That's the plan - I imagine I could end up using it for quite a few other future designs. I better make sure I accommodate for the various pin-headers that the different designs use, including some of those fancy flexible/strap leads (as per the photos you showed above)

BTW, the offer on the 6.5 digit meter for calibration of the voltage refs is still there if it's any use to you.

John

Originally Posted by pldaniels

That's the plan - I imagine I could end up using it for quite a few other future designs. I better make sure I accommodate for the various pin-headers that the different designs use, including some of those fancy flexible/strap leads (as per the photos you showed above)

Ah yes, had forgotten about that - thanks.

I was planning on getting some PCBs done for the new Linear branded refs that I have, putting 1.25, 2.50 and 5.0V into the same box (all isolated of course, running off a 9V drycell).

Paul.

Hopefully the 24-bit ADCs will arrive today and we can start to get a real idea on how far this chip will go before the noise takes over. I'm planning for 20-bit resolution which is a full order of precision higher than the 16-bit.

I've also sent off an order for a PCB I drew up last night to be a graphics LCD driver, again with an AVR mega on it, though I might get away with a mega48 (4K should be more than ample I hope to hold 2 sets of bitmaps for digits, 20px and 8px high). The board won't arrive for probably 2 weeks or more.

Paul.

Oh yes, I forgot to say that after I had routed everything, I barely had 2 or 3 pins left over on the gfx driver board, so there's certainly no way I could have done both the LOM aspect and the gfx-LCD aspect on the same chip in this case, at least not without moving to a bigger AVR (mega32 or such)

24 bit ADCs arrive... still waiting on some 'odd value' resistors to arrive (different supplier).

Good news everyone, I received my Kelvin leads today... they've got shielded BNC connectors on them but I think I'll use that to my advantage for the high precision unit.

Hmmm.. neat. Are they pure Cu or Cu plated tips?

John

Originally Posted by pldaniels

Good news everyone, I received my Kelvin leads today... they've got shielded BNC connectors on them but I think I'll use that to my advantage for the high precision unit.

Not sure, looks like Au plated to me. Still fortunately at least with the nature of the 4-wire setup it's not a big issue, so long as they don't corrode severely.

Here's the high precision one showing the resistance down to 0.0001 resolution, I could in theory go to 0.000001 (that's 0.01mR) but it'll be dominated by noise and probably explode the memory requirements of the AVR (as I'll have to deal with 64-bit math)

So when's production likely?

Originally Posted by pldaniels

Here's the high precision one showing the resistance down to 0.0001 resolution, I could in theory go to 0.000001 (that's 0.01mR) but it'll be dominated by noise and probably explode the memory requirements of the AVR (as I'll have to deal with 64-bit math)

Not sure when production will start... or if it will.

Just gone insane tonight as I did -something- and now the math is all wrong, I'm out by 1 bit somewhere and it's driving me bananas! Perhaps I should plug in a 16x2 LCD for the short term and fill the rest up with raw bit-dumps

Well, I got it back up and running "normally" again. I did have to use 64-bit math which ballooned the firmware size from ~4K to 7.9K. I can say already that beyond 0.1mR resolution is incredibly noisy - even the 0.1mR resolution is something you would want to take much longer term samples of.

A wiser man than me once said, "The best solution is the one that gets the job done - just."

0.1mOhm from something this simple is pretty impressive. Better than that and we are starting to talk Lab gear and big money surely?

John

Originally Posted by pldaniels

Well, I got it back up and running "normally" again. I did have to use 64-bit math which ballooned the firmware size from ~4K to 7.9K. I can say already that beyond 0.1mR resolution is incredibly noisy - even the 0.1mR resolution is something you would want to take much longer term samples of.

I adjusted the firmware a bit more and made the uC go to sleep during the time that the ADC is doing its magic (about 140mS/sample), it's cut down the noise a little bit and I suspect once I mount this board into a metalised chassis and load up the front panel with BNC connectors with common grounds etc etc it should become useful.