We use cookies to provide our visitors with an optimal site experience. View our privacy notice and cookie notice to learn more about how we use cookies and how to manage your settings. By proceeding on our website you consent to the use of cookies.
I am using the MAX17526A current limiter in an embedded design. The input voltage to the current limiter is +30.5V. The schematic below shows my configuration of the part. My current limit setting, with R2 at 6.8k ohms is ~5.5A. I have experimented with the part in “auto retry” mode (as shown in the schematic) as well as continuous mode where the ‘CLMODE’ pin is floated.
The FLAGN pin is connected to a pull up resistor to 3.3V. This is done on another section of the schematic that I didn’t attach.
The load that this part is connected to “VOUT_LIM”, is a combination of inductive motors, piezo transducers as well as several buck converters, various microprocessors and integrated circuits. So, the load definitely has both real and reactive components.
During initial testing with the part, I attempted to use the MAX17526A eval board connected to a BK Precision programmable load. I discovered that the part’s behavior once the current limit threshold was reach, was not at all what I expected. This unexpected behavior was the same for either ‘continuous’ or ‘auto retry’ mode configurations. When the current limit threshold was reached, the part’s output current and voltage would drop to ~0 levels respectively. This would persist, until the load was nearly completely removed at which time the output voltage would return to the expected ~30.5V.
So, I then simply replaced the programmable load with a high-power resistor. Once the resistance was lowered to a sufficient value to cause the part to current limit, the results (output current and voltage) were as expected. The part’s output voltage would reduce to maintain the set current limit value… as I expected.
So, I am just wondering why the part does not behave as expected when using it with a programmable load versus a purely resistive load? I am guessing the answer has something to do with the active nature of how the programmable load develops its apparent load at any point in time and the interaction of this with the control loop in the MAX17526A part attempting to limit the current. I have also tested my programmable load independently, connected to a power supply and it is working fine. So, it’s only when I attempt to use the programmable load as the load for the MAX17526A when the results are not as I would expect.
I appreciate any help in gaining a better understanding of these results.
-longboard
Hello longboard, I have reached out to our internal team for support as this goes beyond my area of expertise. Hopefully, someone can provide guidance on this soon.
Watching the /FLAG and SETI pins may be useful, and further info regarding the specific test procedures used and the expected results would be helpful in offering possible explanations.
Something worth noting is that the device performs foldback-mode current limiting by linear means. Particularly as the configured current limit increases, the window within which the device can do so is quite narrow due to thermal limitations, and once that becomes a factor the part necessarily shuts off the output in order to limit potential for damage.
It is quite possible for thermally-protected parts to enter thermal shutdown before the part warms significantly to the touch, due to the time it takes for heat to transfer from the die out of the package. If charging a capacitance (for example) the initial power dissipation @ 5.5A and 30.5V approaches 170 watts. That’s enough to warm a die from 20 to 150°C quite rapidly, to say nothing of the 10°C thermal shutdown hysteresis.
I’d not be surprised if there was some subtle difference in procedure between use of the programmable load and the fixed resistance, and/or if (as suggested) there is some characteristic of the programmable load that’s interacting with the control functions of the device to yield a difference in behavior.
Long story short though, if one’s expecting this particular part to offer functionality similar to a current-limited bench supply, disappointment is likely to ensue.
I am currently monitoring the /FLAG pin, but you make a good suggestion regarding also monitoring the voltage on the SETI pin.
Regarding the voltage on the SETI pin… the part’s data sheet mentions the following:
The voltage on the SETI pin provides information about the IN current with the following relationship:
However, I am a bit confused by what is meant by I(in-out) in the equation above?
You mentioned that the device performs foldback-mode current limiting by linear means. Is this the same functionality as what the part’s datasheet describes in the comment below:
When the device current reaches the programmed threshold, the controller inside the device prevents further increase in current by modulating the internal NFET resistance
You make a good point regarding the part’s thermal shutdown. This may very likely be playing a role in the behavior I am experiencing from the circuit when powering the actual load it is intended for. I suppose reducing the current limit threshold could be a way of increasing the time required for thermal shutdown to kick in.
I agree with your explanation for why the programmable load is problematic as compared to a fixed resistance. Does a current-limited bench supply reduce output voltage when current limiting in a manner that doesn’t insert a series resistance in the current path as I believe the MAX17526A is doing?
While somewhat stilted,an explanation is given in the prior paragraph. Basic idea is that the SETI pin sources a current 1/25,000th that of that flowing into the IN pin on the device, and whatever voltage that makes depends on how much resistance one puts in the way.
Yes: “linear” here is in contrast to “switching”, and refers to burning off an excess input as heat to obtain a desired output, while a switching scheme involves switching the input on & off to obtain a time-averaged result equal to the desired output.
Using the 29°C/W thermal resistance figure given, with a 5.5A current limit setting the part’s only good for about 800mV worth of voltage fold back (at room temperature) before the thermal protections kick in.
The preponderance of bench supplies these days are switch mode devices. The few linear models that remain can usually be recognized by their massive heatsinks or the burn hazards they pose for lack of such.
Thank you for the clarifications and additional information. I see what you are saying regarding the statement in the prior paragraph:
The device read-out of the current flowing into the IN pin. A current mirror, with a ratio of CIRATIO, is implemented, using a current-sense auto-zero operational amplifier. The mirrored current flows out through the SETI pin, into the external current-limit resistor
I think the part that still confuses me with the Iin-out calculation, is the relationship to the NFET’s series resistance that is being controlled to limit the current. Wouldn’t the part’s input current equal output current? I think the series resistance of the NFET would generate a voltage drop to reduce the current to the current limit threshold?
Is the 800mV of voltage foldback that you mentioned at the output of the part? If so, I think 800mV would represent the drop across the NFET operating in the linear mode? This would then represent a power dissipation across the NFET of 5.5A * 800mV or 4.4W. Then 4.4W * 29°C/W = 127.6°C. But this is lower than the 165°C for the thermal shutdown of 150°C for the thermal foldback. So, I am a bit confused by the 800mV amount of voltage fold back that you mentioned and may not be thinking of this correctly.
Thank you again for your help explaining these details.
That depends on which “output” one refers to. The current through the IN and OUT pins will be more or less the same, however the current from the SETI pin (also an output) will be ~1/25,000th of that. Choosing a name such as ISETI-OUT may have been a less confusing choice than IIN-OUT.
Correct.
The 127.6°C figure represents the expected amount by which the device will increase in temperature over ambient. Add a 25-ish degree ambient to that, and one’s in 150C territory.
Hi @heke.
When we did our testing with the programmable load we were using CC mode.
I agree with your comment regarding the impact of ambient temp on the heat transfer from the part.
Thank you
In that case the results reported initially are pretty much exactly what should be expected.
The impedance of a CC load drops to zero if the source feeding it is current-limited to a lower value. That means the '17526 would attempt to absorb all ~170-ish watts supplied to the circuit once the limiting function engages, which would promptly send it into a thermally-limited state.
Thank you again for the clarifications and information. It makes sense what you are saying regarding the impedance of the programmable load dropping to 0 if the source feeding it (MAX17526A) current limits. I agree that in this case the MAX17626A would then attempt to absorb ~170W being supplied.
Below is an O-scope image showing the MAX17526A part driving the actual load (not the programmable load). The /FAULT line trips several times as we reach the 5.5A current.
I was able to calculate ~5.5A with the I(in-out) equation, which was a good confirmation that the system is working as expected. The /FAULT line would go low when Vseti was ~1.5V, which equates to close to 5.5A. I also included a block diagram of the test setup.
Yellow trace = /FAULT on MAX17526A
Blue trace = Vseti on MAX17526A
Purple trace = output of current limiter
Green trace = output of ideal diode IC, which is after the current limiter IC
Then later in our testing the MAX17526A reached the thermal shutdown point as can be seen in the plot below. The part appeared to be oscillating in this thermal shutdown state, which is something we are working to prevent but reducing the current draw as well as increasing heat sinking.
So, I now have a much better understanding of the MAX17526A part’s behavior during current limiting. We need to improve the heat sinking for the part as well as try and reduce the current draw.
Thank you for your help and time with my questions.