[Ovmsdev] OVMS v3 First Board Layout

[HONDA S-2000]

I've used the MAXIM MAX1595 to produce 5V in many products. I don't know if that would help your situation, but I agree that sticking with the "MCP2551" is probably the best choice.

However, I just noticed that Microchip mark the MCP2551 as Not Recommended For New Designs. They're recommending the MCP2561 instead. The good news is that it is also a 5V part, so there should be none of the 3.3V versus 3.5V issues.

A buck converter from 15V to 5V could be quite efficient, but I don't know what you would do if it turns out to be noisy. I suppose that a passive RC filter bank could smooth things out if your buck converter is willing to produce the extra voltage that will be bled off in the passive filter. That's not as efficient as possible, but it might be as clean as an LM regulator and still more efficient than an LM.

By the way, if a chip needs 5V, then it usually requires at least 4.75V as a bare minimum. The MCP2551 requires at least 4.5V minimum. One thing to be aware of is that USB only guarantees 4.01V on the VBUS if you're on the far side of a passive hub, so I've never considered it safe to run a 5V chip directly off of the USB VBUS. Sure, it will probably work 99% of the time, but I don't like ignoring part of a specification. I mention all of this because I use the MAX1595 in USB Devices that include 5V chips, because I don't want to get caught with less than 4.75V in some situations. In that case, it might be easier to produce the 5V from the 3.3V supply, and then use the diode "or" circuit to feed the 3.3V regulator. The MAX1595 works by using a capacitive voltage doubler, then it bleeds that off. I've actually run the MAX1595 off the 3.3V instead of the VBUS just to avoid bleeding off the excess. 3.3V in becomes 6.6V internally, whereas the nominal 5V VBUS becomes 10V internally. I now use the MAX1595ETC50 in QFN12 packaging because it can handle the additional heat. The data sheet gives equations to calculate the wattage given your circuit needs.

Brian Willoughby
Sound Consulting


On Mar 29, 2017, at 11:18 PM, Mark Webb-Johnson <mark at webb-johnson.net> wrote:
> My concern is that historically we’ve had a lot of power wastage in OVMS v2 (and v1 before that). We used LM style power regulators, and those burned off the 12V->5V difference as heat. Kind of like to Raspberry Pi A and B (but in our case 12V->5V is being burned off, not 5V->3.3V). Cheap, clean and simple, but very ‘lossy’.
> 
> Moving to a 12V->5V pre-conversion, then 5V->3.3V from both USB and 12V sides, seems a simple way to go, but what sort of impact on efficiency does that sacrifice?
> 
> The other concern is noise. For the GSM modem (and presumably ESP32, although I haven’t seen so much concern raised there), we need to keep things very clean.
> 
> Cost wise, the 3.3V transceivers are about 3+ times the price of the venerable MCP2551.
> 
> Regarding the extra connector, the idea is to provide a base module with base functionality. Then have a plug-in architecture (using SIP strips) that will allow an expansion board to be added. That expansion board should be able to connect to the microprocessor, do whatever it needs to do, then expose itself on the extra connector pins. The idea is to have two of these - one for the optional modem, and one for general expansion.
> 
> Regards, Mark.
> 
> P.S. In the above I say 12V, but it is more like 14V - 15V in modern cars.
> 
> On Thu, 30 Mar 2017, Mark Webb-Johnson wrote:
>> I’m trying to finalise the OVMS v3 final board layout, with the factory in China. We have some questions and seek your opinions:
>> 
>> CAN transceivers / power
>> 
>> Overall, the OVMS v3 system runs at 3.3V. We have two power supply sources: USB (where we use a 5V -> 3.3V regulator), and +12V vehicle power (where we use a +12V -> 3.3V switching power supply, to be as energy efficient as possible). Diodes are used for reverse-polarity protection as well as coping with the situation where both usb and vehicle power is applied simultaneously.
>> 
>> Our problem is with the CAN transceivers. I’m used to the MCP2551 (been using it for a decade or more), but that is 5V so greatly complicates the power supply arrangements at the +12V side. We can switch to something like the SN65HVD233 transceiver that works at 3.3V.
>> 
>> But, I am concerned about comments I am reading about 3.3V CAN transceivers and their inability to meet the ISO11898 dominant condition requirement of 3.5V. From my understanding, these 3.3V CAN transceivers get around this by driving CAN-L to 1V, to still get the differential of about 2V (recessive condition?). My concern is compatibility.
>> 
>> What do people think about this? Any recommendations?
>> 
>> External Connectors
>> 
>> The idea is to retain the existing DB9 connector, with the same basic pin arrangement:
>> 
>> DB9-M   Signal
>> 3       Chassis/Power GND
>> 2       CAN-L (primary)
>> 7       CAN-H (primary)
>> 4       CAN-L (alternate CAN)
>> 5       CAN-H (alternate CAN)
>> 9       +12V Vehicle Power
>> 
>> That leaves pins #1, #6, and #8 free for expansion uses. It gives us compatibility with existing OVMS cables.
>> 
>> We would then add a second connector. The suggestions here are DB15 normal density, DB25 normal density, or DA-26 high density. My preference is the DA-26 (as DB25 is the old parallel printer style connector and very bulky). As well as power lines, expansion cards could wire to this connector to expose external inputs/outputs.
>> 
>> What do people think about the DA-26 connector? I’m suggesting a female version (as power is carried there, and I don’t want the pins to get pushed together for a short).
>> 
>> Note that we’ve also got a micro-usb socket, as well as space for GPS and GSM/GNS antennas.
>> 
>> Other than that, we are good to go. Things have stabilised now with Espressif, so we can proceed with building developer boards.
>> 
>> Regards, Mark.