Voltage drop data logger in automotive applications

cork_ie

Joined Oct 8, 2011
428
Are there many different fuse blocks (1000s) or is the number of different fuse blocks a much smaller number? My thought being is to create a daughter board that the fuses are plugged into and the entire unit is then plugged into the vehicles original fuse block. This device could be used for the same type of vehicle or perhaps several vehicles using the same fuse block.

The device could also have onboard intelligence, to mux measurements to an offboard device. Or perform data logging as in the original post.

It’s the monolithic nature of this device that is its attraction. Removing and replacing dozens of fuses is simplified to one operation.
Fuses that are not powered with the ignition and accessories turned off are not going to cause a parasitic drain, so we really only need to be concerned about the fused circuits that are permanently powered. Some of these will be heavy main fuses distributing power to sub circuits and these sub circuits will have further fuses depending on the circumstances. In modern cars some circuits have no separate fuses and have electronic "Fuses" - i.e. intelligent power distribution modules. Every manufacturer has a different name for their particular module but they all do the same thing. Most vehicles are unlikely to have more than 20 permanently powered fused circuits , (I am sure someone will find one with 21!) , so we need only be concerned with these. In addition there can be 5 or more separate fuse locations scattered all over the car - under the hood (UK eng. bonnet), in the passenger area and likely in the trunk (UK eng. boot) area too. So not an easy task to connect all these up and measure tiny voltages across fuses and hope to get accurate results. A 1 Ohm resistor plugged in to the suspect fuse sockets will do a better job & will not materially affect the circuit if the current is small. While not a "precision shunt" it is more than accurate enough to diagnose the problem quickly and cheaply without undue complication.
In the end of the day the object is to locate the circuit with the excessive parasitic drain and repair it. Multiplexing is unfortunately going to complicate matters as each input will have to be separately identified in any logfile to locate the the circuit with the drain.
 

Hymie

Joined Mar 30, 2018
1,277
To confirm whether a glove box, boot (trunk) or bonnet (hood) light is remaining illuminated – use a phone camera (on timer) to take a photo from within (with flash off).

I first came across the above suggestion from Honest John, the motoring correspondent for the Daily Telegraph newspaper.
 

danadak

Joined Mar 10, 2018
4,057
If you use the PSOC its DelSig goes to 20 bits, but absolute accuracy you would
have to do an end to end error budget to ascertain that looking at INL, DNL, G error,
drift, offsets. But this application I think does not need high res to function as a diagnostic
tool. So relative accuracy larger consideration, and dynamic range.

If you found you needed much more absolute accuracy you could consider doing CDS
(Correlated Double Sampling) to get rid of several errors.

So being able to do the design pretty much in one chip will also help by not compounding
errors associated with many parts in signal path and board noise coupling.

Note PSOC has a +/- .1% VRef on board, one more part of A/D system taken care of if
that meets your accuracy goals.


Regards, Dana.
 

Reloadron

Joined Jan 15, 2015
7,501
In my vision of the test rig, the normal fuses are completely replaced by test shunts. These would have to still provide a fuse-like function for safety, but would have a bit higher resistance by design so that they show a higher voltage drop, like maybe 100mV at 1A, or 0.1Ω. The shunt would be bypassed to a regular fuse when the car ignition is on, taking the test rig out of the circuit during normal operation and bringing it online only when the car is switched off.

I didn't say it was simple, but I don't see why it wouldn't be doable.
Well so much for me and the blatantly obvious. I get points for getting all hung up on voltage drop over a fuse. Can't believe I flat out missed that and I should have gotten at least a clue from Danadak's post.

Yank the fuse and shove a shunt in there using an adapter. The downside is no fuse but a fuse can be placed inline (series) with the shunt. Just a matter of figuring out the maximum current to measure. There is still a need for amplification of the shunt signal but things become much easier. That makes things very doable. Even then if a good A/D converter and data acquisition unit were employed an amplifier could be dropped. Just a matter of cost and what someone wants to spend. Budget?

Ron
 

Thread Starter

sid1202

Joined Jul 25, 2016
25
Data acquisition will present a new set of challenges. How often should the data be sampled? Once a second? Ten times a second? One-Hundred times a second? Where do we put the sampled data? Using a uC (micro-Controller) to acquire the data is fine but there are some considerations.
The rate could be user defined with the range the uC's limit's and an SD card on board to store the data is common.

The following images reflect data taken at an ambient room temperature of 70 degrees F (21.1 C). The meter is using a 4 wire kelvin measurement on a standard One Ohm precision resistance.
Generally speaking, a resolution of 0.1 mV is fine since 95% of the time a parasitic draw kills a battery in 12 hours or less, the draw is substantial enough. For the odd cases where the is multiple small current draws you need to setup shunts like cork_ie mentioned.

Any voltage drop across these fuses, especially higher current rated fuses will literally be down in the dirt and noise. Using a 10 bit Analog to Digital converter as in most micro-Controllers with a 5.0 volt reference is going to give us 1024 quantization levels or the best resolution we can hope for is 5 volts / 1024 bits = 4.882 mV which is as good as it gets and does not compensate for any A/D errors which there will be. It becomes obvious any DC Voltage Drop across even a 2 Amp fuse will need to be amplified just to use any form of A/D conversion and even if we went with a better A/D which starts adding to cost we will need amplification and precision amplification. This does not even begin to consider the temperature coefficient curve of the fuse. That curve needs to be figured into the software routine for the data acquisition system.
I had planed to use a 3.3 volt uC that has 12 bit ADC's but it is about the same.

Less any consideration for the temperature coefficient of the fuse, assuming there is none, here is what we would get. A two amp fuse with a fixed resistance of 0.042 Ohms and 2.0 Amps of current will have a voltage drop across it of about 8.4 mV. If we amplify that X100 we get about 0.84 Volts which is close to doable for measuring. But if we now look at the 35 Amp fuse we get .003 * 35 = 0.105 Volts and we amplify X100 we get 10.5 volts. That is a problem using an A/D with a 5 volt reference.
That is why I was trying to find a similar circuit that a DVM usues, it already handles all of that at a decent price. Basicaly instead of displaying the measurement to an LCD I need the information sent to the controler so that the software can decide what to do with it.

Earlier I mentioned using an amplifier with multiple gain settings and that is why. Now looking at what danadak mentions:

As we increase the ability of our A/D we increase the cost. There is quite a bit to all of this to accurately know how much current a fuse is passing and when we toss in the temperature coefficient of the fuse alloy it adds to the cost. Using voltage drop across a fuse is not going to be cheap. Very doable but far from inexpensive to do it accurately and correctly.

Ron
I think that somewhere is the middle should give the precision needed and keep costs reasonable.

Sid
 

Reloadron

Joined Jan 15, 2015
7,501
Yes, it is and it's not a uC exactly it is a A/D converter. I say it's a hybrid because the chip is made as a A/D converter with the display drivers built in. It also has a low end of 200 mV and when measuring voltages as in a meter the input voltage ranges are divided down for 200 mV full scale. The Intersil ICL7106 and ICL7107 are high performance, low power, 31/2 digit A/D converters. There is no micro-controller involved in the design. The chip is just a LSI (Large Scale Integration) integrated circuit, there is no programming to it.

The advantage in your application of a uC is the ability to collect data and store the collected data and be able to program a sample rate of your choosing. The rate you sample the data and where you choose to put the data is what you want and how you want to go about it. The collected data needs to be formatted so at a point in time you can go back and read the data.

Danadak back in post #43 mentions using PSoC (programmable system-on-chip) which is a family of microcontroller integrated circuits by Cypress Semiconductor. Much like the Arduino family of microcontrollers made by Atmega and the PICAXE line is made by Microchip. I think you will find the guys who work daily with these things (micro-Controllers) develop a fondness for certain pet chips they choose based on their application needs and they become proficient at programming them.

You also need to decide if you are planning to design a board and once designed and all the layout work done send the files to a manufacturer and have the board made. That or look at a turn key off the shelf modular solution.

Ron
 

danadak

Joined Mar 10, 2018
4,057
If you know C language programming these days a lot different then times past.

Specifically because most modern tools have API libraries for onchip peripherals
where you call f()'s, pass variables to/from them. Much more productive than years
past where one actually had to write the peripheral driver, mostly at ASM level.

For sure pretty much all designers have "pets". But as a field engineer I saw many
professionals move from one architecture/vendor to another, because specifically
their responsibility was to make their companies product competitive. Its a very
difficult decision to make, HW and SW considerations just a part of the overall de-
cision. Add to that time to market, vendor support and reputation, IDE change impact
on error free and effective design to name a few.

Regards, Dana.
 

Thread Starter

sid1202

Joined Jul 25, 2016
25
Yes, it is and it's not a uC exactly it is a A/D converter. I say it's a hybrid because the chip is made as a A/D converter with the display drivers built in. It also has a low end of 200 mV and when measuring voltages as in a meter the input voltage ranges are divided down for 200 mV full scale. The Intersil ICL7106 and ICL7107 are high performance, low power, 31/2 digit A/D converters. There is no micro-controller involved in the design. The chip is just a LSI (Large Scale Integration) integrated circuit, there is no programming to it.

The advantage in your application of a uC is the ability to collect data and store the collected data and be able to program a sample rate of your choosing. The rate you sample the data and where you choose to put the data is what you want and how you want to go about it. The collected data needs to be formatted so at a point in time you can go back and read the data.

Danadak back in post #43 mentions using PSoC (programmable system-on-chip) which is a family of microcontroller integrated circuits by Cypress Semiconductor. Much like the Arduino family of microcontrollers made by Atmega and the PICAXE line is made by Microchip. I think you will find the guys who work daily with these things (micro-Controllers) develop a fondness for certain pet chips they choose based on their application needs and they become proficient at programming them.

You also need to decide if you are planning to design a board and once designed and all the layout work done send the files to a manufacturer and have the board made. That or look at a turn key off the shelf modular solution.

Ron
I've been looking at @danadak 's suggestion's ever since it was made. I find the possibilities very attractive, but I'm have a hard time wrapping my head around it. There is some training available. I might postpone the project and start from the beginning with the PSoC training first. I'll also need to better understand how the opamps and mux's work. I find understanding opamp's very challenging. Every time I think I got it I seen other information or uses of the opamp that contradicts what I have learned lol

I've also been finding other applications for this tool. I would have to decide the level of precision actually needed. A medium precision tool would be good to find an area of interest, but would not be precise enough to pin point some issues. A higher level of precision would fix that, but would increase the cost significantly. A compromise will need to be made either way.

My initial assumption that reading a voltage drop between 2 point in millivolts is out the window... lol

Sid
 

danadak

Joined Mar 10, 2018
4,057
Last edited:

Reloadron

Joined Jan 15, 2015
7,501
The link in danadak's post is pretty good. Now I have no idea where all of this is going to go but if you get into using micro-Controllers you will see quite a bit about: This series of videos demonstrates how to use Cypress' PSoC® 4 family and associated development kits. The development kits are pretty cool tools. You will see the term SDK used which is a Software Development Kit and it is common to see code samples which can be useful.

What you are looking to do is actually very doable. After some thought I would just take the shunt route rather than trying to measure the drop across individual fuses.

Ron
 
Last edited:
Top