Okay, I've finally tested with HiJack and am getting the same results as Toby.

I've modified the AF debugger visual slightly so that in the WAITING state it also displays the number of frequencies in the AF list, and the signal strength (you will remember that this state is achieved by the range of signal strength readings equalling zero, so only need to display the one value).

The AF list is populating itself fine, I got a population of 10 alternate frequencies after just a few seconds of radio 1 (98.5) - However, all the signal strength readings seem to be returning zero.

This is intruiging since the standard full screen radio info mode seems to display a sensible value for the signal strength of the current station. A rather roundabout route is used by the code which retunes the radio for a fraction of a second during gaps in the audio and measures the signal strength, so it could be something there.

I'm currently investigating further and trying to determine what differs if HiJack is present. No disrespect to ML intended, but I think that due to lack of RDS in his country he will find this one very difficult to debug.

Toby

Toby - did you use mlord's test Hijack (v329+s) for this?

I'm wondering if it could be timing-related? I know the tuner has to sample within a very small time window, so is there something Hijack could be doing that interferes with the sampling? You (Toby) might be able to test that theory with a longer sampling time (though it would probably be unpleasant to hear, it may tell us something).

Otherwise, all I can think of is memory usage; I'm not sure that would make a difference, but I suppose it's possible...

Mmm... thanks for trying this, Toby.

The biggest difference is that Hijack is based on a newer kernel than the stock empeg.. I wonder if a driver didn't "port" perfectly?

Cheers

Are the dev-info and/or FAQ gnomes gathering this?

I dunno. I thought a screen named "debugger" would be a temporary thing, and I don't like to FAQ-document temporary stuff if I can avoid it...

The biggest difference is that Hijack is based on a newer kernel than the stock empeg.. I wonder if a driver didn't "port" perfectly?

I wouldn't be shocked or amazed if something didn't port properly. There was a lot of stuff that I had to patch in by hand when doing the upgrade to 2.2.17. At least it is easy to test, if it works with HJ 299 and breaks in 300 it pretty much has to be the kernel upgrade. So can anybody confirm or deny the problem occurs with hijack < 300?

-Mike

Okay, I'm testing v221 tonight (the oldest I have), since others have tried v329+s.

Just tried HJ v285 and it's the same as v329+s WRT AF.

Okay, so the trigger has been there a while..

Here's a special Hijack kernel, with all of the tuner serial intercept code removed, just in case something's getting chewed up by that stuff. Or maybe delayed too long or something.

http://rtr.ca/v329+t.zImage

Again, this is for mk2/mk2a only.

Cheers

v329+t works fine for me in AF mode!

Good. You're the only one to grab the binary from my server thus far, so I'll wait a day for more folks to see and grab it.

If all goes equally well for others, I'll do a little hack job on my tuner hacks.

Cheers

Ugh.. I just had a closer look at the code.. bad, very bad. Definitely guilty. I'll see if I can fix it. Gimmeasec..

Thanks, that sounds really promising (especially given that you don't have RDS available to test this on).

Mmm.. the BBS was inaccessible here for a while, but back now.

I've uploaded Hijack v330 to the Hijack site. Give it a try and let me know if it fixes the RDS issues.

If not, then turn on [hijack] trace_tuner=1 and capture a startup trace for me.

Thanks

Before AF wouldn't work at all but with v330 it works fine now.

- Trevor

Woohoo, AF worked this morning with v330 installed!

That has fixed the problem.
At least the AF debugger info mode now does what was intended.

Would be interested to hear any feedback from stalk users - I am finding stalk response time extremely variable under 2.0, don't think this is a hijack related issue. Sometimes the stalk response time is perfectly acceptable, at other times it can take up to five seconds to respond to a stalk command, at other times it doesn't respond at all.

Toby

I've uploaded Hijack v330 to the Hijack site

Sorry, it's a bit difficult to try and follow the same subject on two threads, I'll try this tonight.

v300 works for me.

I'm a little concerned by the loss of fake_tuner (remember, I needed this to avoid the tuner "disappearing" as I start the car following about 5 seconds of engine heater) - do we know if it's absolutely incompatible, or are you trying to narrow it down to choose between two causes?

What's the tuner disappearing thing?

Does this relate to power lost due to the car starting?

If so I have a great solution for you - At least it works for me - I know this is not strictly related to this thread so sorry:

I got three UPS batteries (12v at 7Ah) and wired them in parallel, then took a relay and connected one end of the signal to my vehicle body, and the other end to one side of my battery light (This is actually the signal path back from the alternator which indicates it has started producing charge). The actual terminals of the relay just join the three UPS batteries to my vehicle main battery in parallel.

The result of this setup is that whilst my engine is not running the empeg is completely isolated from the main battery. My battery light does not go out until the engine has finished starting and I apply a little bit of revs. So I never get the audio cut out, or the annoying battery symbol whilst my starter motor is pulling all the available current - I imagine it would prevent odd behaviour from tuners due to power fluctuation too although I have never experienced this problem.

Of course, the other great thing about this setup is that I can leave the empeg in standby mode in the dash without having to worry about it draining the main battery if I leave it for extended periods - Just need to find some deep cycle UPS size batteries and the solution would be perfect.

I assume you meant v330 rather than v300 in your post.

I think it oughta be safe enough to put the fake_tuner stuff back in place, once all of the RDS/AF issues have been solved. For now, want the tuner path to be as simple as possible in Hijack, so that we don't waste any more of Toby's precious (and appreciated!) efforts. So for the interim, you'll have to test the RDS/AF stuff by pulling out and reinserting your player after having started the engine.

And could someone please volunteer to cable up a laptop and capture a trace with RDS/AF doing something?

[hijack]
trace_tuner=1


That'll let me know if what I fixed really fixes everything.

Thanks

In reply to:

---

I assume you meant v330 rather than v300 in your post.

---

Yes, sorry for the slip.

In reply to:

---

I think it oughta be safe enough to put the fake_tuner stuff back in place, once all of the RDS/AF issues have been solved.

---

Ah, that's fine by me. I can live with waiting for the player to start before moving the key from "accessory" to "preheat" position for a short while, but not for the rest of my life!

In reply to:

---

And could someone please volunteer to cable up a laptop and capture a trace with RDS/AF doing something?

---

I'd volunteer, but I don't have a portable terminal, or a long enough serial cable. Sorry!

And could someone please volunteer to cable up a laptop and capture a trace with RDS/AF doing something?

Is this what you are looking for? Note that I've used a crappy, unpowered antenna while logging this data. I've had my empeg tuned to a station which transmits songtitles, and gets ~20 AF in the AF debugger.

Ive had the same problems with the stalk, as far back as i can remember with v2. Still problems in the final aswell, although it seems slightly better for me.

Ive also noticed that sometimes Vol+ doesnt respond at all in tuner mode. a few mins later and its fine again. Strange.

I once tested b13 without hijack installed and had the same problems so dont think its hijack.

And could someone please volunteer to cable up a laptop and capture a trace with RDS/AF doing something?

Attached is a bootlog where source=radio (the radio channel did not have any RDS data).

This tuner log was recorded starting from a non-RDS station and changing preset to an RDS station with lots of AFs.

>Is this what you are looking for? [tuner trace]

Yes, exactly, thanks!

I wanted to see the exact sequence for the f0 and f1 RDS switching commands, to make sure Hijack wasn't eating any of them. Looks fine, and safe.

Good!

Everytime this thread pops up, I keep hearing in my head an old Jewish man with a head cold saying ``Hooray!''

Too much Futurama, I guess. Anyone else?

U got it! Hooray 4 all those that made this a reality. We thank U.

"Hooray! Hooray for Zoidberg!"