Frankenso injector drivers

[AndreyB]

Damn, not thermistors so custom code
Adding two just in case

[puff]

Same ADC, with up to 3v voltage, linear characteristics, no custom code.

[AndreyB]

linear characteristics, no custom code.

Thermistor is definied by pull-up resistance and three point curve based on resistance.

What pull-up resistor value do you propose to use and what three points on the resistance curve would work for a linear sensor which is not a thermistor?

[puff]

Seems simple. Straight line, not curve (two points are enough), resistor value is also calculated based on the voltage values from the data sheet. What's resistor value for battery voltage meter? Basically, same stuff.

[kb1gtt]

I believe no pull up required, it will drive to a voltage. If the MCU has the internal pull up enabled, that's not a big deal, the TMP can still get to target, as the internal pull ups are very small current.

[AndreyB]

I believe no pull up required

I know. The pull-up question was related to me explaining that I am not sure how to use existing thermistor firmware logic with this device and insisting that a bit of new code would be needed.

[tomiata]

Here is more temperature data after doing a 300 mile drive with the thermocouple in the same place. Ambient temp was 22 to 28C for the first two files, and cooler on the 3rd file, temp dropped to about 16C. I didn't look at the data, but watching the meter occasionally, the highest I saw was about 67C when ambient was around 28C cruising at 75 at about 4200rpm.

Happy New Year

[kb1gtt]

Looks like temp2 was the worst of all, with a max logged temp at 68.2C, with a max differential temperature of 46.2C. You can see that is has clearly heated up fully, so heat soak is no longer a concern. Can you provide a bit more background information about run? Can you tell us elevation, barometric and humidity, etc as claimed by something like wunderground.com. I'm just wondering what other parameters we could measure and log about this test run.

The design was for a max ambient of 85C, so with a rise of 46.2C, your max temp would be 131.2C. This is above the UL preferred 105C, but below the 150C thermal shutdown. So far so good, but it would be better if you could add a heat sink. While you should be fine to run this as is with out a heat sink, a heat sink wouldn't hurt. I kind of wonder how this would change if you were in the mountains (less air), doing long hard uphills pulls (more heat), or with dust build up after some time has passed (thermal insulator). However you have about 20C before you run a risk of thermal shutdown, so you would likely be fine.

I love it when a plan comes together.

temp2.png

Attached are the quick bits of processing I did to the csv files.

[tomiata]

The starting conditions for the temperature measurments were from McKinney, Texas around 3pm Dec. 31, 2016
2:53 PM 66.9 °F 44.1 °F 44% 29.77 in 10.0 mi WSW 6.9 mph - N/A Clear
ambient in the car was 22 to 25C

2nd file starts at about 4:30pm, but there is not another weather station along the route, conditions from starting location were:
4:53 PM 64.0 °F 43.0 °F 46% 29.78 in 10.0 mi SSW 3.5 mph - N/A Clear
ambient was around 25 to 28

3rd file starts from around Texarkana, AR at about 6pm after sundown, and temp was dropping over the next 1.5 hours
5:53 PM 53.1 °F - 52.0 °F 96% 29.82 in 5.0 mi East 4.6 mph - N/A Overcast
ambient was around 20 to 18C

It got a little warm in the car in the first 2 hours and I turned on the fan to bring in outside air to the floor and a/c vents. The frankenso case was open and had a little are moving over it. With the case closed I expect the temp would be a few degrees warmer.

I attached the msl log files here. The 3 files should correspond to the 3 temp reading files hopefully, I think a grabbed the right files. I really need to get my rtc going so the log files have correct times.

Is the int_temp field the CPU internal temperature? It shows an interesting curve over time. I need to check it, but I think the drop in temperature could be when I had the heater fan blowing on the floor.

int_temp.jpg

What about gluing a 1" square heat sink over the top of the mosfet's, and maybe a small fan blowing into the enclosure. In the texas summer this thing might get pretty hot.

[kb1gtt]

I would think a piece of aluminum taped or screwed to the metal lid which lightly touches the injector drivers would work well. I wouldn't bother with heat sink grease or anything like that. Just get a piece that is the correct thickness to fill the gap between the screwed on lid and the injectors with a small air gap. 1 or 2 self taping screws could hold it to the metal lid good enough.

[AndreyB]

Is the int_temp field the CPU internal temperature?

Yes it is!

[AndreyB]

So far so good, but it would be better if you could add a heat sink.

Can we please not glue anything before the summer?

I would love to see the data for unmodified board in the TX summer. With a bolt-on we can measure both with and without data, with a glued-on not so much.

[tomiata]

I will hold off on gluing anything down. I think I can attach a metal tab to the side of the case easy enough to get the effect that kb1gtt described. I have some adhesive thermal pads I could try also to get the tab to contact the package without being permanent.

I'll try this later on. Right now I have my frankenso on the bench for the power cycle testing set up.

[kb1gtt]

I wouldn't bother with thermal bonding. If you put a piece of metal close to the chip, you'll have a very think layer of air. Because this air gap is so small it doesn't present much of a thermal resistance. In this application, I would say it's not required, what would help is getting the metal on the top of the chip, such that the top of the chip can conduct heat away. Right now most of the heat is leaving via those thin leads that are connected to the copper plane.

[RustyGargoyle]

Would hate to overheat those drivers causing your injectors to intermittently stop working and taking your engine into lean heaven.
Lots of Rice tax on a maita engine.

I think it would be a better idea to cut a hole in the case or 3dprint a new cover. and slap a fan pointing at the board.
Putting a chunk of metal "near" the drivers would technically have an effect. But not as effective as physical contact.

Sort of like magnets. You could also steal a few sil pads out of something and use the case to radiate the heat threw the metal and effectively threw the case.

[kb1gtt]

The hysteresis of the thermal protection would prevent lean conditions. It would cut full until it cooled down several degrees. So complete engine stop for probably 30 seconds to a minute or so if you overheat.

Fan's have pro's and con's. They are moving parts, so they can fail, as well they pump dust into the enclosure, which eventually adds a layer of thermal resistance. Using a piece of aluminium is not very expensive, and would have high levels of reliability.