HP G7 no signs of life DAR18DMB6D0

**BlueMidnight** · 03-30-2019, 11:42 PM

Re: HP G7 no signs of life DAR18DMB6D0

Okay, so you are getting the 3V and 5V switching regulator power rails. On this board, that will happen as soon as there is power on +VIN. That still leaves a lot of things that need to happen before PU2 will come into play.

First, AC adapter must be identified correctly. It contains a resistor of a certain value between 19.5V rail and the center pin that depends on the adapter wattage. This forms a voltage divider with a 12k resistor on the motherboard. AD_ID signal voltage will therefore depend on AC adapter wattage. 45W = ~0.44V, 65W = ~0.57V, 90W = ~0.68V. If adapter is damaged (e.g. center pin is broken), then the ID voltage may be glitchy or missing.

Note: I have seen many cheap 3rd party AC adapters that are marked 65W but contain a resistor value that incorrectly identifies them as 90W or more. Not so great if they happen to be plugged into a notebook that actually wants to try drawing 90W.

Some of the other things that need to happen include:
- 3V power to RTC circuit (VCCRTC / +RTC_CELL)
- RTC clock signal present at clock crystal Y4
- Power to EC firmware chip (U11 on VDD pin).
- Power to EC (embedded controller chip U5 on VCC pins)
- EC taken out of reset state (3920_RST# goes high)
- PCH receiving power (U24 chip on VCCSUS3_3 pins)
- PCH taken out of reset state (RSMRST# goes high)
- RTC clock sent out from PCH to the rest of the board (on SUSCLK pin)
- DNBSWON# signal from EC to PCH should follow the same sequence you measured on NBSWON1# (3V, down to 0V when power button pressed, then back to 3V).
- PCH sends SLP_S5# high, signaling transition out of S5 power state.
- EC sets SUSON high to turn on appropriate power rails. (this is one step that obviously isn't happening for you)

Note: 32.768kHz clock signal is measurable with a frequency counter, multimeter Hz setting, or an oscilloscope. Some multimeters may have trouble testing the crystal directly, but should be fine with SUSCLK (if the board has made it to that step).

**mcplslg123** · 04-01-2019, 06:32 AM

Re: HP G7 no signs of life DAR18DMB6D0

You are right. Coils showing low resistance are all part of vcore circuit and so that's fine. In this mobo its usually G5934 IC which fails.So if you have one, you can try replacing it.

**TVMexicana** · 04-02-2019, 03:23 AM

Re: HP G7 no signs of life DAR18DMB6D0

This is excellent advice, here's what I found. On the adapter, the center pin and positive have continuity, 12 ohms. 18V is injected into the AD_ID line. The issue is that the resistance is too low for the rest of the voltage divider to pull low.

I found resistor R304 to be burnt. Or maybe not burnt but looked oxidized and dead. I measured it and at first it measured 60ohms, after another measurement it measured in the megaohms. I went ahead and replaced it. After doing so I plugged it in and found some nasty boost converter sounds, followed by bit of smoke from the circuit I repaired. I kicked myself over it and then felt better once I realized I only replicated what had already killed this thing in the first place.

Adapter voltage 18V was sent via AD_ID into the divider circuit and to voltage rail 3VPCU through D6 (which is one of the components that got hot) and into AD_Type through R310 . 3VPCU is still functioning. I cannot understand why D6 is here. At first I thought that it was meant so that if any Adapter ID voltage above Vf of D6 was present (.6V) it would go through this diode and get shunted somewhere along the 3VPCU line if it was greater than 3.3V. Analyzing 3VPCU I can't find anything that seems like it would protect this line from over voltage. 3VPCU originates from Q16 and it looks like just a pass resistor from 3VS5. At the same time however 3VPCU is still functioning and if anything got shorted from the overvoltage as a result of this mishap i would expect this circuit to be pulled to ground. Seems like D6 would be a sure way to roast a bunch of components in the event the ID pin of the adapter shorts to positive? Am I missing something here? At the same time nothing seems severly shorted as if the 18V that would have gone through this line did manage to shunted sparing all the parts expecting only the usual 3.3V. Then again that's probably a case of a -not- dead short somewhere.

Do you guys think it's worth continuing further?

EDIT:

I modified the Adapter Type circuit with a shunt circuit to give me about 0.5V to go to the AD_Type line but still have no power.

I find this really weird. It seems like 18.75V went to a bunch of stuff that is only meant for 3V but at the same time nothing looks obviously fried and when I replicated the fault in my attempt to fix it, the only thing I saw smoke was the Adapter type circuit.

**TVMexicana** · 04-02-2019, 04:59 AM

Re: HP G7 no signs of life DAR18DMB6D0

I rewrote my above post to make it a little bit more understandable, however I can no longer edit the previous post. If a mod sees this please just replace above post with this one.

This is excellent advice, here's what I found. On the adapter, the center pin and positive have continuity, 12 ohms. 18V is injected into the AD_ID line. The issue is that the resistance is too low for the rest of the voltage divider to pull low leaving 18V on the AD_Type line.

I found resistor R304 to be burnt. Or maybe not burnt but looked oxidized and dead. I measured it and at first it measured 60ohms, after another measurement it measured in the megaohms. I went ahead and replaced. I didn't think it through as I wasn't realizing my first half of the voltage divider was at the adapter plug and it had a short instead of a resistor which would only replicate the fault. This is because at first I thought this resistor had gone bad from the glue of the sticker that goes under the RAM sticks. Anyway, I plugged it in and heard some nasty boost converter sounds followed by some smoke coming from the area I did the repair on. I kicked myself over it and then felt better once I realized I only replicated what had already killed this thing in the first place anyway.

Adapter voltage 18V was sent via AD_ID into the divider circuit and to voltage rail 3VPCU through D6 (which is one of the components that got hot) and into AD_Type through R310 . 3VPCU is still functioning. I cannot understand why D6 is here. At first I thought that it was meant so that if any Adapter ID voltage above Vf of D6 was present (.6V) it would go through this diode and get shunted somewhere along the 3VPCU line if it was greater than 3.3V. Analyzing 3VPCU I can't find anything that seems like it would protect this line from over voltage. 3VPCU originates from Q16 and it looks like just a pass transistor from 3VS5. At the same time however 3VPCU is still functioning and if anything got shorted from the overvoltage as a result of this mishap i would expect this circuit to be pulled to ground. Seems like D6 would be a sure way to roast a bunch of components in the event the ID pin of the adapter shorts to positive. Am I missing something here? At the same time nothing seems like 18V actually blew anything up on this line.

Do you guys think it's worth continuing further?

EDIT:

I modified the Adapter Type circuit with a shunt circuit to give me about 0.5V to go to the AD_Type line but still have no power.

I find this really weird. It seems like 18.75V went to a bunch of stuff that is only meant for 3V but at the same time nothing looks obviously fried and when I replicated the fault in my attempt to fix it, the only thing I saw smoke was the Adapter type circuit.

**BlueMidnight** · 04-02-2019, 05:22 AM

Re: HP G7 no signs of life DAR18DMB6D0

Note: I wrote this before your latest edit. The one thing I'll add is that microcontroller inputs, like the EC chip on motherboards, tend to have a 6V max rating. If it ever saw more than that, then it may well be damaged.

---

That 18V you measured is odd, and since there is some ambiguity about how you measured it, I will try to cover all the bases.

If you unplug the AC adapter and measure the resistance between the center pin (ID) and the inside wall of the barrel (positive), you should see ~500k for a 45W adapter, 390k for 60W, and 330k for 90W.

If you plug the AC adapter into the wall but not into the motherboard, and then measure the voltage from the center pin (ID) to the outside of the barrel (ground), you should measure a voltage less than the full adapter voltage. This is because of the internal resistance/impedance of the multimeter.

The most common multimeter impedance is 10M ohm, though some are 1M ohm. This normally makes no significant difference in the measurement due to the high values. But when you are doing a measurement like this, the multimeter itself will form the lower half of a voltage divider. And since the resistor in the AC adapter is such a high value, and there is no other lower-resistance path to ground other than through the multimeter, that voltage divider does change the reading by a noticeable amount. For example, if the AC adapter has a 500k resistor, and the multimeter is 10M, then you would read about 18.5V on the ID pin of a 19.5V AC adapter.

The last way you could measure it is while it is plugged into the motherboard. Here we have a couple possibilities for the 18V measurement. Either the top half of the divider is too low a resistance, or the bottom half is too high. Don't forget to check that second possibility. If, for example, there were high/infinite resistance through R306 to ground, then you would have a situation very similar to what I described in the previous paragraph.

Yes, D6 is what handles over-voltage. It's not uncommon to have diodes going from a signal line to both ground and the appropriate power rail, to protect against high voltage or negative voltage. This line just has the one. Other boards/brands have a zener for the ID pin circuit instead, but not this board apparently.

In any case, having a voltage that high is a big problem. A little static zap is one thing. But a sustained 18+ volts could force R304 to dissipate a couple watts or more in the worst-case scenario.

**TVMexicana** · 04-02-2019, 06:04 AM

Re: HP G7 no signs of life DAR18DMB6D0

I think I confirmed that it's the adapter's barrel. This is actually a replacement adapter with an adapter head that goes on the end of the cable from the brick. My measurements were done by removing this head and measuring the pins on the exposed end and the center pin.

I think something must have frayed inside it. One end of the multimeter is on the center pin of the barrel and the other is on the pin of the other end which corresponds to the positive connection. When I squeeze or apply pressure to the adapter head, the resistance between the two changes a lot. I got it to go as high as 240 ish ohms and as low as around 8 ohms. Cheap adapter killed the motherboard. It seems.

I still don't get how this over voltage protection works if we're sending it towards a critical line that I can't see any shunts on. I'm almost curious about replicating the fault yet again to see how far the voltage goes on +3VPCU and find out if it's being shunted at all.

R304 definitely took a good part of the beating. As I said in my edit it appeared oxidized though not charred. I see a lot of stuff feeding off +3VPCU, why are we sending the over voltages there?

Anyway, my little modified adapter ID circuit with a shunt regulator circuit using a diode is giving us 0.5V which is telling U5 (if it's still alive) that we have a 90W regulator and still no power. If it's still worth checking around what else would be your next move? I'm willing to go at it for a bit longer just for the learning experience and for curiosity's sake.

Attached are pics of the adapter head and the shunt regulator I have going to emulate the ID voltage.

Attached Files

**BlueMidnight** · 04-02-2019, 06:24 PM

Re: HP G7 no signs of life DAR18DMB6D0

Try not to overthink it. Go back to the basics of how resistors and diodes behave:

- Resistors: Constant resistance. Voltage drop and current depend on the circuit.
- Diodes: Constant-ish voltage drop. Resistance and current depend on the circuit.

A diode can basically be thought of as a variable resistor that tries to keep its voltage drop constant by lowering its resistance. The voltage drop does increase a bit as current increases, but we can usually consider it constant when trying to analyze a circuit in a general way.

So what do we know about this circuit? Well there is 3.3V on one side, then a diode and a resistor, and then 18V on the other side (assuming the 18V was measured with the AC adapter connected). We also know the forward voltage drop (Vf) of the diode. Therefore, we know the other side of the diode must be 3.3V+Vf. It cannot possibly be anything else as long as there is enough voltage to forward bias the diode (or reverse bias a zener). That is a fundamental characteristic of diodes. The entire rest of the voltage between Vf and 18V is therefore dropped by the resistor.

So when you work it through like that, you see that the problem disappears. Very little current goes onto the 3.3V rail after all, and its voltage never changes. The power is dissipated by the resistor. For example, if the 100 ohm resister were dropping 13.5V, then the current flowing would be only 13.5 / 100 = 0.135A. But that would still be 0.135*13.5 = 1.8W of heat for the poor resistor. The diode has a much easier time due to the lower voltage drop across it.

If something happened to lower the resistance of the resistor (like burning up and carbonizing), then the diode would pass so much current trying to get its voltage drop correct that it would fry itself too.

As far as what to try next... I would increase the voltage on AD_TYPE a little bit. 0.5V should identify it as a 45W adapter. You could go as high as you want really, as long as you're not forward biasing the diode. Going to 1V should confirm it nicely.

But the EC could certainly be bad. Or there could be an additional problem elsewhere. If the board still isn't happy with AD_TYPE sorted out (though it won't be 100% happy with no RAM installed), then check my first post for things to measure.

**TVMexicana** · 04-03-2019, 07:53 AM

Re: HP G7 no signs of life DAR18DMB6D0

Okay so I did the checks.

Power to VCCRTC is a CHECK
RTC at Crystal Y4 is an X NOT present, tested with both multimeter and frequency counter.
Power to EC firmware chip is a CHECK pin 8 on U11.
Power to EC VCC is a CHECK
EC Taken off reset 3920_RST# is a CHECK, this pin is high.
PCH receiving power is a CHECK
PCH Taken off reset is an X this pin is low
SUSCLK pin unchecked as Y4 is not present
DNBSWON# is an X as this does not seem to pull down along with the power switch. When measured on Pin 90 of U5, a modified line stemming from this signal, there is a .2V to .3V drop when pressing power.
PCH sends High signal SLP_S5 this is an X
EC sets SUSON as high, this unchecked as i guess it's pretty obvious it wont do that.

I did however have something quite interesting happen. The first thing is that I discovered that if I press on the power button prior to connecting power, when i connect power and wait for about 1 second, the power LED flashes very briefly. Then when I was measuring the PCH SLP_S5 signal for some reason the LED started blinking on and off. For a period after this I could actually turn the motherboard on by pressing the button and this LED would stay lit but would not turn back off even if I held it down. A couple times it would turn on then blink. I managed to measure the power rails that feed the CPU while this was happening and those where still off, there was also no fan spinning or any other sign of life. It stopped working for some reason and I didn't manage to capture any more measurements of what happens while this is occurring. It seems however something is alive on the motherboard at least.

**BlueMidnight** · 04-03-2019, 04:18 PM

Re: HP G7 no signs of life DAR18DMB6D0

Hmm I'm not really sure what was happening with the SLP_S5 measurement. If you had accidentally shorted EC pin 25 to pin 24, that's just ground, and shouldn't cause any change. But maybe shorting it to pin 26 would have done something, depending on what voltage is on that pin.

In any case, the clock signal missing is a big deal if true. Some tools may have trouble seeing it or may only see it on one of the two leads of the crystal. But it helps having tried a particular piece of equipment on other 32.768kHz crystals on other boards, so you know how they behave. Some of them get overwhelmed with the 60Hz power hum and will only display that, for example. An oscilloscope is always the best way to check it (as long as your probes don't have so much capacitance that they stop the crystal of oscillating), but not everybody has one of course.

If the clock signal is really missing, then the crystal may be bad. It's a standard frequency, and replacement crystals are easy to find on other boards, online, etc. But, of course, the PCH and components surrounding the crystal must be good as well.

If the clock signal is actually there but you're just having trouble measuring it, then there may still be something wrong with the AD_TYPE signal. Or maybe that one pin on the EC is fried and it can't even see the signal. Or I suppose something else could be wrong with the EC or its firmware. Etc.

A problem related to either the clock or EC could certainly prevent RSMRST# from going high.

**TVMexicana** · 04-03-2019, 06:50 PM

Re: HP G7 no signs of life DAR18DMB6D0

Do you think it might be good to just wing it and replace the EC? I try to not replace components unless im absolutely sure they are the issue but if it's cheap enough I might go for it.

I'm fairly confident about my frequency counter getting the measurement right but I'll put it in line with the oscilloscope.

BTW just noticed you are also in Chicago. Pretty cool!

**BlueMidnight** · 04-04-2019, 06:42 AM

Re: HP G7 no signs of life DAR18DMB6D0

Since the clock has to work before the rest, I would start with that. Provided the caps aren't shorted, the crystal is the most likely culprit. The rest of the power-up sequence is never going to work without it.

Those other components did burn up though. And I am a bit concerned that there seem to be 2 unrelated issues on the same board (if the RTC circuit isn't working). If it comes down to it, replacing the EC is good soldering practice if nothing else. But I would still resolve the RTC issue first.

Yeah, I don't think I've seen anyone else from around here.

**TVMexicana** · 04-04-2019, 01:11 PM

Re: HP G7 no signs of life DAR18DMB6D0

Good thing I tested with the oscilloscope. I found the clock is ticking but some weird stuff is happening.

When the board is off the PS and EC is running off the back up battery, our clock produces a very clean signal as seen in the first picture. When I plug in the PS the signal slowly degrades to the slop you see in the second pic. It takes about 2 seconds for it to go from nice sine wave to this. It looks like PS noise but I checked the PS and it doesn't seem to produce a particularly dirty signal.

Forgive the dirt on the scope, it had been stored for a bit.

Attached Files

**BlueMidnight** · 04-04-2019, 04:58 PM

Re: HP G7 no signs of life DAR18DMB6D0

Hmm.. it still bugs me that there would be 2 separate problems with the board, so I'm trying to explain away the clock thing or find a root cause for everything. Although certainly it's possible to have multiple problems. It's just not the most likely scenario.

Some random thoughts:
- The loading that the probes put on the circuit should always be considered. For example, "x10" probes will load the circuit less than "x1" probes. However, I can't explain why plugging in the power supply would change anything here. And a clock frequency in the low kHz range normally isn't so sensitive.
- The ground probe location is crucial. It must be connected to ground and only ground. Since it is almost certainly connected to earth ground in the oscilloscope, it will cause a direct short to ground wherever you place it on the board under test (if the board is plugged into the AC adapter).
- The clock circuit works by taking the output of the crystal, amplifying it, and then feeding it back into the other pin of the crystal. This makes it resonate at its preferred frequency, just like pushing a person on a swing with a steady rhythm. So, while both pins of the crystal will have a signal of the same frequency, one pin will have a bit stronger signal than the other. So check the signal on both.
- I wonder what happens if you wait to connect the oscilloscope until after the board has been plugged in for several seconds. Will the clock signal appear for a moment, or will it already be turned to slop?

If it continues acting the way it's acting, I'd probably try replacing the crystal and see what happens. They aren't expensive (or can often be found on scrap boards) and are easy to solder, so it's no big deal. And then after that, probably the EC. If nothing else, it will be good practice. Hopefully you have a medium width sharp blade tip for your iron and some good flux. Then you can scrape the tip down multiple pins at a time and not end up with solder bridges.

**TVMexicana** · 04-04-2019, 08:53 PM

Re: HP G7 no signs of life DAR18DMB6D0

Let me order up both the parts and we'll see which one is causing the issue. My bet would be the EC, when I worked in communications gear something like this I'd pin on the circuit the crystal is tied to unless the crystal was more of an integrated oscillator with power going into it. If the oscillator inside the EC is damaged it could be leaking some kind of noise onto the crystal which is what we measure once it is doing more functions or once higher voltages are applied. The other side of the signal looks about the same but the bottom peaks are clipped which is another thing I saw often in communications gear. It too turns into crap. Frequency is also pretty dead on. 32.7678khz is around what I got.

I do wonder if some of the noise is caused by a ground loop type issue, I'll try some other tie points. I'll also check the outgoing clock signal from the EC, if that is even a thing when this thing is acting dead.

Another thought that just popped into my head is whether or not in these circuits the clock signal is multiplied, for example 32.768 turned into a harmonic such as 65.536khz. In trippler circuits a lot of harsh signals can be seen.

As you said usually low frequency stuff isn't fussy about load down from probes and the meter is set to X10. I'll try another probe just in case.

**BlueMidnight** · 04-04-2019, 09:25 PM

Re: HP G7 no signs of life DAR18DMB6D0

Yeah, that sounds right. Just like pushing someone on a swing, the half-wave signal is the sequence of pushes that keep it oscillating.

Ah but don't forget that the RTC clock is controlled by the PCH. Which is what is so confusing. If RTC were controlled by the EC, it would just be another data point in favor of swapping the EC.

Hmm the datasheet for it indicates that the RTC circuit is sensitive to fluctuations on VCCRTC. How clean does that rail look when the power supply gets plugged in?

Wait a minute.. come to think of it, the RTC is going to get powered by the 3V linear regulator output of PU3 while the AC adapter is plugged in, but by the CMOS battery when the AC adapter is unplugged. Perhaps the explanation lies down that path somewhere.

**TVMexicana** · 04-04-2019, 09:32 PM

Re: HP G7 no signs of life DAR18DMB6D0

Oh wow I ended up forgetting this is the PCH and not the EC that this ties too. Thanks. I'll look into the voltage rail.

I also ordered the crystal and EC but only just now realized that this is a KB3930 on this board and not KB3926. Will have to cancel the order for that. The only KB3930 i see is in china and will take a bit to get here. Know of any sources?

**BlueMidnight** · 04-04-2019, 09:50 PM

Re: HP G7 no signs of life DAR18DMB6D0

And that power rail (+3VPCU) is connected to both D6 and the EC. So, while the AC adapter is plugged in at least, that would create a potential means of an issue around the EC propagating to the RTC.

Not sure about any special source for the KB3930. The big suppliers (e.g. Digikey) don't seem to have it. eBay does, but who knows which sellers are good or not.

**TVMexicana** · 04-04-2019, 10:55 PM

Re: HP G7 no signs of life DAR18DMB6D0

It seems the issue was a ground loop problem. I tried a shorter ground lead on the probe and this got rid of the issue. I noticed that I was now collapsing the signal even when the PS wasn't plugged in at all and would change when I moved the lead around. With the shorter lead the signal is kept clean all along.

What would the next best move be this having been confirmed?

**BlueMidnight** · 04-04-2019, 11:23 PM

Re: HP G7 no signs of life DAR18DMB6D0

Ah that's good news. Things make more sense with that out of the picture.

I suppose a non-contact probe would be the best way of all to test the clock signal. There was actually a recent EEVblog video about how to make your own for cheap.

Anyway, we seem to be left with the EC issue. If you have good AC adapter voltage and a good voltage on AD_TYPE, then it's really up to the EC to take the next step. So, replacing it seems like a good idea. It's not a sure thing, but nothing is.

Corrupted firmware can cause the power-up sequence to stall there, but since there are actual fried components, I wouldn't bother messing with that.

Can't think of anything else right at the moment.

HP G7 no signs of life DAR18DMB6D0

HP G7 no signs of life DAR18DMB6D0

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Related Topics