EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: Mark on November 03, 2023, 03:57:48 pm
-
This circuit (attached) takes a 0-1mA input and converts it to 4-20mA loop powered at 24V.
Some BOM items:
IC1 = LM358N
CR2 = 1N5300 (1.30mA Current Regulator Diodes)
Q1 = HARRIS D44C2 or SOLID STATE D44C8
CR3 = LM329BZ (6.9V Precision Voltage Reference)
As you can tell from the scan, this is old. It has worked for decades, but now it draws around 90mA when powered up from a fast-rising supply.
A slow ramp up to 24V does not show the issue.
I don't want to muddy the waters at the moment, but does anyone have any pointers?
I had thought it might be latch-up or output phase reversal but I'm not sure now.
thank you.
Edited to clarify: this circuit thas been in production for decades without issue. Only recently the problem has been observed on newly built circuits. Looking at the parts list, a few things have gone obsolete over the years and replaced by equivalents (mostly same part number, different manufacturer)
Alternatives are listed for IC1:
NSC LM358N
ST MICROELECTRONICS LM358N
FAIRCHILD SEMICONDUCTOR LM358N
-
Check the value of C2 - it looks as though it's there to slow the rate of rise of IC1's supply. Most OPAMPS go through funny states as they power up.
-
Check the value of C2 - it looks as though it's there to slow the rate of rise of IC1's supply. Most OPAMPS go through funny states as they power up.
I tried adding another 1.0uF across the op amp but no joy.
edit: even another 10uF doesn't fix the problem.
Even if it worked, wouldn't extra capacitance slow the response of the circuit (to the 0-1mA input)?
-
Any overshoot on the power supply?
-
Any overshoot on the power supply?
No overshoot. The 90mA issue is observable down to about 17V with a fast rising edge on the power supply.
With <17V it starts up with the expected 4mA.
-
As you can tell from the scan, this is old. It has worked for decades, but now it draws around 90mA when powered up from a fast-rising supply.
A slow ramp up to 24V does not show the issue.
I don't want to muddy the waters at the moment, but does anyone have any pointers?
I had thought it might be latch-up or output phase reversal but I'm not sure now.
If the physical device worked for decades, but now it doesn't, then you are looking for a mechanical failure like a bad solder joint, or a component failure like a capacitor drying out.
Your approach should be to carefully examine all solder joints with a good magnifier and reflow any suspect ones. Then, start replacing components one by one based on the probability of failure. You may need to remove them from the board and test them, or just replace them regardless. Start with the cheaper components that are more likely to have age related failures. Semiconductors probably work forever, so they would be last on your list.
-
Before doing anything. Take some good images of the board both sides. You can repeat the problem so that is really good. If you have a thermal camera see whats heating up, otherwise use your finger. Give the board a good cleaning but don't submerge is spray iso and blow if off. Don't test until dry.
Test between every change.
-
Did it work for decades with the same fast rising power supply or this is new?
-
Thanks for the reponses, I edited the first post to clarify that the circuit has been in production without issue for decades until now.
I'm wondering if we have a bad batch of LM358N.
[attachimg=1]
-
If I reduce the supply voltage I get to around 18V when this happens:
The op-amp supply voltage spends time on a plateau at 7.6V before reaching maximum voltage of 17.4V.
I'm using a signal generator pulse to drive a transistor to power the circuit on and off (from a bench power supply).
When I see the phenomenon in the image below, the behaviour is very temperature sensitive.
Slightly cooling the LM358 op-amp makes the op-amp voltage rise correctly and the circuit takes 4mA as it should.
Slightly warming the LM358 makes the plateau extend. When warmer, the op-amp supply voltage stays stuck at 7.6V and the circuit takes ~90mA.
-
Thanks for the replies and ideas so far.
Does anyone have any suggestions of what to measure or test next?
Testing parts? Trying different op-amps?
-
You only have a few possible current paths. If you hack in a few shunt resistors (about 1 Ohm) then you can measure where the current is going though, and you can either confirm or exclude things like latchup.
Do the inputs of the opamp ever get higher then it's supply voltage? That can be a reason latchup occurs. And for a generic LM358, try some other brands?
Note: I did not look very closely at your schematic because it's picture quality is so bad.
-
What is a resistance value between +24Vin and +Vred?
-
What is a resistance value between +24Vin and +Vred?
0-600 ohms. I am testing it with 9 ohms.
-
From the TI LM358-N datasheet:
The input common-mode voltage of either input signal voltage should not be allowed to go negative by more than 0.3 V (at 25°C).
It does not say what happens when the input(s) exceed -0.3V, however, your table of voltages suggests that Vout of IC1B is not what it should be., i.e IC1B is not functioning as intended. CR5 - 1N4933 - does not prevent the input voltage from exceeding the -0.3V level. Perhaps a schottky germanium diode would be better?
-
Thanks for the replies and ideas so far.
Does anyone have any suggestions of what to measure or test next?
Testing parts? Trying different op-amps?
Check the input voltages of IC1B.
Here is a nice scope shot of the inversion effect in vanilla opamps
https://e2e.ti.com/support/amplifiers-group/amplifiers/f/amplifiers-forum/447545/opa-output-inversion-during-input-overdrive
(https://e2e.ti.com/resized-image/__size/1230x0/__key/communityserver-discussions-components-files/14/6114.2.png)
-
What is a resistance value between +24Vin and +Vred?
0-600 ohms. I am testing it with 9 ohms.
Then what is "-Vout" pin? :-//
-
What is a resistance value between +24Vin and +Vred?
0-600 ohms. I am testing it with 9 ohms.
Then what is "-Vout" pin? :-//
+V RED and OUT-(Black) are the two pins available for customer connection.
A load resistor of between 0 and 600 ohms is connected between +V RED and a +24V supply and is used to convert the current back to a voltage.
-
Ok. So, you may use 1000 uF capacitor across the rails. It will give us Trise= 9 R * 0.001 F = 0.009 s = 9 ms.
Or use e.g. 47 R so Trise= 47 R * 0.001 F = 0.047 s = 47 ms.