Adjusting output voltage of DELL brick

[elektrinis]

So I have a "slim" 130W 19.5V DELL brick that I want to adjust to 24V.
I have identified a feedback circuit and a resistor divider which sets the voltage. It is very primitive and should work... However I am unable to go past 21V - it switches off after a split second if higher voltage is specified.
I had a deeper look and could not find anything else on secondary, so there must be some kind of protection on primary side?
Looking for some advice from those who had some experience... Thanks.

[amyk]

The output caps are probably rated for 25V, and 24V is a bit too close for comfort.

You'll need to post some more details about the circuit (e.g. type of IC used, partial schematic, etc.) if you want more information.

[Hydron]

If it's using a flyback topology then it's possible to determine the output voltage from the primary side by measuring the waveform on the transformer/inductor. The supply is probably doing this for over voltage protection in addition to the secondary side sending.

[wraper]

There might be an auxiliary winding used for powering SMPS controller. There is no simple mod in that case.

[elektrinis]

Here are some photos.





Primary driver is LTA804N.
On a closer look (after blobs were removed), I have also inspected an SO8 chip LTA806N, which I could not identify. Initially I thought it is a device ID chip and did not consider it, just disconnected it's power. That, however, did not do anything. It is connected to an extra secondary winding and so far I can't find a link to feedback circuit or how it could do what it's doing.
Need your input on this... Thanks.

P.S. Yes, caps are 25V. I want to push my luck on this one.

[wraper]

Apparently LTA804N is equivalent to TEA1751LT. https://eu.mouser.com/datasheet/2/302/TEA1755LT-1152141.pdf

Protection features
 Safe restart mode for system fault conditions
 Continuous mode protection using demagnetization detection for both converters
(NXP Semiconductors patented)
 UnderVoltage Protection (UVP) (foldback during overload)
 Accurate OverVoltage Protection (OVP) for both converters (adjustable for flyback
converter)
 Mains voltage independent OverPower Protection (OPP)
 Open control loop protection for both converters. The open-loop protection on the
flyback converter is latched
 OverTemperature Protection (OTP)
 Low and adjustable OverCurrent Protection (OCP) trip level for both converters
 General-purpose input for latched protection, for use with system OverTemperature
Protection (OTP)

[elektrinis]

Thanks a lot, now I have something to grab on.

[Gyro]

I modified another brick a while back. Not a Dell but I found exactly the same symptom, dropout when trying to adjust beyond 21V. I found that slightly decreasing the value of the MOSFET current sense resistor (Rsense2) in the datasheet schematic allowed me to reach 24V.

The current sense resistor is normally made up of several paralleled low value smd resistors side by side, so it was easy for me to piggyback an extra resistor. As I remember, I dropped the overall value by about 10-15% (obviously as little as possible).

I also replaced the output side electrolytics with higher voltage, slightly lower value (to maintain can size) ones as I wasn't planning to use anything like the maximum current rating. It really is a sensible precaution.

I'm not saying that this is in any way an ideal mod (the peak current through the MOSFET can obviously go higher now) but the brick has been running reliably (not unattended) since.

Please be very cautious when playing around with the brick. The primary side capacitors can really (and fatally) bite for a considerable time after it is unplugged, particularly if the output isn't loaded - and when it has gone into the 'drop-out' state. There will be residual charge even when the output has dropped to zero. You have been warned!


Edit: In light of your success below, mine was certainly a different SMPS IC (no PFC for a start).

[elektrinis]

The power of community strikes again!
Thanks to wraper...

There was a 82k resistor on pin 4 of LTA804N. Since original voltage was 19.5V and it was tripping at 21.0V (+8%), I have replaced it with 100k, which also gives a rough OVP of +8% on top of 24V.
To actually adjust the voltage itself you need to change a feedback divider next to optocoupler. Mine had two parallel resistors (11k+2M), so you need to replace the 2M one to 43k to get 24V. I only had 39k on hand, so the result is 24.16V, which I am happy with.
For those following this in future: Keep in mind that increasing output voltage increases overall power. This PSU likely has some reserve, but to do it properly you need to adjust overcurrent protection.
To be honest, LTA804N is a nice 2-in-1 chip.

This will be one happy (and silent) 3D printer...

By the way these power bricks are a bargain... They go for 10€ these days.

I modified another brick a while back. Not a Dell but I found exactly the same symptom, dropout when trying to adjust beyond 21V. I found that slightly decreasing the value of the MOSFET current sense resistor (Rsense2) in the datasheet schematic allowed me to reach 24V.

The current sense resistor is normally made up of several paralleled low value smd resistors side by side, so it was easy for me to piggyback an extra resistor. As I remember, I dropped the overall value by about 10-15% (obviously as little as possible).

I also replaced the output side electrolytics with higher voltage, slightly lower value (to maintain can size) ones as I wasn't planning to use anything like the maximum current rating. It really is a sensible precaution.

I'm not saying that this is in any way an ideal mod (the peak current through the MOSFET can obviously go higher now) but the brick has been running reliably (not unattended) since.

Please be very cautious when playing around with the brick. The primary side capacitors can really (and fatally) bite for a considerable time after it is unplugged, particularly if the output isn't loaded - and when it has gone into the 'drop-out' state. There will be residual charge even when the output has dropped to zero. You have been warned!

Thanks for the tip. I have some experience designing >1kW SMPSs back when TL494 was still a thing.
I am not sure how current shunt could allow higher output voltage, but there are some weird chips out there...

[Gyro]

I am not sure how current shunt could allow higher output voltage, but there are some weird chips out there...

Glad you sorted it so easily.

Yes, mine was a 23842, presumably a UC3842 derivative current mode controller. No overvoltage comparators, just undervoltage. After tweaking the voltage feedback, the only thing left to play with was the current sense. 

I guess it just wasn't building up enough flux with the less than optimal turns ratio.

[TimNJ]

Just a word of caution. You can probably get away with this on most flyback converters. But, trying to do this on another topology, say LLC resonant, you might not get such good results.

[wootwoot]

NOTE: I am going to necropost here, as EEVBlog is an active forum, and this thread is one of the main hits that come up while looking for info on adjusting voltage of the LTA804N and LTA806N chips that are frequently in these 19V laptop power supplies.

In my case, I have one of these Smart Air power bricks that I use with a couple of older HP laptops.

Along with a 100-250V AC input, it also has a 12-32V DC input that can be used from a car battery or any other DC source, and this was useful when I worked in automotive development, where originally obtained the device (don't tell them I took it). But - The most interesting thing is it has a button for charge or no-charge, a function it achieves by dropping its voltage from 19V to 15V. Apparently, you used to be able to get access to DC power on planes, and powering a laptop without charging the battery was a safety feature. I like using it with the docking station, as I can avoid having the laptop I'm using from sitting there getting charged for days on end, which leads to less wear and tear on the batteries. I want to adjust a normal power supply to 15V so I can leave it connected to the dock permanently, and take the smart air charger out with me.

A bloke did a good teardown on one of these, and partially mapped out the circuit used on the centre pin: Nerdpedia - "Smart" Laptop Charger

The 103AIW chip is an ST TSM103/A Dual OpAmp and Voltage reference

It looks like it is used in many supplies between the transformer output and the optocoupler. Here are some examples

This bloke has done an another great blog post and identified what the LTA804N and LTA806N chips do:
The LTA804N is similar to a LTA803N which is equivalent to a TEA1751 GreenChip III. Another source mentioned the LTA804N is equivalent to TEA1751LT HV start-up DCM/QR flyback controller with integrated
DCM/QR PFC controller. The 803 might be equiv. to the T variant. (there is also a good application note)
The LTA806N he says is similar to the TEA1791AT Synchronous rectifier controller which may explain why it was not involved in voltage regulation higher up on this thread, although not sure why it still ran with the power disconnected.

Anyway, I attempted to modify one of these big bois: HP 230W HSTNN-LA12 609946-001

Here is a photo of the board. Ignore the variable resistor. I took a bum steer and noticed a voltage divider on VOsense of the LTA804N. I tried adjusting it and it did not achieve anything. Voltage feedback is performed on the secondary side, which is much safer to play with. Unfortunately I did not get any full board pics without the variable resistor

It had a topology like all of those above: A LTA804N on the primary side, 2x LTA806N on the secondary (presumably following the circuit in the TEA1791AT datasheet) and a TSM103/A for voltage regulation that looks like it feeds back through the optocoupler (that had a chip marking of 103AIW). Seeing that the 2.5V reference is connected internally to one input of one of the op amps of the TSM103, it follows that the voltage divider that sets the target voltage would be around the other input on pin 2. And indeed it was. To clarify, the voltage divider is configured so that when the supply is at it's itended voltage, the voltage it puts on pin 2 of the chip is 2.5V. In this charger's original configuration, these values are R1=36k, R2=5.23k.


Putting a variable resistor on this meant I was able to successfully reduce the voltage down. Past 16V however, the voltage started fluctuating. With a load plugged in, it would not sustain voltage. The load was pulling down the smoothing capacitors, and it became apparent the converter was turning off completely. Putting a scope on the transformer windings confirmed this, all pulses stopped. Dipping into the TEA1751LT datasheet and the application note I read that if no output is detected within a certain time the chip resets. I had difficulty probing the LED side of the optocoupler, even using two channels as a differential channel, but the TSM103/A is run off the output rail, so I thought that the dropper resistors for the optocoupler may now be too large for the lower output voltage. I traced two dropper resistors on the optocoupler totalling 300 ohm. I reduced this to around 232 ohm with another variable resistor and this seemed to fix the issue, allowing a load to be supported on 15 volts. There was some small voltage fluctuations (+/-0.25V) when no load was connected, and I could hear an audible change in the 'hum' coming from the converter, coinciding with the fluctuations. Again, reading the datasheet I could see it has a number of modes it can run in for PFC control and I figure something might not be well optimised for low load 15V operation and it might be jumping between these. With a load, it is solid at 15V so I didn't bother going any deeper.

My project failed when I could not get the centre smart sense pin working. I have a few laptops, but I tested an EliteBook 2570p, the run about laptop, and the Z Book 15 G2 workstation. The travel charger works fine on both laptops, charging on the charge setting, and powering the laptop but immediately dropping charge rate to 0W (viewed in BatteryMon) and a non-charging icon in Windows when on 'charge only' mode. On the modified charger, the Z Book seemed to work, but it would think it was 'charging' untill the charge rate dropped eventually to zero. The EliteBook would not draw load unless the voltage was adjusted back up to over 18V.

I measured the centre pin on an array of chargers using a 300kohm and a 470kohm resistor and estimated the centre pin pull up resistance to be around:
65W charger: 493kohm
90W normal charger: 378kohm
90W Smart Air charger at 19V: 378kohm
90W Smart Air charger at 15V: 377kohm
230W charger:103kohm

I changed the pullup resistor shown in NerdPedia's diagram from something like 100k to 360k (with what resistors I had) but it still failed to change the behaviour on the 2570p.

It was clear however that a circuit similar to the one shown above was present in the 230W charger, but I was running out of time to have electrical components all over my flat.

As a last ditch, I cut the traces to the sense pin and attached it with only the 360k resistor. The 2570p still did not draw anything, so I gave up and threw the carcass of the 230W charger in the rubbish.

These chargers are everywhere though, and I might try again. Maybe a smaller 90W brick will be more stable at 15V. Next time I'll fully trace out the centre pin circuit of my smart air charger and the target and hopefully get it working.