[info] Mary Shelley's Frankenso

[OrchardPerformance]

As there is no real guide to putting together the Frankenso board in kit form I thought I would start a project thread for my kit build which can later be edited into a Wiki page for the board assembly showing the pitfalls of being brave and cheap and building the kit yourself.
I will add to this thread over the next few weeks as I cudgel my way through.

So this thread is "Mary Shelley's Frankenso - The story of how I built a monster without any uncontrolled electrical discharges and hopefully no villagers with pitch forks".

I ordered my kit form Frankenso earlier today and the STM32F407VG Discovery has just arrived from Ebay. I have a couple of package pictures and some side by side with the ArduinoMega2560 for comparison.

Challenge 1 was getting it out the packaging, they are really well wedged in there!

The real first challenge has actually been the USB connection on the STM32 Discovery board, it is a USB Mini B, which is something I don't think I have used since about 2012 and the Discovery does not come with a USB cable so it is worth bearing in mind when ordering one that you will need both a USB micro B cable and a USB Mini B cable which not everyone will have to hand.
After a bit of searching I found one and hooked the Discovery up.
This strikes me as an odd choice, do we know why there is a mismatch between connector types?
Is this also going to mean that it will be best for me to carry both kinds of lead in the car at all times?

[AndreyB]

A build write-up is very much needed so I hope this would become a sticky thread once it has amazing content

This strikes me as an odd choice, do we know why there is a mismatch between connector types?
Is this also going to mean that it will be best for me to carry both kinds of lead in the car at all times?

I can see some safety and usability benefits - it's clear which one is for flashing the chip and power (mini) vs which one is for talking to the firmware via some tuning software (micro)

[OrchardPerformance]

I want to make an update to this as its been a week but from the tracking details my board is still stuck in customs (@Russian - I must post the tracking details, they are hilarious, it crossed the Hudson twice going to queens to end up flying from the airport less than 8 miles from where you live ).

What I will do instead is later today do a round up of the kit and equipment I am planning to use to put this together and a bit on the practice boards I picked up to get comfortable with SMD soldering for the guys that are begging from nothing.

[OrchardPerformance]

I said I would do a kit and materials round up so I had better get on with it as I just discovered my board has now cleared UK customs.
Also this is going to be take 2 as I just lost a ~1000 word post due to being signed out.

First off solder and flux:
https://www.rapidonline.com/warton-meta ... ste-547774
https://www.rapidonline.com/warton-meta ... le-87-0720
These two have to be the cleanest and most user friendly solder paste and flux that I have used in nearly 14 years of hobby work.
It may be a little tough for those in the US to get but for the euro guys it is well worth it (well until we crash out the single market right? )
The solder paste is super easy to tin pads on the boards and leaves no mess, it seems to have a pretty high surface tension as well which means spikes and bridges are reduced.
The flux is great, it is the only decent flux I have used with no real nasty fume that is still able to tin the dirty aging wires on an older car.
They supply direct customers with a kuretaki watercolour brush pen for filling with the flux and that makes it real easy and neat to flux the pads and small components as you can just paint it on.
Grab component with tweezers, paint the pads and solder.

My normal goto soldering Iron for vehicle work is an old gas powered Iroda soldering iron but for fine SMD work it is like trimming your beard with a chainsaw.
As the little electric indoor iron was not going to cut it I picked up a W.E.P. 952D+ 2 in 1 solder station.
https://www.amazon.co.uk/WEP-952D-SAFE- ... B01MQRLA4L
This is a temperature controlled iron and hot air rework station. The iron comes with multiple tips and so far has proven pretty good, I barely need to use the hot air for 0802 sized parts like the Frankenso uses and when I have it has been possible to remove and replace things with no damage quite easily.
The only thing I could say might be an issue is I think I detected some voltage on the iron tip, I need to look into this.

Finally, with all this new kit and nothing to solder I thought it would be wise to pick up some practice boards.
https://fluxworkshop.com/collections/i- ... unsoldered
I picked up 3 of these for ~£13 (~$17), they use the same 0802 SMDs on the flashing LED side as the Frankenso uses and unless you can read Chinese then all you get to work from is a schematic, perfect practice for the Frankenso Kit.
They also have some smaller and more complex parts on the back with test pads so you can get some finer practice in.

I did also pull apart an old TV box and have a play with the rework station, turns out one of the most common TV boxes uses some MAX IC chips that are the same package sizes at the Frankenso so it was nice to be able to easily remove and replace these with the kit above without any real trouble, even managed to lift and re-ball a BGA Samsung flash chip, no idea if it will ever works again but I might have a broken USB stick to try it on...

My board should be hear next week, just in time for some leave from work so I can dig into the hardware a little bit soon.

[OrchardPerformance]

My board arrived on Friday (after some battles with parcel force, apparently you cant pay customs the same day as it clears customs and I mysteriously asked for a later delivery. )

I have started populating the board so I will start dropping details into this thread, for now each post is going to be the brief notes that I am keeping as I go but I will add photos of the stages to the posts once I am further along. There may also be the odd question mark in here which hopefully we can resolve as we go along, this is going to be warts and all.

Before starting order 0805 package 10uF 10v and 4.7uF 50v capacitors as these are not in the basic SMT kit.

[Some photos of the kit as delivered. Note - Black plug is from a 316 ECU and not part of the kit. ]









I went for the MAX9926 circuit first as it is supposed to be the hardest and if I was going to screw this up I may as well only screw up a board and 1 chip.
I can confirm it is the hardest.

put max9926 down first, pins are very easy to bridge but several pins are actually connected to ground plane.
left pins 11,12,13 bridged on mine as they are all connected to ground and thus the bridges make no difference?
https://datasheets.maximintegrated.com/ ... AX9927.pdf

[Paste and tinning. Note the marks where I tried to seperate what I thought were bridged pins and they had continuity only to spot later they were actually connected via the traces ]




C101 and C102 - marked 102 (10*100 or 10+"00")
https://www.radio-electronics.com/info/ ... rkings.php
https://circuitdigest.com/calculators/c ... calculator

R111 + R112 = 4K7 strip = 4700 ohms = (472) on resistor - might not be needed unless signal is bad, BMW M50TU = 1x VR 1x Hall

R113 + R114 = 10k ohms (103)

D801 + 802 LED - Line to -ve little dash points to positive, be careful of LED orientation, check ground with meter if needed

C103 - 10 uF (added after photos due to not being in the kit, see note about ordering these at the start).



R102 R103 R104 R106 VR input primary - 4700 ohm (472)

R107 R108 R109 R110 VR input secondary - 5600 ohm (562)

W1003 - do i need to put 0 ohm res in the JMP?

Need to add pull up or down to cam VR to suit the BMW M50TU cam sensor

Also one other question, some of these chips have bare pads on the board to solder down pads under the chips, obviously for heat reasons but so far none of the chips have pads on the bottom, they are all plastic so, uhhh, no can solder?
I have left them solder free unless the chip has a contact pad to solder down, I assume this is the correct choice?

[OrchardPerformance]

Next I did the 8 channel high/low drivers
This meant that I had to do a few additional parts from another circuit due to limited access later on.

U601 U603 = TC4427 https://www.mouser.com/ds/2/268/20001422G-967549.pdf dot = pin 1, check orientation on board by ID pin 3 as ground using multi meter.

C601 C603 = 0.1uF (104) - Make life easy for yourself and do these before U603 or access is tricky



R605 R608 = 100 ohm (101)

LD604 and 3 others that are badly marked = LED, beware R604 is backwards vs the other 3 LEDs actually LD603 LD607 LD608

At this point the KICAD files were needed to read the IDs in this area



R606 R607 = 3300 ohm (332)

R614 = 3300 ohm (332) - very hard to read on board and KICAD

R615 = 3300 ohm (332)

R613 R616 = 100 ohm (101)



R602 R603 R610 R6111 = 3300Ohm (332)

LD601 R601 LD602 R602 LD603 R603 LD604 R604 = LED alternating with 100 ohm resistor (101)



C602 C604 = 1.0uF (104)

R625 = 22 ohm (22R)

R626 = 1000 ohm (102)

R627 = 22 ohm (22R)

U602 U604 = TC4427



Populate the few remaining spaces around the Hi/Low drivers to finish that area off.

R624 R628 = 1000 ohm (102)

R617 R619 = 22 ohm (22R)

R618 R620 = 1000 ohm (102)

[kb1gtt]

W1003 depends on the setup. Hall vs VR. If hall most are pull-up some are pull down. You need to know what you have, them you can install the proper jumpers.

Yes no solder unless you have the solder pad(s). Via's under the chip are to minimize thermal issues.

[OrchardPerformance]

Thanks, thought so.

Just to check, the parts marked DNP are not to be populated by the board manufacturers, I assume this also applies to us home building?

[OrchardPerformance]

Next off put the knock sensor interface area in, few substitution questions on this one.

U161 - D161 - TPIC8101 - Knock sensor interface
http://www.ti.com/lit/ds/symlink/tpic8101.pdf



R172 R172 - Do not populate

R167 R168 R163 R164 = 39000 ohms (393)

C612 C613 = 3300pF (332)

R161 R162 = 1 mohm - Do not populate

C168 = 0.022uF (223)

C165 = 0.1uF (104) - Very close to the Via, however the via appears to be the 5v for the cap anyway.

C501 = 4.7uF (475) Not in kit, on order but easier to do at this point

R170 = 5600 ohms - not on the swap out list, is this value ok instead of 5000ohms?

R171 = DNP

C169 = DNP

C167 = 22pf (220) - not on the swap out list, is this value ok instead of 24pF?

R166 = 2.2m ohms (225)

R177 = 680 ohms (6800) - - not on the swap out list, is this value ok instead of 820 ohms?

R174 = DNP

[OrchardPerformance]

Just gone through the power supply area deliberately leaving the solder through parts until later. One thing to watch here is that it is easy to slightly scorch the LEDs with the hot air gun, a couple of mine have gone a bit yellow where I got them a bit warm so I might have to change them out.

Power supply area:

U1001 = LM2596 http://www.ti.com/lit/ds/symlink/lm2596.pdf - This needs its heat sink soldered down.
Quite difficult to get down with such a large pad, tinned the pad, used hot air + iron to fully melt, placed U1001 down and then continued to heat with hot air until stuck down.

D1004 = Huge D4 diode - Same trick but easier

C1002 C1003 + Big yellow Caps, board should be nicely warm from the last 2 parts to make this easy but hot air still needed as limited access to pads.
Lines on Caps are +ve

Q1002 = Large mosfet - same as large diode



R1002 = 1000 ohms

D1005 = LED

Q1001 = MOSFET

D1001 = tiny SMD diode - Fine line still on the end towards the big Capacitor, non line side on the Q1001 trace side.

R1001 = 3900 ohms (392)

Leaving solder through electrolytic capacitors for later on.

[OrchardPerformance]

To finish off the SMDs on this side of the board populate area below hi/low drivers

R622 = 1000 ohms (102)

R621 R623 = 22ohms (220)

R176 R175 = 10k ohms (103)

This leaves us with only jumpers or solder through components to do on this left hand side of the top half.

[OrchardPerformance]

12 channel injector drivers next

R408 R409 = 1k ohm (102)

R401 R410 = 22 ohm (220)

Q401 -> Q404 = https://www.st.com/resource/en/datasheet/vnd14nv04.pdf Chamfer on package IDs the orentation
Pin 1 = LH chamfer side and pin 1 goes to resistors on small trace
These are damn tough, the iron looses too much heat to the ground plane and the hot air wants to blow them away all the time.
Managed by putting the non ground pins down first and then warming the ground plane with hot air while using a large tip in the iron on the pins.
Suggest maybe a spur of the ground plane to reduce the heat sink effect on these pins to make assemby easier?

D413 = Diode

D414 = LED



R404 R405 = 1k ohms (102)

R403 R406 = 22 ohm (220)

R427 = 1k ohms (102)



Q405 -> Q412 = Sams as Q401 -> Q404

R413 R417 R416 R421 = 1k ohms (102) - use hot air due to ground plane

R414 R418 R415 R419 = 22 ohm (220)

R412 R421 R424 = 1k ohms (102)

R414 R422 R432 = 22 ohms (220)

R426 = 22 ohms (220)

R425 = 1k ohms (102)



Something to note here is there was a minor defect on the board that stopped Q410 from working, the trace from R422 to Q410 was cut just where the trace meets the solder pad for the resistor.
If you look closely at the picture you can see a white line to the left of the R422, this is actually where the PCB trace is broken.
It is easily fixed by tracking some solder across from the resistor to the leg of the IC.

[kb1gtt]

DNP = Do No Populate. You typically don't want this but it often allows potential features. So you may want them.

It's great to see your play by play assembly. I think it's a great start to a wiki page.

[OrchardPerformance]

Pretty much what I thought, I will build this one up "stock" without any of them and then if needed enable things later.

Trying to go through it in the most logical order that allows the easiest access to all the parts and flag any issues or missing bits so it can be wikified later on.

I have been snapping photos which I will go back and add to the posts later but need to upload and then edit them first.

I'm assuming any substitutions and comments made so far are ok in terms of build, and resistor changes are nearly within tolerance of a cheap resistor anyway.

----------------------------

Next up CAN controller.

U471 = SN65HVD230 http://www.ti.com/lit/ds/symlink/sn65hvd230.pdf chamfer side down, line indicates pin 1.

C472 = 4.7uF - Not in kit

C471 = 0.1uF (104)

R427 = 1k ohms (102)

R473 = 120 ohms (121)

-----------------------

Begining on other side of board

starting with FT232RL

U531 = FT232RL http://www.ftdichip.com/Support/Documen ... FT232R.pdf

D701 D702 = LED

R701 R702 R703 = 1k ohms (102)

C701 702 = 0.1uF (104)

Out of bits to complete this - need to order diodes and LEDs

[AndreyB]

Out of bits to complete this - need to order diodes and LEDs

Which diodes?
As for LEDs I guess I need to start putting a bit more LEDs into the kit. At the moment it's 15 while we need more like 12+8+x so maybe 25 or 30 to be on the safe side.

[OrchardPerformance]

I could be being blind but I have populated the mosfet for the power supply and the 12 for the analogue inputs and cannot see any more in the kit and there seems to be a couple of spots one could go still, there's 2 spots next to the FTDI chip but I assume that is optional for a built in STM32 and I have only put it on as it was in the kit.

The LEDs is mostly my fault, I lost one, broke 2 and cooked one with the hot air but you might be right about the number as there are quite a few more slots than I have LEDs for.
I have a load of red 0805 red LEDs on the practice boards I did so I might just whip some off those with the hot air.

At the moment the only things really missing are the 4.7uF and 10uF caps and a few resistors where I have substituted instead.

[AndreyB]

there seems to be a couple of spots one could go still, there's 2 spots next to the FTDI chip but I assume that is optional for a built in STM32 and I have only put it on as it was in the kit.

pictures or even better component reference #s? Which package/kind of the diodes are we talking about just to be sure we are on the same page?

[OrchardPerformance]

Q701 and Q703 next to the FTDI chip.

[AndreyB]

Q701 is a mosfet. There is no Q703 - there is D703, that one is a diode.

[OrchardPerformance]

Q701 is a mosfet. There is no Q703 - there is D703, that one is a diode.

Ok, I thought the Q was a mosfet which explains why I have an extra mosfet.
Will still have to grab another diode from somewhere though.

The mosfet bag had 2 mosfets but was marked 1x
Diode wise - one bag of 12x which has filled out the analog spaces but I will have to go digging for another unless there is a few more hiding in the kit somewhere.

[OrchardPerformance]

Analog inputs

U201 U202 U203 = MCP6004 http://ww1.microchip.com/downloads/en/D ... 21733h.pdf

D210 -> D320 = 12 Diodes

C211 C251 C291 = 0.1uF (104)

R224 R225 R214 R215 R234 R235 R244 R245 R264 R265 R254 R255 R274 R275 R284 R285 R304 R305 R294 R295 R314 R315 R324 R325 = all 1500 ohms (152)



R213 R223 R233 R243 R253 R263 R273 R283 R293 R303 R313 R323 = 100 ohms (101) Pretty much every second pad going across the strip between the diodes and ICs

C210 C220 C240 = 1uF (105)

C230 C250 C260 C270 C280 C290 C300 C310 = 0.01uF (103)

C320 = 0.1uF (104)

Due to staggering of resistors the easiest way to do this is to place 1 at a time folloing the Kicad

R212 R222 R232 R242 R252 R262 R260 R272 R282 R280 R292 R302 R300 R312 R322 = 10k ohms (103)

R210 R220 = 2700 ohm resistors swapped for 3300 ohms. (332)

R230 R240 R250 R270 = DNP

R290 R310 = 1k (102)

R230 = 3900 ohms (392)

next row

R211 R221 R261 R281 R291 R01 R311 = DNP

R231 R241 R251 R271 = 560k ohms (564)

R321 = 10k ohms (103)

R430 R431 = 1k ohms (102)

D421 D423 D425 = LED

R432 = 1k ohms (102)

D427 = LED

D422 D424 D428 = Diodes

R433 = 1k ohms (102) - Hard to get in by the LED

D426 D432 = Diodes

D429 D431 = LED

D430 = Diode

R435 = 1k ohms (102) - Now out of 1k resistors, so hit ebay again and got 200pc for ~£3 delivered.

[OrchardPerformance]

Updating this with the progress over the last few days:

Continued after 1k Resistors turned up, note my 102 resistors are marked non standard 1001 althoguht result of marking is the same.

R434 R436 = 1k (102)

D433 = LED

D436 = Diode

R437 = 1k (102)

D435 = LED

D436 = Diode



Jump to other side of analog input area

D415 = LED

R438 = 1k (102)

D419 = LED

R429 R428 = 1k (102)

D416 D420 = Diodes

D417 = LED

D418 = Diode



--------------------

Jumping to other parts:

pot

RV101 = Potentiometer

R99 = 1k (102)

R630 R632 = 1k (102)

R629 R631 = 22R (220)



sd card

C23 = 1uF (105)

C24 = 0.1uF (104)



SD card holder


-------------------

Odds and ends rear side

C1302 = 100nF (104)

C1402 = 100nF (104)

C1202 = 100nF (104)

C1102 = 100nF (104)

R1003 = 100k ohms (104)

Battery mount - Watch for polarity



Odds and ends for front side

C357 = 0.1uF (104)

C358 = 4.7uF (475)



--------------------

Solder through components

L1001 L1002 L1003 = Inductors, no polarity?

C1001 = Big cap (not solder through, sue me)

D1002 = Small solder through diode

D1003 = Mid solder through diode

U5 = button

D12 D13 = Solder through cam sensor Diodes

Fuse holder - No polarity

Crystal Oscillator - No polarity

Big 64 pin ECU connector



Discovery board raisers

LCD screen raisers

12v switch jumper

----------------------------

[OrchardPerformance]

Edit - Revoked this post, Think it is better to separate the build of the board and the specific bits for the vehicle install as things got a little more complex once I had got it into the car.



This is how I have mine wired up and jumpered for the BMW, I have picked up some Toyota plugs, repined them to match the MX5 plug and play layout and made a loom adapter from those plugs to the 88 pin BMW plug.
While this seems a bit convoluted it means my frankenso board is a standard setup configured for the MX5 according to the current wiki page.

------------------------

[OrchardPerformance]

In this post OP finally begins to deliver...

Going to start dropping photos onto some of the posts to highlight what was soldered into what areas.
I have spotted some of the photos are not idea and I might edit them a bit to square them up or retake a few where practical, possibly some could use a snip of the Kicad to make things totally clear.

Also, you can see my soldering improve as the photos go on...

[kb1gtt]

Nice to see the progress. I would suggest doing the U1001 and other similar large heat sink chips sooner rather than later. You often have to hold your iron on those for a long time to heat up the copper plane. This commonly causes other chips solder to melt. Doing these larger chips first is a good idea. As well, once you heat up the copper plane enough to get those chips to solder, it takes allot longer than other soldered devices until it goes solid again. that copper plane holes a bunch of heat and it's internal, so it doesn't cool off very fast.

[OrchardPerformance]

To be honest the ECU is pretty complete, I was going to finish off a bit of the adapter loom at the weekend and try to get some noise made but have not had time.

I agree on the bigger power supply parts, that copper plane does take a lot of heat, in the end I got them on using the iron and the heat gun at the same time but I did scorch the plastic on a couple of the LEDs a little though.

An amendment might be needed to put those on first and then work away from them when laying the resistors and LEDs near U601/U603

[AndreyB]

To be honest the ECU is pretty complete, I was going to finish off a bit of the adapter loom at the weekend and try to get some noise made but have not had time.

Did you ever make any noise?

[960]

A tip for you, and anyone else that worries about solderbridges. Haven't you heard about Desoldering Braid?

I have used that for as long I can remember, and it removes bridges and exessive tin from whatever.

I usually use the pull method for IC's.

You apply tin at the first pins, and just pull the iron over the row.

If you have applied a little too much, just remove it with braid, and pull again.

You can also use the fill/braid method, and just fill the whole row with as much tin it takes.

I usually do that with used IC's.

Just search youtube for the methods.

Good luck

[OrchardPerformance]

Thanks 960, its something I normally have to hand but at that time I didn't, I should use more patience and waited until I had some. However the bigger problem was me being unable to spot that the traces were connected elsewhere and thus would appear to be bridged even if they were not.

Russian - I didn't try the ECU on the car in the end, instead I hooked it all up to a simulator setup with LEDs fro the fuel and ignition outputs, variable resistors hooked up to simulate the CLT,TPS, IAT and MAF and the crank signal provided by Daqarta.

This let me find a few issues with the setup and essentially debug it before going anywhere near the car. On the plus side everything went pretty smoothly with that and I didn't make any major mistakes in the assembly.

One thing I did find is that I struggled to get the ECU to register a crank trigger pulse with the Daqarta setup and put it down to it being too small a signal. (only ~1v) I stimulated the ECU with the jumpers on the Discovery and did all my testing with that.

Bad news is the issue seems to persist once I moved the ECU to the car, I cannot get any crank trigger signal in the signal sniffer. I'm not sure if this is a problem with my settings or with the MAX chip.
I have a Speeduino extension board with the same max chip which I can get a signal from with the output from Daqarta.
I assume that without a response on the engine sniffer when cranking the car there is no signal of any kind getting to the STM?
Is there any procedure to debug this circuit?
I can play signals from a secondary sound card via an isolation transformer to replicate a crank sensor but I can only achieve ~1v when doing that, should that trigger the MAX chip?

[AndreyB]

I assume that without a response on the engine sniffer when cranking the car there is no signal of any kind getting to the STM?
Is there any procedure to debug this circuit?

In case of miata pretending to have Hall sensors or really having Hall sensors, you just GND the input wire and that should change the output. In the console you have triggerinfo with a counter, in TS you have "debug mode" for trigger which should event counter. I touch board input with a GND wire and usually each touch is registered as 10-20 events, that's part of my QA process for assembled boards.