Jump to content
EwinRacing Flash Series Gaming Chairs

High_Voltage

T|I Advanced Member
  • Content Count

    109
  • Joined

  • Last visited

  • Days Won

    19

High_Voltage last won the day on December 15 2018

High_Voltage had the most liked content!

Community Reputation

44 Semi Elite

1 Follower

About High_Voltage

  • Rank
    T|I Advanced Member

Recent Profile Visitors

The recent visitors block is disabled and is not being shown to other users.

  1. Yes. You can get more details from the laptop's schematic (touchpad and its connector are in page 50). As for the webcam, I kind of expected that you would want to keep it but decided to mention it anyway. Another suggestion would be just integrating a USB hub for a bunch of devices that don't need high data throughput ( such as touch panel, keyboard, webcam (at least not at that resolution), mouse, etc.). I've seen things like that done, too. Although it's quite impressive what people managed to achieve with this approach, you will be surprised by how absolutely tiny this connector is and how many pins it has (I assume you haven't bought the board yet). I would say not realistic, but you never know! I'm not sure if this is even necessary. PCI-E is a very sophisticated protocol and is capable of conducting self-calibration routine on every power-up. It sends a pre-defined packet of known data and calculates the delays and timings for every lane necessary to send/receive data correctly. Although I'm not sure exactly how much variation in wire length this procedure can compensate for. I would say try keeping your wires the same length anyways, make sure to use a separate biaxial high-speed cable for each differential pair (can get those by cutting cheap mini-SAS to SATA adapters. Each of those shiny blue or white cables can carry one pair.) and make sure their shields are connected to ground. Then maybe it'll work just fine. I believe soldering directly to the motherboard shouldn't be much of a problem as the connector on the motherboard side uses only through-hole mounting (holes can be cleaned and used for wires easily), as opposed to the Ultrabay device connector, which uses a row of through-holes and a row of SMD pads. Just make sure you have a way of safely desoldering the connector from the motherboard.
  2. @astrosynthesist That sounds like a neat idea. I also dream of things like that from time to time before realising how atrociously bad the battery life is going to be. Battery issues aside, the Y510p seems like a nice choice for such project, especially for its display which I found to be quite nice, despite it being a TN panel. In comparison, one of my friends got themselves an Y50-70 (the next generation in Y series), also with a full HD TN screen, and the colours were just horrible when compared to my Y510p side-to-side. Also, I love its keyboard even more than the Thinkpad ones, but guess that doesn't matter in case of a tablet. On the USB ports and m.2 cards I have to say that both Svl7's BIOS mod and mine include whitelist removal for WLAN slot. You have to remember however, that the m.2 slot in this laptop in mSATA-only and doesn't have PCIe lanes or USB routed to it. Therefore the WLAN card you've found will just not work. If you want an extra USB port, you can get a couple from webcam and touchpad connectors. I don't think it is realistically possible to print a substitute connector via 3D printing. Even if you manage to print the plastic part very accurately (maybe, using UV resin printing), you'll still need the metal contacts which are small and peculiarly-shaped. Maybe, @Swung Huang could sell you a connector? He used to make adapters for everyone here and surely still has some spare parts laying around. If not, I guess you can always desolder the other end of the connector from the motherboard and just solder to the pads directly with some high-speed cables (as I guess it would have to go via cable to some docking port anyway).
  3. @astrosynthesist I guess fan control in this laptop is something we've all dreamt of at some point. The default curve is just so incredibly bad causing throttling and system overheat even under moderate loads. You're right, in order to get control over the fan, we would need to create a custom EC firmware, which I guess isn't worth anyone's time. Or maybe, reverse-engineer existing one and append it with extra code to receive our commands via SMBUS, which is more doable but would still be very difficult due to how poorly-documented these EC chips are (I guess, all the datasheets are provided to companies under various NDAs). When I was still using my Y510p, I constantly had this idea of a neat hardware mod in mind: CPU fan uses 4 pins: power, ground, PWM for speed control, and strobe to measure RPM. You could, for example, cut the fan cable and insert an ATtiny85V or other small chip in there, powering it off fan's power and ground wires, and cutting into the PWM wire so that motherboard's PWM goes into the MCU and the MCU then controls the fan with its own PWM. Then you could load a primitive program with a look-up table (LUT) for output PWM value vs input PWM value, essentially getting yourself a custom fan curve. A fun and interesting idea which I never had enough time to explore... As for software side of the fan control and the dust removal routine, I remember seeing a very in-depth investigation into this on some forum a long while ago. They wanted to check whether there's a way to control the fan speed or is it just an "on-off" thing to rev it up to the maximum. It turned out to be the latter. @LeapingLamb The easiest way to get the connector is to desolder it off your original Ultrabay GPU (it's not like it's useful for anything nowadays, anyway). For removing it you can try hot air or/and diluting solder in the joints with low melting point alloys like Rose's metal. Maybe, some workshop can do that for you (just make sure to let them know the part you want to keep is the connector, not the board). Also, mind sharing details on the project you're working on?
  4. The program is extremely simple. If I recall correctly, it just replies to a certain read on i2c bus with some arbitrary value which was supposed to represent the temperature. You can easily get an attiny25 or even attiny10 to do this, no need for Arduino environment at all. What @Swung Huang has then discovered is that you can get actual i2c temperature sensors (e.g. TMP175 from Ti) which can be configured to the right address and happen to have the right register in the right place for the PC to be able to read temperature without the need for a microcontroller at all. But it seems like he had some troubles with it not working on some PCs?
  5. @astrosynthesist There should be a .ino file in the hardware mod archive. That's just a text file with source code.
  6. Hi, @astrosynthesist It is honestly quite bedazzling that people still keep showing interest in this old computer and hacks that we designed for it. I really appreciate your determination to create an open-source version of the adapter, but still can't see what stops @gerald or @Swung Huang from sharing a proven design thus saving everyone's time. To determine SATA port position you could use boardview of the laptop's PCB: https://mega.nz/file/fEZgWIwJ#8EZh7pqszrpnmLwSw3WOvV1r6truK6bIjqumGfm26xY You will need to find a cracked version of a program called Tebo-ICT view 4.0 to view this file. On the other hand, a caliper might just do the job... I think your main challenge with the PCB design will be ensuring the high-speed differential pairs for PCI-e 3.0 are calculated and routed properly. As I understand, this also requires a 4-layer PCB, and a special type of PCB material with tightly-defined dielectric permittivity, but not sure if this is a must. Anyway, there are plenty of application notes from Texas Instruments and others, as well as official PCI-e specification from Intel to have a look into. You might also find my old unfinished KiCAD project useful: https://mega.nz/file/XBIizAwJ#eAMfDliC_GCHppsXWRL4BLuBNUGC3jOMBx5re_wc7KY It has a schematic symbol and PCB footprint for the connector already designed, so in case you can consider switching to KiCAD (which is free and open-source) or find a way to convert it for Altium CircuitMaker, this might save you from having to draw it from scratch. The project had a slightly-different idea of using high-speed wires from SAS cable (silvery-blue ones) to extend the interface out of the laptop, therefore you will find a footprint of cable attach points instead of a PCI-e slot in the project. Anyway, good luck with this, and if there's any questions, feel free to let me know. P.S. Converting the unfinished project to PCB adapter design should be really easy, as KiCAD already has built-in schematic symbol and PCB footprint for desktop PCI-e slot connector. No, it doesn't.
  7. @camwhaler I think this should be possible, as the multiplexers are controlled from the PCH, and not from EC. Might have a look into that a bit later. However, why not consider modifying the hardware? Forcing that multiplexer into a desired state is a matter of desoldering a couple of resistors. Edit: For example, removing RH161 and RH171 should force HDMI output to be permanently connected to Nvidia GPU.
  8. This sounds very much like a hardware-related problem. Not sure if it actually matters what engine the game you're playing is using. Seems like the GPU just crashes or malfunctions in some other way intermittently. I would start with un-plugging both adapter and the GPU itself and blowing the connectors with plenty of compressed air before reassembly. Additionally, you should check the soldering on both ends of the adapter for cracks or other defects. Then, I would recommend to try replacing the power supply. It would be my number one suspect at the moment as you've mentioned that more demanding games at higher settings generally crash more. Moreover, other people here have also seen intermittent issues due to what turned out to be a bad PSU. I hope that helps. Please let me know about any progress with the issue.
  9. I would immediately forget about any solutions without power supply: While it's true that the GTX 1650 runs at just 75W and therefore doesn't need any additional power connectors, 75W is still a lot. Even though the GT 750M is rated for 50W TDP, while GTX 1650 is 75W, I think it would be still possible to run it without a power supply, but this would need some additional work: The stock ultrabay gpu receives most of its power from the 19V line, which you will have to convert to 12V using an efficient and powerful enough step-down converter that will have to be integrated onto the ultrabay adapter's PCB. The PCIe slot on the original ultrabay adapter actually receives power from the 24pin motherboard connector, which means you will still need a power supply for it to work.
  10. They are mostly identical, except mine disables a shutdown if Nvidia egpu is inserted. I think this only affects ultrabay though, therefore you may be successful with svl7's one using an exp gdc.
  11. So, If you haven't already, you could try such options in BIOS as: 1) Disabling Wake-on-LAN 2) Disabling the Always-ON function on the right USB port If that doesn't help, you can check whether the other supply voltages on the adapter power off correctly (there's also 5V and 19V). If they do, you can use them to detect power-on and off instead.
  12. Hm, I don't see any difference to my own schematic I made following Gerald's designs. Can you measure PSU_ON voltage with the laptop on and off? Does it always stay 3.3V, or could it be that it drops, but not completely to zero? In order to set aside any suspicion on BIOS settings, you could try with original BIOS and an AMD card, for example. Or with Nvidia, and just wait until the laptop powers itself off-
  13. Try disabling wake-on-LAN if it is enabled. Also, if you share your final schematic, I could have a look.
  14. This is a good point as well. Originally, the CPU and GPU heatpipes run isolated and only meet at the radiator of the heatsink. If you are running an eGPU setup, your dGPU is idling, therefore you can harness heat transfer capacity of the second heatpipe to reduce CPU temperatures. I've definitely seen it done somewhere (maybe even on this forum). The guy has smudged thermal paste between the two heatpipes, allowing the second one to participate in heat extraction from the CPU, and reported a very significant decrease in CPU temps. I would imagine that this idea can be improved on by using better thermal compounds, such as thermally-conductive epoxy (very reliable but brittle and not very high thermal conductivity) or thermally-conductive silicone glue (the white rubbery stuff you find inside electronics). The ultimate solution would be to try soldering them together with some low melting point solder, like indium (157°C) or Rose's alloy (94°C).
×
×
  • Create New...

Important Information

By using this site, you agree to our Terms of Use. We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.