Finally… the Jupiter 8 panel returns from silkscreen!

It’s been a long Hard road full of many bumps and turns. But the day has finally arrived… Cora’s panel was finished!

There really was only one chance at getting this right. After Coras powdercoat failures there really wasn’t going to be yet another opportunity to powder coat her again if the silkscreen went wrong. That’s a lot of pressure on her…and a real narrowing of the chances of a positive outcome. If a mistake is made, you cannot remove paint from a powder coated surface once it’s on, without leaving paint streaks, or the chemical used in an attempt to remove the paint turning the shiny powder coat surface milky…ie damaging its finish. Either the silk screen was going to go on once and correctly, or I would need to sadly dump Coras original panel and have a whole new one laser cut. That is what it basically came down to, because she just couldn’t handle another metal sanding. I don’t think that I could handle it either! And this decal application stage is where the MOST could go wrong. So much of it was out of my hands. I had done my homework and due diligence with the decal file, there was nothing more that I could do. It was now down to Peter, the man with over 40 years experience in silk screen printing to metal that I put my trust into.

Peter specialises in printing to antique and custom car panels. Decal paint onto metal… exactly what the job requires. He’s usually busy adding racing logos to the sides of unusually shaped vertical car doors. flares to bonnets etc. This is why he had larger screen printing frames that could clear taller pieces underneath than most other workshops. And Coras panel sits nearly 10cm high with the rear panel kink pushing it up considerably.

This height is the reason why 2 other workshops were unable to do the job. It seems the modern screen printers use a method and equipment that only has 9cm clearance. To add extra degree of difficulty, the panel to be printed on is not flat when sitting, but on a downward angle adding further technicalities. Printing to the smaller rear section was even more cumbersome.

Peter wanted to wait until after the hot spell of weather as he was worried about paint runs, and paint drying speed being affected and looked for several days in a row that he was guaranteed not to be sidetracked. It was an agonizing wait I must say of several months. I had to do several gentle nudges along the way. Coras panel also couldn’t be baked in any way between prints to expedite the drying process, which is a common technique, … all of that special filler in the rust and weld holes would likely create problems for the powder coat if cooked again. So would be a slow process for the workshop also.

Peter said “your job was one of the toughest I’ve had to tackle in 40 years.”

I felt quite lucky that he was prepared to take it on. There’s a reason why there is only a couple of people in the world that will agree to doing the job of a silkscreen print to a Jupiter 8 panel. Because others take a look at what’s required and if they have enough experience, they see the headaches and potential problems. But I hear you say “Plently of people print to other synth panels, so what’s the big deal with a Jupiter 8?”. Well the main problem is the massive WIDTH of a Jupiter 8 panel that’s required to be printed on (with the clumsy angle problems coming in a close second). Over 100 cm wide.

A single clear specialized film must be made of the illustrator art file. The film is as long as the whole panel. In the silkscreen world, this film is considered REALLY LONG (read: a massive headache). Here’s a pic of the film in the workshop that was used:

This film is then used to create a “negative” on a large metal frame that can swing up and down on a bar using rear hinges like a sandwich waffle maker , with silkscreen stretched across the frame and the image produced using a specialized chemical “emulsion” process and the film:

So where the clear film has “black” writing .. the silkscreen above has instead just clean see-thru silk where paint can soak through when applied, creating that image on whatever is below. On the silkscreen in the picture above it looks light grey, because that’s the color he used on Cora and the paint soaked through. The clear “dead area” parts of the original film are turned into areas on the silkscreen that are solid/blocked (green on the pic above) so paint can’t travel through. To save time and money, all of the graphics needed for the job are printed to the one silk screen. The front including a different layer for each color, the rear and the side mod panel were all committed to one screen.

This printing system has been used for decades. Theres no doubt it was the identical one that was used on Cora originally in the Roland Japan factory. It’s also used for mass T-Shirt production. Here’s a summary:

But here’s the problem… with a Jupiter 8, the frame needs to be so wide that when they put the film across the silk and tighten it to make it stable, the long strip of film stretches. Out of shape. Causing the image to be wider than original. Then of course it’s stretched across a flexible, stretchy silkscreen. For something small to print, the distortion can be as little as around 0.5 mm or so. But across over a span of more than a metre, you get up to 5 mm shift! Ouch! And it’s impossible to predict how much this variation will ultimately be, and then try to compensate on the illustrator file because … every time it’s different. Different film thickness/ brand and printer (that can introduce their own size variation), different tension, different silk, different length, it’s just a massive floating target. A minefield.

Jeff from Custom Synths warned me about this. He had wrestled with the pain of this reality before, from his own experience when doing a Jupiter 8 so I was well prepped when I got the exact same lecture from Peter. It will never be 100% perfect. Especially for the first panel being printed. When Roland was decal printing Jupiter 8s back in 1981 they no doubt threw out the first dozen or so panels as they printed and tested and tweaked their films (at $200 a film… nothing for a big corporation but a lot for me!) and therefore new silkscreen every time, getting the alignment eventually right for the bulk run. Going into this, if all goes amazingly well, Coras panel (the first and only one being printed) might get 90% there. Prepare yourself in your head, I was told by both. There will be something you will have to live with. Either a smudge, a color problem, an alignment problem, a decal mis-print such as a missing letter, a panel scratch or damage if it slips off or the screen is dropped, anything was possible. As far as alignment, the decals most likely to suffer we’re going to be the ones to the far left and right of the panel, they would be the toughest to line up due to the distortion issue.

It’s really, really hard.

As much as I had planned on making the whole process a simple one with a guaranteed single pass of the decal per color due to my time spent making an accurate transfer paper print and file, Peter still in reality had to print in multiple passes, a section at a time and moving across the panel. A total of 5 passes just for the front grey. What he did is “mask off” sections of the screen and then just squash paint through for a certain area using what the industry calls a Squeege (rounded rubber strip..attached to some wood to hold onto) rather than a traditional paintbrush, wait for it to dry, do a re-alignment on Cora, gently remove the masking tape that covered up the next lot of decal from being printed, and mask over what you’ve just done so it’s not double- printed. Peter used a small squeege about 10cm long, small enough to be able to just squash the paint through a small selected area. Very easy to make a mistake and forget to mask something you’ve already printed. Very easy to smudge something that has not dried properly yet.

Cora has 2 colors on the front panel (grey and orange) and 2 colors on the rear (grey and cream). Peter called me in on Day 2 to help get the orange color right. We made up several different color batches …

And tweaked and tweaked. A bit more red… a little more white, etc. Every time, running several test prints on the spare metal sheet I had powder coated for this exact purpose when Cora was done:

Color is another floating target. The same orange paint printed onto silver raw metal looks different to when printed onto a dark background. You need to test .. test .. test. And wait for it to properly dry every time as the shade and reflective properties change as it dries!

Here is Cora with the grey passes finally down but the orange still to be run:

That last miraculous powder coat done by Ed and Sash was looking stellar .. she was shining. I started to see a light at the end of the tunnel it was 60% there!

There were a couple of decals that were around 1mm off. I mentioned them to Peter. The sweat on his forehead and the look on his face pretty much gave me my answer… this was as close as it was going to get. He could start again if I REALLY wanted to, but only if she was powder coated again. Remember, it took 3 powder coats to get Cora right to where she was before the decal. Chances were that another sand would just destroy her, or if I was lucky would need 100s of hours of work to strip and re-fill the surface, and it could then very easily take 3 powder coats again to get it right back to here without bubbles and likely $100s of dollars and months of further waiting. Only to possibly have the decal still be 1mm out for a few places in a reprint. Araraghhhh!!

I briefly went back to the tracing paper print I had made months before that Peter had sitting there to see if any variation was my fault. And to my horror, the whole tracing paper sheet was out of alignment also… when lined up on the left, it was out by a whole 5mm on the right!! Huh? What the hell! Peter just shook his head and said “this is what has thrown us as well …you see what we are up against, take a look at this…” and held up a ruler against my printed tracing paper ruler. Indeed it was out by 5mm. “But Peter, I checked this over and over…”. He replied “I’ve seen this before .. the damn tracing paper has SHRUNK over the few months since the print.” Indeed, it had.

And when a tape measure is laid down next to my printed measuring tape, the reality of the shrinkage compounding across the distance of over 1m is apparent:

So many moving targets it’s just crazy.

We still had the orange layer to go and also the rear printing, we were only 60% of the way there. But the orange layer was a biggie. It’s the color and decal design that uniquely just screams JUPITER 8. It needed to line up with the other printed grey decal in so many places and also with the two LED holes in the panel on the left. It could still all go pear shaped…We decided to forge on and see how this next stage went. Would this be the next potential stumbling block that brought the whole process to an end?

The orange layer needed to be printed in several separate sections too.. and Peter again pulled it off. His 40 years experience was paying returns. Now we were 70% there. However now there was something that didn’t look quite right, the orange was not “popping” quite like expected. Peter explained that the original 80s paint used by Roland was Lead-based. That made it thicker and heavier.. but also poisonous. Lead paint has been banned now for many years. To replicate its look however with modern paints is yet another challenge and it rarely looks identical. Granted, we had used more of a matt reflective paint when the original was probably closer to gloss also. Peter suggested an option (be it a risky one) which was overprinting a second orange layer. If originally Cora had required only a single orange pass of the screen, and it had originally lined up perfectly, this might have been a viable proposition … the lineup of the second print needed to be perfectly the same to look good. But the orange had been printed in 3 stages across the panel with positioning tweaks, to make it line up best with the 5 different stages of the grey layer below that had been printed before it. To now print a second layer of orange over orange in 3 more stages would have been dicing with death.

After some head scratching, deliberation and weighing up the risk it dawned on me that part of what I felt was missing was a shine from the orange layer. “So Peter … what if we overprinted a clear GLOSS layer … would that work? Is that possible? It would help pop the orange, but in the event that there’s a slight mis-alignment it would not look so bad because it’s just clear paint that would be mis-aligned, not actual more orange paint…”. He responded “Hmm…leave it with me”. So I went home. For an anxious nights sleep.

The next day he had a smile on his face. And behold.. well Hot Damn if he didn’t nail the second layer with gloss PERFECTLY! Peter was quite proud of himself from that quiet smirk. Now the orange looked right!

Now we were 80% there. Yes there were small overall 1mm alignments this way and that way in the orange that in a perfect world I would have adjusted. With all the variables that Peter was working with, the result was outstanding. There was every chance that some of the slight variations could even have been my fault in the file. But we were entering into a world of diminishing returns and it was getting harder and harder to turn back. She was very close.

But the next question was …would there be a mistake in the rear panel print that might still cause a scrapping of it all?

Well with 2 layers (2 colors) and that clumsy and difficult angle to print on, anything was possible. Another day later …and Peter came through.. no smudges and the color and alignment perfect on the rear.

Peter got 90% there with the Roland logo and now we were 95% there with the correct rear plug decal print.

There was simply not enough of an excuse to turn back now. I was simply going to have to live with the last 5% small variations. One of those variations (the J in JUPITER on the logo being slightly too close to the next letter) was actually my fault it seemed when I looked closer at the file…. something I had missed several times over during my beta test prints. Plus I had purposely changed some of the decals from the Roland originals, Cora was never meant to be identical to how she was before, for instance I commandeered the far right panel button away from a rarely used “Tape Verify” function to be assigned as a far more critical “MIDI SETUP” for the Kenton midi mod (a much more elegant solution than the ugly red push button drilled into the side key bed rest of other Jupiter 8s), and for the Arpeggio switch I modified the “Ext” sync label to “Ext / Midi” (this is the position for the Kenton Midi mod for the Jupiter 8 Arpeggio to sync to the Midi clock) and I signed Coras name on the rear panel and the date for prosperity. So she was never meant to be just the same as before anyway.

After another day of drying it was finally time to “Come get her.”

Since Coras face panel has been home, I’ve sat and stared at it now for 5 days, waiting for the paint to properly cure. Literally watching paint dry. There’s something about the layout and color of a Jupiter 8 panel that is just so beautiful and mesmerizing. It’s a work of art, not just a synth panel. The Roland designer was a Frikken Genius. My guess is that with all the sliders and knobs and buttons returned to the panel, any slight inaccuracies that are apparent now will sink more towards the background. Peter explained that the paint doesn’t actually harden until the 4th day so i could take it home but the panel couldn’t be touched or twisted. When you stand back and take in the full picture, the result is pretty outstanding. Especially when you think about the poor rusted, flakey scratched dented mess she was before….

Time to start putting the panel back together