DIY PSU Build for Flashing

[aboulfad]

Voltage stable at 13.40V with my old flaky multimeter !!!

[jwzimm]

Originally Posted by aboulfad

jwzimm , I think you read diagonally!

Isolating DC ground is safer and more flexible, I have nothing else to contribute to this

I did not read it incorrectly. My point is that we are using these supplies in a very different way than they are. His point is that the grounds can be at different potentials leading to a shock. This is true if the two supplies are not tied together. My point is that this is impossible with the method we are rigging these. We are tying the two DC grounds together thereby guaranteeing that they are at the same potential. It is physically impossible for the two chassis to be at different potentials.

As I have said, there is likely not risk in isolating the grounds but it is not needed.

[aboulfad]

Voltage potential also exists between the DC +ve and -ve output, a DC PSU isolated or floating design is safer and more versatile regardless of application as by not grounding to "chassis ground" (which AC ground is wired to), you have no power behind the DC outputs and there is far less risk of problems happening when shorts or internal damage happen when connected to a DUT. (Noise, interference, AC current flow...)

I agree with many if the things you said. It's rare that I engage in a back and forth discussion, but given that this is a PSU DIY, and there is potential for risk of personal/material injury when modding PSUs, anyone can do whatever they want on their PSUs, but I would not recommend that things are not needed with certainty unless one is a subject matter expert.

[aboulfad]

Guys, I give up on this PSU I thought with thicker shorter cables that the voltage drop will be smaller, yes it was initially, but with time, as I was recoding my car, I had my ignition ON, lights, but no AC fan, the battery initial voltage of 13.45 (~100%) dropped after 1.5 hrs to 13.25 (~80%, measured of course wo PSU connected).

The voltage that showed initially with the PSU connected at the battery was 13.40 and dropped/stabilized with load to 13.35V.

So somehow, the battery is still supplying current and draining...

[HDDMark]

Originally Posted by aboulfad

Guys, I give up on this PSU I thought with thicker shorter cables that the voltage drop will be smaller, yes it was initially, but with time, as I was recoding my car, I had my ignition ON, lights, but no AC fan, the battery initial voltage of 13.45 (~100%) dropped after 1.5 hrs to 13.25 (~80%, measured of course wo PSU connected).

The voltage that showed initially with the PSU connected at the battery was 13.40 and dropped/stabilized with load to 13.35V.

So somehow, the battery is still supplying current and draining...

I may have missed it but did you jumper pins 7 and 9 with a pot in between?

I just rewired mine this morning this way using three 1k pots in series. Based on my observations. I believe this thing is working much better now but I also shortened my leads to about 3 feet. I'm still using the 10 AWG wire.

The car had a starting voltage of around 13.31 volts and I had the PSU set to 13.6 volts so similar to conditions from when I flashed the car. This time however, with the ignition, fan, and headlights on (forgot the stereo) the PSU was now putting out 15.45 amps with 13.3 volts at the car's terminals whereas with the previous setup it was only putting out about half the current (~7.5 amps). I only ran this test for about 10 minutes so I can't speak to the long term results but it's looking better.

It seems now that with a voltage drop of about 0.3 volts between PSU and car it's putting out 15+ amps. I know it's not a linear relationship but it seems that it should be able to put out 20-30 amps with a voltage drop of less than 0.6 volts. If true, the PSU should be suitable for keeping the battery voltage above 13 volts during flashing. I think that a moderate but reasonable voltage drop is required with this PSU to get it to pump enough current.

[aboulfad]

No I didn't do it yet, tomorrow morning, but you also shortened your cable!Your results are promising as were mine even before I redid the cable, but the test of time is the one that will show really what's happening. One way to know if the battery is draining, like I did, is to measure starting and end battery voltage with some load for one hour or more. Another annoying thing, I couldn't figure out how to keep the stupid ignition ON all the time, even with door open, low beam on, it kept shutting off after 15mns...

[HDDMark]

Seems to me that as the car's electrical needs increase, the voltage at the battery needs to decrease so that the PSU increases it's current to to match the current the car is using. It's just a question of how big the voltage difference needs to be for a given current. Once the voltage drop stabilizes it shouldn't continue to drop. I think that's the way it should work. It would be interesting to know how the voltage drop occurred. Did it drop quickly at first and then stabilize or did it just continue to drop over the entire 1.5 hrs.

Using two PSUs in parallel would roughly half the voltage drop for a given current output?

I'm thinking that using sense wires at the car's terminals would be even more effective and allow for more current output at a much less voltage drop.

[packetpilot]

Originally Posted by aboulfad

No I didn't do it yet, tomorrow morning, but you also shortened your cable!Your results are promising as were mine even before I redid the cable, but the test of time is the one that will show really what's happening. One way to know if the battery is draining, like I did, is to measure starting and end battery voltage with some load for one hour or more. Another annoying thing, I couldn't figure out how to keep the stupid ignition ON all the time, even with door open, low beam on, it kept shutting off after 15mns...

Turn on standing lights (1 o'clock position of the switch), problem solved. :-)

[aboulfad]

HDDMark , ok here is so more useful information.

What you say above has some merit, my friend who is a seasoned EE and is amazing with circuits called me and we discussed this.

The starting and stability of the PSU voltage and cabling is key to ensuring that the battery isn't being drained. He said the PSU voltage has to be higher than the nominal battery voltage (13.2V for our battery) in order for the PSU to supply all the needed current (and of course there's the path of least resistance, considering the car's battery cables are AWG4 ...)

If that is not the case, then both will contribute until a balance is reached, where the battery voltage is close to or lower than that of the PSU. This is all great, on the condition that the PSU is able to maintain the voltage above 13.0 (BMW 2016 rec) for demands up to 30-40Amps.

The initial 13.45V battery measurement most likely indicates surface charge, that with time and load the battery voltage dropped to the nominal voltage of 13.2, so most likely the battery wasn't really draining... but I finished my coding session and didnt go further.

But for the interest of circuits and science, I will hook up my car tonight and leave it ON with load for few hours and see ! I think all the pieces of the puzzle are starting to come together. I will also increase the PSU output to 13.65V.

[aboulfad]

Originally Posted by thisisdave

Turn on standing lights (1 o'clock position of the switch), problem solved. :-)

I can't recall if I tried that, few weeks ago, it'd stay on for like an hour, but now post full 58.3, I tried different light setting and the ignition would switch off after 15mn or so. What's the longest you kept your ignition on? Position one clock is parking lights, maybe I'll try that for my long run...

[HDDMark]

OK,

Since in my haste I shortened my leads and reconfigured the PSU to use pins 7 and 9 instead of pins 3 and 9 without seeing how much improvement the shorter leads would yield,
I decided to to a bit of investigation.

Experiment 1 - Original Long Leads:

I hooked up the long leads I was using originally and found similar results as before rewiring my voltage control through pins 7 and 9. The battery started at about 13.32 volts (no PSU).
With the ignition, fans, and headlights on and PSU set at 13.6 volts, it was putting out an indicated 7.68 amps.
This was similar to before changing the voltage control to pins 7 and 9. After the test, the battery voltage with no PSU attached was 13.17.

Experiment 2 - Shortened Original Leads:

The battery started at about 13.25 volts (no PSU). With the ignition, fans, and headlights on and PSU set at 13.6 volts, it was putting out an indicated 15.46 amps.
After the test, the battery voltage with no PSU attached was 13.28 volts. It seems that the shorter leads were probably the greatest contributor to the increased output of the PSU
and it actually increased the battery voltage relative to the voltage before the test.

It seams that short, fat leads are the way to go to reduce the voltage drop under load. Again these were short tests since I'm impatient.
When I flash a few more of the ECUs on my car I'll let you guys know how it works.

[aboulfad]

I just borrowed a clampmeter and the measured DC voltage is lower than my multimeter by ~0.2V ... Jeez how could I not think to calibrate my instruments so now I have to figure out which is the accurate one and back to the drawing board. I am currently in the process of draining my battery a bit... First mistake: trust an old multimeter ...

[aboulfad]

Updated post with more experiments. Coupled with my friend's explanation, Mark's hunch and my bro's clamp meter, many posts and weeks later I understand something and measure it... The only other experiment missing is to characterize Vpsu under varying larger loads without voltage regulation, but I think I reached my limit for experiments (one would need a load resistor ...). I would to thank all you guys and terahertz for keeping me busy and learning few things!

Terms:

OCV: open-circuit voltage, the voltage measured across a battery leads, without any load (in modern cars, OCV is quasi CCV as there's always a slow discharge for various stuff)
CCV: closed-circuit voltage, is measured when a charge or discharge is applied
Nominal voltage: the reported or reference voltage of the battery (13.2V for LifePO4, 4x3.3V)

Info 1: the PSU with the higher voltage and lowest path of resistance will supply the majority of the current for a load. This is key, as you will see from the experiment below!

Experiment 0:
First I wanted to drain my battery to Vocv=13V, which is roughly 30-40% state of charge. After a trip of ~35mns, Vocv=13.65, Vccv=13.45 (IgnON, low beams, radio, fan max).

At T=70mns, Vocv=13.13, Vccv=13.02
At T=130mns, Vocv=13.02,Vccv=12.98
Ibatt=23A (IgnON, LB, Fan)
With Vpsu=13.48 (using the clamp meter), Ipsu=23A !!!

NOTE:ISTA/D "state of charge" ABL reported 30% which is close to what I calculated using the Lithium battery state of charge curve.

Experiment 1:
I alternated few times between two different Vpsu of 12.5 (no pot) and 13.48 (pot between pin 3 & 9), and measured voltage, current at PSU (Ipsu) and battery in trunk (Ibatt), And the results are impressive and confirm Info 1 and also the newly recommended programming voltage for LifePO4 terahertz

Vpsu=13.48, Vccv=13.3 stable (T=60mn), IgnOn
Ipsu=17A

Vpsu=12.5, Vccv=13.2 and dropping within mins to 13.15 , IgnOn
Ipsu=0A I didn't believe my eyes, I checked Ibatt=23A!!! It was almost like a switch, with Vpsu being lower than Vbatt, the battery was supplying most of the current. Ipsu sometimes fluctuated but still less than few amps.

Info 2: surface voltage exists even with LifePO4 batteries, and under load after some time, the battery will show nominal voltage.
Experiment 2: charge and measure Vocv, depending on your lithium charger it will be 13.6 to 14V. As soon as a load is applied (ignition on, lights), Vccv will drop quick to nominal voltage.

Info 3: Our LifePo4 batteries discharge voltage are very flat, 70-80% of the stored energy remains in the flat voltage profile
Experiment 3: evident by constantly monitoring Vccv and remaining quasi flat for few hours!

NOTE: How to calculate voltage drop (Vdrop) due to cable resistance, Relevant posts #59 and #84: Vdrop=R.I=(2*3.5ft*0.0006Ohm)*(23-26)~0.1V
NOTE: It took close to 10 hours to recharge fully the battery with the CTEK lithium charger.

[aboulfad]

Pic 1: my concoction to monitor remotely my multimeter using an old iphone4 with AirBeam app
Pic 2: current with PSU connected, ignition ON, fan, low beams
Pic 3: current at battery + in trunk

[packetpilot]

Originally Posted by aboulfad

I can't recall if I tried that, few weeks ago, it'd stay on for like an hour, but now post full 58.3, I tried different light setting and the ignition would switch off after 15mn or so. What's the longest you kept your ignition on? Position one clock is parking lights, maybe I'll try that for my long run...

I'm sure I've had my car's ignition on for than 15min in 'standlicht' mode (parklicht, or parking lamps, in the european sense of the term, is for the below, despite these being called "parking lights" in NA), before and after my 58.3 update.

[stealth70mde]

thisisdave: This is a beautiful picture.

Besides, I have seen parking lights on one side like this on a Merc E class this winter when a neighbour parks it on the street. I found that it was cool and didn't know our BMWs could do a similar thing...

I guess I have to finish reading the manual...

[aboulfad]

Originally Posted by thisisdave

I'm sure I've had my car's ignition on for than 15min in 'standlicht' mode (parklicht, or parking lamps, in the european sense of the term, is for the below, despite these being called "parking lights" in NA), before and after my 58.3 update. ...

The pic you show corresponds to "Roadside Parking Lights", see pic from manual, those can ONLY be activated with ignition OFF. Those as in your pic illuminate the halos, while the parking lights illuminate the fugly Amber on my Xenon car. ( stealth70mde see pic below for instructions)

In your earlier post, you mentionned Parking Lights (position 13:00), and those are the ones we are discussing. Their behaviour in order to keep ignition ON with engine off is a bit fuzzy. See all attached pics from manual. At one point in time, I had the ignition ON (engine off) for more than an hour with some combination of various operations but I can't recall how.

Recently when I was doing all the experiments above and coding, I tried with the Parking Lights , but the ignition would go OFF even with the light switch on position 13:00.

[aboulfad]

Pics for lighting... Come to think about it, this should be in my other thread

[terahertz]

I remember reading somewhere that the newer firmware levels turn off after 15 minutes while the older ones didn't if you set the lights a certain way. If you start the tal execution, it won't turn off. Let me see if I can find where I read that.

Edit: It was in the flashing pdf:

Appendix B
On new cars (7/2014 F30) the ignition will not stay on even if you turn light switch. After 15 minutes or so
ignition will turn off. This is not a problem though, just remember to cycle ignition before svt soll
calculation and before starting TAL processing (actual flash). Ignition will stay on as long as the TAL is
running.

[aboulfad]

Originally Posted by terahertz

I remember reading somewhere that the newer firmware levels turn off after 15 minutes while the older ones didn't if you set the lights a certain way. If you start the tal execution, it won't turn off. Let me see if I can find where I read that.

oh even if this is partially true, then I am not crazy but that would be a bummer if its true, because then during long coding sessions, you'd need to recycle the ignition and hope that the PC doesn't loose connectivity ...

[terahertz]

You aren't crazy. I added the info to the post above.

[tom @ eas]

Originally Posted by aboulfad

oh even if this is partially true, then I am not crazy but that would be a bummer if its true, because then during long coding sessions, you'd need to recycle the ignition and hope that the PC doesn't loose connectivity ...

Fastening driver's side seatbelt during programming process will prevent this from happening.