Archaeology Resistivity Meter

Great thank you.

Hello Ken,

mk2 block diagram.

Several changes,

The DAC is driven by the FPGA

The ADC has changed type to a TI single channel audio type (much cheaper (about £3.30) and OK for our purpose I think.)

The ESP chip provides Bluetooth and WiFi, chosen because its widely available, dead cheap and there is a lot of app material on the web.

The battery type is defined as Sony NP-F970 equivalent, 5AH gives about 35Whrs, and I estimate a WiFi and display on power

consumption of 3W so the available batteries should give at least 10 hours per charge. I bought 2 and a charger for about £40.

The display is ideally a Riverdi 4.3" which uses the Bridgetek controller and has SPI interface. You can get a bare bones one for about £36

but the nice one with the touch screeen and flat glass front (like the PI display) is about £50. Or you can use a sub £5 quarter VGA type from

China - but I won't be writing code for it

Any illegible and unguessable features, please ask.

MK

The current is measured as explicitly as in any other system, in the circuit in thread 44 it is sensed by R3 and R8, currrent in the load has to flow through these resistors.

The feedback is applied immediately and continuously to adjust the drive voltage to maintain the current.

If the amplifier runs out of voltage "headroom" this is detected by the voltage monitoring and the operator can be warned, the measurement tagged as an error (if it was taken)

or the current demand adjusted.

The Beck meter uses a similar system with much simpler (and less accurate) current sources. It compares the max output voltage detected with a reference and illuminates

an LED if the output is too high. All the operator can do is change range - and reduce the current by a factor of 10.

There is a potential problem with this or any other system that if the current is set too high at the start the amplifier may run out of headroom in a different position. The

operator will have to balance the lower noise of a large current against the risk of having to reduce it paert way through a series of measurements.

I don't know how comparable resistance reaadings at different currents are.

MK

Hi We had an old commercial meter, but I can't for the life of me remember the brand. It was very similar to the Beck design, except it had a built in logger. It had the red fault light, just as you describe. It would always illuminate with no contact on the probes, but would go out when soil contact was made. If it didn't we had to adjust the current. This was never a good idea half way through a grid as it would change the calc. Our current meter has no controls for setting the current, so I can only assume there is some form of auto ranging. We do occasionally get out of range reading warnings and nothing is recorded. In this instance you just try again or mark the grid location as a dummy.

shabaz  wrote:

 

Hi Dave,

 

Once the high-voltage is detected, it could flag an alert on the display (or sound an audible alert, I guess that needs to be on the block diagram if there is a requirement for this), to indicate that there is likely a probe fault, or a lower current setting may be required, because that condition should only occur when the probes are not in the soil, or all other reasons for the resistance being too high, making it not possible to pass enough current, reaching the supply limits.

Thanks Shabaz,

This is one aspect where I fear either I don't understand of what's being proposed from an abstract basis, or that there is a gap.

The survey mechanism in the main use case (manually advanced frame) (see post 161 for more details).

1) The mobiles probes are inserted into the ground by dropping the frame

2) The instrument takes a reading

3) The instrument beeps to indicate to the operator the reading has been taken

4) The operator lifts the frame clear of the ground, advances to the next location and - if not at end of a grid - drops in the next location as at step 1.

So, the measurement cycle is probes in ground for perhaps 1/2 to at most 1 second; then 'handfuls' of seconds as the frame is lifted, re-positioned and re-plunged into the earth. There is a meaningful measurement window of say 0.7 seconds, then say 3-4 seconds before the frame is in contact with the earth again. During descent through the vegetation, there will be intermittent contact, and similarly during the extraction after a measurement. So, there will a period of totally open circuit (possibly the majority of the time), a period of noisy connection on the way down, a period in the earth to take a measurement, and then another noisy connection as the frame is lifted. (if the use case is a wheeled or other cart, the same phases will apply but maybe not the exact same timings).

If an alarm sounds every time the voltage maxes out then the alarm will only be silent when the probes are in the earth and are making a valid measurement.

Dave

michaelkellett  wrote:

 

The current is measured as explicitly as in any other system, in the circuit in thread 44 it is sensed by R3 and R8, currrent in the load has to flow through these resistors.

The feedback is applied immediately and continuously to adjust the drive voltage to maintain the current.

If the amplifier runs out of voltage "headroom" this is detected by the voltage monitoring and the operator can be warned, the measurement tagged as an error (if it was taken)

or the current demand adjusted.

The Beck meter uses a similar system with much simpler (and less accurate) current sources. It compares the max output voltage detected with a reference and illuminates

an LED if the output is too high. All the operator can do is change range - and reduce the current by a factor of 10.

There is a potential problem with this or any other system that if the current is set too high at the start the amplifier may run out of headroom in a different position. The

operator will have to balance the lower noise of a large current against the risk of having to reduce it paert way through a series of measurements.

I don't know how comparable resistance reaadings at different currents are.

 

MK

Thanks Michael.

OK, I understand the V=>I output stage has feedback, fully understand. What I was trying to understand was how the control processor would know when current was actually being delivered and had reached some kind of stable state, so that it could initiate a measurement cycle.

The control process is in a simplified state machine

1)Try to inject commanded current

2) When current is being satisfactorily delivered, initiate a measurement cycle and beep when completed. If contact with the earth is believed to have taken place but current can't be injected satisfactorily, raise distinctive* audible alarm to user as one of the probes on the frame might have landed on a stone or piece of pottery etc., so the operator will re-plunge to re-attempt the reading.

3) Monitor current and after period of no current, start again at (1)

You really really don't want to just tag reading as an error in the data as if you go back you really need to repeat the whole grid as ground conditions will almost certainly have changed.

* It needs to be a distinctive alarm as the operator is used to dropping the frame, hearing a beep, lifting it and moving on. In using current commercial kit, the alarm may not be that different so the first you know is the operator gets to the end of a line and instead of getting a double-beep to indicate end-of-line, you only get one beep. You don't know which one was duff, and subsequent ones may well have been offset, so you have to delete the line, walk back to the start and repeat the line. The problem is that muscle memory takes over, it may well be the two thousand three hundred and twenty seventh time that afternoon...

More on headroom in a moment, but I would urge not to produce an instrument which uses marvellous and with high, indeed exciting, potential, to emulate an instrument that has limited resolution and would be a downgrade for almost all, if not all, users. If this can be brought to fruition it could not only address the issues Ken original posed (usability and cost) but once basic facilities in place to match exiting commercial kit, then there should be scope to open up so many more avenues.

Dave

Hi Dave,

The design is very 'software-defined' so the logic around alarm sounds is not hard-coded.

However, having any sound capability, or not having a sound capability at all, is hardware-defined, so needs to make it onto the block diagram (or some documentation) because currently there is no such requirement recorded.

It's a computer which happens to have the right connections and buttons/display. What it actually does is mostly software-defined. Any procedures on what alerts occur, and when they occur, are all coded in software and are easy to edit/remove/add. But the physical speaker/sounder is needed on the block diagram, if sound is likely to be a requirement.

Hi Dave,

I can help here, since these are mainly software-related questions:

how the control processor would know when current was actually being delivered

On the 'source' side, there is hardware circuitry to alert if the demanded current cannot be delivered, because the voltage will raise to a limit if that condition ever occurs (barring damaged circuitry - everything has a failure rate, and there can be a test procedure or perhaps even a self-test to identify that - it's quite easy to do that in software, perhaps prompted on the display, prior to using the instrument). The software will receive the alert.

and had reached some kind of stable state, so that it could initiate a measurement cycle.

Again, super-easy for the software. All measurement instruments based on source/sense will wait for the reading to be stable by taking multiple measurements rapidly. The precise algorithm is implemented in software.

The control process is in a simplified state machine

1)Try to inject commanded current

Understood.

2) When current is being satisfactorily delivered, initiate a measurement cycle and beep when completed.

Understood. That means sound capability needs to be added in hardware.

If contact with the earth is believed to have taken place but current can't be injected satisfactorily, raise distinctive* audible alarm to user as one of the probes on the frame might have landed on a stone or piece of pottery etc., so the operator will re-plunge to re-attempt the reading.

From your description there doesn't seem to be any relevant (to this requirement) sensing inputs missing compared to existing systems, so there's no risk of a downgrade, only an upgrade (since the behaviour can be better implemented in software). The logic could sense instability or no current flowing in the measurement for a defined period of time, and if either of these occur to raise the distinctive audible alarm. If this doesn't meet your requirement, more input is definitely needed here.

3) Monitor current and after period of no current, start again at (1)

Understood.

It needs to be a distinctive alarm

Understood. The hardware needs to implement sound capability with some flexibility.

not to produce an instrument which uses marvellous and with high, indeed exciting, potential, to emulate an instrument that has limited resolution and would be a downgrade for almost all, if not all, users.

Please can you specifically point to the downgrades so they can be addressed?

If this can be brought to fruition it could not only address the issues Ken original posed (usability and cost) but once basic facilities in place to match exiting commercial kit, then there should be scope to open up so many more avenues.

That makes sense, agreed.

Hi in using our current kit we have on the display as soon as the probes are lifted 'PLACE PROBES AT ...' The unit is sensing that there is no contact with the soil on the current probe.

Hi Ken,

That makes sense, that's a fairly straightforward condition to detect.

However, I'm wondering if a sensor should be added to assist, It's a cheap (in terms of hardware) thing to attach, to get better sensing of the physical state of the equipment (e.g. if it is physically moving or raised). Hi michaelkellett it sounds like a MPU-9050 (or similar, since I think that's obsolete) might be useful : ( or at least easy-to-access pins on the I2C bus. A bit of feature creep though.

EDIT: just to throw another idea out there, maybe it is better to just have a generic socket, e.g. PMOD or (perhaps even better since there's a huge range) Mikro-E Click, rather than a MPU.. sensor. Then people can experiment with add-ons such as additional sensors to adapt the design further.

Re headroom in the current injection:

Instruments typically offer range of constant-currents to inject - such as 0.1mA, 1mA, 10mA.

When you are about to start actual measuring, best practice is to take a couple of test readings at the most distant points of the (typically three) grids that you intend to measure from one C1P1 fixed probe location, and effectively check that you're maybe not using more than, say half your range to achieve that. Hopefully then you won't hit a range problem mid-grid.

Just to answer Michael's question about changing the injection current during a survey.

It is absolutely a fact of life that if you survey a non-trivial area, conditions will change. What you strive to do is to ensure that any changes are between grids, rather than mid-grid. The normal/expected discontinuity occurs when you relocate C1P1 after three grids or whatever. (Good) processing software will look for step-discontinuities between grids and will adjust them out. The not-uncommon error is that the operators forget to advance C1P1 in time so they run out of cable part-way through a grid. If you just relocate them, the readings are inevitably different due to distances and also possibly different soil conditions. You can then see operators doing the hokey-cokey with the probes - one watching the res on the screen whilst their oppo moves the probes in and out; or sometimes the operator will try (if the machine allows) try fiddling the calibration. The problem is that a mid-grid change in measured resistivity is usually associated with an underground change, so you can't risk auto-adjusting it; whereas a similar change between all twenty or forty points on each side of a grid boundary are usually associated with a change in the reference conditions.  That is usually C1P1 relocation, but could also be a change in injected current.

I have seen a change mid-grid due to the operator running out of range, and it was repeatable; the effect wasn't enormous, and if the data points also all log the conditions such as the constant-current used then the discontinuity could be a documented/visible event - but it would be better if it didn't occur; and if it did occur, it shouldn't be a silent auto-ranging, the UI should prompt the operator along the lines of 'Max range exceeded - change current now mid-grid? or change current and re-start the grid?'

Dave

shabaz  wrote:

 

Hi Ken,

 

That makes sense, that's a fairly straightforward condition to detect.

However, I'm wondering if a sensor should be added to assist, It's a cheap (in terms of hardware) thing to attach, to get better sensing of the physical state of the equipment (e.g. if it is physically moving or raised). Hi michaelkellett  it sounds like a MPU-9050 (or similar, since I think that's obsolete) might be useful : ( or at least easy-to-access pins on the I2C bus. A bit of feature creep though.

Shabaz/Ken,

Its not an instrument dynamics device or sensor, its whether or not current is passing C1 to C2 ...

Dave

shabaz  wrote:

 

Hi Ken,

 

That makes sense, that's a fairly straightforward condition to detect.

However, I'm wondering if a sensor should be added to assist, It's a cheap (in terms of hardware) thing to attach, to get better sensing of the physical state of the equipment (e.g. if it is physically moving or raised). Hi michaelkellett  it sounds like a MPU-9050 (or similar, since I think that's obsolete) might be useful : ( or at least easy-to-access pins on the I2C bus. A bit of feature creep though.

Shabaz/Ken,

Its not an instrument dynamics device or sensor, its whether or not current is passing C1 to C2 ...

Dave

Hi Dave,

That's understood. This isn't to replace detection using the probes. It's about the potential to make the software more intelligent about precisely when it sounds an alert, for instance if the user decides to take a few seconds longer before setting the instrument down.

Yes agreed, but as Shabaz says a fairly easy condition to detect.

Ken