 |
My homemade RV propane detector (see original post) has been working more or less okay since it replaced my ancient, broken Ten-Tek 800a. There’s been a couple of developments since then. |
 |
First, I found another Ten-Tek 800a unit on ebay (sold as new-old stock, still in original box) which I bought because I’d spent enough time with the old one, unable to get it working, that I was still curious about what had caused the original failure. With a working unit side-by-side with the broken one it should be possible to find the fault. Particularly because, secondly… |
 |
I’ve got an oscilloscope now. It’s the Rigol DS1054z, not a high-end unit but it’s a nice 4-channel scope for a good price. Very handy tool to have when trying to debug electronics and it should be plenty fast/capable for checking the old detector. (And to track down some random crashes/resets I’ve been getting on my homemade detector caused by noise on the supply voltage.) | |||||||||
| First test, does the new (well, really old actually, but still in its original box) Ten-Tek 800a unit from ebay work? Hook it up and power it on… Yes, it’s game on. Well, that’s a good start. | ||||||||||
 |
Before getting down to anything substantial I was curious to compare power consumptions of the factory Ten-Tek 800a compared to my homemade kit. Took two readings for current draw for each unit: with and without the propane solenoid activated. And the scoreboard says: | |||||||||
(*) With spikes to about 0.750 A, when the unit sends voltage spikes to pick the solenoid every couple of seconds. So, my unit draws approximately half the current of the original. This is mostly due to the modern buck-supply 7805 substitute in my unit, compared with the linear regulation in the older Ten-Tek 800a. |
||||||||||
 |
Anyway, if you’re always plugged in to shore power you might laugh about a 0.29A current draw. Why care about whether your propane detector chews 0.17A or 0.29A (Approximately 2 watts and 3.5 watts, at 12V)? Even if you’re boondocking but have got acres of solar panels or run a generator 24×7 you can probably ignore the puny draw of any propane detector. On the other hand, if your mode of operation is summer boondocking, in the same place for days at a time, with no solar panels or generator, then it starts to matter. | |||||||||
Next item. I’d experimented with different solenoid voltages for operating the propane supply solenoid before I’d had a working unit to compare against. Now, with a working unit, and even better an oscilloscope, this is easy. You can click on the scope screenshot on the left to get a higher-res version. First surprise is that the Ten-Tek unit doesn’t try to output a single pulse to open(‘pick’) the solenoid, followed by a maintenance voltage to hold it open (which is the approach I’d taken with my build). Instead it outputs a steady train of pick pulses. I guess this has the benefit that if the solenoid gets closed by some bizarre event, it can be re-opened by the next pulse. Not sure if this was a deliberate feature or it was just easier to output a continuous series of pulses. Either way it uses a pick voltage of about 11.2 V and a hold voltage of about 2 V, sending a pick pulse every 2.5 seconds or so. Note: we’re controlling gas, not electricity; a gas appliance probably will barely notice such a short interruption at the supply valve. |
||||||||||
 |
I’d sort of been expecting an epic struggle to get the broken Ten-tek unit diagnosed but once I’d put the two units on the bench side by side it didn’t take much poking with voltmeter and oscilloscope to find a burned out (but surprisingly intact looking) 1 ohm resistor in part of the main supply circuitry of the broken unit. I notice that the replacement unit has a rather heftier component in this location (as well as some other minor changes in other places) which makes me think it’s a later revision. | |||||||||
 |
I drop in a new 1 ohm resistor, of hefty heftiness (1 W) so it shouldn’t burn out in a hurry. Now I’ve got two working Ten-Tek 800a units. You might be able to make out the small red arrow in the pic showing the new resistor. Good to go. | |||||||||
| But wait, there’s more…
I was tempted to put my now-repaired original Ten-Tek unit back into service because it would look way nicer than my ugly homemade unit. And, well, I was getting some annoying intermittent troubles with my homemade unit: it would sometimes reset when the main blower motor of the heater furnace kicks on, and more often when the motor turns off. But, because my unit chews less power (and is rather more configurable), I figure I’ll try fixing its shortcomings and keep using it. (We’ll just ride quietly past the part where I’m using a homemade, shoddily calibrated unit to control a critical safety shutoff for explosive gas.) |
||||||||||
|
Putting the oscilloscope on the homebrew detector’s 5 V rail shows the horrific spike when the furnace blower motor shuts off. It’s about a microsecond of complete hell that manages to blow through the regulator and the filtering caps on both sides of the regulator. Not sure if there’s supposed to be some sort of choke/capacitor in the furnace to try to contain this mess, but I figure my time is better spent getting my circuitry to be robust rather than persuading everything else in the vehicle to play nice. I do some puttering around with increasing the size of the main electrolytic filtering capacitors. Problem is, the higher the value of the caps, the less they can block high frequencies. In the end I managed to fix the problem by putting much smaller ceramic caps as close as I could to the AVR chip power pins. Sure, this is recommended practice to avoid just this kind of random noise problem so the caps should have been there ab initio, but I’ve been too used to my previous projects running on bench supplies rather than the rough and tumble world of vehicular 12 V. | |||||||||
|
So, with the new capacitors in place, a bunch of turning the furnace on/off fails to reproduce the random reset problem. I declare victory. Yay, victory! Ugly, homemade victory! | |||||||||
Future work:
- I’d like to replace the Figaro TGS 813 with the lower power version: Figaro TGS 2610. Discovered the existence of this component when a friend of mine had his much newer Safe-T-Alert propane detector fail and saw this sensor unit inside. I looked up the spec sheet and figured it would be a lower-power substitute (always good to use less power in an RV environment).
- Still need a better case/panel. It looks too 1970s homebrew chic for my tastes. On the plus side, reworking the hardware will let me separate the control panel from the sensor unit, which could be a much nicer setup.
- Would like to redo the MOSFET circuitry that controls the solenoid voltage. It’s
working fine but would probably work more efficiently with a MOSFET that had a lower activation voltage, or by using an opto-isolator to apply 12 V to the gate instead of the 5 V that I’ve got from the AVR chip - any sane man starting from scratch would use a $2 3.3 V USB serial adapter from china/ebay, rather than the 9-pin DB9 port with the extra level buffering IC that I’ve got in my circuit. After all, who the hell has got a computer with a serial port anymore, and who carries a serial adapter in their laptop bag? Easier just to have a usb cable sticking out of the back of the unit for any reprogramming and testing in situ.