Archaeology Resistivity Meter

Looks good.

Do you think we need a spec next ?

Might be worth hanging on a couple of days to get comments on the block diagram and thoughts so far.

Things to think about on spec:

Operating temperature range

Endurance on one fully charged battery

Max output current and voltage (if 20mA RMS and nearly 60V pk-pk won't do it's going to get a lot more expensive )

Input common mode range (max +/- 14.5V including signal) ref instrument internal ground.

Max signal combined signal and common mode +/- 29V differential

Input impedance - could be 100M if wanted but I think this is a bad idea - should be lower , maybe 1M - problem is that you want the option

of input caps for AC coupling. If you have 100M input Z then you get -3dB at 10Hz with a 160pF input cap, but its impedance at

150Hz is 6.6M which will make things noisy. I'd like the capacitor impedance to be no more than 50k at 150Hz = 22nF, with 100M input that gives

a time constant of 2.2s so it may take ages to settle, I'm tempted to go for 10M or even lower input resistance to reduce that effect. (NB, the settling time

of the AC coupling RC is nothing to do with the normal settling time of measuring - its to do with how long it takes the system to settle after a large DC

bias is applied to the input.) Ideas / comments etc welcome. (AC coupling will be switchable.)

Input protection - static discharge (8kV ?, 16kV ? - full lightning ? (please not !) - any ideas from the field.

Operator controls - buttons - membrane key pads are expensive - nice push buttons are also expensive -  any ideas.

Would a touch screen be any good in the field ?

Well there's a good collection of random thoughts and comments.

MK

Looks good.

Do you think we need a spec next ?

Might be worth hanging on a couple of days to get comments on the block diagram and thoughts so far.

Things to think about on spec:

Operating temperature range

Endurance on one fully charged battery

Max output current and voltage (if 20mA RMS and nearly 60V pk-pk won't do it's going to get a lot more expensive )

Input common mode range (max +/- 14.5V including signal) ref instrument internal ground.

Max signal combined signal and common mode +/- 29V differential

Input impedance - could be 100M if wanted but I think this is a bad idea - should be lower , maybe 1M - problem is that you want the option

of input caps for AC coupling. If you have 100M input Z then you get -3dB at 10Hz with a 160pF input cap, but its impedance at

150Hz is 6.6M which will make things noisy. I'd like the capacitor impedance to be no more than 50k at 150Hz = 22nF, with 100M input that gives

a time constant of 2.2s so it may take ages to settle, I'm tempted to go for 10M or even lower input resistance to reduce that effect. (NB, the settling time

of the AC coupling RC is nothing to do with the normal settling time of measuring - its to do with how long it takes the system to settle after a large DC

bias is applied to the input.) Ideas / comments etc welcome. (AC coupling will be switchable.)

Input protection - static discharge (8kV ?, 16kV ? - full lightning ? (please not !) - any ideas from the field.

Operator controls - buttons - membrane key pads are expensive - nice push buttons are also expensive -  any ideas.

Would a touch screen be any good in the field ?

Well there's a good collection of random thoughts and comments.

MK

Ken,

I'm just catching up, sorry to hear of your loss.

Michael, will post some feedback on various points in a moment

Dave

Thoughts on power sources.

Sony consumer-camera-style batteries are a great source, I've used them for probably nigh-on twenty years for portable kit where they're suitable. My introduction came via a broadcast TV production client for whom I used to design and manufacture equipment used for various motor- and power-sport broadcast, both telemetry/data-logging and video capture. In days of yore, pre solid-state recorders, the client used to use the Sony GVD miniature DV decks in resiliently-mounted housings, and for commonality, I designed my kit to use the same NPF batteries.

Just a couple of issues to be aware of:

Electrical: A few packs had/have inbuilt over-discharge protection, but many (especially non-Sony clones) don't, so the power input will need to sense and then shut-down before over-discharging the pack, as Michael just mentioned (an acquaintance last year came to me with a cheap LED lighting panel powered by NPFs, and he was bemoaning how the batteries were no longer taking/holding much charge despite only having a dozen or so cycles on them...). There needs to be ample warning, you don’t want to have to stop and do a battery change mid-grid as stopping mid-grid for even a few minutes can introduce a discontinuity (the equipment may resume measuring accurately, but the ground conditions may have changed).

Mechanical: Most of those packs don't inherently have complete physical attachment. When the slide onto a holder, the terminals at one end connect with those in the holder and the battery is constrained by a cover or hatch; on the NPF style there are also a couple of slotted-hooks like those on, say, some PC case covers to help keep the battery aligned. Those are usually sufficient to hold the battery in the plane of the contacts and a modest spring pressure usually keeps the contacts connected. However, if the battery slides back out away from the contacts, the connection is broken and the hooks can disengage. Sony recorders with exposed batteries had a mechanical latch which sprang-up once the battery was in and would stop it sliding backwards; but many of the third-party plates don't have that feature. Those plates are great for charging and for low-dynamic power uses, but not so good as-supplied in higher-dynamic environments. Need to remember that an instrument such as this is dropped thousands of times every day – 400 or 800 times per 20x20m grid. Whilst lifting / pulling the probes out before moving them is usually a smoother movement, the probes are inserted into the ground by swinging the frame forward and dropping it with additional downward push – usually it decelerates relatively smoothly as the probes penetrate, but sometimes it jars when a stone is hit.

Dave

Hi Dave,

That's very useful information. I've noticed the same thing, my low-cost charger doesn't have that latch either.

However the battery-plate that I've ordered from aliexpress does have it (at least, it looks like it from the photo), so I'm hoping that works. Still, I wouldn't trust that on it's own with this cheap battery plate, and there may need to be some 3D-printing for a screw-on battery cover perhaps (maybe with foam padding), or some damping ideas, since so many knocks can't be healthy for the battery.

There are better battery plates on Amazon, but the price is higher. Since the battery is a standard voltage dual-cell device, worst case the battery and the plate could be swapped out for a different option when testing too, or constructors may be invested in a different battery system (e.g. Canon) and so they might use that anyway.

Although, not that I'm an end user for this project, but if it were me I'd prefer Sony over Canon because the battery has easier-to use connections (large round sockets, whereas Canon batteries [at least the ones I've seen] use metal leaf connections).

The aliexpress battery plate (search term is Andoer Battery Adapter Base Plate Battery Plate for Lilliput FEELWORLD Monitor for Sony NP-F970 F550 F770 F970 F960 F750 Battery in case the link goes bad) arrived.

It is made as well as any typical consumer item. The battery I tried is held with zero wobble, and the latch mechanism worked to prevent it sliding off. The wire is heavy gauge too, and battery contacts are nicely made.

There are photos below but in brief, I think it is good value, I wish I'd ordered another couple.

The yellow arrows show the slide-on hooks, red arrows show the latch.

The latch spring is visible here:

The battery contacts have springy curved metal like banana plugs:

Some epoxy glue on the back of the pins could be useful maybe, for extra strength. The connector is JST PH.

Shabaz,

That plate looks to lock on the side, that may well do. The other way which the Sony decks use is that you drop the battery onto the tray and slide it in to not only make contact but also the hooks engage with protrusions on the side, then there's a spring-loaded latch which pops-up and stops the battery sliding back so the hooks and contacts remain engaged. I haven't shopped for free NPF plates for a few years now, but I haven't seen any with the positive lock like the Sony GVD and similar decks:

Dave

Power supply management:

One aspect touched upon by Michael already is the need to have a pre-emptive shutdown to avoid over-discharging the battery pack, and I mentioned the need for plenty of warning to try and avoid having a break mid-grid.

It is, I suspect, unlikely that it will be possible to take advantage of the intelligence built into some packs which tracks charge and allows predictions. As I was driving yesterday, the sight of the fuel consumption/prediction display in my car made me wonder if it might be possible to include something basic along the same lines, that could well bring a useful usability improvement in the field if the kit is constrained by its battery pack. The usability could not only be a ‘battery fuel gauge’ but, once one grid had been done, the geology and hence power consumption would be approximately known*, so at the end of one grid and before starting the next grid, the UI could pop up a message “You may not have enough battery left to complete the next grid” so the battery could be pre-emptively swapped or charged.

* i.e. good noise-free conditions where you only needed 20mW (20v to get 1mA), or more challenging where you needed over half a watt (50-60v to get 10mA).

Dave

It's not too hard to measure the current drain from the battery. and its under load voltage. From the current we can estimate

the drain from the battery since last swapped and from the voltage (and possibly ambient temperature) we can estimate (less accurately)

how long the battery has left.

It would be nice if the instrument could recognise individual batteries but I think this would be too difficult to get right.

We do need a good way of telling when the battery has changed  - without some means added specifically to do this the instrument

can't tell the difference between being switched off and the battery being removed.If operators insist on swapping a battery for a half charged

one, we can't estimate its endurance other than by voltage.

This means that we need a circuit that is always connected to the battery terminals and can indicate to the processor if the battery has been disconnected

for more than (for example, 10 seconds). The processor can reset the circuit once it has reset its charge measuring system. The battery disconnect

detect circuit should draw no more than 50uA from the battery when the instrument is off. (8% of a 5Ah battery per year.) It doesn't need to be very accurate,

battery present voltage, > (4 - 5.5 V), battery absent, < (2 - 3.9 V), timing +/- 30%

Suggestions welcome (simple and cheap is good)

MK

Hi Michael,

Maybe an alternative could be for (say) a 4-bit switch (i.e. 16 values), or instead a PCB with scratch-off traces to be taped to the battery, with a plug on the end, to encode 15 or so batteries. Then the microcontroller can always know if a different battery is attached (writing to the microcontroller board existing NVRAM), but also always prompt the user (at power-on) to acknowledge the battery was not changed, if it detects that the same battery as before is connected (this message would therefore be rare for the user to see during the day, if the instrument is only switched off to change batteries, since a battery change to a different battery is auto-detected through the 4-bit input). The acknowledgement would only be needed on power reset if the same battery was detected.

The four or so inputs could be useful for implementing different battery detection or measuring methods in the future, e.g. I2C EEPROM chip with a plug on the end too, or full I2C fuel gauge chip if the battery holder and battery were replaced in a future modification.

Another way could be to allow an option with two batteries, OR'd, and then the microcontroller can always know if either was disconnected and re-inserted, by looking at the voltages on the battery side of the OR.

Both the 4-bit input and the OR circuit are compromises, but have scope for future enhancements if a few I/O lines were left near the battery connections.

Also, (just so no-one can patent it), going to put the idea of it self-measuring the battery impedance out there, using its existing circuitry. The DAC output could toggle a MOSFET to place a load on the device, at (say) 1kHz or any other frequency (square wave could be used for this test since sine-wave accuracy might not be needed), and relays or another method (could even be a manual switch) to connect the sense inputs to the circuit. This could be performed at power-on and in-between grids. It would easily detect the battery state being in the last (say) third of its charge state (experiments with a single 18650 in the diagram below, this was done during BA6010 review).

Hello Shabaz,

While having individual battery ID would be nice it does seem a bit complex.

\there will be connectorised spare IO on the pcb so add ons are possible.

You would still need the battery out detector because the user might remove a battery then charge and replace it without turning the instrument on.

We will measure battery voltage and current, from you graphs battery voltage is as good an indicator as battery impedance.

I've come up with a battery out detector - draws about 30uA, mostly through R5.

Seems OK in simulation.

AT 4 seconds, once the cap has charged up the pulse on INP sets the output of the comparator.  The hystereis via R2 will

keep the cpmarator set until the supply voltage drops too low for R2 to keep the voltage at the positive input above the reference.

The time between battery off and comparator supply low is set by C1 and R5. Most of the current (about 85%)  flows though R5,

but 30uA is quite good enough. You can fiddle around and get to less than half that (C1  to 47u, R5 to 750k, but the timing

will be more variable with comparator current.)

MK

Thanks Michael, some feedback below:

michaelkellett  wrote:

 

Looks good.

Do you think we need a spec next ?

Might be worth hanging on a couple of days to get comments on the block diagram and thoughts so far.

 

Things to think about on spec:

 

Operating temperature range

Endurance on one fully charged battery

Max output current and voltage (if 20mA RMS and nearly 60V pk-pk won't do it's going to get a lot more expensive )

Input common mode range (max +/- 14.5V including signal) ref instrument internal ground.

Max signal combined signal and common mode +/- 29V differential

Input impedance - could be 100M if wanted but I think this is a bad idea - should be lower , maybe 1M - problem is that you want the option

of input caps for AC coupling. If you have 100M input Z then you get -3dB at 10Hz with a 160pF input cap, but its impedance at

150Hz is 6.6M which will make things noisy. I'd like the capacitor impedance to be no more than 50k at 150Hz = 22nF, with 100M input that gives

a time constant of 2.2s so it may take ages to settle, I'm tempted to go for 10M or even lower input resistance to reduce that effect. (NB, the settling time

of the AC coupling RC is nothing to do with the normal settling time of measuring - its to do with how long it takes the system to settle after a large DC

bias is applied to the input.) Ideas / comments etc welcome. (AC coupling will be switchable.)

 

Input protection - static discharge (8kV ?, 16kV ? - full lightning ? (please not !) - any ideas from the field.

 

Operator controls - buttons - membrane key pads are expensive - nice push buttons are also expensive -  any ideas.

Would a touch screen be any good in the field ?

 

Well there's a good collection of random thoughts and comments.

 

MK

Power output: needs to be able to inject chosen constant current (up to at least 10mA and possibly 20mA) at least 50v if not 60 of either polarity into the ground, so in terms of pk-pk 100 to 120v.

Note P1P2 measuring will allow DC option, great, thanks.

Settling times: not sure if I'm fully grasping but the excitation most usually used for archaeological prospection is 137 Hz reversing DC (square wave) so each injection is around 3.5 mS.

Input range: P1P2 measuring needs to allow for the +/- say 60v used for injection, plus margin for that resulting from stray / telluric currents.

Input protection: not certain what would be justified. Have to trust that the user doesn't spear an inadequately-buried mains cable!  One scenario increasingly common in rural areas is presence of electric fencing (typical discharge 7kV upto 25J in 3mS), normally fence network has +ve pulse, any return is via the earth; in theory only return if an animal, standing on the ground, touches the fence; in practice leakage all the time, not so much from poor insulators but from grass etc. touching lower wires. You don't need to be alongside fence-line, return current will take shortest (or at least lowest resistance) path back to energiser's earthing point, so current can flow across fields. Assuming user doesn't actually touch instrument to electric fence, then effect is the extra voltage from return current flow., Have seen about 10v over 50m from this when unwittingly near (other side of the hedge from) an earthing array.

UI: not certain about touch screen - keypad array or buttons may be longer-lived and easier to operate when gloved. I did post earlier about the option, if a 'graphics' or bit-mapped screen was used, that could ease reversal. My suggestion would be a suitable small set of buttons either side of the screen, don't 'hard' label them, but label them as soft-keys as, for example, seen on say a scope screen. Then, if the instrument is reversed to go back down the next line, rotate the screen and hence key labels/functions.

Temp range: suggest aim for same as current commercial instruments.

Dave