reg + reference

That would need drivers, output rectifiers, all that stuff. I can buy a sip7-packaged DC/DC converter for $7 or so, all done, second-sourced, with a very small pcb footprint. SIPs use height to save board surface area.

We have used ISDN transformers for isolated power. They are small and very cheap, under $1.

Each channel can be programmed to...

Output 4-20 mA from an internal supply (drive an indicator or controller) and report voltage

Regulate 4-20 mA with external loop excitation (simulate a transducer)

Measure 4-20 mA as an ammeter

Measure voltage as a voltmeter

Output a programmable voltage and measure current

Be an open

Be a short

Self-test

The ARM per channel lets us run various fast internal PID loops to make all this happen. We'll measure loop current and terminal voltage and use a DAC channel to drive a mosfet, and servo whatever is appropriate. The ARM will also do fast shutdowns to protect us from customer abuse, like putting 50 volts across us in ammeter mode.

12 ARMS running at 100 MHz, plus a 13th to be the VME master, should be enough compute power.

We invented this product last week, during a conference call with a customer, and they want units in 60 days.

John

On a sunny day (Sat, 27 Mar 2010 10:49:58 -0700) it happened John Larkin wrote in :

loads.

Just stick one on the meter and see for yerself? You are the one who always talks about using parts out of their specified specs.

loads.

If the typical TC blows my error budget, and the guaranteed TC is many times worse, there's no way "using it out of its specified spec" is going to help.

John

That's the part. I've got it running now, just starting to take data. Pulsing 7 msec on at 1 mA, 1 Hz, the mosfet ON current seems to peak at about 15 volts, running maybe 1.3 amps. It's not warm, so I guess I'll go up.

Turnon time is about 5 msec to the 1.3 amp flat-top, faster if I drive the LED harder. Drive doesn't seem to affect the mosfet limiting current. Turnoff is fast, 200 us maybe, independent of LED drive level. That's good... we can shut them off fast before the smoke leaks out.

More voltage!

John

Hi, Spehro,

Actually, the dynamic impedance of an LM4040-3.0 is low enough - about

1 ohm - that I can run it from the 1117 output through just a resistor. I'll probably run the 1117 output at about 3.4 volts to make sure the drop in the resistor is reasonable. 400 ohms would put about 1 mA into the 4040.

John

How about something like this?

Best regards, Spehro Pefhany

--
"it's the network..."                          "The Journey is the reward"
speff@interlog.com             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com

Except that these 'voltage refrences' are normally bandgap references, very stable over temperature and voltage but have low accuracy in absolute terms. You'll need to calibrate it to get any decent absolute accuracy.

Mark.

1% tolerance is pretty bad for a precision reference. Here's a better one:

Not avaialble at 3V unfortunately, but +/-1mV at 2.5V is 0.04%! 20ppm per deg C temperature coefficient max.

Mark.

Artemus a écrit :

That's just to make copycats life harder :-)))

--
Thanks,
Fred.

Just a silly comment re the labeling of the pins of the 1117 in your dwg. Art

LOL. Didn't even notice that. I tend to swap digits in long incomprehensible part numbers when I'm tired & in a rush.

Best regards, Spehro Pefhany

--
"it's the network..."                          "The Journey is the reward"
speff@interlog.com             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com

We will have to calibrate each channel. My target is 0.1% accuracy, and I'll be buying lots of 1% sorts of parts. The big concern is TC. The LM4040-3.0 that we have in stock is spec'd at 150 PPM max, with a typical of 15! I think somebody, Analog maybe, may make better

2-terminal bandgaps.

John

Tolerance doesn't matter, as we'll be calibrating channels. The low TC is appealing.

Those floating-gate things are interesting, and I could run the ADC ref at 2.5 volts, but their dynamics looks goosey. Too big a risk for me on this project.

John

Well as i noted in another post 1% tolerance is pretty bad for a precision reference. Here's a better one:

Not avaialble at 3V unfortunately, but +/-1mV at 2.5V is 0.04%! 20ppm per deg C temperature coefficient max.

Mark.

Haven't been following this thread too closely, but these look like pretty nice references and may be of use:

I find NXP documention for the digital/firmware side very good, but the "analog" specifications seem very much an afterthought. ISTR the ADC changed from 1M SPS in the initial publicity to 200k SPS in the datasheet, Its the opposite with the Analog Devices ARMs. They have good analog documentation (and performance) and comparitively underpowered digital peripherals and CPU. They do models targeted at 20mA loop-powered applications.

I would say the same applies to the actual parts, too.

--

John Devereux

Thanks. That means I'm not getting old :-) I also recalled the LCP17xx ADCs to do 1MSPs not 200kHz. Weird. The LPC2000 series ADC can be used up to 500kSPs.

--
Failure does not prove something is impossible, failure simply
indicates you are not using the right tools...
nico@nctdevpuntnl (punt=.)
--------------------------------------------------------------

Larkin

temperature?

has

Could you elucidate on the coldfire being seriously ill? Preferably with= =20 references / links, as it may impact my employer.

it=20

out=20

will=20

Hmm. My experience with wire-wrap has been uniformly good. Also like=20 soldering and crimping, it must be done right. 30+ years ago i was = certified=20 for solder, wire-wrap and crimp.