A7 (3-Pin SMD) Identity?

[Trurl]

I've run into another mystery part on the HP server PSU(HSTNS-PD43) I'm analyzing, and would be grateful for any advice.

Device(A7) as it sits on the auxiliary(side) PCB.
*Measured Body Dimensions: 2.1(L) x 1.25(W)mm


Surrounding Schematic
Notes:
1. Two BAS40 diodes were used in the schematic for the time being to show the A7's inner diodes' orientations according the multimeter tested results -> "common anode"(pin-2) diode pair with pins 1 & 3 being cathodes.
2. Cap values are unknown so they are all marked by "default" to 1uF.


As shown in the schematic above, my tests of the A7 (in circuit) using a multimeter in diode setting(& Ohm setting) revealed the following;
(PL=Positive Lead, NL=Negative Lead)

1. PL on Pin2 / NL on Pin1 = 0.508V (Resistance 1.336k) - (Seems like a good diode)
2. PL on Pin2 / NL on Pin3 = 0.508V (Resistance 1.336k) - (Seems like a good diode)
3. Pin1 & Pin3 connected to GROUND.
 
But I noticed the following as well ...
4. NL on Pin2 / PL on Pin1 = 0.188V* (Resistance 515 Ohm) - (What the hell is goin' on here?)
5. NL on Pin2 / PL on Pin3 = 0.188V* (Resistance 515 Ohm) - (What the hell is goin' on here?)
*not showing as "OL" in the multimeter diode check(leads in reverse) 

I've searched literally hundreds of  diode arrays and noticed that the following types are available.


The closest type from the image above would be the one in "Fig.33 - SERIES(Rev)" as pins 1 and 3 are connected to cathodes, but as the diodes are in series it would not explain the fact that the resistance between pins 1+2 and 2+3 are both 1.336k - same.

I'm wondering if this A7 device is not a diode array and perhaps something entirely different?

Any pointers would be appreciated. Cheers~

[MLXXXp]

Probably a BAV99

https://www.mouser.ca/datasheet/2/308/BAV99-1118535.pdf

[Trurl]

BAV99 (Nexperia) was actually the first device I suspected as its datasheet indicates;
https://assets.nexperia.com/documents/data-sheet/BAV99.pdf

"Marking code: A7* (* = placeholder for manufacturing site code)"
This explains the rotated "3" next to the "A7" in the top photo.

But my diode/resistance tests (in-circuit) with a multimeter(mentioned above) threw me off.

(Please do correct me if I'm wrong but) The multimeter diode tests suggests a "common anode"(on Pin2) diode pair with pins 1 & 3 being cathodes (as I've shown in my adjusted schematic above).

Could it be that testing in-circuit (and the ground connection of pins 1 and 3) is causing false readings on the multimeter? 
 

[MLXXXp]

BAV99 (Nexperia) was actually the first device I suspected as its datasheet indicates;
https://assets.nexperia.com/documents/data-sheet/BAV99.pdf

"Marking code: A7* (* = placeholder for manufacturing site code)"
This explains the rotated "3" next to the "A7" in the top photo.

It's hard to believe that it's not a BAV99 when compared to a photo like the second one on this page (even though they call it a transistor):
https://www.alibaba.com/product-detail/THJ-Power-Transistor-BAV99-A7W-Smd_1600998341493.html

Could it be that testing in-circuit (and the ground connection of pins 1 and 3) is causing false readings on the multimeter?

It's possible. It would be best if you could remove it and test it alone, assuming it's still good.

[Trurl]

Thank you for your support!

I also do think it's a BAV99 and it was the initial part I had in my schematic for simulation. It resulted in oscillations of the voltages nearby. This made sense as one of the horizontal traces connects to the PIC's pin 27 "TCK"(Test Clock Input). The current schematic representation of the A7 on the other hand doesn't do anything of interest during simulation.

Fortunately, I've found an amusing YouTube video demonstrating how to make "Y" shaped 'SMD Desoldering Tips'(from thick gauge copper electrical wire) so I may give it a try (Note: at about 7:05 into the video, Morrisey's alter ego  demonstrates desoldering a 3-pin SMD device:
"How I desolder and process SMD components ..."

[MLXXXp]

Fortunately, I've found an amusing YouTube video demonstrating how to make "Y" shaped 'SMD Desoldering Tips'(from thick gauge copper electrical wire) so I may give it a try

Before you desolder it, you could just make a narrow cut in the PCB trace going to pin 1, right at the pin, to isolate it. You would then be able to see if you get readings that indicate a diode with its anode on pin 1 and cathode on pin 3. If that looks promising, you could then proceed to desolder the part to verify another diode between pins 2 and 3.

You would scrape off some solder mask on the cut trace and re-join the trace with a solder bridge after soldering the part back in.

Alternatively, you could similarly make two more trace cuts to isolate pin 2 and not have to desolder the part at all for testing.

PCB cuts:

[Trurl]

Very interesting idea! That would prevent cooking the 'lil bugger!

I have done such trace cutting on some ICL7107 based volt meters for various tests (to study how the differential inputs behave with certain pin connections etc.) and soldered the cut trace(s) back as you suggested without too much trouble.

I've realized that my analysis of the server PSU in this particular area (where the A7 sits) is not likely to affect how certain resistor placement hacks affect their respective op-amp comparator inputs for feedback voltage and crowbar setting mods. But it's good to be reminded of such options, and I'll be sure to consider it if the need arises.

I'm almost done making the partial schematic of my manual probing & 'tracing out' of the (HSTNS-PD43) PSU's auxiliary PCB, but without access to SPICE files for the more complex ICs and their firmware/software, op-amp and transistor simulations have started to cause "NAN"(not a number) errors (or jumbled flickering text such as "V=-1.#IND")to appear on voltage probe locations. So I've had to disable op-amps and some transistors too in order to view how simple resistors, capacitors, and diodes along the way towards op-amp comparators affect the hacked voltage inputs, which is my main focus. But the A7 does not affect my points of interest either way (included or excluded from simulation). Hence, I'm okay not knowing exactly what it's doing there, for the time being.

Still curious but I'll let someone with deeper knowledge of this PSU to chime in if he or she exists...

In the meantime, I'll finish up my (partial) schematic of the auxiliary PCB and share it here in the forum, perhaps in the 'Projects, Designs, and Technical Stuff' section so the big boys and old timers can share their insights about its feedback & PWM control circuit design. I'm certain it will be of interest at least to those in the RC and DIY auto programming community.

Cheers~!

[Trurl]

Dang! My initial schematic was missing a trace so I'm updating it below marked in red. Also, the A7 is shown as if it were a BAV99 (though my in-circuit tests mentioned above suggest otherwise?).

[MLXXXp]

I doubt that both sides of C9 go to ground. The same with A7 pins 1 and 3.

[Trurl]

I doubt that both sides of C9 go to ground. The same with A7 pins 1 and 3.

I rechecked the points you mentioned and found the following:

1. A7 pin 1 does NOT connect to ground(Thanks for your astute observation). The via right of C10 is actually connected to the shunt resistor("0") below C10, and seems to connect to the other side of the PCB which is difficult to access (due to the big fat cap right up against the length of the rear side of the auxiliary PCB).

2. On the other hand, A7 pin 3(the single pin on top) connects to the via just to the left of the pin. This via is a direct connection to ground. C9's upper contact is obviously connected to this via(ground).

3. C9's lower contact may not be connected directly to the via just below C8 which connects to ground, but C9's lower contact connects to the PIC's pin 25(VSS_2) and this may be why it appears as if C9's lower contact connects with the via below C8.

I've attached the updated and expanded schematic along with its physical layout image below. Again, the A7 is shown as if it were a BAV99, though my in-circuit tests suggest otherwise(as suggested in my first uploaded schematic above).

[MLXXXp]

I've attached the updated and expanded schematic along with its physical layout image below.

You still have both sides of C9 going to ground, which would make it useless. I doubt they would do this.

Are you sure A7 pin 3, and the upper leads of the 5.1k resistor and C9, really go to ground? From the board photo, it looks like they may go to PIC pin 24, under the 0 ohm (4) resistor, instead of where you have that pin going to in your schematic.

[MLXXXp]

3. C9's lower contact may not be connected directly to the via just below C8 which connects to ground

It difficult to tell from your photo but this may be true. The trace may not be connected between the via and the lower lead of C9, as I've indicated in red.

I see what might be a trace running under A7 and the 5.1k resistor. If there is a trace, I'm not sure where its ends go but perhaps from A7 pin 1 to the upper lead of C8, as I've marked in yellow.

[Trurl]

OH... MA... GOT!

Your suggested "sneaky" under-the-parts traces are indeed "for real." HOLY COW!

But that via just left of the A7's pin 3 still beeps in continuity testing with the main ground tab. Also, there is a physical trace between C9's lower contact, C8's upper contact, and the via(to ground) just under C8 (despite what the white line on the PCB might suggest).

I've updated the schematic with your pointed out traces, and also added a ground symbol for the upper right trace's via between C38 & C39 - updates marked in red.

[MLXXXp]

So now you're back to having A7 pins 1 and 3 shorted together to ground, in addition to both leads of C9 grounded. That doesn't make any sense. Why bother having a separate ground trace running under A7 from pin 1 when you've got ground on A7 pin 3? The same goes for having separate traces from PIC pins 24 and 25.

But that via just left of the A7's pin 3 still beeps in continuity testing with the main ground tab.

A continuity test that just beeps doesn't necessarily indicate zero ohms when it does. It could be low impedance but not truly shorted, for example, with an inductor or transformer winding. There's also the possibility that one of the components has gone bad and is shorted internally. I just grabbed the closest multimeter at hand (an Aneng AN8009) and got a continuity test beep with a 47 ohm resistor (which also showed 47 ohms on the display while beeping).

[MLXXXp]

Your suggested "sneaky" under-the-parts traces are indeed "for real."

Just to let you know, there are more under component traces that you've missed on your schematic. For instance, under 01C (2), C41 and 01A (3).