I did not manage to fix this and have now sold it for £40 to a more competent person I hope.
Here for posterity is the text from the repair cafe email and my ebay listing:
I’m going to summarise my thinking/reasoning of what’s going on with the video mixer, and hopefully this can either expedite some of the repair if you take it to the guy that I gave you the number for, or anybody else that may have a stab at it.
All of this is just what I’m inferring based on what circuits I’d expect on the board and the datasheets for the devices that are available. I wasn’t able to find an applicable schematic for the board itself annoyingly.
In picture ‘path_of_composite_signal.jpeg’ It shows the path of the signal from the BNC connector in yellow going to the bias network, into the TW2866 chip, which is the composite video handler. I can see the composite video making its way to the bias network of a few diodes and capacitors circled in red in ‘signal_bias_network_signal_OK_here.jpeg’, so everything up to this point is working correctly even when the output screen shows no composite signal. (Signal can be seen on my scope in ‘composite_signal_at_bias_network.jpeg’ which is what you’d expect the composite video to look like on an oscilloscope)
The signal then goes directly into the TW2866.
I can see the two voltages provided to the TW2866 by IC23 and IC17 (circled in red in ‘roland_v-4ex_3_highlighted_LDOs.jpeg’) and they are both fine, there’s no voltage ‘drop’ where they’re no longer able to provide the voltage that synchronises with the loss of composite video output. I will say however that I don’t know what they should be providing as I can’t find a part number for either of them and Roland wouldn’t provide a part number for either of them, so they may be providing the incorrect voltage for them but I have no way to know.
My overall suspicion of what is causing the synchronisation issue is one of two things. Because I don’t think it’s the voltage supply to the TW2866, I think it’s something to do with the clock synchronisation or with the FPGA that does the processing of it.
In the ‘path_of_composite_signal.jpeg’ image, there’s a small device circled in red which is the crystal oscillator for the FPGA (field programmable gate array) (circled in blue). I think the crystal oscillator provides the signal to the FPGA and maybe also the clock for the TW2866, or the FPGA generates a clock signal for the TW2866. ‘Clock signals’ in this context are a kind of device or signal that provides a ‘tempo’ for all the electronics to work together - if the tempo starts to fail, everything becomes a mess and the processing doesn’t work properly anymore because different devices are no longer working at the same tempo or synchronisation (like the composite video kind of failing and tearing!). I don’t have the right kind of measuring device to see if the crystal oscillator is providing the right kind of signal, and because I don’t know what outputs are coming from where in the FPGA it becomes very difficult to know what a signal should be when I measure it.
In terms of repair, I think either solution is to replace the device circled in red (X2) in ‘path_of_composite_signal.jpeg’, or the TW2866 itself as there is some slight discoloration of the device as in ‘suspicious_discoloration.jpeg’ but it’s hard to tell conclusively if that’s normal. Both of these devices can be replaced easily if somebody has the right tooling to do this kind of work, which unfortunately I don’t anymore. The other issue may be caused by the FPGA (blue circled device) and if that is the cause, I’m afraid it’s a lost cause. Even if you were to replace the device itself, there’s software that would need to be loaded onto the FPGA which Roland would never hand out as it’s their special sauce.
Hopefully all of this gives you and anybody that has a stab at it some idea or guidance of what’s going on with it, and why it was such a complicated pain to try and diagnose - even if it was a bit fun!
Bad Ebay Poetry
Hello video nerds, welcome to this tale of woe, a man of hubris brought down low.
Sold spare or repair, with hdmi working and composite bust, a V-4EX incited his gear-lust.
Emboldened by fixes to pots and through hole bits this miserable creature thought himself fit to tackle surface mount kit.
Months of research, experts sought on forums, in person, via Facebook, even to a repair cafe brought.
Either crystal or microchip was at fault, a tricky job indeed, but the chip was for sale for basically naught.
It would cost £400 for a new PCB and so determined not to pay, he cried out loud “how hard could it be?”
Hot air, flux, magnification, hot iron, ventilation, all were set to play their part in his deep regret.
In short he shorted, with sloppy solder, lifted pads of infinitesimal pitch, weeping and sweating and swearing each moment more desperate than the last.
Until at last defeated he knew this mission was never going to be completed, delusion and avarice had guided his hand, nothing could possibly go as he planned.
Now this beautiful broken mixer sits forlorn in the dark on his hobby cupboard shelf of shame, the man himself forever chastened, never the same.
He learned a lesson most painful: no one is perfect and we all must respect the terrible power of the Dunning-Kurger effect.
-–
So yeah totally failed with my attempted repair and I think actually it was the crystal that was messing up the composite signal.
I’ll supply the buyer with all the info I found in my research, the original chip and another untouched copy of the replacement chip.
Seriously you’ll need to be really good at surface mount repairs to fix what I did.
Mostly buy this if you need the other PCBs or components.
I honestly wish you better luck than I had, not just better luck but better skill too.
-–
Auction starting at 4p, one penny per channel!
