Half a hundred HeaterMeter guts


 

Bryan Mayland

TVWBB Hall of Fame
I've received all the parts from Mouser for 50+ HeaterMeters and I thought I'd share to try and keep up with Peter :)




I say 50+ because some parts were purchased for 50, others were more cost effective at quantity 100. There are two kits left from the last batch of 25 with the "pure green" LCD. After those are gone the next 50 will be "amber". If you prefer the green display, better hurry because 50 kits will probably take us through to next year.

Shockingly, it takes about 20 minutes per kit to count out the parts, cut squares of antistatic foam, bag the parts, box it up, enter shipping information, and print a label. Thermocouple soldering and testing adds about 15 minues per board. Procurement is about an hour per run. It is surprising because I can actually build a whole HeaterMeter in under an hour and a half so next year I'll be looking into replacing the resistors, capacitors, diode, atmega, and shift register with surface mount and see if the amount of time is roughly the same.
 
Looks familiar ;)

Replacing all parts that require bending and cutting (resistors, diodes, transistors etc) by SMD will save a lot of time, not sure about the IC's. How about replacing R0 by a cuttable trace on the PCB?
 
If you get a solder paste stencil made from the eagle drawing, thing's will go pretty quickly, especially the multipin ICs. Pop 'em in the toaster oven and be done in minutes. Just need a way to streamline the LCD install, but that could be done easily enough with a home made jig.

I've been reading up and watching some eagle tutorials with the goal of being able to efficiently convert the current PCB into one that is primarily SMD based.
 
Converting the existing board to SMD is trivial, considering 0805 SMD is way smaller than the through hole parts and you can't really change the footprint of the HeaterMeter so you're sort of locked in to the layout. It's on my todo list but I also want to separate the analog stuff a little better so I keep putting it off. Well, that and I keep thinking about how crappy the B+ is for being a HeaterMeter host.
 
What about using a Rpi compute module for the next version? I know it would involve adding/routing USB, etc, but it seems like it would allow a much smaller HM+Rpi combo.
 
Since the compute module requires a board to be used with, it wouldn't seem logical to have to build that as part of the HM build. It would be easier to find a way to get it to work with the the new B+ board or possibly move it to something like the HLK-RM04 mentioned here, which has the added bonus of on-board wifi. Though, from what I can tell, I don't think we're going to see a shortage of regular rPi B boards for some time, and if that does happen, there's still the model A board.
 
Last edited:
But, that board could just be the HM itself. No need for a separate Rpi board. Yes, it would logically have 2 parts - HM stuff, and the Rpi USB. But that's the only port 95% of people use. I don't hear of anyone using the HDMI or other ports in their Rpi. Yes, a few people use the ethernet port, but that could just as easily be a USB ethernet device.

If you wanted to go nuts, the Rpi compute module has ~48 GPIO ports, and could make a super HM with lots more probes. Of course, you might as well start a new project with all the re-writing of code that would entail. :)

Or, totally spitballing... A Rpi compute module could control more than one HM. Might be a simpler way to get more probe slots than re-doing the internals of the HM code.
 
Well, if people are getting grumpy at dealing with the current SMD state of things, wait till they have to deal with a 200 pin SO-DIMM socket! The GPIO pins won't help with probes, as they need the ADC functionality of the Atmega.

I do enjoy the extra GPIO pins on the new B+ model for my basement RasPi. I'm currently loading it up to control plenty of relays.
 
That's what external ADC chips are for! But, yes, a 200pin socket would be, um, a leap for many.

My HM works so well and flawlessly that I can't help but think of ways to make it more "fun" to debug. Or something like that.
 
Well, if people are getting grumpy at dealing with the current SMD state of things, wait till they have to deal with a 200 pin SO-DIMM socket!
Yeah, this is why the RPi Compute module makes no sense for anyone. If you have the manufacturing capability to solder 200 pin SODIMM slots, you probably have the engineering team that can slap a BCM2708 BGA down just as easily. The Pi isn't any sort of wonder of modern circuitry, it is 99% Broadcomm reference design. The Pi's power is its low price point, easy integration, and its ubiquity which means you have a large community to draw from.

The Pi also lacks a lot of features of the ATmega, notably: realtime processing, hardware PWM, and timers.

The SMD board is on my todo list, but I'm probably going to do a different SMD board with the VoCore board first just to see how well it works.
 
Yeah, this is why the RPi Compute module makes no sense for anyone. If you have the manufacturing capability to solder 200 pin SODIMM slots, you probably have the engineering team that can slap a BCM2708 BGA down just as easily. The Pi isn't any sort of wonder of modern circuitry, it is 99% Broadcomm reference design. The Pi's power is its low price point, easy integration, and its ubiquity which means you have a large community to draw from.

The Pi also lacks a lot of features of the ATmega, notably: realtime processing, hardware PWM, and timers.

The SMD board is on my todo list, but I'm probably going to do a different SMD board with the VoCore board first just to see how well it works.

My VoCore arrived yesterday. It's so cute! Haven't done anything with it yet other than powered it up an ssh'd into it. I got it with the dock since I plan to use it like a mini router anyways.
 
The VoCore looks really good. The number of small scale Linux platforms keeps expanding all the time. Continuity of supply etc. will make design direction choices really tough for Brian!
 
I'm still waiting on my VoCore, which is good because I am not ready to work with it yet. I have a few concerns about it:
-- Speed. Is the speed on par with the Pi? 360MHz MIPS vs 800MHz ARM
-- Memory. How easy is it to run out of RAM. The Pi can serve like 20 requests at once and still have plenty of headroom.
-- Storage. Gotta fit all our stuff into a tiny space.
-- Ease of access. There's just one guy making these things in China, which makes them hard to get.

Advantages
-- Cost. $20 vs the Pi $60 (including wifi, sd)
-- Size. The new Pi B+ is impossible to interface with without making compromises in design. The A+ is alright, but it won't fit a B+ because they are different sizes. This almost fits in the corner cutout I have for the USB jacks.
 
A couple of pics of my VoCore



Note: A Canadian quarter is the same size as a US quarter :)

With the add-on "dock" that provides a microUSB power port, usb, 10/100 network and an microSD card slot.

 
I'm still waiting on my VoCore, which is good because I am not ready to work with it yet. I have a few concerns about it:
-- Speed. Is the speed on par with the Pi? 360MHz MIPS vs 800MHz ARM
-- Memory. How easy is it to run out of RAM. The Pi can serve like 20 requests at once and still have plenty of headroom.
-- Storage. Gotta fit all our stuff into a tiny space.
-- Ease of access. There's just one guy making these things in China, which makes them hard to get.

Advantages
-- Cost. $20 vs the Pi $60 (including wifi, sd)
-- Size. The new Pi B+ is impossible to interface with without making compromises in design. The A+ is alright, but it won't fit a B+ because they are different sizes. This almost fits in the corner cutout I have for the USB jacks.

Speed - The overall speed of the unit seems ok, I don't think this will be a deal breaker.

Memory - Fixed at 32MB, this could be a potential bottleneck

Storage - 8MB of flash is pretty limiting and the developer doesn't yet have working code to properly communicate with the microSD port on the dock. Could potentially use USB as extended storage.

Wifi - The onboard wifi is so-so at best and really lacks range. The tiny SMD antenna doesn't do it any favours and could really use an external antenna. I think this will be troublesome once folks attempt to use the unit outside and their AP/Router is inside.

Ease of use - Since this unit doesn't boot off of an SD card like the RasPi, if you manage to make a config error that locks you out of the network, you've got to establish a serial connection to the unit and bring up a serial terminal to re-config or re-flash the unit, which I had to do. I think this will be too much of a hurdle to deal with for a good chunk of the DIY community that might be "getting their feet wet" with the RasPi

Physically connecting the VoCore to the HM board is also going to present a challenge. If you order the VoCore without a dock, which would be the way it should be ordered for use with the HM board, you'll have to solder pin headers onto the board, and the 1.27 mm header spacing is very tight compared to the 2.54 mm spacing on the RasPi.

My personal feeling is that the VoCore is probably not a good fit for the HeaterMeter project, but if you want a cheap, super small OpenWRT based unit to tinker around with, go for it. I think I might hook up a 5v relay board to mine and turn it into a wifi based Christmas lights controller :)

Kickin it old school @ 57600 BPS

 
Physically connecting the VoCore to the HM board is also going to present a challenge. If you order the VoCore without a dock, which would be the way it should be ordered for use with the HM board, you'll have to solder pin headers onto the board, and the 1.27 mm header spacing is very tight compared to the 2.54 mm spacing on the RasPi.

My personal feeling is that the VoCore is probably not a good fit for the HeaterMeter project, but if you want a cheap, super small OpenWRT based unit to tinker around with, go for it.
I picked up my VoCore from the post office yesterday and did some testing with it last night. I have reached the same conclusion you have.

Speed - More than adequate
RAM - Might be tight and would possibly limit the maximum number of connections
Flash - Not enough to include all the packages that are currently included
Ease of use - Not very. One wrong configuration item and you're going to need an FTDI cable to fix it. The pins are way too fine a pitch for a first-time solderer to pull off without bridging, and if you fry the board good luck getting it out.

There's just too many downsides despite the tiny size and super cheap cost. Still it is pretty awesome as a device. I might try to integrate it with my EV charger or something.
 
Looks like there's a new guy on the block, out of china comes the 'Banana Pi'. A bit more powerful but could be used..
 
I've seen it but it's a different thing that just has the same form factor. The codebase is all different so our code won't work and it costs more. It's nice if you need a little more CPU power but we don't.
 

 

Back
Top