Smart laptop charger

"Smart" Laptop Charger

My employer provides me with a HP/Compaq 8510W laptop. It's a nice machine, but heavy, and not light on power consumption either. It came with a 90W mains charger. I take it with me when I visit the head office 15,000 km away. The flighs are about 12 hours and about 5 hours (up to 14 and 6). On some of these, there is DC power available. I have an Igo Juice 70 (approximately equivalent to this unit, visible on this page). It suited my older laptop nicely, but there isn't (yet) at tip available for it. Plus, I'm interested in how you can charge a 90W laptop when you are only supposed to draw a maximum of 75W from aircraft outlets.

Naturally I searched the internet for answers, but none were forthcoming. A few months ago, another visit was about due (since canceled due to the current price of air travel), so I didn't want to wait for iGo to come out with the new tip ("coming this summer", presumably northern hemisphere summer). Plus, it was not obvious to me that it would actually work, due to the mismatch between what the charger could supply (70W) and the possible power draw (90W). A little more research revealed that HP in fact manufacture a charger suitable for air travel, made for the 8510W (and many other models), and it has a maximum capacity of 65W. So how does this work? Can you only run without charge, or charge without run? There was talk of a button that selected between two modes... what does this button really do? Surely if a 65W power supply can charge or operate this laptop, I can somehow use the Juice 70 charger... but how? This page is about trying to answer those questions.

For reference, the iGo Juice 70 (and presumably the Wall 85 and Auto/Air 85 which use the same tips) use a simple 4-wire system. Two are Vout and ground, which connect directly to the laptop connector. The other two have resistances to ground, which program the voltage and maximum current. Here are the tips I had (one tip was butchered as indicated below; I didn't record the resitances for that tip):

Tip Vout Rv Ri
2 19.0v 86.3K 5.87K
3 19.6v 79.8K 5.2K
4 16.6v 163K 14K
6 15.4v 382K 46K
9 19.7v 78.6K 5.1K
11 16.6v 163K 14K
12 20.0v 74.5K 4.73K
14 20.0v 74.8K 4.73K
15 24.6v 44.9K 2.53K
None ~14v - -

The connector on the Compaq supplied mains charger looks like this:

It was trivial to establish that the outer conductor is ground (1), and the inside of the outer sleeve (2) is the +19.5v to do the charging/running. The third connector, a very thin inner pin (3) is the mystery. It reads a volt or so lower than conductor 2. This connector is unlike any that I have in stock, or would be likely to be able to get a hold of in a reasonable period of time. It would be great to have a plug and socket, so I could monitor what was going on while charging/running, and not have to butcher the existing mains charger. One of the "tips" that came with the Juice 70 has a connector which fitted the laptop, and although it had a centre pin, nothing seemed to connect to the centre pin. With this, I managed to feed 19v or 19.5v into the laptop. It responded... by flashing the charge LED eight times, but did not draw power. Obviously the centre pin was needed in some way.

I thought it would make a nice Nerds Night project to put a CRO on the centre pin and see what pulses or whatever appeared on it from the mains charger. However, that would still not answer my main question, i.e. how to I "tell" the laptop to only draw 65 watts? So I relented and ordered a ED993AA from DigitalYes.com.au for AU$116.50 plus AU$9.99 postage. Conveniently, the unit arrived just in time for Nerds Night. It came with a short adapter lead for early HP laptops which used an ordinary 2-pin power connector. That meant that I had a spare socket for the connector from the adapters, although it might not do anything with the centre pin.

The first part of the project was to dig away the plastic of the Juice tip, which originally looked like this:

until it looked like this:

with convenient clip leads. We didn't use the adapter socket, because the centre pin didn't connect anywhere. A metal sleeve (2) was extracted from a bullet-to-bullet adapter; it happened to have just the right outer diameter to make contact with the positive charge voltage. The smaller connector (1) was used to connect to the centre pin; it was salvaged from an IDE "header" socket. The green is heat shrink tubing to make sure that the inner and middle contacts don't short out.

So now I could intercept the signals from either the mains 90W adapter or the new 65W auto/air/mains adapter, and see what was going on.
Image Image Image Image

The first surprise was that the CRO wasn't needed; it was a DC signal on the centre pin. This contrasts with the serial EEPROM signal used by Dell laptops that use the same (or a very similar) connector. For reference (thanks, Dave!):
and the TO92 single pin EEPROM datasheet is here: http://datasheets.maxim-ic.com/en/ds/DS2502.pdf (also thanks to Dave).

That Nerds Night, we determined that the 65W charger output something around 2.5 to 3v on the centre pin, and the 90W mains charger some 5-7v, but the relationship between control voltage and current drawn was not obvious. However, one thing was clear: less than 5v on the control pin causes the laptop to be in "don't charge" mode: the word "charging" disappears, and the batter icon loses the "lightning bolt".

The 90W adapter seems to put about 6.3v on the centre pin, steady. The current ramps up to about 4.7A, and stays there.

The 65W charger (in charge mode) starts with about 5.2v on the centre pin, then at some point it jumps to about 6.8v and seems to stay there. It draws about 3.3A.

Both chargers put 19.2v or so on the centre pin with no load, suggesting that there is a pullup resistor to the 19v rail. I measured about 285K from 19v to centre pin on the 65W adapter, and about 360K on the 90W adapter. Sure enough, a 330K resistor is enough to make the laptop draw current. With the Juice supplying power, it starts at about 1.5A and ramps up to just over 4.2A. This is a little too much for the Juice; the output sags to 17.8V, and the power at that point is some 75W. I'd guess that the Juice is current limiting at that point, or just saturating its transformer or something. I didn't leave it very long at 4.2A.

I've noticed that the current ramps up from 1.5 to as much as is available in about 20 seconds when the centre pin is between about 5.2 and some threshold that I'm guessing is about 6.5V. Note that the 90W adapter stays below this threshold; the 65W adapter starts below it, and ends up above it.

So now I'm thinking that the slow current ramp is a deliberate part of the design. Maybe the adapter is supposed to say "present" (5.2v) then when the current gets to its limit it shouts "enough!" (6.8v). But when I try to simulate this using the voltage control knob on my power supply, jerking the control voltage to about 7.5v always seems to reduce the current to about 1.6A. Reducing the control voltage seems to cause the current to increase quite rapidly, to well over 4A, so I have to stop.

I found that the 90W adapter only got to about 3.3A. By then the charge on the battery was over 90%, where the charge light goes from orange to green, so maybe the battery takes less power there. So I started a big compile; both cores go to around 80% with that. Now the current ramp starts at 2.5A and continues to over 5A, peaking at 5.2A at times. OK, I have to be a little careful, the 90W charger is supposed to supply a max of 4.74A. Then I had another idea: does the centre pin from the 90W adapter go over 6.5v to limit the current? Indeed it does; it also goes to about 6.8v, but only if/when the current exceeds *its* maximum.

So I'm thinking that I need a comparator circuit that measure the current draw. Maybe with manually jerking the voltage control on the power supply I can't get past the threshold fast enough. I was thinking at one point that I could just use a timer, but that won't work because of how the current load without charging varies so much. Another possibility is to just put 7v on the centre pin, and accept just 1.6A of charge/run current.This seems a little low, and besides is no fun at all.

One other thing: when the 65W charger is in charge mode, there is obviously not enough power to run the laptop full tilt. So what happens if you try? It seems that the laptop reduces the core frequency so you can't draw too much power. Unfortunately, I'm yet to find a utility that tells me what the core frequency is. My Acer tells me in Properties of My Computer. The HP has the same soft of information, but it always indicates 2GHz. You can easily tell when the frequency is cut back severely; it's certainly not always running at 2HGz.

Hopefully soon: I hope to build a simple comparator circuit. A single comparator should do, if it can work right down to the negative rail, or right near the positive rail. I'd also be very interested to see if the forthcoming "tip" for the 8510W has all this built in. The tip I pulled apart had a small printed circuit with two surface mounted resistors, so a comparator circuit would not be impossible (just rather unlikely in my opinion, but we'll see if and when it is finally released).

Update 2008/July/30:
I just received an email from iGo informing me that tip S37, which suits the Compaq 8510W, is now available. Unfortunately, it's for the "everywhere 130" model, which can supply 90W (but hopefully not draw more than 75w from an aircraft outlet). I considered getting this tip (it costs US$10 plus postage) simply to reverse engineer it. However, the iGo site isn't international-friendly (except for Canada and a handful of European companies), so probably not. Dick Smith no longer sells iGo chargers for laptops (only the smaller versions for phones and PDAs); these have a different connector with 6 pins), and I can't readily find a supplier for the S37 tip in Australia.

Second update 2008/July/30:
It's dawned on me maybe that all I really need is the 330K resistor to make the laptop draw power, and just resistors as per say tip 3 to limit the current from the Juice to a reasonable level. In fact, the "sagging" to 17.8v was probably the Juice beginning its current limit. Whether the laptop will put up with that or not remains to be seen, but I'm guessing that that's all the S37 tip would be doing, so presumably it must work.

Update 2008/July/31:
The laptop does indeed seem to tolerate reduced charge voltage as another means of reducing the charge current. I found that if I took the negative end of the load from the connector at the end (instead of at 90% of the cable length, where there is a convenient connector that is intended for charging phones and PDAs), the maximum currrent dropped from about 4.24A to 4.00A, which is more in the comfort range of the charger (4.0A x about 18v ~= 72W). Perhaps the iGo Juice 70 uses the negative supply lead as the current sensing resistor, so by taking the load part way along the supply lead, I was effectively reducing the sense resistor. Taking the positive load wire from the end connector made no difference, supporting the theory that the negative lead is used as the current sense resistor.) This was using a standard power tip (I tried two, but the current setting resistor (Ri) on all the 19-20v tips are much the same). I can probably change the current limit by varying the current setting resistor.

Next step: package it all up neatly, and try to set the current limit a bit lower for peace of mind. Maybe I'll also provide a "run only" switch and go all out.

Update 2008/Aug/7:
When I tried to make a neat unit, things didn't work out so well, so I decided to pull apart the 65W charger (the one I own, not the company's), to make sure I know what's going on. First to pull it apart:
There are three screws, all somewhat hidden. Two are under the rubber that it rests on; this is easy to peel back and will glue on again if necessary later. The third is behind the metal foil with all the safety accreditation logos etc. I may be able to flatten that again after reassembly.

With these three screws out, the top and bottom prise off with a bit of effort:
The part in the middle, surprisingly, is just a rubber boot; the adapter is held together by the top and bottom plastic halves. Here is the top half; the run only / run+charge button is actually on the printed circuit board, with just a flexible membrane on the cover:
You can just see the two LEDs; they are the white surface mount rectangles near the top of the cutout in the shields:

Here is the main unit with the rubber boot above; the boot slips over the output cord:

The outer copper shielding solders off without too much trouble, though the heat sure drains away. The inner one doesn't come off that readily, because there is a large quantity of heat sink compound in there:
Ick. Best to leave that alone if possible.

However, the inner shield and its insulator can be prised off enough to get to the end of the adapter, where the output cord is, and where I wanted to trace the circuit:

At the end of the unit, we see the wires from the output cord labelled:

VO+ and VO- are obviously output voltage positive and negative; the other lead is labeled "ID", presumably for "Identity". So this centre pin really is for identifying something about the adapter, probably its output power capability.

With a multimeter, I was able to trace out this part of the circuit:

I also noted a pair of transistors on the top of the board very close to each other, that I suspected formed a flip-flop toggled by the run-only/run+charge button. By carefully running the unit out of its case, I was able to confirm this. This suggests that the IC is a standard switching power IC, designed for a fixed output voltage, and they change the output voltage by changing the feedback of the output voltage to this IC.

So as I suspected, it seems that a resistance of about 330K to the positive rail is all that is needed to get the laptop to draw current. However, presumably when the switching supply is about to current limit, Q27 and hence Q26 turn on, putting a larger positive spike on the control voltage. I'd say that this is the cue for the laptop to not draw any more current, either by charging the battery less, or by slowing the CPU, dimming the screen a bit, and so on. When this happens, the adapter will no longer be current limiting, so the positive spike becomes less noticeable almost immediately. Depending on how well the laptop is able to limit current draw at that moment, the adapter may or may not be in a state of partial current limit, perhaps explaining at least in part the confusing variety of voltages on the control pin after the current has ramped up a bit.

So I'm still hopeful of a simple circuit that avoids a comparator and current sensor. When making my first attempt, I'd forgotten that the output voltage changes from 19.5v for run+charge to 15v for run-only; in order to accomplish that with a rocker switch, I'll need a socket for the 4-pin cable from the Juice adapter. That will allow me to set the current limit a bit lower as well.

Update 2008/Aug/10
I'm able to confirm that after all the above, the following trivial circuit works just fine:

The current limits itself to a shade over 3A, possibly because of the 46K Ri in tip 6. (Though it happily tried to put more than 5 amps into a 2.2 ohm resistor. and well over 3A into two 2.2 ohm resistors in series, so maybe it's the laptop only requiring 3A). Using wPrime, I found that the CPU was not slowing down at full load. It's tricky reading the CPU frequency; when running batteries, the cores seem to run at 60% speed, until there is a significant CPU load. Running CPUz seems to be enough to trigger running at full speed, so CPUz always reads 2HGz. Ironically, wPrime gets its CPU information by shelling out to CPUz, so it may just be CPUz's GUI that triggers full speed.

I decided against a switch for run-only verses run-and-charge, since I'll want to use it mainly on board aircraft, when charging isn't possible. I found that the connector at the end of the Juice's cable isn't 0.1" spacing as it first seemed, so it would be difficult to make a connector for it, so that rather limits me to a fixed voltage using one of the standard tips. I found that a standard DC socket wasn't reliable enough, so I ended up putting a self tapping screw into the connector's end with a solder lug under it. It seems very sturdy that way.

I'm glad it's working now, especially since I dropped the HP power supply when I was tracing the circuit (some leads dragged on me when I went to get up). I thought perhaps a ferrite transformer might have shattered, and indeed when I powered it up, there was a sickening "fut" and the mains fuse blew. I didn't have a replacement handy, and didn't think it would help anyway. Miraculously, it still works fine from 12vDC.

Created by Mike. Last Modification: Sunday 10 of August, 2008 21:20:27 AEST by Mike.

Main Index

Switch Theme


eddie, 11:29 AEST, Sat 22 of Jun, 2024: Offline wiki repair is mostly done. It is running on tiki-27-alpha
eddie, 18:14 AEST, Sun 12 of May, 2024: Phase one of wiki repair is to upgrade to V22 (the last version with image galleries). All 150 images in the wiki-up directory are broken and need manual wiki page repairs.
admin2, 19:45 AEST, Sun 28 of Apr, 2024: Tested tiki 26.2 on a raspberry pi. Tiki 26 does not support image galleries so I have to try again after migrating images to file galleries.
eddie, 13:04 AEST, Thu 10 of Aug, 2023: Offline tiki 26 upgrade went badly. Waiting for 26.1. Will limp on.
System Administrator, 18:45 AEST, Wed 26 of Jul, 2023: Recovered from lockout but unable to upgrade to V24

Last-Visited Pages

Online Users

13 online users