I would be happy with that. I don't do much precision measuring on days when the temp gets above 35 in Melbourne and I'm assuming you are not talking about self heating here?

0.01% error in a 10 mOhm measurement is 0.001 mOhm which is far smaller than your 0.1mOhm precision unless I have mised something obvious (as usual )

John

Originally Posted by pldaniels

John,

I can produce the 24-bit units with the current boards that I have, the trouble is I won't have tempco on there (unless I install a daughterboard). Will have to get a few of them in stock before I build of course . Without the tempco there'll be a swing of about 10ppm/'K from 25'C ( so, at 35'C, there will be about a 0.01% error just from the reference, ignoring other temperature related cumulative effects - however those should be nulled out due to the nature of the way I'm doing the testing ).

Paul.



Paul.

More a case of myself perhaps overstating the severity of the issue

I'll see if I can do some more work tonight.

Finally sourced some reasonably priced Kelvin test leads... will be about 20 days before they arrive though.

New "high precision" version of the boards arrived today. These are the ones that'll take a 24-bit sensor amongst a few other things to really 'tweak' the results.

Waiting patiently.

Originally Posted by pldaniels

New "high precision" version of the boards arrived today. These are the ones that'll take a 24-bit sensor amongst a few other things to really 'tweak' the results.

Partially built the the new high precision board, the 24-bit ADC is still flying its way to me so that'll be a week to go alas.

I was testing the 13-bit temperature sensor though, it works very nicely and has a 0.0625'K resolution, rather handy.

The temperature sensor will be used to let me determine the 'true' resistance value of the reference resistor as the ambient temperature changes.

Here's a photo of the current setup, the LCD is at an odd angle simply to let me debug things with ease.

So do we get a free precision thermometer thrown in with the LOM? Might not be such a silly idea. That kind of accuracy at room temp is pretty hard to find. Assume would just need a small switch and some clever Paul programming to show one of two alternate displays.

John

Originally Posted by pldaniels

Partially built the the new high precision board, the 24-bit ADC is still flying its way to me so that'll be a week to go alas.

I was testing the 13-bit temperature sensor though, it works very nicely and has a 0.0625'K resolution, rather handy.

The temperature sensor will be used to let me determine the 'true' resistance value of the reference resistor as the ambient temperature changes.

Here's a photo of the current setup, the LCD is at an odd angle simply to let me debug things with ease.

John,

I was planning on concurrently displaying the temp + resistance... and alternating to show the current as well, let's see... 8 chars... 5 for temp, 1 for current, (0~9mA) mmm it won't look good with temp + current on one line so I'll just alternate

The sensor works very nicely, I can hover my finger over it by about 15mm and it detects it within a few seconds as the airtemp warms up around my finger (assuming no breeze/airflow).


Paul

Found some domestic stock of the 24-bit ADC that I need, quite costly indeed but at least I should have it on Wednesday to test.

I've had to also bump the microcontroller up to the next higher level (8K flash, 1K RAM) in order to cope with the extra math (dealing with 32-bit integers, things get large very quickly).

I am pleased though at how nicely the design is working out.

The display is the next thing I'm not "entirely happy" about. 8x2 LCD is great, it does the job wonderful but it's not very large. A physically larger 128x32 pixel display would be a better option and then use bitmapped characters/digits. I could use a 8-digit 7-segment LED display but the cost is about the same and the power consumption is much higher, not to mention you lose a lot of potential display options such as putting the current and temperature on concurrently.

Paul.

Hmmm... not sure about that.

Wouldn't an obvious application for this be to have a continuous readout of resistance as you move around a circuit/trace looking for cracks etc?

Alternating readings would drive me bats under most circumstances. I would prefer to be able to lock on to one that is prominantly large for continuous display.

John

Originally Posted by pldaniels

John,

I was planning on concurrently displaying the temp + resistance... and alternating to show the current as well, let's see... 8 chars... 5 for temp, 1 for current, (0~9mA) mmm it won't look good with temp + current on one line so I'll just alternate

The sensor works very nicely, I can hover my finger over it by about 15mm and it detects it within a few seconds as the airtemp warms up around my finger (assuming no breeze/airflow).


Paul

That would be nice. I've just been testing a little Wattmeter from AEO-RC and I'm REALLY impressed by how much more usable and readable the display is, even at a small size, with different size digits.

See attached.

John

Originally Posted by pldaniels

The display is the next thing I'm not "entirely happy" about. 8x2 LCD is great, it does the job wonderful but it's not very large. A physically larger 128x32 pixel display would be a better option and then use bitmapped characters/digits. Paul.

That looks much like a 128x64 display. I can get away with 128x32, I can use 20x16 pixel high main digits and 8x6 pixel high sub digits (for current/temperature). Of course, the trouble there is now that you really do need an entirely separate display board with a separate controller

Well, these guys did it all with one board and an Atmega8L.

Originally Posted by pldaniels

That looks much like a 128x64 display. I can get away with 128x32, I can use 20x16 pixel high main digits and 8x6 pixel high sub digits (for current/temperature). Of course, the trouble there is now that you really do need an entirely separate display board with a separate controller

Not sure if I can do it on mine... certainly cannot do it with the current board because I'm using a 4-bit LCD interface (gfx displays want 8-bit) and I've run out of I/O pins due to chewing them up with the :-
buzzer
calibrate interface
mode-select
auto-ranging
SPI bus + ADC and temperature-sensor control

Ultimately it'll be 'simpler' :COUGH: for me to make a separate SPI/serial controlled LCD controller board, that way I can move it around in the case (thinking bigger cases here) and only need 3 wires to run it (data, +, -).

And at least it would be re-usable for other designs??

J

Originally Posted by pldaniels

Ultimately it'll be 'simpler' :COUGH: for me to make a separate SPI/serial controlled LCD controller board, that way I can move it around in the case (thinking bigger cases here) and only need 3 wires to run it (data, +, -).