chet

On the schematic in the very first post of this thread which says 7/15/06 rev 2, the resistor in the line coming from RXD appears to be 33K. I don't know anything about electronics, but each of the lines above that has a 3.3K. Is the bottom one supposed to be 3.3 as well, or is it 33? Like I say, I don't know anything about electronics but I'm going to try to build one of these. Also, has anybody already put together a layman's diy on building one of these in case certain things need to be pointed a particular direction or how to tell which lead from a transistor is what?

PDriver

It's a 33k resistor for current limitation. The other one are pullups to drive L and K to Vcc.

PDriver

PDriver

@maurice97C2S

Pin 4 is signal ground. nothing to do which OBD type is on board.
You have to update your cable to reach pin 3 or pin 7 with the
incoming lead on pin 7. can you open your obd-connector?

PDriver

maurice97C2S

Hi PDriver

I've got both ends cracked open .. the only connectors populated OBD end were 4, 5, 7, 15, 16 ...

On the 9 pin end 1-5 are connected to the circuit board, hard to trace, there seems to be nothing on 6-9, which doesn't seem to match up with Ray's circuit ..

Any ideas ?? I thought it must be pretty similar, but not sure what gives .. just a reminder, it waas sold for OBD2 use ...

Thanks for any assistance, cheers, Maurice

raycm

"certain things need to be pointed a particular direction or how to tell which lead from a transistor is what?"

http://www.fairchildsemi.com/ds/2N/2N3904.pdf

The pdf with transistor specs. Retrieved from a goole search.

If anyone has a schematic drawing app, your welcome to make a readable drawing.

Ray

cowtown

Here's a rough draft of my attempt at the V2 schematic - it may have errors.

I'm also working on a bare circuit board layout, because I'd like to get a board or two made just for the coolness factor. A cheap single-layer board should be enough.


EDIT - this schematic is obsolete, so I'm removing it. See below for newest --Colin

tj90

Cowtown, if you are making the board at a fab shop, have you cosidered taking orders from other R-listers to defer the costs..... I would be interested in a board...

raycm

cowtown, thanks for getting involved.

For a pcb I suggest a fuse for the +12 Volt line. To current limit any user misconnects, etc.
I'll take another look with a scope at the voltage levels of OBD side.

This circuit would be more robust with the Q3 OBD receive being replaced by a 2n3906
circuit. A guess is adding one transistor and a few resistors to existing circuit. I would
have to work out the specific parts.

Thanks again,

Ray

raycm

The pcb:
Transistor symbols are reversed. Long flat should face right (OBD).
RS232 pin 5 trace needs to connect to trace it crosses, 0Volts.
RS232 pin 2 trace doesn't connect to trace it crosses, 0volts.
The RS232 symbol is drawn not looking at solder pins. It is a female connector. Looking at the solder side pin one would be at the bottom where you show pin 5.

Ray

cowtown

Ray,
Thanks for taking the time to look at the PCB. I'm going to need some peer review because I don't have any background doing PCB work. I'm posting the latest that (I hope) incorporates your suggestions. This one's my attempt at making it a single layer board with no wire jumpers, and more compact footprint.

I don't quite understand your sentence about the DB9. I thought the square solder pad on the PCB indicated 1 (?). Maybe I'm still reversed there.

TJ90: I'm willing to get a batch made or provide the CAD file to anyone who wants it. I'm thinking I can siamese two of these together on one board and get a small run produced, but I don't know what it would cost yet.

Colin




EDIT - this schematic is obsolete, so I'm removing it. See below for newest --Colin

raycm

"I don't quite understand your sentence about the DB9. I thought the square solder pad on the PCB indicated 1 (?). Maybe I'm still reversed there."

I'm wrong. The pcb layout of the connector is correct. You might check it again.

If you give this schematic to a pcb designer along with a parts list she'ed verify. The connector has mounting tabs, the connector must stick out from the pcb far enough to clear other stuff on the back of the PC when plugged into the PC. The board should be sized to fit some box. Radio Shack has had small boxes. The worst problems in laying out pcb's are these mechanical issues. You can easily replace a trace with a wire but a connector that is actually mirror image of the pcb means a redesign. I'm saying the physical layout is most important.

I'm attaching a schematic revision. It adds another transistor. I tested and it works. This circuit allows better OBD pullup resistor values, and makes the OBD interface more robust. Also this circuit would allow assembly to the previous rev 2 circuit if some problem occurs.

Ray

Lorenfb

"This circuit would be more robust with the Q3 OBD receive being replaced by a 2n3906 circuit."

It's highly questionable about the value of added the PNP circuitry:

1. ODB3 load is same; ~ 33K, actually greater, i.e. 33K || 33K/ (3906 Beta).
2. Drive to the Q3 NPN is the same ~ 33K.
3. Real improvement comes from the 1K pullup versus 3.3K.

Best solution comes from:
1. Using a FET for Q3 (less parts), or
2. Using a NPN emitter follower to drive Q3 from Q2 (less parts
than the added PNP circuitry), or
3. Replacing Q3 with the 3906 being driven by Q2,
With a collector pulldown (~ 1K) on the new 3906 (Q3) collector
with a 10K base drive to the 3906 from the collector of Q2.
This results in better Q2 & Q3 rise times. Only possible problem
is inadequate "Lo" for RS232 signal because of 1k pulldown versus
active old NPN Q3 pulldown.

Additionally, better 12V supply filtering is needed:

1. 100 series resistor from ODBII +12 with a 14-15 volt zener to ground
at the capacitor side. This will prevent transients/excess voltage from entering
the DME or PC.
2. 10 uF cap used instead of .22 uF.

Notes:
1. Any NPN or PNP can be used since the forced Betas are less
than 10-15 versus actual Betas usually > 100; e.g. PN2222/PN2907, MPSA20/MPSA70.
2. PCB info indicates incorrect transistor orientation, i.e. view is
trace view of PCB. It's only correct if components are inserted from
the trace side. Even if the view is "thru the board" it's incorrect.
The standard pin orientation for transistors (not Euro types) is
E B C viewed from the flat side & bottom view. Thus, both PCB
views (trace & component) should be shown.

raycm

"It's highly questionable about the value of added the PNP circuitry:
1. ODB3 load is same; ~ 33K, actually greater, i.e. 33K || 33K/ (3906 Beta).
2. Drive to the Q3 NPN is the same ~ 33K.
3. Real improvement comes from the 1K pullup versus 3.3K."

The PNP addition allows the 1k pullup, because its base will switch at any voltage below ~6 Volts or whatever the resistor divider is set for. The previous circuit required the OBD signal to go below 0.5V. With 1k the DME and others no longer pull the data line to <1V. (Which is why I used 3.3K previously) The OBD suggested 510 ohm pullup resistor is worse in that regard. If you look at the diagnosis section of the 993 manual it shows the expected voltage waveforms of that setup.
The answer to your above comment is: Added circuitry is required to get the real improvement.

"Best solution comes from:
1. Using a FET for Q3 (less parts), or
2. Using a NPN emitter follower to drive Q3 from Q2 (less parts
than the added PNP circuitry), or
3. Replacing Q3 with the 3906 being driven by Q2,
With a collector pulldown (~ 1K) on the new 3906 (Q3) collector
with a 10K base drive to the 3906 from the collector of Q2.
This results in better Q2 & Q3 rise times. Only possible problem
is inadequate "Lo" for RS232 signal because of 1k pulldown versus
active old NPN Q3 pulldown."

You may well be correct that other active components may have some advantage. Frankly I have the transistors I used which makes it easy for me to build and test. A few resistors one way or the other doesn't matter. I obviously tried to limit the resistor values. Not using a PNP and adding another NPN would certainly be nice.

The RS232 low level is a definite concern. Spec is below -5Volts.. My Q3 does only ~ +0.3V. Most RS232 IC's are designed to recognize logic levels so it works. But a resistor pulldown means slower rise/fall times between the crucial 0.3V to +3V switching levels. My answer is a Collector/Emitter to 0V is best for receive data. And with my circuit, Q3 emitter and the 33K can be lifted from 0Volts and connected to a - V source. This would provide the correct +/-5 Volt levels. RTS and DTR are candidates. This change would rely on the scantool type app to keep RTS or DTR low at the proper times. But this custom mod allows a solution when someone finds their serial port needs +/- Volt levels. I should add the RS232 interface does not have a -V or +V pin.

"1. Any NPN or PNP can be used since the forced Betas are less
than 10-15 versus actual Betas usually > 100; e.g. PN2222/PN2907, MPSA20/MPSA70."
Yes, I agree. The resistor values and diode type are also not critical.

"2. PCB info indicates incorrect transistor orientation, i.e. view is
trace view of PCB. It's only correct if components are inserted from
the trace side. Even if the view is "thru the board" it's incorrect.
The standard pin orientation for transistors (not Euro types) is
E B C viewed from the flat side & bottom view. Thus, both PCB
views (trace & component) should be shown."

Yes cowtowns first layout had transistors backwards. Please check out his next layout. I also agree the layout can be confusing. I think I'm looking through the pcb from the component side. The silk screen will be on the component side and the traces on the solder side.

What about the RS232 connector? A right angle female connector soldered into the pcb may allow plugging the pcb into the back of a laptop without adding an RS232 cable. I believe we are stuck with a cable on the OBD side. I also suspect a RS232 female to female cable will cost as much as the interface pcb.

I believe this interface board will work with some OBDII apps. and can therefore be used on 1996 and later anythings. Some apps require buying their proprietary interface and some don't. This means one laptop, one interface, 2 apps and your all set. The 964 and early 1995 993 would need to have 2 auto interface connectors/ cables. Later 1995 993 use the OBDII connector. cowtown asked who can use this interface. 964 & "1995 993" with Dougs scantool can use it. Buy an OBDII app and 996, etc can use it. The nice thing with Porsches is later models may be able to use the Climate Control, etc trouble code reads.

Loren thanks for the input.
I wont quit my day job. :-) investing.

Ray

cowtown

EDIT - 8/18/06 - This is not the latest. See the end of the thread for updated schematics.

Lorenfb

"The previous circuit required the OBD signal to go below 0.5V. "

That's easily solved by JUST using a base-emitter resistor for Q3
which has been added:

Q3 Vbe (on) = 12 volts X Rbe / (Rbe + 33K) = ~ .55 volts

Thus, there's no need for the PNP. With the PNP, the threshold for the Q3 is
basically > = 6.0 volts at the PNP emitter. By using a 3.3K at the base of Q3,
the threshold (on) for Q3 is about 6.0 volts which is the same with the added
PNP. This lowers the effective Rbe of Q3 & thus lowers its switching times.

"My Q3 does only ~ +0.3V. Most RS232 IC's are designed to recognize logic levels so it works."

That's why a FET works better (lower on voltage) AND a higher gate threshold > 2.0 - 3.0 volts.
They're not expensive and in a TO92 package.

"And with my circuit, Q3 emitter and the 33K can be lifted from 0Volts and connected to a - V source."

A simple charge-pump circuit off of the TXD line generates the needed negative
voltage (3 - 5 volts).

"A right angle female connector soldered into the pcb may allow plugging the pcb into the back of a laptop without adding an RS232 cable"

That "asks" for a broken pcb long term, given the way people insert/remove cables/connectors.

Note: Fuse has limited/no protective utility given the circuit values,
i.e. a fuse doesn't stop transients.