coil on spark, bench setup: noise issues

[puff]

it was cheap clone of saleae. the outputs were:
01
23
45... I connected those wires to #0 and #2. channels #1&#3 were empty. I connected these wires directly to the discovery board. Brown, orange and yellow channels weren't connected to anywhere. But I believe that noise of unconnected inputs is okay.

[kb1gtt]

It appears that the pulses keep changing the their pulse width for some reason. Perhaps analog issues. I recall the console can allow a test mode which should remove this kind of analog issue, as it would hold a fixed pulse rate and fixed pulse width. Can you get it started with the console and see if it keeps running after you are disconnected? Getting a consistent pulse train will be a first step in tracking down noise issues.

[puff]

I can't do anything with the console for some reason. It just keeps disconnecting. Can I do the same with tuner studio?

[puff]

weird stuff. on the dashboard I could only see the voltmeter and needles moving. I disabled all the input pins, saved configs, but still the voltmeter is alive. I then rebooted the board, and the CLT also resurrected. Is it a bug?

[kb1gtt]

Using the console with out spark still fails? If that is true then I think you have other issues to deal with. Do you have a discovery board that has not been connected to the spark plug jig?

[kb1gtt]

I see some noise spikes that would be kind of expected if you have some energy coupling where it shouldn't couple. These land on specific transitions points.

Noise1.PNG (10.87 KiB) Viewed 14160 times

However the below spikes are less obvious where they are coming from. Perhaps they correlate to the dwell from this ignition module.

Noise2.PNG (10.89 KiB) Viewed 14160 times

From where I'm sitting I simply don't have enough data to be able to track down the issue.

[puff]

Yes. Using devconsole I am constantly disconnecting from it. With tunedstudio everything is fine (but for yesterday, when I couldn't make it save config. Although, I hadn't updated the project from early September, but flashed the new firmware yesterday (to check if reconcile was fixed) - may be that was the reason tunedstudio didn't let me save the changes..
What other data you need? If I only knew for sure which pins need to be and might be safely grounded. .. plus I need to figure out at least what's up with saving configs..

[puff]

No, I don't have an extra discovery board.

[kb1gtt]

Does your discovery need a firmware update? In ST-LINK Utility, you should have a pull down "ST-LINK" --> Firmware Update.

Can you check that and see if you should have a new firmware installed. This will update the firmware in the discoveries programming chip, it does not change the firmware in the STM32F4 chip. Just that other littler chip. I'm not sure, but this might make the console play nice. Hmmm, perhaps you need to install that jumper. I seem to recall that with out the Frankenso board with it's external FTDI chip, you need to specify the serial stream is on the proper channel. Perhaps that's why you can't get the console to work.

[puff]

I just updated the firmware and it didn't help Could you please be more specific, which exactly jumper should I have? You mean that which was used previously to flip the ports between devconsole and tunerstudio? As russian explaine, it is not required any longer...

[kb1gtt]

I was thinking the jumper was no longer required. It's been so long sense I looked at it and I've got both ports, so I wasn't remembering if it was removed by hardware or software. Not being able to connect the console is certainly a red flag issue. Are you trying this in windows? Perhaps you are trying Linux. Do you have any ideas about why it can't connect to the console when it's just the discovery and not other stuff connected?

[puff]

osx. I would say this happened with the upgrade to El Capitan (not sure though). But which is really strange is that Tuner Studio works fine (despite using the same? java framework for those conections?). I'll give it a try with the obsolete version of osx soon.

[kb1gtt]

I know lsusb has issued with windows and Ubuntu where for some reason they decided there was some kind of security issue. The result is that the communications thread needs to be run with super user permissions. AKA in Ubuntu pretty much any program that uses a serial interface needs to be launched from sudo and in windows (especially x64) it needs to be launched with the "Run as administrator" option. I understand that programmers who know about this issue can some how make installers that deal with udev rules, or something similar on windows. Such that not all lsusb based programs need super user permissions, but it's a very common issue, and I understand the fix in windows is different than the fix in Linux. So it's common it works in one and not the other.

Perhaps you are experiencing the same permissions issue with this JAVA serial implementation and osx. I'm grasping at straws, but who knows perhaps that's the problem.

[puff]

I am aware of that lsusb problem, but it seems lsusb is not used in osx.
i tried it with sudo, tried it using a newly created user account. will try it on another machine next week. it could be so that my discovery board is faulty. but then why tuner studio works fine then? it could be the serial library used in devconsole and not used in tuner studio. or that ring buffer (but that is unlikely)

[puff]

one more quesion.
I have a lukey soldering station with an integrated power supply (12V 2A). when I am using it to power up my ignition module - it doesn't work (probably, because of the shortage protecion of the power supply -- the amermeter needle jumps and the module constantly turns on and off - or is that because of the noise on this SMPS?.
but when I power the module from the car battery charger (which is basically a transformer with a rectifying bridge) - it works fine, and the ampermeter shows less than 0.5A current consumption.

[kb1gtt]

The coils typically have a 10A to 20A current draw when they are charging up. However you typically charge up for very short periods of time, such that your average draw is low, but your short term draw's are much higher. This is what is probably causing the 2A supply to have issues. It sounds like it can not provide the 10A to 20A needed in those short periods of time.

[puff]

So the fuse I need for that circuit (for 2 such modules) should be rated for 20A-40A?
I am still thinking of how to wire up everything: should I have 8 or 16 wires to my injectors from the ecu case? which fuse should I use for the ecu? which gauge wire should I use to power it up? Should I have relays inside the ECU, or it's more preferable to put them outside?

[kb1gtt]

Several questions, perhaps not related to spark coil stuff.

Fuses have trip curves, they can sustain much larger currents for short periods of time. The rated current is what it can handle for a prolonged time at a specific temperature. See this trip curve, for 1 second, the 2A fuse can sustain 3A, and for 0.1S it can sustain 8A.

fuse_trip_curve_blade.PNG (72.31 KiB) Viewed 14026 times

Also there is a fatigue issue known as I2t (That's amps squared time) which is a common reason for failures after several cycles. See page of the below link. If you want to properly select a fuse, you really need a scope with a current measurement probe. If you do not have such a scope, make sure to have extra fuses on hand just in case it blows a fuse.

http://www.littelfuse.com/~/media/automotive/catalogs/littelfuse_fuseology.pdf

For frankenso I suggest it uses a typical blade style. From page 9 of the schematic it suggests to use a 1.5A fuse. However that was for the one mounted on the PCB, which generally has a small wire installed. The below 2A would work for an external fuse.

http://www.littelfuse.com/~/media/automotive/datasheets/fuses/passenger-car-and-commercial-vehicle/blade-fuses/littelfuse_mini_datasheet.pdf

As for wire size, that depends largely on the upstream fuse you select. The injectors are going to draw about 1A, so can technically get by with tiny stuff like 24AWG or 22AWG. However I would suggest following the NFPA79's suggestion to use a min of 18AWG, as smaller is easily mechanically damaged. In terms of electrical sizing this is very much over sized, which is only suggested to get more rugged mechanical properties. If you use
18AWG, then I would suggest the wire be protected with no larger than 15A fuse. If you use
20AWG, I would suggest no larger than a 10A fuse, and if you use
22AWG I would suggest no larger than a 7A fuse, and if you use
24AWG use no larger than 3A fuse.
Those fuse sizes are based on this table for chassis wiring. http://www.powerstream.com/Wire_Size.htm

Relays outside the ECU are a good idea. Relays can generate a bunch of heat as the contacts burn in. It's a good idea to keep that potential heat away from the ECU.

16 wires for 8 injectors is a good idea, as it allows you to choose were the GND wires run. That's 2 wires per injector, so one wire for + and one wire for the - of the injector. This can be a handy option when trying to steer noise away from things. Alternatively the negative side of the injector can be connected to the engine frame and ground through the engines ground strap. I generally prefer to avoid using the engine's ground strap, as it causes problems when it fails. I would generally run the injector wires to the battery, and I would twist them with the injector + wires. When the wires are together like this, the - current has a tendency to conduct in the wire it's twisted with. This minimizes the antenna and minimizes the RF emissions from the injectors.

[puff]

as far as i remember, injectors have common plus (probably from the relay located somewhere close to the ecu), and the ecu switches their negative side. So probably one relay should be used to power injectors, another - for a fuel pump and the third one - for ignition modules. Trying to imagine how to solder these 8 wires to a single point on the relay.

1A for all 8 injectors? Besides, relays also draw some current.

[kb1gtt]

You are correct, I had it flipped in my head. + to injectors and Frankenso is low side drive.

Each injector is going to draw about 1A, so 8 injectors would draw about 8A perhaps 10A. Using some kind of wiring distribution block can allow multiple connections to one relay point. Those cost a pinch of $ though. If you hare looking for low cost you could stack several ring terminals on a single post, then one of the wires heads back to the relay, while the other go to the injectors.

This one includes fuses
https://powerwerx.com/atc-ato-blade-fuse-block-10circuit?gclid=cju0scbqrc8cfygmhgodr-afdw

http://www.ebay.com/itm/4-6-8-POS-POWER-DISTRIBUTION-BLOCK-COVER-LEAD-WIRE-CAR-TRUCK-MOTORCYCLE-/331769773125?hash=item4d3f04fc45:g:7~gAAOxyaURTgOe9&vxp=mtr

http://www.homedepot.com/p/SPT-8-Way-Terminal-Block-Bus-Bar-15-TB01/206639364?cm_mmc=CJ-_-1319015-_-10368321&AID=10368321&PID=1319015&SID=1100381112632&gclsrc=ds&gclid=CNmXuuzprc8CFVTr6QodiRgBag&cj=true

[puff]

i got the idea.
just wondering, why you say that 1.5Amp fuse should be sufficient for frankenstein, while it switches in total 8Amp of injectors. Is that because of the short duration of those inj pulses?

[kb1gtt]

See below link for a better understanding of how the fuses should be wired. I once created this system schematic for the Frankenstein board.

http://rusefi.com/forum/viewtopic.php?f=3&t=360&p=7540#p7540

[kb1gtt]

I just updated and added wire information here
http://rusefi.com/wiki/index.php?title=Manual:Hardware:User/en#Route_the_wires_and_harnesses

[puff]

what are the inj pulses at full throttle? batched or simultaneous injection. During this short period the current through frankenstein would rise to 4A-8A, so the fuses should be slow enough...
I purchased these connectors:
https://ru.aliexpress.com/item/GX16-8-pins-16mm-aviation-connector-female-plug-male-socket/32299454607.html?spm=2114.03010208.3.381.5ESkuS&ws_ab_test=searchweb201556_8,searchweb201602_3_10039_10057_10048_10047_10056_10055_10037_10049_10059_10033_10046_10058_10032_10045_10017_10060_10061_10062_10064,searchweb201603_2&btsid=dda3b064-e5cb-413e-b992-71189ac43ee9
at the bottom there is a table, mentioning some 'soldre inter diameter' of 1.15mm for my connectors.
18 AWG is just 1.0237, but it seems to me it would be too thick for it. thinking of some teflon wires.

[kb1gtt]

That connector is typically for signal level, not power levels. I do not know how well it will function for power level signals, or what it's contact resistance will do over time. Keep in mind that a contact resistance of 0.1ohms can cause a tank circuit which can cause your injectors to see hundreds of volts when being powered via 12V. Also I suggest connectors with a latching feature. This helps ensure you have full penetration of the connector. Also I would suggest you get a connector where you can solder the pin to the wire, then install it into the shell. These cup solders joints tend to damage the shell, which causes contact resistance issues.