Don't Miss

Samsung G3 Series 3.5 Inch USB-Sata Adapter Repair

By on May 13, 2021
USB Sata Adapter Repair

Recently my busy computer user friend Marco asked me to check upon his no longer working Samsung G3 series 3.5 inch USB to Sata adapter board.

Because I at first had no clue what he had brought me I started by examining all passive components. Made some photos and measured all values. Here follows the list I made:

list of components

Also C52 above U4/G5795 was nc. And position L1 left next the dual mosfet was nc too. (nc=>not connected).

After a quick examination I found no bad or broken resistor nor any bad capacitor component. And I started to check the active components being the step down DC DC converter G5795, the 93LC46B1 serial eeprom, and mosfet AP4957 with transistor Q3 with smd code DAYS.

I couldn’t find any info on the G5795 converter, maybe again because of infringement of some firm’s copyright?, but I assume that the given datasheet of a G5796 resembles the exact same functions. Because that type was constantly given at any Google search. Next photos show this board.

usb sata board repair

usb sata board fix adapter

Naturally I also tested the USB to Sata adapter board by inserting a brandnew Sata 260GB SSD hdd and by connecting an 12V DC adapter to the power input jack. And connecting a USB cable to one of my USB ports of my Quad-core desktop computer. But as expected nothing happened. My computer still didn’t see the attached SSD drive.

So I started to measure the output voltages of the step down converter but assuming it should at least give a 3.3V DC output, there was nothing measured. Also the coil with marking 100 wasn’t giving any output voltage. Below an extract of the assumably pin compatible G5796 datasheet:

G5796 datasheet

And after a quick ‘hot spot’ test, by just using my fingertips, I noticed that the G5795 chip was getting abnormally hot. Which explains why the bigger square controller IC with marking JMB509 couldn’t work and also the sop8 93LC46B1 serial eeprom couldn’t be read by the JMB509 either.

Also finding a datasheet of this JMB509 was impossible, probably again because of some copyright issues with Samsung, who sells and probably designs these small adapter boards? And testing this chip was simply not possible other than just checking it by trying to fix the culprit on this defect board.

And I did a quick transistor Q3 measurement by using the diode beep test on my DMM. And I also assumed that the dual 7,4A 20/30V P-channel Mosfet was still intact.

Next I desoldered the sop8 serial eeprom from the board to check its function by trying to read its content by using my new GQ-4×4 USB programmer and a smd to dip socket adapter. Next screen copy shows that the 1K eeprom was still just completely fine! The screen also shows the successfully extracted 1K bin file.

About this new GQ-4×4 Universal Programmer : I bought it because I needed to make sure I could read,copy and save the calibration data of the battery backupped SRam’s in my Tektronix oscilloscopes. I managed reading all 8192Bytes of the 4464 in my Tektronix 2465A with my TL866II

and a special made 28 pins cable with IC test clip, but it gave verify issues reading the Dallas 1225y or the FM16(w)08 memory chips. Why I needed a better programmer which is the new GQ-4×4!

Although both programmers still not support all of these SRam or battery backed up Dallas memory chips and neither the Ferro Magnetic FM16(w)08 and the larger 18(w)08 (for the 2440 Tektronix oscilloscopes). GQ-4×4 does support the Cypress FM16(w)08 Rams but fails to support the almost identical (both firms are now one?) Ramtron FM types. And also the 18(w)08 types altogether. But perfectly supports all Dallas 3V lithium backupped modules! And sadly also the RT809H, and the TL866II still need many more upgrades supporting all those types too! But if the FM types are not available they possible still can be read as a Dallas DS1225 memory type. The RT809H however can’t read any of the Dallas DSxxxx memories! Why we simply need several universal programmers to fill the device supporting gap! (Why I already count now having 5 of those different universal programmers, and also several different smd to dip and other IC adapters).


Next picture shows part of datasheet of this 3-wire 1K serial eeprom that according to its specs is capable in providing at least 1000.000 secure read/write operations! And a retention time of over 200 years!

And the datasheet also shows several different chip pin configuration settings. The nice thing of using an universal programmer is that we normally do not need to bother with using the right setting because we just insert the chip into its smd to dip adapter with pin 1 in the right direction and we are good to go! And if we do not get any good read or verify we simply check and adjust our chip or chip setting.

As I already wrote in the earlier given component list, the likely 3.3V output of the 100 marked coil goes to the source 2 pin of our AP4957 dual mosfet. And a page of that datasheet follows.

93lc46a datasheet

Next an extract of the AP4957 AGM P-channel dual mosfet datasheet. It is obvious to assume that the D2 output is giving no voltage out because there is no output voltage at pin 3 of the step down IC G5795. Strangely these G5795’s are still plenty available on Aliexpress although there is no datasheet available. Like happened in another 230VAC to 5V USB adapter repair I wrote that was about a defect unknown ‘data protected’ sop6 PWM controller chip from a firm that used these in their special power unit devices.

ap4957agm datasheet

On 10 May ’21 I ordered five of those G5795 3.3V DC DC step down sop8 chips for about +3 Euro including shipping on Aliexpress.

I am almost certain that replacing this ‘specs unknown’ chip will fix this small USB to SATA adapter board! And the herein given information will no doubt be most useful for other Samsung G3 series 3.5 inch Sata converter owners.

Update: On this link afterwards the exact datasheet for the G5795 was found. It is indeed pin compatible and has the same function but works on a lower 340KHz frequency, instead of the 500KHz of the G5796 and higher types. Strangely the datasheet of this IC is always only one page, and there should be more!

In the meanwhile my friend got in contact with Samsung to claim a new adapter board. But all they thus far said was that it shouldn’t have happened so soon and that it was quite unexpected. Anyhow, no matter how the service of Samsung turns out to be, as soon as the new G5795 DC DC step down converter is here, I expect that his old Sata board will be working like a brand new one!

albert from netherlands

Albert van Bemmelen, Weert, The Netherlands.

Please give a support by clicking on the social buttons below. Your feedback on the post is welcome. Please leave it in the comments.

P.S- If you enjoyed reading this, click here to subscribe to my blog (free subscription). That way, you’ll never miss a post. You can also forward this website link to your friends and colleagues-thanks!

Note: You can read his previous repair article in the below link:



  1. Paris Azis

    May 13, 2021 at 1:24 pm

    Hello Albert,

    This article is an excellent piece of work!

  2. Albert van Bemmelen

    May 13, 2021 at 7:07 pm

    Thanks Paris. I remember that you also have used a special bought eeprom programmer you needed in the past for programming a Bios or other boot chip in some problem device. Those are indeed great for fixing boot chips problems in laptops and alike but mostly are limited to the sop8 or dip8 chip sizes. And since the latest eeproms already are exceeding the 42 pins dip size a universal programmer is a must. The new 44 pins eeprom memory chips already exist like the 27c322. And even if we have a universal programmer that supports many chips and chipsizes those now require an adapter too. Good thing is that we can make these ourselves to expand our universal programmer with it. The pcb can be bought for $5 and they will ship 5 boards but the shipping costs are sadly very high (about $23 USD). See here:

    And recently some pcb boards were ordered for making the Jim Williams Avalanche-fast rise-pulse generator. I got 3 of those boards free shipping for just about $6.65 USD at:
    So if we have a pcb design that we want them to produce, OSH Park is probably a better option.

    • Albert van Bemmelen

      May 14, 2021 at 6:01 pm

      Warning!!! Better never order anything from the above mentioned PCBway website!! Because after I ordered those boards in worth only $ 5USD, and also already paid $23 USD on shipping costs, DHL today asked me completely unexpected to pay an additional fee of 43,85 Euro which is absurd,completely insane and criminal!!

      • Albert van Bemmelen

        May 19, 2021 at 11:17 pm

        After my above mentioned just $5 USD costing PcbWay order became extremely overpriced on insane and wrong DHL import charges, after DHL already was paid $23USD for shipping costs, I refused the parcel and asked DHL and PcbWay what in-the-hell had happened?
        DHL after 5 days answered that I could ask for a special form to state my complaint, but I gave up on DHL and PcbWay and in the meanwhile opened a dispute on Paypal and on Pcbway for a full refund. And today after many emails and bad communication I finally got my S30USD money back minus a small fee again to Paypal. So I hope never having to deal with money ripping DHL or PcbWay ever again!

        • Albert van Bemmelen

          May 24, 2021 at 3:12 pm

          Update: In case anyone also is planning to build the OSH Park avalanche fast rise pule giver, they do not work with the 2N2369A transistors. The transistor must be a now no longer sold or manufactured 2N2369 type. Apparently only with those the pulser works producing fast rise pulses in the kHz range! With the wrong 2N2369A types used they only avalanche at random at a very low about ~2Hz frequency and are completely useless to use to trigger your scope on for the purpose the Jim Williams Pulse generator was designed! Instead therefore better build Kerry Wong's avalanche pulser that probably also works with the 2N2369A transistors at a higher Uce breakdown supply voltage of 120V DC. See here for more info.

          • Albert van Bemmelen

            May 24, 2021 at 5:49 pm

            PS: Today I tested even a marked 2N2369 without the A extension that still didn't work in my 2 build OSH Park board Jim Williams avalanche fast rise pulsers! And without those exact selected 2N2369 transistors that no longer are available, the OSH Park pcb circuit never will work! It did cost me a lot of invain ordered parts exact according to the given BOM part list. So better build another pulse generator as given on Hackaday or Kerry Wong's design that works with many different transistors! Even with the ones that completely failed in the OSH Park circuit!

    • Marco Hekers

      May 31, 2021 at 11:36 pm

      Well done, Albert.
      The drive works and is seen again.
      It's also accessible again, like it should.
      Thank you...

      Greets, Marco

  3. Albert van Bemmelen

    May 13, 2021 at 9:07 pm

    PS: The 27c322 is already a dip42 size for the extra added addressline(s) to be able to address all memory cells. One adressline more doubles the capacity of memory in our eeprom. Future eeproms will be even bigger. The mentioned eeprom adapter board expands the to 40 pins parallel programming limited socket of most universal programmers. The eeprom adapterboards can also be bought readily made in smd size and dip adapter size for the GQ-4x4 and the TL866II. See here for the TL866:
    or here for the way too expensive overpriced GQ-4x4 ADP-054 adapterboard!

    And here for the smd adapter ADP-019

  4. Parasuraman S

    May 13, 2021 at 9:59 pm

    Most went above my head!

    • Albert van Bemmelen

      May 14, 2021 at 4:58 am

      When you fix one of those modern LCD TV screens by just removing some copper tracks, you lost me too dear Parasuraman! My knowledge gap on fixing modern TV's is big, and I rather just build and fix smaller devices that do not take up so much space anyway. Which is one of the reasons why I never liked being an electrician. Always working in large buildings and constantly on the road for yet another big project.

  5. William Shaw

    May 14, 2021 at 2:12 am

    Very nice work and interesting too!
    You have a vast knowledge of data chips which I'm sure started when you got that Commodore128 in the background of your picture (1985?).
    And as for "universal programmers", maybe they should call them "somewhat universal programmers"!

    • Albert van Bemmelen

      May 14, 2021 at 2:08 pm

      My first computer was the first personal color computer the Commodore VIC20 which only had 3,5K Ram and used expensive ROM packs for the games, William. But the first computer I worked on was the one that my father bought and I still keep somewhere, the TRS80 from Tandy which only had about 14K Ram. That computer also was used on the secondary technical School in Roermond I later went to. And my personal computer collection has grown since then. And I never throw any good working device away, although I probably do not frequently use them anymore. Knowing that a old dated processor in 1980 contained just about 10,000 transistors and now fits 300.000 times in a today's microprocessor that has 3 billion transistors, it is easy to understand why they become kind of obsolete now.

  6. Andy Shiekh

    May 14, 2021 at 2:57 am

    I can but admire your tenacity; how much does the board cost new?

    • Albert van Bemmelen

      May 14, 2021 at 1:51 pm

      I simply do not know Andy. Probably just getting a new board may be a problem now too? And my friend probably knows what the now defect device had cost? I have to ask him.

  7. Waleed Rishmawi

    May 15, 2021 at 3:55 pm

    greetings. that is a lot of information to read and understand but I enjoyed the article and learned a lot from it. thanks for sharing. have a blessed day

    • Albert van Bemmelen

      May 15, 2021 at 7:26 pm

      You too Waleed! Glad that you liked the article!

  8. Yogesh Panchal

    May 17, 2021 at 3:20 pm

    Nice write up,
    I think external SATA casing is more economical.Now future repair jobs becomes more complicated because each and every module in electronics are you have RT809F Serial ISP BIOS Programmer?? ......just for review.

  9. Albert van Bemmelen

    May 17, 2021 at 7:45 pm

    I have the improved RT809H nand programmer that can do what the RT809F can do and more, Yogesh. Plus the Genius G540, The TL866/866II, The Xeltek SP-L programmer, and the new GQ-4x4. Plus the Elektor Gal programmer, several Eprom programmers and some other devices like a STM32 programmer. Plus a few USB to serial adapters, and a few self made USB programmers like the USBasp and others. And of course also some of the necessary USB to FPGA interface adapters for programming Xilinx, Altera (Now Intel) and Atmel (Now Microchip) FPGA and also AVR chips. And also a DIY Wireless Xilinx FPGA programming interface with a programmed Raspberry Pi 3B+ board that works in the Xilinx Web development version starting with the Vivado devices. (Interface is not supported to work in the older Xilinx FPGA development programs for Spartan and so on versions).

  10. Andrea Del Corso

    May 19, 2021 at 8:33 pm

    Come sempre un ottimo lavoro,congratulazioni...e grazie!
    As always a great job, congratulations ... and thanks!

    • Albert van Bemmelen

      May 20, 2021 at 7:31 pm

      Grazie Andrea per i complimenti!
      Dutch translation: Mijn dank Andrea voor de complimenten!
      My thanks Andrea for the compliments!

  11. Albert van Bemmelen

    June 1, 2021 at 1:12 am

    Update: On 31 May the culprit U4/G5795 sop 8 chip was finally received, immediately soldered onto the interface board and the USB to Sata adapter with my 2.5 ich SSD hdd tested. It worked perfectly and my friend was happy his device could be used like a new device again!


Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.