Phase controller - initial planning

[persivore]

I've been thinking about designing an open source phase controller for the community for a while, and I've finally got some time free to put a design together. All software, firmware and hardware designs will be publicly available.

The purpose of this thread is to see if there's enough interest in the community for such as design, and to see if there are any specific features people would want before I spend too much time on a design. If there's enough interest, I might consider getting a batch of bare PCB's made up, and possibly a limited number of completed units.

Some initial specs:

4*20 character LCD
USB interface with windows config/monitoring software
Standalone operation with LCD/push button interface
Evap, condensor, heat exchanger and ambient temp monitoring via K-type thermocouople or RTD.
4 PWM fan control channels
Programmable thresholds for shutdown under fault conditions, turning on compressors in multi-stage cascades and powering up the system.

If there are any features people want to see in a phase controller, post them here. I'm aiming to start the hardware design at the beginning of next week if there is enough interest.

- Tom.

[sdumper]

In my opinion phase controllers are a bad idea.


- People would rather control the starting and stopping of the phase unit manually
- They add cost to a unit which is already too expensive for most people
- Many controllers are already in existence
- They add additional complexity to the wiring
- They are prone to short circuit
- for cascades PID's are cheap and reliable

[wdrzal]

I thinks it's a great opportunity to teach electronics to the community. I also think selling all the individual parts in a un-assembled kit would sell well. Since it's hard or expensive when you buy in single/small quantities. I think a lot of guys would love a kit with some building support.

[bazx]

tom i would love to use one on my unit

i have a ton of specs i would want on such a controller i will drop them in here or pm later

[kitfit1]

I would welcome a design for a controller as well, along with a hardware source.
I have a feeling that the reason we don't tend to use them much now, is because all the ones i have seen only work in Win XP.
Open source code capable of running in Win7 64 would be a real bonus, configurable for Cascades would be good as well.

[sdumper]

Wow I took the temperature on this idea wrong. I had no idea so many folks were still using these.
Please disregard my earlier comments...

[wdrzal]

Wow I took the temperature on this idea wrong. I had no idea so many folks were still using these.
Please disregard my earlier comments...

I think many early custom models were poorly designed with even poorer customer support. While more work and cost,people are used to(and like) things running automatically.

As most things today,they are a combination of mechanics run by electronics.

I think for cascades that operate colder than -80c, a warm up mode,just to open a bypass solenoid every so often(programmable) to allow the evap to warm to-40c for a few min. as needed to allow any oil there to thaw and return.

also a timer for total system run time,like a hobbs meter.......for feedback for builders.

[persivore]

Thanks for the feedback everyone - it looks like there's enough interest, so I've started on the design.

I've attached some first drafts of circuit diagrams which I still need to check properly (i've already spotted a few mistakes), and then its on to PCB design. If anyone has any feature requests which might need some specific hardware, could you post them ASAP, since its much easier to add at this stage than editing netlists once I've exported the design to the PCB package.

Here are some specs for the hardware. I've gone for a 2-PCB design to keep the high voltage/current switching away from the low voltage control side for safety.
- Coldfire MCF5213 microcontroller (I refuse to use PIC's )
- USB interface via FTDI USB-UART converter (programming USB stacks is evil )
- 4 PWM fan channels with tach monitoring
- 4 K-type thermocouple inputs
- Ambient temp monitoring via RTD and 3 additional RTD channels (should be more accurate than a thermocouple with greater temperature range
- 20x4 LDC with software variable contrast/backlight brightness
- motherboard power-on via optocoupler once a programmable threshold has been reached
- Control of up to 7 line-connected loads (compressors/solenoids/mains fans)

The design will largely be surface mount to reduce size and make build time faster. With low cost equipment and a few hours practice, surface mount soldering isn't that difficult.

- Tom.

[ultralo1]

How about an option to control the second stage comp slaved to the firsts stage HX temp?

What price point are you looking at?

[-EC-UnRockStar]

why you doesnt use a ATMEL AVR?

[persivore]

I'm going to put in programmable temperature thresholds to turn on multiple stages depending on HX temperature. Some basic hysteretic evap temp regulation would be possible, but I don't expect compressors will like constantly being turned on and off. Variable frequency/voltage control of compressors would probably work better to control rotor speed, but would make the design much more complex.

As for the price - I would expect parts to cost around £50-£60 GBP.

How about an option to control the second stage comp slaved to the firsts stage HX temp?

What price point are you looking at?

The choice against an AVR is similar to my aversion to PIC's - I find that they're greatly over-rated, compared to other options in the same price area. PIC's and AVR's are great if you're relitively inexperienced, but there are much better options available which offer better performance (although processor speed isn't really critical in this sort of application). Some of the higher-end AVR's do compare similarly to the lower-end Coldfire microprocessors/controllers however.

The Coldfire series is based on the 68k architecture, so they will run stripped down versions of Linux (although I won't be running Linux in this design), the development tools are good and the assembly language is high level enough to easily write efficient code for anything that is timing critical.

There are some good open source JTAG/programming adapter designs available for the Coldfire too, which can be put together for a reasonably low cost (these do exist for PIC's and AVR's too).

why you doesnt use a ATMEL AVR?

- Tom.

[[XC] Lead Head]

PWM fan control would be awesome for many. Just throw some chunky fets in there that can handle probably 25-30 watts of fan a piece.

Will you be using standard mechanical relays, or solid state relays?

One more thing to consider is shielding, the old G-Froster controllers were plagued by interference problems. Perhaps your design won't? Will be worth testing though to be sure.

[SexyMF]

I've made my own controller before. I designed it so it could be made on a single side PCB using SMT and non SMT components (so a reasonable soldering iron was enough).

Here is what I found:


- Dual layer boards are beyond peoples manufacturing capability
- Open source is useless if your target user can't make the hardware. Expecting them to take a dual-layer board design to somebody to cut is a stretch.
- Open source is useless unless your target can build the programming tools
- Thermocouples are prohibitively expensive due to the CJC circuitry
- Thermistors are cheap and work (albeit with more complicated software). But they are generally limited to -50C. i.e. only good for single stage controller.
- 4x20 Blue backlit LCDs look great

- Easier the assembly the more likely people will attempt it. (I designed mine with no throught hole, so no drilling required)


Without quantity purchasing I can't see you getting the price-point you suggest given the SSR's, CJC's, Thermocouples, op-amps going into your circuit.

You will also have to simply split the project into single stage or dual/multi stage controller. The specs are just to different.

[persivore]

I'll be using mechanical relays - They should cope better with abuse switching inductive loads. Thanks for the info about the interference problem - I'll add in some supression across the relay contacts to catch spikes when the load is switched and do plenty of testing before releasing a final version of the design.

PWM fan control would be awesome for many. Just throw some chunky fets in there that can handle probably 25-30 watts of fan a piece.

Will you be using standard mechanical relays, or solid state relays?

One more thing to consider is shielding, the old G-Froster controllers were plagued by interference problems. Perhaps your design won't? Will be worth testing though to be sure.

My initial prototypes will be on double sided PCB's made at home in the kitchen sink, so it shouldn't be too hard for people with a bit of electronics experience to produce their own PCB's - The hardest part I find is lining up the 2 sides properly. I'll do a run of profesionally made PCB's if there is enough interest for people that can't etch their own boards.

Cold junction compensation will be done in software based on the absolute ambient temperature from an RTD/thermistor to convert the relative thermocouple temperature to an absolue at the sensing end. The RTD's I've been looking at have a temperature range of -200 to +600c and are typically cheaper than K-type thermocouples.

After some consideration, I've decided to change the microcontroller to an ARM 7 based device. Most of these have internal hardware bootloaders programmable over UART/USB, so a dedicated programming cable won't be needed. Plus it means I don't have to write bootloader code to allow firmware updates .

I've made my own controller before. I designed it so it could be made on a single side PCB using SMT and non SMT components (so a reasonable soldering iron was enough).

Here is what I found:


- Dual layer boards are beyond peoples manufacturing capability
- Open source is useless if your target user can't make the hardware. Expecting them to take a dual-layer board design to somebody to cut is a stretch.
- Open source is useless unless your target can build the programming tools
- Thermocouples are prohibitively expensive due to the CJC circuitry
- Thermistors are cheap and work (albeit with more complicated software). But they are generally limited to -50C. i.e. only good for single stage controller.
- 4x20 Blue backlit LCDs look great

- Easier the assembly the more likely people will attempt it. (I designed mine with no throught hole, so no drilling required)


Without quantity purchasing I can't see you getting the price-point you suggest given the SSR's, CJC's, Thermocouples, op-amps going into your circuit.

You will also have to simply split the project into single stage or dual/multi stage controller. The specs are just to different.

- Tom.

[ultralo1]

Can you make it simple enough for me to use? I type with two fingers.

I am interested but I am an electronics tech turned into a refer head, I had one semester of Q basic programming 20 years ago.

[[XC] gomeler]

How difficult would it be for someone to swap in SSRs and program it to act like a PID? I'm looking for a fancier PID for my cascade I have modest high level language skills so I might be able to hack it up a bit myself. What language do you program the controller in? C with a special compiler?

[persivore]

If you can find SSR's in the same footprint as mechanical relays it will be as easy as just swapping the parts, otherwise it will mean a new PCB design for the relay/power board.

Most of the control firmware will be written in C, with a bit of assembly for anything thats critical on timing. I've not decided on the compiler yet, but it will be something free. If you've got some experience of embedded system programming, it should be pretty simple to modify the code. Even with just some X86 C experience it shouldn't be too hard.

How difficult would it be for someone to swap in SSRs and program it to act like a PID? I'm looking for a fancier PID for my cascade I have modest high level language skills so I might be able to hack it up a bit myself. What language do you program the controller in? C with a special compiler?

- Tom.

[SexyMF]

The RTD's I've been looking at have a temperature range of -200 to +600c and are typically cheaper than K-type thermocouples.

Alink to these RTD's would be appreciated