Crikey! Off the shelf Howland Pump

Just exactly what's being "pumped" there? Answer: nothing.

Perhaps current?

Maybe that's where the OP got it...his original reference calls it a current source by way of a differential voltage to current transducer. It's neither here nor there. I'm just not seeing any pumping going on.

Those look pretty cool, opamps and precision R's. How best switch them? Relay's? Analog switches for the

George H.

I have always heard of (usually constant) current sources referred to as a "current pump".

Electricity is a fluid, sloshing around inside the wires.

So a current generator is a pump.

If you know how your boards are going to be oriented, mercury-wetted dry re ed relays are the switches to use. They have to be mounted within 15 degree s of being vertical (otherwise the mercury droplet doesn't start off where it should) but contact resistance is low - as about 0.1R - and stable (whic h can't be relied on in regular reed relays (or at least not to 0.1R).

They also last a factor of ten longer than regular reeds - 10^8 operations rather than 10^7 - and don't bounce (due to the damping effect of the mercu ry film).

You may get offered orientation-insensitive mercury-wetted reeds - I certai nly was - but they seem to be difficult enough to make that nobody kept on offering them for more than year or so.

That OpAmp configuration was purportedly first devised by Brad Howland (of MIT), and came to be called the "Howland Current Pump".

Bloggs is our resident village idiot... never anything useful to say. ...Jim Thompson

Note the absolute tolerance of the resistors is +/-25% which is ok as long as it doesn't have to match to any other resistor.

That also means that the transimpedance, the basic gain of the Howland pump as shown, can be that far off.

Digikey wants about $4 for the LT1995. My little negative resistance current source thing, posted in another thread, can do 1% for about 25 cents. That's arguably a better deal by about 400:1.

The LT1995 seems to have a fatal flaw, the same one some ancient opamps tended to have: if you rail the output, it can draw a bunch of supply current and destroy itself. That's a _major_ blunder. They should download LT Spice and check for mistakes like this.

You'd think that IC designers would, by now, have a checklist of common hazards, and bring it to design reviews. We have such lists for our board-level design reviews.

At least they admit it, in boldface type. That's good.

It's been around since 1922 :-)

I like the giant H on their lawn

James Webb Space Telescope uses a 1.000 mA Howland pump for all passive telemetry channels.

I used 0.01% RNC90 resistors and a space-grade Op amp (but I have to dig through notes and schematics to remember which Op amp -- it's been almost 10 years since the design).

-Tom Pounds

Does that "Space Grade" op-amp come in a sealed vacuumed package?

And while we're at it, those 0.01% Resistors, what orbit position around the sun were they calibrated at and what time of the year ? I am sure the vacuum of space was also considered when the engineers were sipping the socky. Or was that cooking the sushi ?

Jamie

Why do you think .01% resistors are so strange? They're mainstream enough that DigiKey carries them.

I didn't say they were strange, it's just the whole message content as it started, it kind of sounded like an off the way, take a drink and spit out all over the wall, post!

Jamie

That does describe most of what you post here. Even krw does better.

I don't want to out do you with the idiotic twisted content you try to present here and make others beleive, that's your job and you are damn good at it. I wouldn't take that title away from you.

Jamie

The idiotic twist is in your understanding of the content, not in the stuff I present. The title you want to give me is your choice, but it's a trifle idiosyncratic, and best understood under the heading "Jamie proves himself to be an idiot, again".