|
|
I guess I should repeat the test with just twisted ends.
Yes, the EM3 is set up for eight Type J (CHT) inputs in addition to the eight Type K (EGT) inputs. However, with considerable error if the EM3 input module is not kept at a constant temp (e.g. 25°C).
The is because the MAX6675 chip, that is designed to convert the small voltages from thermocouple wire into a digital signal, is designed for type K. It measures the voltage on the wire, then adds 41uV/°C for the temperature on its pins before converting the input to 4 bits/°C, using 41uV/°C. Called cold-junction or ice-bath compensation. For a Type J thermocouple it should instead be compensating with about 52uV/°C. So the hotter the input module, the less the CHT channels will show. At 150°F input module ambient, it'll show about 20°F too low if calibrated at 80°F. Unfortunately I located my input module in my left cowl cheek extension where it's getting that hot in spite of just have added a blast tube. Anyway, Type J is really only 28% more sensitive than Type K. So that's why I have Type K wire on order and will replace my Type J wire.
Finn
On 6/4/2021 5:19 PM, Charlie England ceengland7@gmail.com wrote:
Well, if all the probes have exactly the same error and they're operating in the same environment, maybe it isn't that bad, especially if you know the approximate error.
Any chance that the silver solder is adding a junction? My engineer neighbor talks about just twisting the wires and hitting them with a TIG torch (no filler)
Also, can you set up the EM3 for type J? Lower temp range so they might be more accurate down around 200 degrees F than K probes.
Charlie
On 6/4/2021 11:25 AM, Finn Lassen finn.lassen@verizon.net wrote:
I got the digital meat/cooking thermometer I ordered. It actually indicates 212°F in boiling water.
Strapped a home made thermocouple (Type K wire with ends twisted and silver soldered) to the tip and put it in boiling water.
Had to change the EM3 scale factor by 4% to make the EM3 read 212°F.
The MAX6675 chip does have max cold junction error of +-3°C and +-2.25°C temperature error. I guess just my bad luck that it adds up to 4°C or more at 100°C. Not quite the accuracy I was going for.
Finn
On 5/21/2021 9:47 PM, Charlie England ceengland7@gmail.com wrote:
That should make it fairly simple to roll your own; you can silver solder the junction into the fitting after you weld the junction.
On 5/21/2021 8:27 PM, Finn Lassen finn.lassen@verizon.net wrote:
The EM3 likes grounded thermocouple wires, however not really sure it's required.
They go into balanced inputs. Actually there are four 8x1 multiplexers in the EM3. They have clamping diodes to protect inputs.
Finn
On 5/21/2021 4:36 PM, Charlie England ceengland7@gmail.com wrote:
Don't forget to check on whether the EM3 wants grounded or ungrounded probes. I forget which is which, but one type of 'meter' will accept either style; the other won't.
On 5/21/2021 2:43 PM, Finn Lassen finn.lassen@verizon.net wrote:
When I made my own engine monitor for my RV-3 back in 1999 I found it easiest to use sensors that output linear 0 to 5V.
So when I got an EM3 from Tracy, I modified it to use such sensors (LM35, TMP35, TC03 and similar ICs). My reasoning was that they were spec'd to accuracy of a couple of degrees over temperature range of interest.
I wondered why the EM3 had 8 EGT and 8 CHT inputs. (Actually in the EM3 the 8 CHT inputs also go into the same MAX6675 chip designed for Type K thermocouple (EGT) wire.)
After several failures and weird readings of LM35's I'm now beginning to see the wisdom of Tracy's approach. And the LM35CAZ is not cheap either.
If you buy a roll of type K thermocouple wire (less than $1/foot), even if it's off a bit compared to an ideal Type K wire, all measurement points should be accurate relative to each other. And you just need to calibrate one input for absolute accuracy and use those calibration factors for all the other inputs.
I've spent a lot of time trying to figure out why the coolant output from my right radiator is so much lower than the left's. I even made 3/4" cu tubing inserts with 1/4" wells in them for the ICs to make sure I was measuring actual fluid temps.
After swapping the hoses to the radiators, the coolant out (out from engine) sensor now inexplicably reads lower by more than 20 deg F than the in-engine sensor. Thought it might have been air in coolant near sensor, but lower readings remain after multiple engine runs,
For the CHT inputs, I had bought Type J thermocouple wire, reasoning that it'd be more sensitive than Type K. Well, 50uV/C is really not much better than 41uV/C, and being fed into a chip that expects Type K really makes it much harder to re-correct (the MAX6675 chip corrects for ambient temp for Type K not J).
So, I think I'll rip out all sensors, order a roll of Type J wire, make and install new sensors: 1/8NPT plugs drilled for 5/32" brass tubing with wire twisted and soldered to the pinched-shut brass tubing. (My old sensors were 1/8NPT plugs drilled for 1/4" cu tubing that the LM35's could fit in.)
Quite a bit of work, but hopefully will pay off in terms of reliable temperature readings. Without those I don't think I can optimize the cooling system.
Hope that others may learn from my mistakes.
Finn
---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus
-- Homepage: http://www.flyrotary.com/
Archive and UnSub: http://mail.lancaironline.net:81/lists/flyrotary/List.html
-- Homepage: http://www.flyrotary.com/
Archive and UnSub: http://mail.lancaironline.net:81/lists/flyrotary/List.html
-- Homepage: http://www.flyrotary.com/
Archive and UnSub: http://mail.lancaironline.net:81/lists/flyrotary/List.html
|
|