Basic Note about Encoders before I ramble on: Encoder’s can have a step rate of a few pulses per 360 deg of rotation to thousands of pulses. The Mechanical encoder I use for volume control has 16 and the optical encoder for the FEDrive has 128. Depending on the application, some mechanical encoders can be a dog to control without correct denounce software “and” hardware.
Anyway, after a few days of testing my new FEDrive system, the need for rotational speed sensing “auto variable encoder step-rate” became obvious.
Two problems arose; for example scrolling song lists can be too fast to read at times. Easy solution is to turn the encoder slowly, however, if a Menu is being scrolled then it takes a bit of concentration to scroll slowly, and that is undesirable when driving.
To alleviate the problems above, I had an automatic two speed scroll rate implemented for Menus and lists; this was fine until it came time to “press” the encoder knob to select a song from a song list. Song lists are scrolled at a faster speed and in doing so it became very hard to just “casually” press the Encoder knob without moving it to another track, it was also difficult for a passenger to do this with any car movement. Worked ok on the bench though but then it wasn’t moving!
Anyway, after a lot of trial and error I managed to implement continuously variable encoder rate relative to rotation speed. I included two adjustment sliders in the Encoder setup menu to tailor this to almost any encoder pulse rate.
Now, when you slow the encoder, or stop turning the encoder in order to select the desired song, any slight movement of the encoder is rejected because the required number of Encoder pulses to trigger a valid step rapidly increases with slowing speed.
It really works well, scroll speed moves up or down smoothly with varying encoder rotation speeds and pushing the encoder as carelessly as you like causes no jump from the highlighted track.
The next task is to redo the Hardware USB controlled Volume and Mute unit that sits at the input to the Amplifier. I need to get positional information back from this control in order to implement automatic speed and noise variable sound level for the FE.
The overall impression of the Car-PC is that it now feels like it was meant to be part of the vehicle. There was a time when I begin to question whether it was worth the effort, expense, and monumental amount of time I was/am putting in to this and there were so many silly things that I never thought of that kept cropping up, but dam its working beautifully now.
My enthusiasm for doing the hard part which I hate has returned. That is: Running new cables and adding a few extra, relocating and moving the PC into a new case, and finally, making it look nice.
Oh and making a few videos.
M4-ATX and an attempt to find/mod a supply that generates less or no spurious noise (interference).
Since I’m rebuilding the case and PC layout I decided to take another look at the M4 and locate an audio noise that is “absolutely” coming from the M4, however due to other issues I have with the firmware of this device I decided to look at an alternative first.
During the course of my initial tests, when trying to make the M4-ATX usable near a GPS and DAB radio in weak signal locations, I tried a small low cost supply from Jaycar.
This is a 205 watt unit, it’s nicely made and it did indeed produce very little RF interference.
However it had at that time two limitations:
1. The 12v input is not regulated or boosted – and is a major reason for its low noise.
2. It has no start-up or shutdown controller.
The second problem is not really a problem as the M4-ATX PC interface and shutdown controller sucks!
It will take me all of a day to make a microprocessor interface with voltage monitoring and controller settings available in a small windows application that blows the M4 dumbness out of the water.
The first problem allowed me to find a way to regulate the +12v supply using some form of linear regulation – NO switching regulator to generate noise. I will not be boosting the 12v rail as I run the PC from a small aux battery AND boosting the rail requires some form of switching convertor.
Yes I know +5v and +3.3v rails use a switching regulator, however the voltage is low and the power is low in relative terms, it’s that switching +12v @10A (or more) regulator/booster that is the real noise maker.
There are problems using a standard linear regulator due to the voltage drop across the device when the input voltage approaches or drops below the design regulated output voltage. And NO, I’m not interested in some exotic circuit that you may have to overcome that.
I had an idea for a simple low cost (some would call it crude) system to keep the voltage at around +12v +- 5% from an input of 11.5v to 15v. However I wondered just what tolerance my MB would take before it spat the dummy (Signalled power good failure or just stopped).
The CPU and most circuitry in a modern MB are 5v or below. I assume that the input to the switching regulators for most of these devices is usually the +12v supply, therefore I could see no reason why it could not run at much lower supply. Searching the NET was a waste of time with usual folk law and totally dumb/wrong information trotted out as fact.
The test on my MB was surprising, I don’t know how low I can take the +12V rail but I decided to stop at +8.4 volts. The PC boots and runs faultlessly at that voltage.
NOTE: I should point out that there could be a “Major problem” in taking the 12v rail low. The CPU and other circuitry can be drawing high currents and the switching regulators for these devices will be drawing more and more current as the input voltage drops. It’s possible to go outside the design limitations of the MB and damage it.
I reasoned that if I kept the +12v to a max of +12.4v and I feel I could safely run the MB down to 11v. Actually I think it would be fine down to 10.5 volts.
This relaxed the very fine regulation requirements and allowed me to try an idea for a controller consisting of 3 x 30A diodes in series, each shunted by a power MOSFET. The FET’s are controlled by a small voltage monitor and switched progressively on or off as the input voltage changes.
Each diode has a forward drop of almost 800mv for a total drop of 2.4 volts in three 800mv steps.
With this circuit the output voltage swings between 11.7 and 12.4 as the input voltage varies from 11.8 to 15v at currents from 2A to 20A.
At an input voltage of 10.9v the output is 10.7v.
This circuit causes absolutely no noise, the total dissipation at 15v input and 10A is around 24 watts, at the normal running current of 3.3A its 8 watts. This is with an input of 15V which it will never see due to wiring length and the Vehicle regulator.
At a nominal 13.6V input at 10A the dissipation is around 16 watts. At 12v @ 3.3A output for 13.6v input (PC runs at that all day) power dissipation in the circuit is around 5 Watts.
This is a very simple low cost circuit and should be almost indestructible. Combined with the Jaycar PSU, it results in a complete absence of interference in the Audio, FM, DAB and GPS spectrums.
It has been running the Car PC all day at 14v input and is cold to touch.
I have been away for a while with health problems (ongoing) but I’m just getting back to the install now.
Before stopping work on the install I was looking at different power supplies in a bid to reduce the last of the noise from the system and to trace a strange problem where the PC would randomly fail to resume from STR, in doing so it would go into a “loop” of continually trying to boot and instantly failing (but only when trying to resume from STR – Hibernate is fine). Power had to be removed from the PSU/MB to resolve the problem.
The loop problem turned out to be a Mother Board design problem as other MB’s don’t have this behaviour and the behaviour follows the MB.
The random failure to resume from STR turns out to be a complex interaction between the MB and the PSU, in some installations it may never show up, but if you get the right conditions of input voltage change, noise when cranking, direction of wind and moon cycle, it can happen. Other PSU’s caused this to happen quite frequently, the M4 was better than other supplies I tried.
It really is a chicken and Egg syndrome as the failure is caused by either the PSU thinking there is a problem and dropping +5v Standby OR the MB is (for some reason) signalling a power good fault and resetting the PSU and therefore dropping +5v Standby. Any interruption to +5v standby voltage will cause a failure to resume from STR.
With a MB that does not go into a start-up loop there is no problem, just a delayed cold boot.
During the course of these tests I came to the conclusion that the M4-ATX (when placed in an RF shielded enclosure with feed through caps on all leads) works really well and is causing no problems.
One problem with a trunk install can be heat. The existing system draws only 36 Watts, however after 3 hours of driving on a hot day the air inside the boot is gradually heated by the system and although the temps are within design limits there would be a real problem with a trunk full of luggage on a hot day.
To try and overcome these problems I decided to redesign the case and rebuild the PC for the third time, and as it turned out, the final time. I wanted full control of layout, separation of modules and full control over a common ground plane without compromising for lack of space.
The MB was updated with a new model from Gigabyte. This new MB with the same processor and ram draws just 24 watts when running (the older model was 36 watts) Peak start-up current is halved from 8A to 4A and the system heat-sink is now a much larger unit.
This complete system including 8” LCD and all devices powered draws a total of 35 watts from the supply.
These pictures show the case on the work bench with the Lid open and closed.
The case is made from thin sheet aluminium (low cost) and aluminium angle bracket. Both items were available at the local hardware. The thin aluminium sheet is glued to a ply top and bottom panel. The angle bracket is cut and screwed to the ply top and bottom panels.
The lid is secured by three hinges and is electrically bonded with a thick flexible copper strap.
When the lid closes, the angle brackets “snugly” overlap with the top lid brackets resting on 3 large limit screws. This makes a very strong light case that, when mounted in the trunk, supports any weight of luggage that is placed/thrown on top.
Case with back seat down.
The forward facing vent has a filter and fan which cannot be blocked by anything thrown into the trunk.
Case open with back seat down.
Location of vent with arm rest.
From the Trunk.
Picture of the earth strap added to the DAB-FM radio module that cured the last of the noise that I had been trying to eliminate. The design of this DAB-FM module does not correctly ground the RF shield, adding the thick green earth cable directly to the case removed all noise and improved the reception so much that this receiver now outperforms a high quality dedicated car receiver. This connection is an RF ground not a DC ground; it must be “short” and thick. Do not confuse the effect of this earthing with audio or DC ground loop earthing principles.
I added an Aerpro amplified roof mounted AM-FM DAB antenna, this unit has 3 lead connections, FM-AM coax, DAB coax and a separate power lead. The aerial input to the receiver has to be switched when changing from DAB to FM and this is done automatically by my DAB-FM radio application through the FE software. I used this high quality RF relay as I’m in a fringe area for digital reception and wanted as low a loss connection as possible.
This aerial is fantastic, all DAB stations received, even under a 6 x 6 meter steel carport behind a steel panel lift door. Mounting the aerial through the rear car roof and getting the leads back to the receiver turned out to be one of the easiest mods I have made to this vehicle.
Picture of the Aerial changeover relay; it’s taken from an old mobile radio I junked.
It’s the yellow block thing mounted to a bracket.
The Fe-Drive and control panel with icons and labels. It’s hard to get a picture because of reflections from the finish when using a flash.
Now all I have left to finish is the centre dash panel. This is a picture of the prototype centre dash backing panel in black, once again the screen is washed out because of the angle and camera flash.
I have gone from a small system behind the dash (the first build) to a larger case mounted in the boot and now to this very large case mounted in the boot in a recess in the floor behind the back seat. This is a fabulous location in this CAR. The new case draws air through the rear seat centre arm rest and exhausts air into the boot spare tire well. It runs as cold as, even on the hottest of days.
The larger case is a joy to work on and make changes to hardware. Simply lower the back seat and lift the lid, no screws to undo, no reaching into inaccessible places. Most connections are too sockets mounted along the front of the case and it literally takes under 2 minutes to remove the PC from the car and have it running on the bench.
The boot can be filled with luggage chucked in anywhere without any concern about blocking vents or throwing anything on top of the PC.
Almost forgot to include a picture of the Alpine amplifier mounted where the old PC case was located.
Simply Awesome! 007 car in disguise.
Thanks, it’s been a long road to this point. I’ll never try and squeeze commercial entertainment electronics (MB, PSU etc) into a small case again. This layout in a bigger case is so robust and accessible, makes fault finding, mods, upgrading software and components a breeze.
BTW, I had a look at your tablet install, really nice work, so many ways to find a solution that meets our needs.
Cool Thx! How do you like your Transflective monitor?