Instrument Amp and C-load

Hi all, had a bit of a weird instrument amp oscillation today. (Oscillations were intermittent and triggered (sometimes) by the heat relay in my soldering iron switching on or off.) The int-amp (AD620) is looking at a voltage pulse on a 50 ohm resistor. Today I added a long (DB9) cable between the

50 ohm resistor and circuit board, and this resulted in the intermittent weirdness. Not a clean oscillation, but more of a rectified wave with AC coupling. Freq. 1 MHz (or 500 kHz depending on how you count.) Here's a 'scope shot.

formatting link

Anyway after much fussing I figured out it was the rather long coax cable I had hanging on the output of the int-amp* that was causing the problem. I added a 100 ohm series resistor and this cured the problem. But I'm wondering if there is some other (better) fix. (Short of adding a buffer amp to the output.) Is there some way to slow the int-amp down? A little resistance in the reference connection to ground?

Thanks

George H.

*just to be clear the coax is distinct from the long db9 cable added to what is the input to the int-amp. I needed both to have a problem
Reply to
George Herold
Loading thread data ...

So, no RF filtering at all, where the cable comes into your thingy?

Not that you normally expect op-amp inputs to act as outputs, but alas, real op-amps are hardly as "OP" as the textbooks would like them to be.

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
 Click to see the full signature
Reply to
Tim Williams

IIRC the AD620 is one of the many amplifiers that doesn't like driving a capacitive load. See AoE pg 264 for alternatives to the

100ohm output resistor (though your method is the simplest, especially for this amp).
Reply to
Frank Miles

I tried some simple RC filters on the input... it did nothing to the oscillations.

George H.

Reply to
George Herold

Right I've done that type of stuff (AoE3 pg 264) with opamps, but there's no way to get into the 'inner workings' of an instrument amp.

I guess a buffer is the best, but I would rather not re-spin the pcb. George H.

Reply to
George Herold

Some amps are unstable when loaded by a range of caps, and can be cured by loading them with a huge cap. Another fix is to load them with a cap+series resistor, or a cap with some ESR, like a tantalum or aluminum.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

Yeah, I've got a little circuit board I can hang on the output, RC low pass. It's laid out for a 10uF Y/Z ceramic. (Not an elegant solution.)

Say I just want it to work for a range of typical coax lengths. No more than six feet. (less than 1 nF!)

If I put the right inductor on the reference output...? Won't that decouple the HF?

OK maybe a board re-spin... got a favorite buffer driver? Low DC offset... 100uV. (I'll most likely pick some opamp in stock.)

George H.

Reply to
George Herold

A different instrumentation amplifier is a possible solution

formatting link

Every now and then you find somebody who has tried to save money by using a cheaper instrumentation amp than AMP-01 and can't understand why the cheaper circuit oscillates.

--
Bill Sloman, Sydney
Reply to
bill.sloman

A 100-ohm resistor is my go-to solution for driving some cable with an opamp output.

Hah, The RIS-767 test-set board I recently sent a few s.e.d. blokes, has two instrumentation amps. These were for an Arduino to measure the in/out currents. But there are two wiring points, so this signal can be grabbed for off-board use. R Legg connected them to his 34970A logging voltmeter, and got noisy results, until he added series resistors to the INA outputs.

You know the op-amp story, Zout = Rout/A, where Rout is usually about 200 ohms and A is the open-loop gain. This gain is falling with frequency (it's unity at fT), so Zout is rising with frequency. Zout looks like an inductor, so we add a capacitive load, and whammo!

A fancy two-resistor-plus-cap circuit would be nice, but as you say, that's not possible with an INA. A simple resistor to dampen the Q works just fine.

--
 Thanks, 
    - Win
Reply to
Winfield Hill

On a sunny day (1 Jun 2017 06:31:51 -0700) it happened Winfield Hill wrote in :

The theory why the opamp oscillates is correct. But also when driving a coax we should ALWAYS drive it with its characteristic impedance, and terminate it with that at the other end. So the [ideal] amp should have a gain of 2 (to compensate for the voltage divider thus created) and as it is zero ohms out if it is a GOOD amp, a resistor in series that _matches the cable impedance_.

Whatever is in the 'perfect amp' at that point does not even matter. And then there are also the _reflections_ from an improperly terminated 'long (says OP)' coax, that may severely change what you expect. End of lecture.

Reply to
Jan Panteltje

Termination at the far end is optional for lowish frequencies. Any reflections get absorbed. And G = 2 isn't required and, very important, full output swings are fine with ordinary op-amps.

--
 Thanks, 
    - Win
Reply to
Winfield Hill

On a sunny day (1 Jun 2017 07:37:16 -0700) it happened Winfield Hill wrote in :

You almost sound like an audio guy :-)

Reply to
Jan Panteltje

Source termination (50 ohm drive resistor, open load) works at any frequency. And the net voltage gain is 1.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

Thanks Win, I mucked about a little with some series inductor but that didn't work. I was wondering if the 100 ohms made the output stable for any C-load. And now I'm dreaming of a 'twiddle box' with various R's and C's.

George H.

Reply to
George Herold

Hi Jan, Well I use coax and bnc connectors for all sorts of DC/ audio signals. 50 ohm is not really needed. In this case I think the fastest edge I see is 10-100 us. No ringy dingies on the 'scope display.

George H.

Reply to
George Herold

On a sunny day (Thu, 01 Jun 2017 08:34:41 -0700) it happened John Larkin wrote in :

That is true, but there is something like SWR, this site gives nice examples:

formatting link

For maximum energy transfer to the load, the load must have the same value as the source, and as the cable impedance.

And your opamp would not like it / could be destroyed by a bad SWR. That fault condition is common in transmitters, that is why I have an SWR alarm beep on my SWR meter, antenna connections has been blown lose several times in storms.

Reply to
Jan Panteltje

On a sunny day (Thu, 1 Jun 2017 09:04:12 -0700 (PDT)) it happened George Herold wrote in :

Well if you use 50 Ohm coax why not drive it with a 50 Ohm resistor / impedance? I would not dream of connecting it directly to the output of an opamp. For DC... OK.

Reply to
Jan Panteltje

For signal transmission, we don't need maximum power transfer. It is nice to have a baseband gain of 1.000.

Very unlikely.

An instrumentation amp is not an RF transmitter!

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  
 Click to see the full signature
Reply to
John Larkin

Yes. Series terminate at the source end. That should keep the in-amp happy. Leave the other end hi-Z.

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  
 Click to see the full signature
Reply to
John Larkin

If the source impedance is not just resistive but has a reactive component, you get maximum power into the load if the load is the complex conjugate of the source impedance, i.e. the same real part but the negative of the reactive part.

Sometimes you don't want to extract the maximum power from the source, e.g. the mains power socket in your house may have a source impedance of a fraction of an ohm. You don't want your light bulb or electric heater to have the same impedance as the source, even though that would extract the maximum possible power (for a very short time!). Even if it did not blow the fuse, the efficiency would be only 50%, whereas with a mismatched load having a higher impedance than the source, the efficiency can be much closer to 100%. Even for radio transmitters, it is more economical if half of the power is not being dissipated in the source impedance. If the antenna is well matched to the feeder cable then there is no reflected power into the transmitter and its output impedance does not need to match the characteristic impedance of the cable. If you really do need to match both ends of the feeder cable, you can use an isolator or circulator at the output of the transmitter to absorb reflected power. Those are weird, with magnets and ferrites and stuff.

Reply to
Chris Jones

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.