weird ways to damage a ad620 chip

[DiTBho]

I don't know if this is the proper section of the forum, but the context is very unusual and informal, like a chat.

Yesterday my colleague was testing four ad620 units bought on eBay because found for cheap. Farnel lists a single chip for about 10 euro (plus postage), on eBay four units were purchased for 25 euro including postage.



The AD620 is a high accuracy instrumentation amplifier that requires only one external resistor to set gains of 1 to 10000.

Datasheet of AD620 says that this resistance (called Rg) must be calculated like this:

Rg = 49.4 kOhm/(G-1)
G = gain

As far as I see from his breadboard, he used 47 Ohm resistor, so gain is about 1000. Reference voltage is 0V, there is a single power supply 5V, and shorted V- V+ to the ground through a couple of 10K resistors.

Results? He said two chip units are damaged by these experiment, but I don't see why 

I have here a working circuit that uses a ad620 unit to measure the output of "KMZ10B magnetoresistive" sensor, which has a Wheatstone bridge. In this case the reference voltage is about 2.5V.

I removed my ad620 (which works) and tested the two chips used by my colleague: and booom! It's true, those two tested chips really don't work! Their Vout doesn't move from 3.8V, so probably there is a damaged input section or something.

But remember? He has four chips, and it's more interesting that the other two untested chips do work!

Seriously I don't know if and how his testing could damage a chip.

[DiTBho]

I have already tried to simulate his testing scenario like the schematic attached in the previous post, but TINA (it's like PSPICE) doesn't report anything weird.

[NiHaoMike]

My guess is something not reflected in the schematic such as ESD.

[Ian.M]

The AD630 is *NOT* a single supply instrumentation amp.  He's violating the permissible Vref input range, which must not approach either rail closer than 1.6V from it.

However there's a high risk that expensive chips from Ebay are duds to start with - ESD damaged, failed production test, remarked 'pulls' or even outright fakes - so his test jig, although incorrect, may not be responsible for the damage.

[DiTBho]

In my circuits, I never used Vref shorted to the ground. The data sheet doesn't warn you avoid doing it, but it tells to avoid high capacitance to the output.

Probably ESD, I don't know if he worn an anti static bracelet, for sure the ad620 has high impedance inputs, not sure if it also has strong protective diodes.

X-FILE   

[DiTBho]

The AD630 is *NOT* a single supply instrumentation amp.  He's violating the permissible Vref input range, which must not approach either rail closer than 1.6V from it.

The ad620 can work with single supply, but for single supply voltage of 5V, the max output swing is only from 1.1V to 3.8V.

Look at the schematic, I don't see how connecting a couple of grounded resistors to the two inputs could violate the input range and cause a damage.

[DiTBho]

Anyway, as solution I have already ordered from Farnel a couple of AD8422 which has a wider operating range, and this time I will assist him personally.

I can't even imagine what could go wrong this time 

[JohnnyMalaria]

Yesterday my colleague was testing four ad620 units bought on eBay because found for cheap. Farnel lists a single chip for about 10 euro (plus postage), on eBay four units were purchased for 25 euro including postage.

Maybe that's why they were cheap?

[DiTBho]

Maybe that's why they were cheap?

it seems so 

One month ago I was surprised to see an envelop containing five "IXTH140P10T" sold for 35 euro including shipping. The IXTH-boy is a strong P-MOSFET, 100 V, 140 A, 0.01 ohm RDS_on, TO-247 package, a strong device for which Farnel asks 26,04 euro + postage for two of them.

Recently I have personally tested them. Five of five in perfect working order without a visible defect up to 10A of load.

Good! Happy ending! Or just lucky? Who knows, but for sure you are perfectly right, I have already different and different times to avoid buying electronic things on eBay, unfortunately for my colleague there is no worse deaf than those who do not want to listen.

[Ian.M]

The AD630 is *NOT* a single supply instrumentation amp.  He's violating the permissible Vref input range, which must not approach either rail closer than 1.6V from it.

The ad620 can work with single supply, but for single supply voltage of 5V, the max output swing is only from 1.1V to 3.8V.

Look at the schematic, I don't see how connecting a couple of grounded resistors to the two inputs could violate the input range and cause a damage.

He's also got Vref grounded which is only permissible (for normal operation) if -Vs (his Vee) is at least 1.6V below ground, but that's also tied to ground.  However I doubt that would cause damage as if it could, there'd be limits for Vref in the datasheet's ABSOLUTE MAXIMUM RATINGS section.

[NiHaoMike]

One month ago I was surprised to see an envelop containing five "IXTH140P10T" sold for 35 euro including shipping. The IXTH-boy is a strong P-MOSFET, 100 V, 140 A, 0.01 ohm RDS_on, TO-247 package, a strong device for which Farnel asks 26,04 euro + postage for two of them.

Recently I have personally tested them. Five of five in perfect working order without a visible defect up to 10A of load.

Good! Happy ending! Or just lucky? Who knows, but for sure you are perfectly right, I have already different and different times to avoid buying electronic things on eBay, unfortunately for my colleague there is no worse deaf than those who do not want to listen.

Easiest way to detect fake MOSFETs is to check the RDSon, then checking if it can withstand its rated voltage. Most often, a fake would have a much higher RDSon. The rated voltage check then weeds out those that are cleverly remarked parts with similar RDSon but a lower voltage rating.

[DiTBho]

Easiest way to detect fake MOSFETs is to check the RDSon, then checking if it can withstand its rated voltage.

Yup, the low RDS_on is exactly the reason he bought those MOSFETs, so it was the first thing I checked during my tests

[DiTBho]

He's also got Vref grounded which is only permissible (for normal operation) if -Vs (his Vee) is at least 1.6V below ground, but that's also tied to ground.  However I doubt that would cause damage as if it could, there'd be limits for Vref in the datasheet's ABSOLUTE MAXIMUM RATINGS section.

Thanks a lot for this clarification!

[DiTBho]

I have hust tried the other two AD620 units my colleague bought from eBay. These two were in perfect working order with my circuit (Vee=-5, Vcc=5V, Vref=2V) so I have tested them with my colleague's procedure, and it seems they are very fragile. Connecting Vref to gnd showed two effects

• it has forced the chip to eat 12x more current, I measured 12mA instead of less than 1mA
• it has - somehow - damaged the chip

I won't repeat this test with the two new fresh chips ordered from Farnel. I am also moving from AD620 to AD623, this one is better for my colleague's needs.

[amyk]

The increase in power consumption suggests latchup.

If you still have the damaged ones, there's some other forum members who might want to get die photos...