Archives for October 2018

New FDA Guidelines Affecting Your Medical Device Development

10 min reading time

Bob Marshall Sings “The Regulatory Blues”

(You’ll want to hear this.)

Reading Time: 10 minutes


I had a pre-submission meeting
so I went down to Silver Springs.
I took documents, and drawings,
and lots of other things.
They said, “Go home, boy.
Wait until the telephone rings.”
I got the Regulatory Blues.
About to lose my mind.
I’m waiting on some data
no one can seem to find.
Now my 510(k) submission
is running three months behind.
I got a call last week.
Said they were coming for an inspection.
Someone down at the hospital
got a real bad infection.
Now they’re telling me
I got to recall the whole collection.
Tried to renew my registration.
I just sent it in the mail.
Because every time I went online,
their website would fail.
Now they’re putting me in handcuffs,
and taking me to jail…

Joe Hage: I’ve been doing this for six years, Bob. That was a first. Thank you very much.

Bob Marshall: Well, thanks very much [laughs].

New FDA Guidelines

So, I’m going to talk to you today a little bit about new FDA guidelines. Some things to think about, some of it from experience and exposure to the industry.

We’re going to cover two things: We’re going to look at the 2017 guidance, some of those in detail, that were issued in fiscal year 2017 – some of those impacts and things to think about and learn from. And we’re also going to take a sneak peek at some things that are coming in this fiscal year, that we’ll also have to adapt to.

Real world evidence

The first thing I want to talk about is real-world evidence. This is an interesting thing because we have FDA, in a very progressive way, starting to understand that every time we bring a new device out, we don’t have to do a new trial. There’s a lot of data, right? We’re moving toward a whole world of Big Data.

I think that’s a really good thing. But if you look at the standard and when they came out with it — I actually did an article on this because it’s almost talking out of both sides of your mouth — it’s like saying that there are situations where we could look at this and we may have data that is the same as or even better than what you could come up with through a trial, but at the same time, nothing is going to change our evidentiary requirements. So that’s an interesting challenge, to deal with both sides.

But, I have seen it effective. I recently talked to Stryker, and it was interesting that they had recently submitted a product – it’s part related. It’s dealing with stroke, so heart and brain related – and, in that process, they did not have to do a clinical trial related to the function of the catheter for that specific application. And it was cleared within 30 days.

But part of the issue was they had good bench data. They had pre-clinicals, but they also had real-world data from very similar other applications that they could lean on. So, it worked out very well for them and, I’m sure, others potentially.

It’s something we have to get used to, which is a positive. It doesn’t mean every device that may be of a Class II level or Class III is necessarily going to require a full clinical trial. So, there’s still a lot to learn in testing this, but it does seem that they’re pretty open to it. So, it’s certainly something to think about.

Reporting age, race, and ethnicity data in clinical studies

For a long time in clinical trials, we’ve not done a good job covering gender, ethnicity, and other issues in having a good population to represent the end users. And, they put out guidance to talk about how they want to see this being different. And, I think it certainly is time.

I think we’re always a little worried with trials to get a little too involved with different splits of patient population. But in the guidance, they give us pretty clear guidance. And, it’s really based on that differentiation, potential risks, different types of effects for those different populations.

So, it doesn’t mean you necessarily have to do that, but you have to think about what those effects might be if you’re dealing with something different for men and women, “how does that work?”

Some of the examples they used in the guidance were interesting. There were several studies that had been done for, I think, knee implants. For a long time, generally, the patient population was men. But as women were having them, they were having many more complications after being implanted for some time. And they really weren’t properly represented in the initial clinical studies related to that implant. So, it’s definitely something that is important to cover all of those issues.

Drug/device classification issues

We’re blurring lines everywhere within the regulatory space. And this is one that has been there for a while. I do think the Office of Combination Products at FDA does a great job. But, then again, part of the reason is because, ultimately, it’s not their problem.

All they’re going to do is guide you in the direction to go. And at times when I’ve had the involvement with engaging them, they were extremely helpful. So, I think that’s a real good part of this.

But, in reality, they did try and come through and deal with some of the issues around that specific classification and “how should we do that? How do we deal with that?”

They really realized that some previous guidance they had given was going to be problematic, specifically related to things around chemical action. They initially had said, “There can’t be any chemical action or it’s not a device.” And so, they’ve taken those couple of things and really, specifically in this guidance, backed those things off. It’s good because there were definitely concerns that those things could be problematic, certainly, if someone gets pushed down a drug path when they’re actually a device.

That’s a small part of it. So, they’ve looked at a better way to determine whether, “Is that chemical action almost coincidental or is it the main part of how the product achieves its results?” So, I think that’s an interesting piece to look at in trying to understand what we have here.

Pre-submission process

How many folks have been involved in a pre-sub? [Show of hands] And did you generally find it to be helpful and positive? Yes. Okay. Hands are up.

So, it’s interesting — and I wrote about this, too — because there seem to be two camps in this. There are folks who see it as a natural follow-on to transparency at the agency and the desire toward collaboration in the industry.

But, some folks are just hard-set against it. It’s like, “Don’t go down there. Don’t ask them any questions. Just do your best to interpret the regulations and go.” But as it gets ever more complex, it’s difficult to do that. And so, I really think there’s an opportunity there.

Personally, I’m a big fan. Prior to doing what I do in writing for Med Device Online and organizing the content, I was a consultant for seven years and, in the device industry, for 15 more before that.

I’ve been there a lot of times and I think they were all positive. I can’t think of anything that went bad. One or two times, on our end, people we were working with and representing didn’t do some things that were probably the best for their desired outcome. But in general, it’s been a good process, I think.

But it’s interesting that folks I generally see not recommending this process are usually folks that are ex-FDA. I don’t know if that’s just a small data set, but it seems like maybe they’ve seen the inside and they’re like, “Don’t mess with that. Don’t waste your time.”

And I think that’s unfortunate because I think there’s a lot of good there and, as I said, I think that the results have been great.

It’s important to keep in mind, though, they gave us very clear guidance on what that pre-submission should look like. And a couple of the things that are really important – the device description is critical.

You’re explaining something new or different, at least, to the FDA. They don’t have any frame of reference on this. You may have been developing this thing for two years. And, to you, it’s second nature. You really need to describe it well, so they can understand what it is, and they can get comfortable with whatever it is you hope to do.

So, that part is really key. I think it’s also really important to make sure that you are very good with your specific questions. Don’t go down and ask them, “What kind of clinical trial should we have for this?”

We can all guess what the answer to that would be, right? There’ll be so many sites and so many patients that we can’t afford it.

So, it’s important to frame that carefully. Do your homework. We come up with a reasonable plan, “We think, if we have three sites and we have 30 patients at each site, that we can demonstrate the intended outcome for this product. Do you agree or disagree?” It’s always important to put those into a yes-or-no type question.

I think it’s a process that we’re all getting used to. I think there’s a lot of value in it. It’ll be interesting to hear the continued experience of the group with it.

De Novo submissions

These are coming more and more forward. These have been around for quite some time. In the late ‘90s, early 2000s, they were getting, maybe, 10 a year, And the general word on the street was, “Just stay away from it. It’s just a wormhole. You’re going to get lost in the process.”

But the idea is, if we have something that’s not classified, and we don’t really know where it belongs, it’s an automatic Class III. But there are devices that are clearly not major risk. They’re minor or moderate risk. And so, this is the way to do that.

It’s a little bit quicker in theory. FDA is trying to say they’ll get them done in 150 days and working the percentage of them being done that way up over the next five years. It’s a good effort to do that. Much better than having to go down the PMA path, if it works for you.

They put that guidance out and they gave some guidance about what to have in that submission. It’s important to pay attention to these things. They don’t do it for no reason.

It is a newer process, but the numbers are starting to come up. Sometime into the 2010-ish timeframe, we were getting toward 25 a year.

We’re actually starting to get the list of recent clearances. I get a lot of requests from people that say, “Hey, you know, we’ve got our new device released.” And they want Med Device Online to cover it.

There are actually, now, some de novos that are coming in. And they were under the radar, I think, because they happened infrequently. So, it’s a new pathway and it’s important to keep an eye on that.

Breakthrough Devices Program

So, breakthrough devices are unified here. We had some different access pathways in the past, but they’ve laid this out and explained – which I think was necessary – how we can use this.

I’ve seen a lot of people wanting to get in the program, obviously, because it gets quick attention and rapid response. And that’s good in two ways. As we were talking earlier, the idea is that you want to help people, and a company wants to get in business.

I put up a couple of examples here. These were actually cleared through the expedited access pathway which is now replaced by Breakthrough. But it is happening.

It’s good for patients everywhere and it’s good for the businesses, but it has to truly be that type of need that is going to do some very necessary things and help folks, so certainly a good thing to have.

Submissions involving cybersecurity

We’re all very aware of a lot of things that are going on. The incident recently in Atlanta caused a lot of concern just on the commercial side. But we all worry about those types of things in the medical device space.

So, here, they’ve given us guidance of what they are looking for. That really means that you need to evaluate your product or systems vulnerability, understand what risks there are and what things you’re going to do about it.

And the interesting thing is, it doesn’t end there. They really want you to discuss how things will be updated over time because we know software changes all the time and is easily changed. But, you have to think about this impact as well as the functional impact that it could have.

They’ve given some really clear guidance as to what they want to see when that is a part of your product. And the big thing is, they want to see that you are thinking about that part of the product. If it’s connected, as so many devices are today, we really need to think about potential cybersecurity threats.

Medical device accessories

Medical device accessories, for a long time, have been an interesting challenge. It was, if you were an accessory to a device, then you were a device and you were in the same class as the device you were connected to.

But they’ve really come out and turned on that a little bit. And I think that’s actually a good thing for industry because they’re trying to look at the accessory on its own merit – what really is this thing doing?

When I saw this, the first thing I thought of was the simple example of a respiratory mask. The mask is typically and traditionally classified by what it is used with. It’s not used by itself, but it can be used on a CPAP to help somebody. It could be used on a ventilator that is helping to keep them alive.

It depends on what your use is. It can be a Class II, it can be a Class III. So, those kinds of things make it important to be able to consider it on its own and not connected to the device.

You need to evaluate its risk level and how it could potentially interfere with the function of the device. But, it’s nice to see them open that door for us as well. So, those are the key pieces there.

Coming distractions

We’ll do a quick look forward here, relative to what I call coming distractions. So, these are the 2018 draft guidance documents. The interesting thing is the list of draft is, for this fiscal year, much longer than the ones for final.

And the ones that will be final guidance are specific and complicated, but I won’t read those to you. You can obviously see them. But I’ve got a couple of quick comments on them.

The Third-Party Review has been bandied about for quite some time. The program has been around for 20 years or more. They did put out some guidance in 2016, and it was a draft, but they’re coming back around next year to visit it again. So, I don’t know exactly what the bee is in their bonnet, but that one seems to be something that’s staying there.

They’re going to actually come around, again, with the pre-submission. They’re going to talk more about Q-Sub, probably give us more direction on it. So, I think that’s interesting as well to look at.

The abbreviated 510(k), which has not been used a lot except by companies for very simple follow-on products. So, all of these things are important things to be aware of and to stay on top of.

I know everybody’s already got a full plate, but it’s definitely an important thing to do.

So, I’m going to draw it to a close there. That is the end of what I had to present and the end of my time.

Joe Hage: Well, I have a question for you, and this is a way for us to help you.

As editor of Med Device Online, I suspect that you love scoops and good content. I suspect some of you here in the audience have good content to share that you’d like some visibility on Med Device Online? Show of hands? Meet the editor.

Bob, thank you very much. Let’s hear it for Bob.

How Product Design, Technology and Manufacturing Change on Miniaturized Medical Devices

12 min reading time

Our January 2019 Sales and Marketing agenda

Presented by Carsten Horn Business Development Engineer at Maxon Precision Motors – April 4, 2018

Reading Time: 12 minutes


MaryBeth Privitera:: I like to take pictures of things. I observe people for a living, and I observe the surroundings I live in.

Carsten Horn My speech today is not about showing you what great motors we have, and what great gearboxes, and whatever we have. It will be more about the methodology that I figured out over the years – how to design such small medical devices and the impact on miniaturization – and what I learned over the years, what is beneficial and not.

About maxon precision motors

Of course, I need to talk a little bit about Maxon first. That’s part of the story, that you have a bit of an impression of that. I will talk a little bit about some driving factors for miniaturization. There’s a lot around. I can spend hours on that. And if you look on the internet, it will be endless.

Then I will use one example – what I want to use to follow along during the process of the development of a miniaturized version. I will talk a little about the outside in, inside out. Maybe you heard about the development. The design’s always from inside out. I also want to bring the part from outside-in in here.

I will talk a little bit about technologies that make it possible. Miniaturization has a lot of technology that’s behind it that gives us the opportunity to go so small.

Then some of the design principles to follow. Miniaturization is always about tolerances, about sizes, about handling, and some impacts on costs also.

In miniaturization, it is not always the case, if you go smaller, that it is getting cheaper. That’s a misunderstanding, I can promise you. It doesn’t matter.

I will go to some points, and we’ll talk about the good-enough design for many of these things.

Let’s talk a little bit about Maxon. We are a Swiss-based company with about 2500 employees worldwide. About 10 percent of all these employees are engineers in the R&D department. So, we’re mostly driven by engineers, also our CEO and everyone. It is still private owned, so there is no stock available for the company. And our biggest strength is really the development part of ultra-miniaturized stuff.

We have a wide range of standard products. You, maybe, have seen we have a big catalog that’s been around for years. It has about 500 pages, and there’s a lot of DC and BLDC motors, and gearboxes, and encoders, and controllers inside. The good thing on that one is, if you want to make a fast prototype, we can put these things together and you have very fast prototype running because everything fits together.

I need to admit also if this thing has more than 400 pages now, that’s only 20 percent of our turnover. 80 percent of our turnover is strongly customized or at least modified. We make much more than just only these motors.

Typically, we design, for customers, mechatronical units.

Medical Technology Reliable, Compact, Efficient

I need to say that about 60 percent, in the U.S. it is even higher as about 70 percent of our turnover is medical. So it’s medical devices. It’s typical that you find us in applications like drug pumps, atherectomy devices, surgical robots, in surgical instruments, and on and on in many devices out there.

I think one of the major reasons is the reliability of our components, and the quality of the components that we have. And this is pretty much a leading factor. And, also, the power density that we can generate, and the sizes that we generate on motors, which leads me to the main topic of my speech today, which is the miniaturization of products out there.

Driving factors for miniaturization

I think the major driver, from my experience, is the patient. And everything that we do, that we develop, we should always have him in our focus. Smaller devices, if you talk about ambulatory stuff, so the thing that you wear around with you, like a drug pump, stuff like that. So, these things need to get lighter and smaller. The reason for this is it is much more comfortable.

So, what we see at the moment, the push on the drug delivery devices is the patch pumps, for example. You wear them like something you glue on your arm, and you don’t feel them. They shouldn’t bother you. You should be able to do your normal sports, your swimming, your daily business like you want, and they shouldn’t bother. That’s the whole goal of the story.

But, this means you need to make them extremely small. For the pharmaceutical industry, it means you need to have really high concentration on your drugs because the viscosity of the system is increasing.

Other examples – surgery tools, the insertions or the wounds that you generate or the number of insertions you make should be minimized. We have heard about the cost in hospitals and stuff like that.

So healing is a very big point of all that is minimally invasive. This is the driving factor for getting people to recover faster and getting out of the hospital.

Going there, you need to minimize. We are working on micro robots and miniature robots that are working within the body and doing the surgery, not from the outside anymore. That will be the future, and you need to miniaturize for that.

Other things like the example you see over here – This is a heart support system.

With all of that size – this thing has an outer diameter of six millimeters – you wouldn’t be able to help patients with that one. This is a small pump that supports your heart system. So it’s just a heart support system, but it gives you the opportunity that the heart can heal. Stuff like that is driving the miniaturization. And it’s always about the patient.

Micro ambulatory drug pump

I will use an example, which I will follow through my speech. This is a small-pump system. This is an ambulatory pump. There is a special pump that is named the double-chamber pump. So, you have two nice chambers on that side. There are pistons down there.

While the pump itself is moving, you have a pumping mechanism and the valves that are switched over there. So you have two ways of switching it. This gives you a continuous flow, pretty much, so you don’t have a lot of pulsing in the system while you control the speed of this pump mechanism precisely.

And, of course, this is just a prototype. What is shown here is just showing a little bit of the pump unit because the requirements are very tough on that. The footprint should be extremely small, as small as possible, because it’s ambulatory. The power consumption should be low because, otherwise, you’ll make the batteries big, and you don’t have any advantage out of that. And this unit needs also to be sterilizable. So, it needs to be cleanable after a while.

Some more of the requirements: I think one of the most important points when it comes to designing such units – and this is always the first step in the design process, as far as I can tell from my experience –­ is build up a strong mathematical model. Try to get from the outside requirements that you have from your customer and understand these requirements precisely.

I think this is one of the key things – I can’t explain it often enough to my engineers – really understand the requirements.

There’s a really wonderful speech from Meghan Thorne from Medrobotics. You maybe have heard that. She had been on the MD&M show in Chicago. I think she will be on BIOMED in Boston once more.

It’s really interesting because she was the girl who goes into operation rooms to look in on the surgeons and finds out that 70 percent of all surgeons are doing the operation with socks, no shoes, and she was wondering why.

You need to understand this sort of application, and you need to really write them down. And now the tricky part is, transform these external requirements into internal requirements that you can design from the inside out. To do that, I strongly recommend, build up strong mathematical models. Try to get the requirement and transfer it to an internal function.

Back to my example from the pump. So, you need to understand torque/speeds. I just used here two physical sizes. There are the forces for the pump itself and a little bit of friction for the valves themselves. So, you need to understand how this is applied, what you will really see to understand that, and how the forces and torque on the motor on the slide will look like. Because this gives you the option to understand what are the peak forces and the average forces.

What this is determining later is the size of every drive unit. It doesn’t matter if it is ultrasonic or with an electric motor or something else, you need to understand it.

Also, you should understand the motion exactly, how this thing is moving, because there is a switching point – we talked about that – where the valve is switching. So the question is how precise you need to be at that point.

Outside In => Inside Out

Build up a really, strong mathematical model. This is very simplified here, so evaluate it and try to transfer it. This has a lot of impact later on your design. I use that all the time.

I need to be honest. I worked for 20 years with this sort of mathematical models and increased them over the 20 years. I built them up, and I have a lot of templates, so I’m pretty quick with that.

It gives you really the inside out. If you start to design and change parameters and change your model with it, you always have your requirements in your hand. You always see the impact. You’ll immediately see, “Okay, my efficiency’s going down, my current is going up. This has impact on my electronic side.” For that reason, I think it’s pretty important.

It also helps you later on in the change process, if you do changes. And change is normal, expect that. The customer will change requirements.
I don’t know how many pumps I have designed, from syringe to linear peristaltic, all the way around. I have never had the project where they haven’t changed. We always had changes in the requirements. Always. So, expect that it will happen.

I need to say that because I’m from Europe, we have more the understanding of, “Oh, golly. We get a specification, and then we do the design and off we go.” And that’s not the reality, to be honest.

Technologies – Design from inside out

Design from inside out. If you have the strong, mathematical models finished, then start with a technology assessment. But, as the design engineer, automatically in your brain, you’ve already started designing. So, you have something in your mind – “that’s how it will look. I’m going to put this there.” You know, the mechanical part. Also, the electrical part, it’s the same thing – “how I put the controlling mechanism okay. If I put the encoder on the output, I need to handle the–” All of this is going on. That’s normal.

But why try, then, to first do an assessment on the technologies? Because this has an impact, later on, on your tolerances and sizes. You can’t decrease parts with the tolerance. If you miniaturize something, the part is getting smaller. But you can’t decrease the tolerances in the same percentage. It’s not possible.

I have an example for that, a simple one. Take a look around on technologies that are available. It’s not only the mini motors or micromotors from Maxon, there are more around. For example, take a look at the clock industry, what they use for technologies, for production, down there.

Take a look at how connectors are produced, how you can integrate stuff easily, what is available on that. Define, for yourself, a technology base. I think it’s pretty important.

Then comes the big point – if you have figured all that out, then you start the design, really, from inside out. And I believe, in this technology, there’s no way around.

If you start from the outside and try to get to the inside, this is always going to be a nightmare. Start from inside; get to the outside. Really understand the power dimensions or the power that you need. For example, for the drug parameters, determine the sizes of the magnets, the winding technology that you can use.

Or using the ultrasonic motor, determine the frequency that you need to have and the diameters that are necessary for that. Start from inside to out.

If you have concerns, regarding some parts, of strengths and stuff like that, FEM analyzers is the way to go. Start FEM analyzers. And, if you have electromagnetic circuits here, do simulations down there. Dynamical problems can also appear. Not in this case, to be honest.

So, for example, Also, build up a risk analysis in parallel. Always, on the bottom part, take a part to say, “Okay, I have a technology risk. I have a design risk, whatever it is.” That is typically the input for your design FMEA that you need to do anyway.

Design principals – technology to make it possible

I talked about that already – you can’t size down and expect that you have the same tolerance fields.

So, for example, if you have a shaft with one millimeter or have a shaft with 10 millimeters and have 10 microns on it as the tolerance, which is possible, and now you just shrink them down ¬– and the shaft is looking the same way, by the way – it’s going down to one millimeter. You will see that your tolerance is, unfortunately, getting to 0.1 micrometer. Then that’s not possible anymore.

The dimension is changing, so expect, on small parts, lots of relatively large tolerances that will exist.

Also, production – for example, if you produce a shaft with one millimeter, it’s something different than producing a shaft with 10 millimeters. All this world is changing.

If you have small parts like that, unfortunately, dust is going to be a problem. If you have small gear wheels, they have only the diameter of a millimeter.

At the point of dust, it’s a big animal. It will stop the working of your gearbox. It will stop working because it will be blocked.
So, you have requirements like clean rooms, unfortunately not because of medical, but because of dust. And this is impacting all of your designs. Be aware of that, if you go really small down there.

Prototypes – design from the inside out

Then we’ve heard about a lot of prototypes. Rapid prototyping, very important part. I am the biggest fan of that.

Unfortunately, when it comes to small parts, there’s not so much technology around. The tolerances are not there. So, I need to talk to this guy. We need something that is for 1 mm parts, super precise,that will have only micrometers of tolerances.

Then we would use it immediately, and much more than we do at the moment. We use it, but we always know it’s a little bit of pain that we are working to.
If you build a prototype, I think one important point is to check your mathematical model. Verify that it is working, that it is correct, because this gives you a strong and robust design later on.

Try to separate functions. I think this is also very important. Don’t try to combine things in one part. This is a misunderstanding – trying to integrate functions in one part and thinking that the cost will go down and it will be much easier.

This is where you get your problems. I promise you, from experience, it will be that part. Because if you change or optimize one function on this part, you always impact the other functions. So you can’t optimize that.

So keep one function for one part, if possible. I think you can optimize one function, you can optimize the next function and so on. You can optimize your product by that. If you integrate it and you start working on that part, it has impact on the next one. It is a running wheel, and you never come to an end, if possible. I know that it’s sometimes not possible, but mostly if possible, try it.

Also, sometimes it’s cheaper to have two simple parts and assemble them, instead of having one super-complex part where an injection molding guy loses his hair and gets mad with you all the time when he tries to produce it. Just an example.

Try to test your mathematical model on your prototypes, function by function, to see if they are working, separately. Then, later on, if you have mistakes in your unit, you can easily use those results for analyzing where your problem is coming from. I think this is also pretty important.

And always share that information with your customer. Every result you should share because he also has impact, he gets a better understanding from that, too. And, also the impact on that.

Prototypes – Verification of the requirements

Prototypes testing – whatever is possible, step-by-step verification, it really depends. You need to have good equipment for testing like thermal cycles, like vibrations, like low tests, like lifetime test. There are a lot of things that are going on there. So be aware of that.

Check out what you have, what you can do in house, and what you need to outsource. And don’t save money on the testing. You’ll pay it later.

You know what experience is? It’s the number of mistakes you ever made in your life. I have a lot of experience. And this was one of them.

Miniaturization and costs – Good enough is good enough

Okay. Good enough is good enough. A little bit on costs – target costing, also around in the world, I don’t know how popular this is in the U.S., but in Germany, typically, we build up a BOM – a bill of material. Immediately we have the technologies in there, we try to analyze the cost per part.

We also build up the route. We have every step – the minutes of production – down there. And during the development process, we try to reach all that cost.

So, you hold that at a certain point, and then you follow up with your cost and try to figure out where you are.

It’s not only that. You need to fulfill the requirements regarding some technical issues or quality issues. You also need, at the end of the day, to reach your costs. Because, otherwise, this is not a product you should produce as there is no win-win situation for you and your customer. So this is also very important to take into consideration.

Sometimes if you can’t reach your cost point, you need to go back to the starting point, if necessary, and rethink everything because you can’t sell it into the market because the market price is not there. I think this is very important. So you need to follow up on that one.

I think this is concomitant calculation meant here. We also look a little bit at engineering time. That’s also pretty nice, but it really depends on the projects. If the volume is higher, then engineering time is not a big issue because, typically, they are divided by a lot of parts over years. It depends on what is the company policy out here. On smaller amounts, this is much more critical. You should also follow that one.