Resistors For OVP Mod: Carbon vs. Wire Wound - Different Results!?

[Trurl]

I've been testing the OVP triggering conditions(mod resistance in parallel vs set output voltage) for a server PSU using regular carbon resistors with leads and a small 5k trimmer pot for fine adjustments. But I noticed "drift" in resistance(e.g. +/- 0.01~0.02k) depending on room temperature & humidity, that seems to have affected the results to be mixed(i.e. sometimes triggering and sometimes not) for the same initially set resistances.

So, I tried using a more stable large wire wound multi-turn pot, and though the resistance is definitely much more stable, the OVP doesn't trigger at settings that with carbon resistors always triggers. 

Is this due to the wire wound pot's inductance? If so, how is the inductance affecting the OVP triggering results(not triggering)? Or is there some other phenomenon at work here?

Thanks in advance...

[Gyro]

You almost said the answer yourself. Carbon resistors have a very high temperature coefficient [Edit: use metal film resistors instead]. If you're talking about an open carbon pot, the exposed track may well have a humidity coefficient too (unless the humidity is also a function of temperature).

You could try a cermet pot if the multi-turn wirewound is physically too large for you, but check the specs.

[Zero999]

It seems like the problem is the voltage setting is too close to the nominal output voltage. Increase it a little, to avoid false triggering.

[Trurl]

Let me reemphasize my question...

Taking into account that carbon resistors have a "very high temperature coefficient", I purposefully tested at a resistance in which with carbon resistors, the OVP ALWAYS triggers, but with the wire wound pot (set at the same mod resistance) the OVP NEVER triggers. Why?

Inductance? If so, how? Or some other issue?

P.S.
Regarding voltage setting, the curious results mentioned above are the same whether the output voltage is set to 15.5V or 15.0V as examples.

[Zero999]

Inductance would definitely explain it, because the impedance of the resistor is higher, with increasing frequencies, so the voltage across it will increase more, with sharp spikes.

[Trurl]

Inductance would definitely explain it, because the impedance of the resistor is higher, with increasing frequencies, so the voltage across it will increase more, with sharp spikes.

If the resulting voltage (across the wire wound) increases, I'd think that the voltage at the OVP related sensor (opamp/differential amplifier) would increase as well and actually make the OVP trigger ALWAYS. But the opposite is happening - it is NEVER triggering, hence my puzzlement.

Perhaps the wire wound pot is "absorbing" the feedback signal increase in either voltage/current and having a "damping" effect pre-OVP sensing circuit and that is causing it to not trigger?

I'd appreciate any explanations...

[Zero999]

Inductance would definitely explain it, because the impedance of the resistor is higher, with increasing frequencies, so the voltage across it will increase more, with sharp spikes.

If the resulting voltage (across the wire wound) increases, I'd think that the voltage at the OVP related sensor (opamp/differential amplifier) would increase as well and actually make the OVP trigger ALWAYS. But the opposite is happening - it is NEVER triggering, hence my puzzlement.

Perhaps the wire wound pot is "absorbing" the feedback signal increase in either voltage/current and having a "damping" effect pre-OVP sensing circuit and that is causing it to not trigger?

I'd appreciate any explanations...

A schematic and picture of the pysical construction are required to form a solid conclusion, otherwise the best one can do is speculate.

[Trurl]

I probed and traced the following attached schematics of the HP HSTNS-PD43 server PSU's auxilliary(side) PCB(outer side only, of a double-sided multi-layered board). I devised a grid format so that descriptions of locations can be relatively easy.

Color Schematic With Hand Drawn Initial Feedback Traces for Voltage & OVP


Black & White Schematic Without Markings


Actual Auxiliary PCB


The color version of the schematic has red and blue hand drawn lines for (voltage mod / OVP mod related traces - initial feedback).
In C-10 and G-10 grid locations of the schematic, there are terminals marked "+12V+" which are feedback(vias) locations. If the blue lines are followed, one passes the point marked "<OVP HACK" this is where one end of the OVP mod resistor is connected (the other end is connected to ground). This OVP mod resistor has the effect of lowering the voltage at this point and as a result into the  LM393 opamp(s), allowing for boosting the main output voltage from 12.3V to a higher voltage depending on the voltage mod resistor.

For the time being, for testing, I have wires soldered onto the volt/OVP hack points and the other end of the wires to a breadboard for testing with various resistors (carbon, wire wound etc.), and the breadboard's ground connected to the PSU's main output side ground tabs via alligator clip & wire.

In the schematic's D-12 location, there is the LM393's opamp A, and pin 3 (differential input +) takes the OVP sense related input signal voltage. I'm thinking that if the voltage at this input is higher than a certain threshold, the OVP triggers.

If a wire wound pot results in voltage spikes, shouldn't it rather support OVP triggering as opposed to it NEVER triggering?

[Trurl]

Could this(OVP non-triggering with use of wire wound pot) simply be due to OVP triggering threshold being set such that it triggers when the input to pin 3 of the LM393's opamp A (differential input +) and/or elsewhere, gets a significant drop in feedback voltage due to output loading?

If so, it would explain why OVP triggers only when the PSU is connected to a load(simply boosting the output voltage without loading doesn't trigger OVP*). It would also explain why OVP tends to trigger with carbon resistors (and more so at higher loads as results indicate), but doesn't trigger with wire wound pot even under the same output voltage setting, OVP hack resistance, and loading (the inductance of wire wound pot causing a voltage spike and perhaps "masking" the voltage drop that would otherwise be present with carbon resistors).

* UPDATE: At a sufficiently high carbon resistor resistance(e.g. 20.35k) and a high output voltage setting of 15.50V, the OVP apparently kicks in immediately upon power ON effectively shutting down the output of the PSU (before any external loading). The wire wound pot at the same resistance doesn't trigger OVP at all even with about 7A loading.(!)