VDI-RODERICO

[waltr]

Yes, the Boost convert chip on the OLED display died. Temporary fix is very easy. Simply solder wire on the boost output, at cathode of diode and fee with 12V. Display now works.
Actually, Newhaven Display has this as an option on their OLED display that allows disabling the on-board boost converted and supplying 14V from external source.

[waltr]

then maybe there is some pin in the display to control the contrast that you need to configure similary to the ST7567...

Nope, there is not contrast control on OLED.
My current theory is the Boost convert chip died. There should be 13V to drive the OLEDs.
It seems to be a fairly standard 5 pin SOT so maybe another will work.

[roderico]

Just checked my inventory and it is a ZR431L that it the 1.24V version and adjustable from 1.24 to 10V.

That is an idea to calibrate. However, I found that the X-Y plot was much more useful than the VDI Number when in the field and deciding to dig or not.

Bad news is that the second OLED display simply Stopped working the other night. Same happened to the first display but a 'one off' and having a spare didn't seem a big deal. Having a second display just stop working is killing me.
No idea why since no connection or code changed. Just, its working, turn off, turn on -nothing!
Scope says all signals from PIC to display look ok.

then maybe there is some pin in the display to control the contrast that you need to configure similary to the ST7567...

[roderico]

The PIC18F25K22 has the same adc voltage references as the 16f1788, so a good choice it can also run at 64mhz. I saw your photo of the zoren glcd above, nice graphics! I have mine working now also. No problem operating from 3.2V - 5V also added a pnp to switch the backlight on and off. The emitter connected to +5, I used a 100 ohm resistor between the collector and the led anode. I ordered 3 more displays about the time the coronavirus breakout, probably won't receive then until March or April?

the zolen display should work well with the latest schematics, I think PIC16F17xx are the only 8-bit mcu with internal op-amps, but the firmware will not fit in PIC16F1788 (only 16KB) and PIC16F1778 (28KB) has no internal voltage references, the other option is to move to 3.3V (as waltr) and 16-bit but it would break compatibility with KS0108 displays...

[waltr]

I think the minimum for LM431 is 2.5V, 1.24V might be fine, but there is always the resistor divider at the last resort. I just had a look at TGSL schematic (v3) and assuming you connected the vdi to the output of U103a and U103b a possibility could be to calibrate it with J3 in "All metal" in that way you would have a reference when you are using the disc mode (by switching to all metal).

Just checked my inventory and it is a ZR431L that it the 1.24V version and adjustable from 1.24 to 10V.

That is an idea to calibrate. However, I found that the X-Y plot was much more useful than the VDI Number when in the field and deciding to dig or not.

Bad news is that the second OLED display simply Stopped working the other night. Same happened to the first display but a 'one off' and having a spare didn't seem a big deal. Having a second display just stop working is killing me.
No idea why since no connection or code changed. Just, its working, turn off, turn on -nothing!
Scope says all signals from PIC to display look ok.

[Altra]

That would be very good for a v3.x because you would save the MCP602 and the external voltage reference reducing the number of components and costs, however there were also some other suggestions for replacing PIC18F25K22 and IMO a compromise between improvements, required work and cost is needed before selecting a new pic

The PIC18F25K22 has the same adc voltage references as the 16f1788, so a good choice it can also run at 64mhz. I saw your photo of the zoren glcd above, nice graphics! I have mine working now also. No problem operating from 3.2V - 5V also added a pnp to switch the backlight on and off. The emitter connected to +5, I used a 100 ohm resistor between the collector and the led anode. I ordered 3 more displays about the time the coronavirus breakout, probably won't receive then until March or April?

[roderico]

That would be very good for a v3.x because you would save the MCP602 and the external voltage reference reducing the number of components and costs, however there were also some other suggestions for replacing PIC18F25K22 and IMO a compromise between improvements, required work and cost is needed before selecting a new pic. PIC16F1788 only have 16KB of program memory we need at least 32KB.

[Altra]

Been following the thread with interest, many of the newer pic's have built in references

I bought some pic16f1788's and never did anything with them. They have op-amps, 12bit adc with built in references including a 1.024v, 8bit dac etc. It would be interesting to see if you could make a complete vdi meter with the internal op-amps, reference, a few external capacitors and resistors.

http://ww1.microchip.com/downloads/en/DeviceDoc/40001675C.pdf

[roderico]

I think the minimum for LM431 is 2.5V, 1.24V might be fine, but there is always the resistor divider at the last resort. I just had a look at TGSL schematic (v3) and assuming you connected the vdi to the output of U103a and U103b a possibility could be to calibrate it with J3 in "All metal" in that way you would have a reference when you are using the disc mode (by switching to all metal).

[waltr]

Looks good.
Thinking for a 3.3V Vdd one could use an adjustable Voltage reference for the MCP602s.
The LM431 type reference comes to mind. There are others with a 1.24V minimum, just don't remember the part number.

[roderico]

I updated the schematics to include the 2.5V voltage reference, however, this does not scale with Vdd, i.e. for 3.3V you need a 1.65V reference, I also increased the capacitor values and added an optional input for 78L05 (or better regulator) and changed back the values for the MCP602 that are more tested, this latter could be improved with something similar to SWL's mod.

[roderico]

Interesting. Any 'noise' not coming from the coil might be reduced with better Reference of the op-amps and ADC.
Noise coming into the coils is tougher and possibly nothing can be done to reduce it. The two differential amp stages are 'Bandpass' filters centered around 11Hz.

this can explain why it picked up the furnace (10Hz) and not the mains (50-60Hz), however if I remove the coil then mine picks up the AC mains (see picture)

Did get out yesterday and away from AC Mains the ADC center values flicker about 5-6 counts. Maybe better ref, op amps, etc could reduce this but probably not worth the effort since the VDI does work rather well and has been very helpful addition to the TGSL especially in the Iron junk rich farm site.

If there is a 10mV permanent signal or noise then for 5V range the ADC resolution is 5V/2^10 ~5mV then 1-2 counts for 3.3V the resolution is 3.3V/2^10 ~ 3mV then you need 4 counts. Maybe it has to due with the nulling of the coil since it is of the same order of magnitude (10-20mV)

Something I did play with is looking at the VDI numbers with varying the DISC control. At farm sites I keep the DISC nearly all the down and not DISC out foil whereas in Parks I turn the DISC up to remove foil. This does change the VDI numbers but makes sense when thinking about what the DISC does.
With the VDI one can leave the DISC below foil and note a low VDI number, ~10, is Foil. Or if turning the DISC up to eliminate foil the VDI numbers for other target decrease by about 10. Example is Brass gun shells that VDI around 38-45 drop to 28-35.
Nothing wrong here, just noting how the VDI and the TGSL (IDX would be similar) works.

there might be something wrong here, mine does not change when changing the DISC, maybe is the pot or where it is connected in the TGSL

[waltr]

I did a quick test with a TL431 as 2.5V voltage reference and it is a little more stable but the resistor divider should be fine, the flickering comes from X and Y, i.e if it does not come directly from the coil then it is generated in the path to the ADC input, e.g. some filtering might be needed, reduce the gain, etc. nevertheless the diode is a better approach IMO than the resistor divider for the Vrange/2 reference.

Interesting. Any 'noise' not coming from the coil might be reduced with better Reference of the op-amps and ADC.
Noise coming into the coils is tougher and possibly nothing can be done to reduce it. The two differential amp stages are 'Bandpass' filters centered around 11Hz.

Did get out yesterday and away from AC Mains the ADC center values flicker about 5-6 counts. Maybe better ref, op amps, etc could reduce this but probably not worth the effort since the VDI does work rather well and has been very helpful addition to the TGSL especially in the Iron junk rich farm site.

Something I did play with is looking at the VDI numbers with varying the DISC control. At farm sites I keep the DISC nearly all the down and not DISC out foil whereas in Parks I turn the DISC up to remove foil. This does change the VDI numbers but makes sense when thinking about what the DISC does.
With the VDI one can leave the DISC below foil and note a low VDI number, ~10, is Foil. Or if turning the DISC up to eliminate foil the VDI numbers for other target decrease by about 10. Example is Brass gun shells that VDI around 38-45 drop to 28-35.
Nothing wrong here, just noting how the VDI and the TGSL (IDX would be similar) works.

[roderico]

I did a quick test with a TL431 as 2.5V voltage reference and it is a little more stable but the resistor divider should be fine, the flickering comes from X and Y, i.e if it does not come directly from the coil then it is generated in the path to the ADC input, e.g. some filtering might be needed, reduce the gain, etc. nevertheless the diode is a better approach IMO than the resistor divider for the Vrange/2 reference.

[SWL]

Yes, of course .. I?ll wait for the results .. ..