XTR111, saturated output

Hi Stefano,

What is the impedance of the load you're driving with the XTR111 output?  This sounds like a voltage compliance issue with the current source.  The max current you can create from a current source is dependant on the supply voltage and the resistive load.  Basically Ohm's law can't be violated so a quick estimate of the maximum output current can be calculated by dividing the supply voltage by the load resistance.  I_max = Vsupply / Rload.  In reality there are some other terms at play such as the saturation voltage of the PMOS device and the resistive (I*R) drop across the two 15 Ohm resistors in series with the output, but the calculation will give you a good estimate of the max current flow.

So with a 24V supply is it possible you have a load resistance of around 1.24kOhms (24V / 0.0194A)?  This would cause the current source to compliance.  Try using a 250 Ohm load and see if you're able to achieve 20mA. 

If that doesn't fix it then it is possible that your current limit circuit (Q1 + R44) is becoming active.   With only 20mA of output current, the circuit shouldn't become active because the drop across the 15 Ohm resistor will only be 0.3V (0.02A * 15R) which shouldn't be enough to forward bias the base-emitter diode (>0.6V) in the transistor and turn it on.  Are you sure you have a 15 Ohm resistor in the circuit for R44?  If so, try shorting it out or reducing it to 1 Ohm and see if that fixes the issue.  If reducing the resistance or shorting it fixes the issue then you'll have to redesign the current-limit circuit.  Please confirm which PNP device you're using and we'll work the circuit out. 

Colin,

as my circuit already has a 15 ohm resistor on its output side, I did not connect any additional load, and I'm measuring the output current by simply connecting the output pin (IOUT1 on the schematics) to GND through an Ammeter.

I also tried to reduce the resistance on the base-emitter diode of the transistor, by adding a 10 ohm resistor in parallel (6 ohm total), but nothing changed. I don't have resistors smaller than 10 ohm, at the moment, but as I see EXACTLY the same current with 15 or 6 ohm, I guess that's not where the problem is.

The PNP transistor used is KST2907A from Fairchild, as suggested on XTR111 data sheet.

Thanks again for your help.

Stefano

Hi Stefano,

Thank you for running the tests, I'm a little puzzled that the issue wasn't resolved and I agree that your test of the external current limit was sufficient. 

I have a few remaining ideas. 

First, please reduce the value of RSET by 1/2 (1.25k) and see if you're able to achieve greater than 19.6mA on the output.  It should only take 2.5V to create 20mA with RSET = 1.25k, so see if you're able to get 20mA or greater with this different circuit configuration.  This test will help me determine if the issue is with the internal current limit of the XTR111. 

Second, the only connection not shown on your schematic is for the error flag (/EF).  What voltage is this signal pulled-up to, and what resistance do you use for the pull-up resistor?  Does the /EF signal go low when your output gets "stuck" at 19.6mA?

Let us know what you find.

Hi Collin,

we put a 2k resistor in parallel with the existing 2k5, and found that the output current saturated at a value of about 43 mA (this seems to match the data sheet specifications). Eventually we removed the 2k resistor and put only the 2k one, and the current output saturated at 24.3 mA, with an input of 5 V.

This number sounds a little familiar: according to the gain formula, with 5 V input and 2k Rset we should have 25 mA, which is about 0.7 mA more than what we have in the actual circuit. That's the same behavior we have for a 2k5 Rset (there we had 19.4 instead of 20 mA), so I think there must be something somewhere which draws those 700 uA...

Our next step could be removing the protection network, in order to see if the TVSs do generate this problem, unless you have some more ideas. In the attachment you can also find the protection network schematics, just in case.

About the error flag, it's quite difficult to check, because we have two XTR111 on the same board, the two error flags are wired ORed together and pulled up to a digital Vdd which, at the moment, cannot be turned on (because of other problems found on the board), so we cannot say whether they are set or not.

Thanks,

Stefano

Hello Stefano,

Thank you for your continued debug efforts, this new data does help.  At this point it sounds like the XTR111 is likely not the issue and there is something else in the circuit pulling the current as you've suggested. 

That said, I looked at your protection scheme and don't see any issues.  The SMAJ43CA diodes will have mimimum reverse leakage even at 24V and the RB160M-60 has a max leakage of 50uA at 60V reverse voltage so it wouldn't seem like these devices would be causing up to 700uA of leakage.  However as a quick test idea, what if you took an ammeter and connect one end to pin 2 of the JP1 jumper and connect the other end to GND?  This should create a straight path to GND that doesn't go through the protection network.  See if you get all of the expected current when testing as described.  If that isn't convenient or doesn't work then you could just remove all of the protection circuitry as well and see if the issue is resolved.

Let me know what you find and we'll probably be pretty close to finishing this one up.

Hey guys, I have just been looking at the XTR111 and I saw this thread, and thought I would make a comment...

It sounds to me like a simple resistor tolerance issue. In the schematic you have a 2.5k resistor, and you are maxing out at 19.4mA at 5V. Could this be simply because the resistor you have is not actually 2.5K, but a little more? What is the tolerance on the resistor? A 2-3% error will give the output you are measuring.

If you cannot resolve the issue, you could just increase the resistance slightly to compensate. :P

Hi Carl,

the 2.5k resistor is a very high precision one. It's a Vishay VSMP with 0.01% tolerance, so we are sure it's not the problem. Moreover, we found the problem to be present on 4 identical blocks, with exactly the same symptoms; if the resistor tolerance was the problem, we surely would have seen some differences.

At the moment, we are involved in other designs, so this one is left behind; but my suspect is some math error to be found somewhere. I will surely look better...

Regards,

Stefano