Chris, I walked the floor at Diskcon yesterday from open to close. It's a big show. I like to talk and there's a lot to learn. I made the mistake of going to Innovex last as I wanted to be primed with all the questions when I got there. Well, I was almost too tired to understand what I heard by the time I reached Innovex. I will go again tomorrow.
My impression is there are a lot of competitive strategies for wiring heads. This is not a single battle between Innovex and Hutchinson. Even TSA has serious competition from similar technology. I can tell you it is not clear who will win, but TSA has a lot of interest. Often on SI, we seem to think if Hutch looses, Innovex wins. Or if Read Rite looses, Applied Magnetics wins. Well, there are a lot of Asian companies doing a very good job and they don't get much notice to us arm chair quarterbacks.
I know of at least 5 different choices being offered for head interconnects. 3M, Read Rite and Innovex all make flex circuits. Hutchinson (TSA) and Magacomp (CAPS) laminated circuits. I spent more time learning about the last 2. I'll try to get deeper into the differences in the flex circuits next visit.
There isn't much difference in all these offerings to us laymen. Laminated circuits have a stainless steel base. The circuits are placed on one side with a clean back for bonding to the suspension. First layer is polyimide. This is used in all the 5 offerings to insulate. Then comes copper wiring which is gold plated and then coated with a protective layer. The difference between TSA and CAPS is how they apply the copper. TSA lays a sheet of copper and then digs out the empty space using plasma etch leaving the wire pattern. CAPS lays the wires using spluttering, and then cleans it up with plasma etch. One subtracts the copper and the other adds it. Pretty simple.
CAPS is a technology Magnacomp licenses from Fujitsu. CAPS has been used at Fujitsu for 3 years. It is a proven technology, but Innovex and Hutchinson have always claimed it was too expensive. Magnacomp says it's cheaper and easier to make CAPS than TSA. Also, they plan to make it in China. Well, we know TSA has not been easy to make as proven by the huge losses and delays getting production ramped. Maybe Magnacomp is right, especially if 3 years of production have left a mature technology. Magnacomp does not have exclusive rights to CAPS. I didn't talk to NHK. They may also have a plan to use CAPS.
Magnacomp if you don't know it is profitable suspension supplier publicly traded in Singapore. They also buy TSA from Hutchinson and will supply whatever the customer wants.
So, all these circuits, whether laminated on metal or not, are made separate of the suspension and then attached later. That means when Hutchinson sells a TSA to Magnacomp, NHK or others it is yielding the same result to the customer. This was news to me as I had thought they built up the TSA circuitry directly on the suspension.
So what's the big deal about laminated vs. flex? Seems that having a metal base layer changes how the part is attached. Metal is bonded with a laser weld and polyimide is glued. Maybe (???) it's easier to laser weld and the lumps of glue cause the flex circuit to lay on the suspension in a way that isn't as good. Also, Hutchinson claims the metal layer keeps the circuit from buckling.
Innovex obviously believes this is not a problem. In fact, they say the electrical qualities of a two sided circuit are more important. Two wires on a plane will electrically influence each other, a problem all you physicists should understand. This influence is reduced by placing a wire on both sides of the polyimide layer. That allows you to run higher levels of current and talk to the head faster. All the designs for TSA are starting to show a pre-amp on the suspension. The pre-amp allows low voltage on the suspension and then boosts the signal on to the brains of the drive. Perhaps that wouldn't be needed with a two sided flex circuit.
Also important is how the customer has to work with the product. Hutchinson claims a customer that must buy the circuit and a suspension, and then attach them has more work and cost. So, maybe it's cheaper to buy a completed suspension. To answer this, I believe Innovex plans to work with KRP, another independent suspension supplier, to supply pre-boned suspensions. KRP said that the announcement from AKI that you posted confirms there is no obstacle their offering a pre-bonded suspension.
Next issue is attaching the head to the pad. Currently, you have 2 ways to bond heads. I can't explain them, but Innovex has not been able to provide both bonding solutions while Hutchinson can. Innovex says they are qualifying the second method now and expect it to be available in a month. If Innovex solves the pre-bonded suspension issue and they can attach heads anyway the customer wants, this should eliminate any arguments that it's harder to work with flex circuits.
Just to muddy the water, I was also told some customers think it's better to attach the head to the circuit and them both to the suspension. This is said to make alignment easier. So, it's clear to me that there is no single solution.
So, maybe it keeps coming back to cost? If the customer has to attach the circuit to the suspension, that adds cost. Innovex pre-bonds circuits then you have to companies sharing profits. All the other suspension suppliers won't go out of business, so they will either buy from Innovex, Rite Flex, 3M or make their own solution. I suspect any advantage Hutchinson may by developing TSA internally will be lost due to high cost labor cost. I think KRP for sure will be partner a suspension partner with Innovex. Both companies are reputable stand a good chance of success in my opinion.
Another interesting development is Hutchinson isn't, so rumor says, signing new customers for conventional suspensions. They want to pressure a switch to TSA. I think they figure they can get production numbers up quickly now. Asian competition is killing them on conventional suspensions so they want to get out of no margin products which take away capacity to make TSA. So, if you read Hutch is selling less conventional suspensions, it has no bearing on Innovex's older wire sales.
Also, I talked to companies that have recently sold Seagate a lot of equipment that would be used to make suspensions. Seagate has always made suspensions, but now they want to make more in house. That could mean that the issue of pre-bonded suspensions would continue to have no impact on Seagate's strong relationship to Innovex. But don't let that lull you into a false sense of security. I also hear they are testing or using Rite Flex, 3M and Hutchinson. All these customers have no loyalty. They must be won over again and again.
Innovex is currently selling flex circuits to Maxtor to bridge the suspension base plate to the head stack. Innovex hopes to build on this to sell Maxtor flex circuit for heads. They say they have very high interest from Maxtor and they feel all the developments of pre-bond suspensions, and ball bonding heads will further this process.
Hope this all makes sense. I'm not an engineer, and I heard so many things yesterday I really need to revisit these questions to make sure I understand what's going on. Also, for what it's worth Read Rite is all the buzz at Diskcon.
Regards,
Mark
PS Market seems to have renewed interest in Innovex. This is obviously accountable by the AKI announcement and repeated statements of improved sales and reduced channel inventory. Barring Joel's market melt down, I think we have a good investment. |
