Newer Model Instrumentation Amp

Over 40 years ago i designed and built an EEG-type module for brain wave monitoring and training (alpha wave if i remember correctly: OOoommm). Used a single uA709 with JFET input stage; radical compensation on the 709 and RC input lowpass filtering for net result of open-loop gain showing at the peak (pole and zero was widely separated for that). Used dimes for the pick-up pads, 2 9V batteries in case about size of cigarette package; worked like a champ.

Thanks for posting a real-world design, it's nice to see that here.

Items I found "Interesting":

- I seem to recall you mentioning the use of the Supertex parts, not seen them used like this before. Would have guessed the fault current was too high, but in fact it looks like it is only a couple of mA.

- Use of latching relays and their driver. Never used latching ones myself (or relays at all for that matter in low power circuits)!

- There seem to be twice as many gain setting switches/resistors as you need. Is that for CMMR/layout symmetry reasons?

--

John Devereux

Thanks.

Cluck. Peck. It doesn't have much in the way of RFI suppression on the input. Cluck. Cluck.

--
 Thanks,
    - Win

Oh, sorry, I missed L25 and L26. What are they?

--
 Thanks,
    - Win

Look like 0603 ferrite beads 1K @ 100MHz, so it should roll off at maybe 10MHz with the 47pF caps.

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

On a sunny day (Sun, 20 Jun 2010 15:38:40 -0700) it happened Jim Thompson wrote in : tp://jjlarkin.lmi.net/22S490B_ch12.pdf

But you design chips no? So could you design a chip that does all that, maybe Larkin would buy 10?

This would be difficult to do as an IC. Especially if it had to meet my input range (10 mV to 40 volts), overload (250 volts), drift, and common-mode requirements. I even had to trim the CMRR of the INA154, which is already a laser-trimmed thinfilm thing.

Note the +-16 (actually 17) supplies.

But if someone did try it, they may as well include the next page: clamp, 4-pole lowpass filter, 16-bit ADC with reference.

John

It's interesting how few designs actually get posted to s.e.d.

These run pretty consistantly about 1.4 mA Idss. Small-signal Ron of the pair is close to 2K, a bit higher than I'd like. 4K of Johnson noise is about 8 nv/rthz, OK but not great.

These are the cute little surface-mount Fujitsu telecom relays. Since the coils are mostly off, there's no heat source to generate thermal EMFs in the signal path. They are a nuisance to drive, both hardware and software. I guess I could have used non-latching relays, with the activated state kicking in the HV attenuator, so they'd be off on the millivolt ranges. One ADC LSB is about 80 uV.

It improves the AC CMRR and distortion compared to something lopsided. Layout-wise, I'd prefer fewer parts.

John

Right. That should be pretty good at zapping the sort of RF that messes up opamp front-ends. Usually just the bead helps a lot. That's all pretty much guesswork anyhow. Too much C can mess up higher frequency CMRR, so, who knows?

1/3 of the customers insist in grounding shields at the transducer, another third insist on grounding shields at the VME end, and the rest get it right.

John

Young Buck Alarm: Driving Inductive loads directly from CMOS logic is NOT recommended without some kind of protective elements. Allowing CMOS body diodes to be exposed to flyback currents can be hazardous, possibly fatal to the chip if latch-up occurs. Observe data sheet current limits carefully. ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |

            Where is Joe McCarthy when you need him ??

Good advice in general. I've used emitter followers from CMOS for the unipolar latching type, which don't require diodes.

Not sure which ones JL used, but some of those relays only draw 8mA or so at 4.5V, and the diodes are shunted by MOSFETs under normal conditions so it ought to work okay. In pathological conditions (say a sudden drop of Vdd to 0 during the brief pulse when the relay is energized) the diodes would conduct, but something like 8mA maximum is pretty conservative for that particular part- data sheet says 100mA or

50mA.

A 20-cent H-bridge...

What John did is AOK, the SN74LVC2G02 is rated for ±24mA drive, so I'd expect the body diodes can cope. These "simple" gate parts have few, if any, sneak paths for latch-up. Just don't try it with a complex chip.

Yep.

I've recently been involved in designing chips for security cameras.

Besides the usual video functions, there are actuators, for example... iris control and day/night filters.

So I've developed a special-purpose H-bridge for such applications that needs no diodes to cope with flyback.

As usual I can't divulge details until the patent applications are in place :-) ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |

            Where is Joe McCarthy when you need him ??

The relays are Fujitsu FTR-B3GB4.5Z-B10, 4.5 volts at 25 mA. Wonderful little parts, second-sourced by Omron. They will turn on/off in under a millisecond if you drive them right.

Yes, this driver is prefectly safe from latchup, for the two reasons you mention. We've shipped about 8000 so far, no problems.

This is the second time in recent days that this old hen has criticized my design suggestions without bothering to understand them first. Or bothering to be constructive.

John

Do pay attention, John! ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |

            Where is Joe McCarthy when you need him ??

On a sunny day (Mon, 21 Jun 2010 09:13:14 -0700) it happened John Larkin wrote in :

That is not fair, he just gave good advice, to check the currents. He could not know the currents, and to be fair it may work, just as did that blind shear in the blowout preventer for BP. Question is: Will it always work? Some chips can go into latchup if you spike them on pin. I am not saying this is the case in your design, but Jim's advice is sound.

On a sunny day (Mon, 21 Jun 2010 09:10:42 -0700) it happened Jim Thompson wrote in :

You are evading the subject: Where is Osama?

If you look at the datasheet for these nor gates, it says in the headline text, huge font, no latchup at 100 mA. A tiny relay like this is not going to use 100 mA of coil current. And, as Spehro points out, the cmos fets are always on.

This one is OK. We checked.

John

FWIW, Supertex's datasheet says "not recommended for new design." They suggest LND150, which looks identical.

500Vgs ... cool.

-- Cheers, James Arthur

I've made a point about this here a few dozen times over the years. "My favorite" LND150, I usually called it.

Right, 5.6nV * sqrt 2 = 8nV for two 2k parts. BTW, I like to add a 1k resistor to control the current, in a scheme that only needs a single resistor, hence less noise impact. Others have also thought of this; I grabbed the drawing below from a 2002 post by Adam Seychell. Only I use 1k, 200 ohms won't do much.

. ,-------------, . | | . ------- | . connector | | | S D . from o---+ +-+-- 200R--+-+ +-----> to non-inv . outside D S | | | . world | ------- . | | . '-------------'

They simply renamed the LND250, which was the elegant SOT-23 version of the TO-92 LND150, to the LND150K1, and then the old TO-92 part becomes the LND150N3, and they also added a TO-243 part, LND150N8, which can handle more power. Nice.

This is all in keeping with Supertex' longstanding numbering scheme, and it makes my remark easier, because now I don't have to say, my favorite LND150 and LND250. With all the choices, now the LND150 really is my favorite!

--
 Thanks,
    - Win