Ok, this is gonna be a bit of a quickie. Basically, i got this one ridiculously cheap as well, with the intention of repairing it and gifting it to the new singer in my significant other's band, in order to facilitate more productive creative work. Just like the first one i did a teardown on and then modded, this was sold as faulty, but with more... obvious issues, let's call them.
The seller had been considerate (?) enough to include a photo of the issue in the eBay listing, so at least i knew what i was going to be in for. Didn't look like too big of a deal though - one chip, one resistor and one capacitor missing from the board, as well as the USB socket.
That being said, i wouldn't have grabbed this, if i didn't have my modded one to use as a reference. As it turned out, the missing chip was a reset controller / voltage supervisor, keeping the USB PHY and probably the Xmos controller in reset, until the 1V core voltage rail stabilizes, plus some delay.
It took some digging, but the "SDE___" marking on the 5-pin SOT-23-packaged chip meant it was an OnSemi NCP303, more specifically the version with 0.9V as the threshold voltage. The missing resistor was a 47k pulling the reset controller's open-drain output to the 3.3V rail, and the missing capacitor was, at least according to the USB3343 datasheet, a 1uF for its internal 3.3V regulator.
Sadly, it seems i neglected to take any photos from during the process, but i hope a textual description can suffice.
Installing a new USB socket was, as expected, quite the "surgery". Virtually all the plated through-holes had been ripped out, but on the bright side, none of the four traces were on the internal layers of the PCB, so that was one bullet dodged. On the bright side though, the pins of these USB-B sockets are thin metal strips (as opposed to solid cylinders that would fill the holes completely), so i had a thought. Using some thin leftover resistor legs, i bent them into L-shapes, and soldered them to their respective traces (two data lines, 5V and ground) after scraping off some of the solder-resist, and the short side of the L's into the holes for the pins. After clearing out the old solder and pieces of pins, i was relieved to see that my plan would work, ie. both the USB socket pins as well as the "recreated traces" would indeed fit inside the holes in the board.
Double-lucky was the fact that the plastic insert of the USB sockets i had, protruded a bit below the edge of the outside metal shell of the connector, so i wouldn't even need to modify the shell to prevent it risking making contact with the exposed "new traces". I did, though, need to shave off some of the protruding plastic rails to make room for said "electronic prosthetics", let's call them. This, for the purpose of hopefully ending up with the USB socket sitting flush on the board (or as close as feasible).
For extra mechanical reinforcing, the underside got a doubled-up strip of used solder-wick, spanning the distance between the two mounting "claws" of the connector shell, as well as generous quantities of solder. At the end of it all, it turned out quite sturdy. Not sure if or how well it might stand up to whatever abuse got it into the initial shape, but anyway...
Now, for the sake of swiftness, i chose to transplant the voltage supervisor chip to this Saffire 6, and then replace it in my own one at some later point. I will say that it took a few hours in several afternoons, two iterations of the circuit (in simulation), not to mention quite a bit of tweaking component values.
I had several bullet-point criteria the circuit needed to fulfill. I needed it to trigger at or around the same 0.9V as the original chip (which pretty much ruled out a MOSFET-only option); i needed some hysteresis in order to avoid "chatter" near the threshold (which took me in the direction of Schmitt triggers); and last but not least, some way of setting a delay between the supervised voltage crossing the threshold, and the output being activated.
This "version 1" served more like a block diagram of sorts. I searched through the schematics of several dead laptop mainboards i had, but couldn't find a single-gate Schmitt buffer or inverter, and using an entire 14-16pin chip was off the table - i needed this thing to be as small and compact as feasible. But then i remembered i had quite the stock of scavenged comparators, and there had to be some way of adding hysteresis to one...
And a big of google-fu later, i had my answer - a slight bit of positive feedback, as it turns out.
The slightly oddball values are due to them being scavenged from the above-mentioned laptop mainboards and others, so no need to take that part too literally. But even this version ended up requiring a revision - initially, the MOSFET on the output was missing. For whatever reason, due to a series of brainfarts (i guess?), as well as not bothering to measure before delving into this, i had achieved the complete opposite polarity of the output signal. Thus, it required an inversion on the output, hence the MOSFET there. Which is just as well, since the original circuit had an open-drain output itself, so there. I also added the reverse-parallel diode at the input (remembering MOSFET gate-driving for switching power supplies and class-D amplifiers), for a quicker discharge of the "delay capacitor" upon power-off.
Yes, the "cosmetics" are indeed a bit of a dog's breakfast, but hey - what can you expect, for free AND quickly? I ended up sticking the comparator board onto the top of the MIDI optocoupler with some double-sided tape, and the inverting MOSFET with some more tape to a blank spot on the mainboard. Connections were made with resistor legs - between these and the tape, as well as the negligible weight of those things, i don't envision any sort of mechanical issues whatsoever.
No comments:
Post a Comment