Ari Zagnoev - 18 August 2021
I have been posting a brief monthly status update on the Combimouse website. In this blog I will go into more detail.
My preference is for you to post comments on the Combimouse website instead of commenting here.
Let me know what you think of this blog and if you would like me to discuss any other topics or if you have any other feedback.
Things should start speeding up and we will then be more active on social media. At the moment we only have 32 likes on Facebook. I reckon we deserve more than that. As someone has pointed out we haven’t been very active so maybe we don’t deserve more. Soon things will pick up and we will become more active, but we need to get the ball rolling. My preference is to post less with a focus on quality and interesting news.
I have been working on the latest Combimouse prototype since 2013. The aim is to make a prototype that is ready for mass production that can be made at a competitive selling price. Also the aim is to use latest technology. Once the bulk of the development is complete we will then approach investors to help us get into manufacturing.
One of the objectives of my current work is to improve the battery life of the Right Combi. Previously it had a battery life of about one week. I estimate that the current prototype has a battery life of at least 3 months. This is with only one AAA battery which just goes to show how technology has improved. When I look at the low current consumption of the various ICs I am astounded. This is without any attempt at optimisation. Later when there is time I will optimise power consumption and I expect to get at least another month or two battery life, but at this stage it is a low priority task.
I have made good progress and I now feel that with the possible exception of the keyboard modules the end is in sight. The keyboard modules consist of the keys, key mechanism and flexible membrane circuit. During this development I haven’t made many any announcements and haven’t being active on social media. Development is such a risky endeavour with many setbacks and frustrations. If I were to tweet about them it would be appear chaotic. Often things don’t work as expected and do strange things - I would say to myself that this thing has it’s own law of physics - but of course once I find the cause of the problem it all makes sense.
I have been working on a limited budget and have done everything myself. This is not optimum and this has resulted in development has taken much longer than I would have hoped for.
I have redesigned the right Combi plastics. It is now a lot more structurally stable and is manufacturable. I am an electronic engineer with no experience in plastics design however I have researched the subject and have received advice from the company in China that will potentially manufacture the Combimouse - more on this later. I still need to run the design through mold flow analysis software to determine if the design is suitable. I will do this once we have funds and can commit and liaise with the manufacturer. In addition when we receive funds an experienced Industrial Designer will design the plastics.
To keep costs down the plastics must be designed bearing in mind Injection Molding requirements (if you are interested here is an article). The Combimouse design is a three way compromise between strength, weight and cost to manufacture. Tooling for injection moulding is very expensive and will be more than US$100K for the Combimouse. However the design will probably will be optimal for injection moulding and there will be no unnecessary incremental tooling costs. For example, for now I don’t envisage any undercuts.
Software development requires that the human readable code must be converted to digital information that can be run on electronic devices. A compiler/linker is the software that does this conversion. The compiler/linker for the Nordic processor is expensive so I haven’t purchased it and have had to make a number of compromises. This won’t affect the final product.
I have updated all the electronics. There was always a risk that something wouldn’t be suitable or that there would be some incompatibility and interference between sub-systems. As of 18 August 2021 I would say that this risk has been greatly reduced. I now have most things working.
As of 18 August 2021 I have the following:
Dongle - I have had a dongle PCB manufactured that is exactly the same as the reference design from Nordic. It did work but as discussed below I had some problems programming it. I am currently using the development dongle on the development kit.
Left Combi - I have a working circuit board. I have an old plastic housing which should be okay. I haven’t yet integrated the keyboard. I made a mistake in specifying the membrane and so the tail is incorrect and is too short. I might use it but it will mean that I can’t use the real housing. I might have to order a new membrane but I am waiting to see if there are any other problems on the left membrane and possibly problems on the right Combi membrane - I order them together.
Right Combi - I have an old plastic housing (as discussed below I have to make a new one). I have a populated circuit board that likely has no problems (this is a big achievement). I have adapted the reference mouse/keyboard software solution from Nordic and for now everything seems to be okay. The keyboard brake still needs to be tested and I still need to add the keyboard. The mouse works well.
At the moment the proximity sensor on the right Combi takes about 17 seconds to calibrate when it is first powered up - ie. when a new battery is inserted. I spent some time trying to resolve this and I will probably need to contact the supplier to ask for assistance. It is a nuisance during development because I have to wait. It isn’t really an issue for production Combimouse because it only happens when one changes the battery.
The proximity sensor on the Right Combi at the moment has a relatively slow response time. I haven’t measured it but it probably is something like 100 to 200ms from when the right Combi is gripped to when the mouse starts working. I need to look into that and hopefully I can improve it. I do have some ideas.
There are also at least three other risks
1. I haven’t yet implemented the keyboards. I have found a keyboard manufacturer. This has been difficult. Their keyboard design has some limitations which I need to assess. They have sent me sample keyboards and they have made custom membranes to my specification. I have typed on the sample keyboard and my initial feeling is that they are not very comfortable - they require a high activation force and have an uncomfortably hard end to the key press travel. Soon I will work on the keyboard. The keyboard design from the supplier is suitable for the Combimouse because it can be integrated to give a lightweight solution. I will be using some of their components and will need to work out how to integrate it - there is some uncertainty here but I’m sure I will find a good solution.
2. I have the electronics and software mostly under control. I am now working on the right plastic housing. I will then add the keyboard. I have simulated key presses by shorting out tracks. I expect the the keyboard will interface well with the electronics but there is always a risk.
3. I am using the solution from Nordic. I have adapted their reference solution which is a separate mouse and keyboard. I have obviously had to adapt their solution and combine the keyboard and mouse solution for the Right Combi. For now I have got the basics working and it seems to be okay. Hopefully there will be no problems when I am typing at full speed with both the Left and Right Combis working simultaneously which is not what the reference solution is designed to do. At this stage I don’t know but my gut feeling is that it will be okay.
Besides the above there is still a lot of cleaning up and optimisations to be made. The Combimouse will support user firmware upgrades which reduces the software development risk considerably.
Fortunately I only had to have one revision of the PCBs manufactured by a PCB company. Having PCBs made is relatively expensive. I have had to make numerous hand modifications to these first revision PCBs, but for the most part they aren’t too messy and the current board is very usable. I am very happy that I have not had to have many revisions manufactured. Previously I used to etch PCBs myself but with the move to the new technologies this wasn’t possible because the Nordic radio/microprocessor IC is located on the board. This is a small pitched component and it wouldn’t have been possible to etch the board. In addition - previously I would drill holes and solder through wires for vias. These vias could obviously not be under components and this made the PCB design impossible to be etched myself as the latest version is very compact.
There were a number of times when I could no longer program the microprocessor with the latest draft software. This was a problem for both the dongle and the Right Combi.
For the Right Combi that meant discarding the board and repopulating a new one. I am not sure why I had these problems and this was very frustrating. I have discarded a number of boards after first attempting to remove the IC. I have changed a number of things in my development setup and my latest board appears to be reliable with respect to software changes - although I shouldn’t speak too soon.
This was a real problem with programming the dongle when I could no longer reprogram the microprocessor. Probably the cause of the problem was that I thought I could save some money and I did not buy their adapter. I made my own. Eventually after some time Nordic sent me a free adapter and they also sent me a development dongle which doesn’t require an adapter. They sent me these items because we couldn’t work out what was going wrong. I haven’t use their adapter yet but rather their development Dongle - so some unknown still here but I am thinking that it should be okay.
As I have a lmited budge I am making the plastic housings by hand. I am using ABS plastic sheeting and welding with Methylene Chloride. If I had funding I would probably use CNC machining. That is suitable for one offs. Stereolithography is better if more than one are required. 3D printing is not suitable as I discuss in this blog. On 18 August 2021 - my Right Combi house has distorted because I had not stiffened it up when I made it last year. So I am going to hand make another one - that should only take a few days.
I have moved to Melbourne. Previously I had identified a suitable Chinese manufacturer. The design company is in Perth. I have had over ten years experience with this company. However now that I have moved to Melbourne we may need to identify another company - but then again the Perth company is very good and we may stick with them.
The burden of finances has been an issue for me. We do have a number of shareholders but I am the main shareholder and I have to finance everything. This type of development would normally require a significant investment but I have found ways to do it on a limited budget - the downside of course is that it takes longer.
There was also so much to be done but now that I can see the end in sight I feel a lot more confident. I would say that from now on the issues for completing the prototype, are mostly mechanical which are a lot easier to identify and address. Whereas previously, when the microprocessor/radio and the proximity sensor did strange things it was very difficult as they are difficult to fault find.