Latest on 8.33 requirements (merged)

I am not familiar with the GNS boxes but I have a 1 x KX165A (8.33) which is "permanently 8.33" and I don't have any trouble setting 25k frequencies on it.

Is there something weird about the Garmins?

There is a wider issue here which is that of the intercom. Historically, I know from my 120hrs pre-TB20 time on the GA rental scene, few planes have a proper intercom. The installer wired up some bodge whereby the headsets are all in parallel, direct to the radio, so sometimes a particular headset brand interferes with the others, etc. Changing the radio is an opportunity to sort this out, but again at some cost because a decent intercom is about £1000.

How much would it cost to actually make a replacement 8.33 unit?

I am not in avionics, but keep half an eye on the business...

The ex factory cost of an aviation radio is something like £100.

The selling price might be £1200, for a very cheap one. Work back from that. Very few end users will be installing this, so most will be supplied and installed by a dealer. The dealer (installer) gets a discount of about 25%, so he gets it for about £900. If he did not make that margin, or the retail price was say £200, he would not handle the product at all. That is IMHO what supports the very high prices of certified avionics. This is why nobody is ever going to bomb the market price. Everybody could but nobody will.

You may recall the UK CAA talking about "low cost" Mode S transponders, for ~ £600. It never happened. The "low cost" ones are about 10-20% cheaper than the market leaders.

Certification is expensive but not that expensive for an established company which has the expertise in-house. GA avionics (of this sort) have no special software integrity or reliability requirements. There are environmental specs, and on RF gear there are performance specs. Certification is just a dead handy excuse for the pricing.

8.33 is however trivial to do. Every modern radio is done as a digital phase locked loop, where you have a voltage controlled oscillator whose output is divided by a selectable frequency divider (controlled by the frequency knob) and the output of that is compared with a reference frequency. Supporting 8.33 is just a case of selecting different division ratios. There are some tradeoffs in the design which is why e.g. it would be more difficult to make a radio with a 0.1kHz spacing, but 8.33 is probably just a trivial software change in most current designs, where you have a microcontroller reading the knob(s) and setting up the division ratio.

IOW, this is not a market anybody new will be entering

does anyone know if there are any ergonomic or other reasons not to make radios with a numeric keypad to enter frequencies

I think the number of people who have asked that question is exceeded only by the number of people asking if one can do the JAA CPL/IR without sitting the 14 exams

One can only guess why 0-9 buttons have not been provided on radios. Maybe they think there is not enough room.

Keypads have always existed on jets and have recently arrived on "GA" stuff so if you buy a new Cirrus or whatever you get a keypad which is used centrally, to load up whatever it is connected to.

There is one plausible argument though: knobs work better in turbulence. In whatever is defined as "moderate" or worse, the GTX330 0-9 buttons are very hard to use, and one ends up resting another finger against something adjacent to stabilise one's hand, so whatever "that" happens to be gets its paint worn off by peoples' fingernails... That is also a reason to not mount avionics flush. On this project I had the option to mount the EHSI totally flush (it comes with a kit to do that) and it does look really slick like that, but I chose not to because the little buttons on its periphery would be unusable in turbulence.