zubax / opengrab_epm_v3 Goto Github PK

Add test dipswitch routine to drwatson

The title says it all.

[21:06:48] Pavel Kirienko: I can actually add some idication saying whether the signatire was new or reused

[21:11:13] Andreas Jochum: PIO1_3 should J4
[21:11:38] Andreas Jochum: and PIO0_3 should J3
[21:13:34] Andreas Jochum: J3 open, J4 closed

Some people are using high impedance AA Alkaline cells,....

Force 1 second delay when power stage misfires. Misfire = main capacitor is not discharged after firing

It is possible on different hardware that the holding current is reached. Via the cathode pull down resistor on the power stage
It cause the Thyristor to stay on for up to 1 second. Subsequently firing its lower pair will cause the two thyristors to eat to much power and fail as short.

I'll take care of it before we change out the thyristors for the next rev

There's some weird arithmetic in getSupplyVoltageInMillivolts() that doesn't make sense. What were you trying to do? Why +700?

unsigned getSupplyVoltageInMillivolts()
{
    static std::uint16_t old_value = 0;

    std::uint16_t new_value = 0;
    (void)Chip_ADC_ReadValue(LPC_ADC, ADC_CH6, &new_value);

    // Division and multiplication by 2 are reduced
    unsigned x = static_cast<unsigned>((static_cast<unsigned>(new_value + old_value) * AdcReferenceMillivolts) >>
                                       AdcResolutionBits);

    old_value = new_value;

    x *= 5;
    x += 700;                           // Poor man's calibration, it's within 100mV - TODO WTF is this?
    if (x < 4500)                       // Under 4500mV Vref drops, mesurements useless
    {
        x = 0;
    }
    return x;                           // Voltage divider on board, shoud not go here
}

make PRODROPPER=1

First time after boot when Vin is measured it returns 5000ish mV instead of 12000ish. ACD pin V is correct at this time at around 1000mV.

This commit has the syslog print out to reproduce

https://github.com/Zubax/opengrab_epm_v3/tree/pro_epm

Boot
CAN auto bitrate...
Mag on 2 //button is pressed
Mag ON chrg started //mosfet never gets switch and LED blinks in error pattern
CAN auto bitrate...
CAN auto bitrate...

It works fine when pressed again

See the email sent Aug 8:

What I would need to work on this would be:

1. HW with JTAG and optional DEBUG Console connectors and any special cables.
2. Schematics in searchable pdf format
3. The pin map for the EPM product in tabular form that list
   1. SoC pin capabilities (CPU Usage)
   2. SoC pin number (Pin)
   3. Signal name from the schematic (Signal),
   4. The use in the application (Signal in parens),
   5. Point of termination if it goes to a connector. (Board)

static constexpr std::uint16_t TurnOffCycleArray[][2] =
{
{ 450, 0 }, //these first 3 are just there to get from the right top corner of the B/H loop somewhat close to center. If they are executed while not in the top right the magnet will turn on with the opposite polarity
{ 450, 0 },
{ 300, 1 },
{ 180, 0 },
{ 162, 1 },
{ 146, 0 },

It seems that it miss read input or output voltage, triggers an error

When PWM is high or low the EPM constantly charges and turns on/off, we need to add a 250ms break in 500ms break in between to prevent the inductor from overheating

There are some variations on the saturation current of T1 between parts

Reevalute using a smaller PTC fuse, fuse should pop before T1 gets damaged

Increasing minimum command interval

https://github.com/Zubax/opengrab_epm_v3/raw/master/annotationR5C.jpg

Pin numbers on the PCB are correct; Pin number on the picture are incorrect

Mag on 2
Mag ON chrg started
Mag ON chrg started
Mag off
Charger restart
Mag OFF chrg started cyc 0
Mag OFF chrg started cyc 1
Mag on 2
Charger restart
Mag ON chrg started
Mag ON chrg started
Mag off
Charger restart
Mag OFF chrg started cyc 0
Mag OFF chrg started cyc 1
Mag on 2
Charger restart
Mag ON chrg started
Mag ON chrg started
Mag off
Charger restart
Mag OFF chrg started cyc 0
Mag OFF chrg started cyc 1
Mag on 2
Charger restart
Mag ON chrg started
Mag ON chrg started
Mag off
Charger restart
Mag OFF chrg started cyc 0
Mag OFF chrg started cyc 1

When Vout is not rising fast enough charger should quit, it doesnt
When Vout is not reached <1s charger should time out, it doesnt

We have two functions, magnetOn() and magnetOff(). At the moment they are filled with test patterns, we need to clean them up and implement the correct logic. What would be the correct logic?

I imagine it is as follows:

• Charge the capacitor with a pre-defined amount of energy, or to a pre-defined voltage.
• Switch the bridge.
• Optionally repeat a few times.

When the magnet does not need to switch, the cap will remain discharged.

Also, the Charger class has to be rewritten in a non-blocking manner, since we can't block in run(), there is other logic to run too. I can refactor it myself once #2 is resolved.

Temporary fix: Don't use CAN and PWM together,

PWM should be ignored when CAN is being used.

The CAN command is being send over and over again and the receiver only response to a change in command.
This is the typical way to use CAN bus - the same state is being published at a fixed rate or when it changes
This allows to ensure consistency even if the receiving node restarts

There seems to be a bug in somewhere as well....

I can fix the problems with duplicated code (although it would be best to not write it like that in the first place), but I can't fix the functions cycleXXXX_XXXX() not understanding what they are. It seems like you were furiously trying to calibrate something, producing copious nop statements in the function bodies. Could you not just use board::delayUSec()?

Also, disabling interrupts for 1 ms would actually disrupt PWM input handling. I will evaluate our options to use hardware capture unit instead of software processing.

[12:05:07 AM] Andreas Jochum: how does the ID map to the dip switch postions?
[12:05:17 AM] Pavel Kirienko: directly
[12:05:51 AM] Pavel Kirienko: actually this should be changed a bit
[12:06:07 AM] Pavel Kirienko: the highest bit of the dip switch should configure node ID

Temporary fix: Don't use CAN and PWM together,

PWM should be ignored when CAN is being used.

The CAN command is being send over and over again and the receiver only response to a change in command.
This is the typical way to use CAN bus - the same state is being published at a fixed rate or when it changes
This allows to ensure consistency even if the receiving node restarts

There seems to be a bug in somewhere as well....

References:

• http://www.digikey.com/product-detail/en/SM04B-GHS-TB(LF)(SN)%3B-/455-1566-1-ND/807834
• http://uavcan.org/Specification/8._Hardware_design_recommendations/#physical-connector-definition
• pixhawk/Hardware#18

It would be easier if we separate out Drwatson function into
Upload, sign, and verify signature....

@ctech4285 please remove branches that are no longer needed. See this page: https://github.com/Zubax/opengrab_epm_v3/branches. I specifically suggest to remove PET_testing and degauss_debug.