Last year I took the plunge and bought myself a Yamaha MOXF-6. As with the Motif XF series (and later Tyros models) this synth could be upgraded with a Flash Expansion Board to allow the storage of new waveforms.
Having a look around for flash board prices I discovered that Mutec do a similar board to the Yamaha version but lower in price. Further investigation revealed they have three versions of the flash board – 512MB, 1GB, and a whopping 2GB version.
Having decided that the 2GB version was a little expensive for my needs I found the 512MB version on Thomann. The price differences between the 512MB, 1GB and 2GB appeared to be pro-rata to the memory capacity.
Looking at the photos of the 512MB version on Thomann flagged up an idea – what if I got the 512MB version and sourced the Flash ICs and hand-soldered these to the underside. The Flash ICs are similar to the chips I use for the S101 Waveblade Card & EXFLM 16MB but with quicker write-times and higher capacity. The result should be a 1GB card for much less than the 1GB card retail price!
Mutec FMC-05 Stock Photos @ Thomann:
This was looking good! I had now sourced the 4x Flash ICs ( JS28F00A TSSOP) and the other components for the bottom side (33R Resistor Nets & 100nF caps). Off went my order to Thomann and I waited for my soon-to-be-1GB card to arrive!
It arrived in my place of work and as I opened the package I could see a different board through the pink bubble wrap:
My plan was scuppered. Mutec had changed their FMC board design to a single-sided BGA (Ball Grid Array) based layout. They now populate the boards with the required number of Flash ICs according to card capacity: 2 ICs for 512MB, 4 ICs for 1GB, and 8 ICs for 2GB.
This new layout makes a lot of sense in the production environment. The PCB and panel layout is the same for all versions so the same tooling can be used. All the pads in the photo above have solder deposits so they certainly use the same solder stencil for all versions. With the board being single-sided this cuts down production time as a double-sided board has to go through the paste process, pre-inspection, pick-n-place, reflow, and final inspection processes twice.
Based on the cost of the Flash ICs and other non-fitted parts the actual production cost difference between the 512MB board and the 2GB version would not be more than £55. The (Thomann) retail difference between the 512MB and 2GB is £209. You read that right!
My new 512MB card came home and was installed in the MOXF 6. After an initial format it declared I had 512MB under the hood. I was happy with that – until one of our buyers said he could get hold of the 6 Flash ICs needed to bring this up to 2GB. The ball starts rolling again..
If I was to upgrade this board using BGA Flash ICs a number of things would need to considered.
First there is the moisture concern. All IC packages absorb moisture out of the air and this causes no problems in normal use. When an IC is placed through the reflow oven the moisture boils inside the package and destroys the wafer inside. Before I place the BGA ICs on this board they will need dried-out in a moisture cabinet (as with all ICs in a production environment). The Mutec board will need the same treatment prior to going into the oven or the already-installed ICs will suffer the same damage.
Secondly there is the solder pad issue. Luckily Mutec uses the same solder stencil for all versions of the board so all the pads already have identical amounts of solder so no problem there.
Thirdly there is the problem of keeping the BGA ICs exactly in place during the reflow process. If the IC is slightly off the pads it may fall between the pads during the reflow and that’s just a big mess of solder bridges underneath. Luckily the BGA flux I used for this is quite tacky and the BGA ICs were kept in place during the process. All the other pads for the resistor nets, capacitors & buffer ICs were fluxed and the components put in place.
All that’s needed now is a suitable reflow profile for the oven and a steady hand to place the board on the reflow oven rails. As this is a ROHS product a compatible ROHS profile for a similar product was used.
Although the temperatures inside the oven reach 250 degrees the PCB does not get subjected to this until near the end of the conveyor. The actual time the solder is melted is quite short compared to the time the PCB is in the oven. This little drawing below will explain better:
After waiting at the other end of the oven for 10 mins I was relieved to see everything kept in place. Using a stereo magnifier I was able to see that all the BGA ICs were nicely centered and all the other little components took to the solder.
Now just a matter of taking it home and trying it out.
The MOXF took a while to format the board and finally declared 2GB under the hood. I filled the board up with WAV music files (that took a loooooooong time) and each file played back as expected.
The time taken to upgrade the board from start to finish was about an hour. Hand-placing all the little components took longer than I expected. Ensuring the BGA ICs were accurately centered took a bit of time too but that was a priority and the time didn’t matter.
I still have those JS28F00A TSSOP Flash ICs so if an earlier version of the FMC-05 turns up those will be put to good use!