and you have to find another distro.  The worst case is that the product fails and injures someone, and you have absolutely no legal legs to stand on.

Here's UL's page on the listing process: http://ul.com/offerings/manufacturers/preparing-for-your-ul-mark-evaluation/

[John Roberts {JR}]

The low-voltage section would have:
- A microprocessor (Arduino, Parallex, TI, or similar -- I haven't decided yet)
- A display of some type, probably monochrome LCD unless I find a color one cheap
- A button panel
- A small transformer to power it from below

The high-voltage section would have:
- A 50A input, CS plug
- A digital voltmeter IC measuring across each input combination (H-H, H-N, H-N, H-G, H-G, N-G) which would double as a means of warning against a miswire
- A number of 20A breakers, probably 6 or 8, feeding regular 5-20 receptacles
- An inductive coil on each circuit, connecting to above, to measure current
- (Optional, this may be pushing it) A 20A relay on each line to allow remote open/close of each outlet. Perhaps it would even lock in the open state in the event a miswire is detected.

Maybe this is a bit absurd but it sounds fun to build. The programming and wiring all seem pretty simple, but the one thing that gets me is how exactly to run wire between the two sections. For instance, the output of a voltmeter IC to an input pin on the processor. Can you run HV and LV in the same box? I had planned to address this by building it in two sections, connected by a conduit, and only the necessary sensor wires would traverse it. This would minimize exposure, but by design it is impossible to eliminate. Does it even matter as long as it's all contained within a listed box? I know there's commercial solutions similar to this, but they're special purpose and very expensive. Still, that proves it's possible to do it right. Based on the scope of my project, what do I need to keep in mind to do it within code?

Thanks in advance!

You say you have created distros before. Were they agency approved or just DIY? I find the design of the instrumentation more interesting than the agency approval.

Many modern microprocessors have 10 or more A/D convertors so you can roll your own voltmeters.

Relays to switch power seems old technology, I've switched mains voltage in DIY projects from a micro using a triac with that gate driven by an optical fet. So the micro just has to send a few mA to LED inside the opto-fet and the triac switches on/off.

To sense current it seems something like a current probe should work. Maybe that's what you are talking about. Regarding resolution the cheap micros have up to 12 bit capability built in. 

It seems like it would still end up a hot chassis design to probe the voltages (like separate voltmeters would be).

If the design does not expose the end user to shock hazard, and doesn't catch the building on fire it should be agency approvable but that is a significant expense to undertake casually.

JR
 

[Jonathan Johnson]

This blog post that dissects both a genuine Apple iPad charger and a counterfeit clone provides some good insight on the differences between correctly isolating high and low voltage, and incorrectly isolating it when within the same enclosure.

[John Roberts {JR}]

This blog post that dissects both a genuine Apple iPad charger and a counterfeit clone provides some good insight on the differences between correctly isolating high and low voltage, and incorrectly isolating it when within the same enclosure.

In my judgment I would be tempted to do this as a hot chassis design (i.e. not isolated from mains like old TV sets). Only the i/o or things users might touch need to follow spacing and insulation safety rules. A wifi interface with perhaps remote control via that, could finesse that part of the design. I do not know the insulation resistance of a standard LCD display cover but a few hundred volts or more seems reasonable. A second clear plastic escutcheon could be added if needed.   

JR

[Sean Schult]

Thanks for the responses guys! I knew this would be a good source for information, and reread in more detail later when I get home.

To be clear, I intend to use it in a professional capacity (on stage) but do not have any intention to sell it. Therefore being safe is an absolute requirement, but being listed/approved by an agency is not. I didn't mean to imply that I wanted to list it, only that I thought starting with a panel that is listed by itself would be a good step.

The distros I've made before and use are not agency approved, but they are safe. I'm small time and rarely have to meet an inspector, but if one did come to an event I was working, I want it to pass. Not have him go "what the heck is that? get it out of here". I do have access to a panel builder and could make it look like a commercial product. Is there anything else I can do to make it passable? I see DIY distro's in use all the time around here, and don't really consider this any different -- as long as the added complexity of these features does not in any way compromise the safety or function of the distro itself. Perhaps this is what I would need to prove to an inspector? That I took care in this regard -- for instance that the G/N paths remain unswitched, conventional circuit breakers are used so it does not rely on the CPU for overcurrent protection, etc -- basically "if the complicated crap dies, it still works as a basic distro or passes no voltage at all"

[John Roberts {JR}]

Thanks for the responses guys! I knew this would be a good source for information, and reread in more detail later when I get home.

To be clear, I intend to use it in a professional capacity (on stage) but do not have any intention to sell it. Therefore being safe is an absolute requirement, but being listed/approved by an agency is not. I didn't mean to imply that I wanted to list it, only that I thought starting with a panel that is listed by itself would be a good step.

The distros I've made before and use are not agency approved, but they are safe. I'm small time and rarely have to meet an inspector, but if one did come to an event I was working, I want it to pass. Not have him go "what the heck is that? get it out of here". I do have access to a panel builder and could make it look like a commercial product. Is there anything else I can do to make it passable? I see DIY distro's in use all the time around here, and don't really consider this any different -- as long as the added complexity of these features does not in any way compromise the safety or function of the distro itself. Perhaps this is what I would need to prove to an inspector? That I took care in this regard -- for instance that the G/N paths remain unswitched, conventional circuit breakers are used so it does not rely on the CPU for overcurrent protection, etc -- basically "if the complicated crap dies, it still works as a basic distro or passes no voltage at all"

I doubt many inspectors would be willing to bless your DIY instrumentation on the fly.

Perhaps consider making it stand alone to just meter voltages at the output of the distro. If an inspector makes you remove it, the show can still go on.

JR

[Lyle Williams]

I have an ethernet controlled power strip, with relays that control PAR lamps.

The relay board and the ethernet control board are opto-isolated.  The relay board and control boards are driven by different AC/DC converters.  The AC/DC converters are listed products.  All cabling is screwed down with bootlace ferrules on the ends, so nothing can come loose.  All cables are tied into place with cable ties, so if the impossible happens and something comes loose, it can't go anywhere.  There is an insulated shield between the AC and DC sections. 

All AC power into the unit comes through a listed breaker and a listed dual pole RCD/GFCI (so if active and neutral have been swapped by some genius upstream, the protection still works.)

It is inspected, tested for faults annually, and labelled as tested.  Test results are in a log book along with test results for everything else electrical I use. 

If there had been a commercial product to meet my needs, I would have bought it.  That would be cheaper and easier.

[Steve M Smith]

In my judgment I would be tempted to do this as a hot chassis design

If the measurement devices had outputs in digital form it could be both.  Hot chassis measurements and a separate interface isolated with opto-isolators.


Steve.

[John Roberts {JR}]

If the measurement devices had outputs in digital form it could be both.  Hot chassis measurements and a separate interface isolated with opto-isolators.


Steve.

Yup I thought of that and since popular micro's have multiple A/D convertors, one micro could be on the hot side and another micro on the cold side (they only cost a few $ each), isolated by a single opto-isolator passing the combined data in a single digital link. Control data could be passed back the same way to control relays (triacs) and like on the hot side.

Just because you can is not always enough reason to do something, be careful and have fun.

JR

[Steve M Smith]

Yup I thought of that and since popular micro's have multiple A/D convertors, one micro could be on the hot side and another micro on the cold side

That would work.  Link them with an opto isolated I2C.
 
 
Steve.

[John Roberts {JR}]

That would work.  Link them with an opto isolated I2C.
 
 
Steve.

I'm an old SPI guy... but I2C would work and give you some addressing capability. I tend to roll my own SPI routines. For this app I might use two (one for each direction) since opto-isolators are not bidirectional like a hard wired data bus could be. Might look into a self clocking data transfer scheme. Data bandwidth is probably modest. While one master clock could support moving data in both directions.

JR 

[ProSoundWeb Community]