INA measurements done practically does not match datasheet in theory

[fbp]

I have selected an INA848 IC instrumentation amplifier for a project for the following unique features observed in it’s datasheet that matches my requirement.
- Output Amplitude vs Freq graph - 0 to 5Vpp @500kHz with Power rails at +/-5V(Datasheet page 13)
Can go even from 0 to 12 Vpp @500k Hz with power rails at +/-15V
- Gain 1
- Slew Rate: 35V/us
- CMRR: 62dB @500kHz

The requirements I need for my project application are as follows:
- Input constant frequency: 500kHz
- Input Amplitude: 4 - 5 Vpp
- Ampifier Gain: Unity Gain (To have variable gain whenever needed)
- CMRR: Good
- Good Slew Rate

Datasheet: https://www.ti.com/lit/ds/symlink/i...15411&ref_url=https%3A%2F%2Fwww.google.com%2F

I have observed the following limitations in this IC when I test it practically on a breadboard.
- With power rails set at +/-5V, The output sine waveforms starts distorting at input of 2.5 Vpp @500kHz although the datasheet mentions that the output Vpp can go upto 5Vpp @500kHz
- When I test it using +/-15V power rails, the IC gets super hot and I feel that it wil burn off, so I’m guessing it would get defective when the IC gets super hot. It does not seem to match the datasheet. The datasheet mention on page 13 that at +/-15V power rails, at 500kHz the output can reach upto 12V pp but the IC starts getting super hot when I start increasing the the power rails from +/-5V

Please assist me. Is there something I am doing incorrectly during the test setup? I have checked my connection multiple times, but it always gives me the same results. Have tested it with two different ICs, both give the same results.

Any help would be highly appreciated.

 

[wwfeldman]

see page 5, VCM - operating input range
and see figures -2 and 8-3
both indicate that the output tops out less than the rail.

page 13 says Vs = +/- 15V, and Vcm at mid supply, which means about 0V

this is NOT a rail to rail device - the spec sheet never says rail to rail

as for it getting hot at +/- 15V, you'll have to provide a schematic

 

[fbp]

Thank you for your reply.

For the schematic, I am using the EVAL AD8224 Board. I desoldered the Dual INA AD8224 IC, and have implemented the INA849 IC using small wires (since the IC pin numbers do not match) just for testing the IC.

Schematic:
EVAL AD8224: https://www.analog.com/media/en/tec...ation-documentation/854884503AD8224_EVALZ.pdf

My input is from a sensor (which gives me a constant 500k Hz frequency and approximately 3 to 4 Vpp output from the sensors). The positive and negative of the sensor is connected to the input of the +/-INA.

The INA starts getting hot after I increase the input rails from +/-5V

 

[fbp]

I am not sure why I am not able to go to above +/-5V rails on my IC849.

Please assist me.

 

[FvM]

I prefer to ignore the cross-posted same topic thread which should be deleted.

Nothing has been said yet about output load, but that could be the problem. In another related thread, you are mentioning 2 m cable and a transformer driven. Please show an application circuit with real input signal source and complete output load.

 

[wwfeldman]

I am not sure why I am not able to go to above +/-5V rails on my IC849.

then how do you know it gets hot at +/-15V?

soldering surface mount parts by hand is a tricky business.
run a sharp X-Acto knife carefully between each pad, where you added a wire, on your evaluation board
also, run it between the leads of your INA849
the goal is to ensure no pins are accidentally tied together - sometimes solder splashes or whiskers cannot be seen

 

[fbp]

then how do you know it gets hot at +/-15V?

soldering surface mount parts by hand is a tricky business.
run a sharp X-Acto knife carefully between each pad, where you added a wire, on your evaluation board
also, run it between the leads of your INA849
the goal is to ensure no pins are accidentally tied together - sometimes solder splashes or whiskers cannot be seen

When I touch the IC, it starts getting hot exponentially as I start increasing the power rails from +/-5V to +/- 15V

Yes I did check connectivity across each line and ensured nothing is shorted.

I just wanted to ask, if I am reading the datasheet correctly or not.
With a 500kHz input and an amplitude of 4Vpp, can I get the same output of 4 Vpp, if it a unity gain amplifier. This is my main goal to achieve from this IC.

My Vcm is 0 since I have connected my sensor-positive terminal to the non-inverting of the INA and the ground of the sensor to the inverting of the amplifier. (with a transformer between the sensor and INA to ground the common mode noise in the cable wires from the sensor to the transfomer)

Just to confirm, is Vcm calculated as follow
Vcm=(non-inverting+inverting)/2 = (5Vpp - 0)/2 = 2.5 Vpp
Or does Vcm not exist because the transformer is eliminating the common mode noise

 

[KlausST]

Hi,

usually the datasheet says to some traces "to be as short as possible" this does not mathc the current solution.

Why don´t you post a schematic of your wiring? Textual description is not useful, it leads to misunderstanding and guessing.

For debugging ... as a first step: Just use a DVM and measure each INA pin (directly at the INA) with respect to PCB_GND.
Report each single pin voltage.

Klaus

btw: Using PNG for photos is not a good idea.
This is less than 5% of file size ... losing no thread realted informations.

 

[fbp]

I have attached the schematic below

 

[danadakk]

pin 5 and 16 both show +Vs, is that correct, no negative supply to board ? In you schematic.....

Regards, Dana.

 

[fbp]

There is a negative Vs, it on Pin 8 or 13. Sorry I didn't mention it in the above the schematic

EVAL AD8224: https://www.analog.com/media/en/tec...ation-documentation/854884503AD8224_EVALZ.pdf
The IC I used here is INA849

 

[KlausST]

Hi,


also pin#6 and pin#15 show the same signal "REF_A", so they are short circuited.
If pin#5 is not GND it misses a capacitor.
I did ask for your (underlined) schematic. For a good reason.

I´m too tired for games.
Good luck.

Klaus

 

[dick_freebird]

What is needed here is less advice, than bench debug skills and discipline.

As to the power dissipation "problem" the 4V P-P output (if I skimmed the
original complaint well enough) required the same output power but the
output stage, if you think about it, sees the same current but the 3V Vds
(or Vce) at +5 becomes 13V at +15 and that's all loss. By now you should
have a number for your output current at both high and low peak. Do the
math for the two cases.

If that "2m cable" is driving the 'scope as shown then presumably you have
the internal or a bullet 50 ohm termination. Presuming output centered on
'scope ground (and do you know for sure there are no weird ground offsets
or currents? I trust nothing particularly about the reveal of the power scheme.

So back to the suffering output stage and its 2V (ideal) to 4V (single ended
positive ground referred) you could have 40mA (2V) to 80mA (4V) and
the power shed internally is best case 120mW at 5V and 520mW at 15V.
Of course your laser-focus attention to datasheet detail tells you this is
any kind of OK for a small outline op amp trio in still air, yessiree Bob.

Many instrumentation amps use a A=+2, A=-1 sum and this means the
input difference cannot exceed Vrail/2 (and less any specific headroom
requirements if not RRIO). Your input signal you said was 2.5V? Bonk.


Seems to me you would be happier with Vector board and DIP packaging
until you get the plan to work. You don't seem to be challenging frequency
or isolation, just looking to get the sensor reading right. Right? I say ditch
the eval board with extra cheese and just hand build it, signal flow will
tell you the best arrangement and likely less spaghetti than now (wire
coupling no bueno, neither).