120 Degree Phase Shift Osc.

an

e 0 (PST)) it happenedBillSloman 0800 (PST)) it happenedBillSlom=3D

T),BillSloman

(PDT),BillSloman

00 (PDT),BillSloman

hering to

e I was

ness to be

ssertion until

to lie about it.

hat,

sort,

iling

me serving

ll

Only according to your idiosyncratic criteria, where not agreeing with your demented misinterpretation of what I wrote counts as lying.

from

rom the

t

ur

im is

aid

e got

icro

can

turered

ompetent that

the datasheet

o

find the

estion is

pdf

e

the web.

=A0 There's a distinction between mistakes and lies - though this is the kind of subtle distinction which does seem to escape you.

r n

n't even

my.

t

t came out

But not the part described by the data sheet you posted.

-- bill Sloman, Nijmegen

Look again. They're not ClearCom. ClearCom is only the distributor for the entertainment sector. We do our own sports marketing.

Huh? Just another variable (that won't) in the stew.

(PST),BillSloman

(PST),BillSloman

(PST),BillSloman

(PST),BillSloman

(PST),BillSloman

(PST),BillSloman

(PST),BillSloman

(PST)) it happenedBillSloman

(PST)) it happenedBillSlom=

:

wrote:

(PDT),BillSloman

(PDT),BillSloman

(PDT),BillSloman

bothering to

was

to be

assertion until

lie about it.

serving

More lies.

the

manufacturered

incompetent that

datasheet

the

question is

web.

More lies.

even

Wrong again, AlwaysWrong.

a

Oh do go on. I just showed you something fun--how just adding one resistor makes a much purer sinewave.

It's not like this is a real application. If so, I'd have just used a PIC like I said in my first post. This is all just for kicks.

-- Cheers, James Arthur

e a

s.

!

And I pointed out that you didn't even need to add the resistor to get the same effect. And that you hadn't added the right resistor.

Of course. We are just brain-storming around the OP's requirement. "Just using a PIC" isn't a particularly specfic suggestion - granting the range of PIC's available, it covers a multitude of very different solutions, each with its own strengths and weaknesses, and every last one of them uses more transistors than is strictly necessary, not that this is a serious objection.

-- Bill Sloman, Nijmegen

Ok -- what I meant was "I'm noticing those units look identical to the ones on the Clear-Com web site." :-) They've never zoomed in nearly close enough to get a good look at the labeling.

Trust me, there are plenty of people calling themselves "programmers" who think a wireless headset needs an RTOS. :-)

The last widget we did here with a TI DSP... the DSP guy intended to use DSP/BIOS but couldn't get it going in time. On the current project he did get it going -- he's mentioned that the documentation for it is rather spartan; it's easy to have it not work if you don't read very closely (as it turned out, in the first project there, he was using a timer that DSP/BIOS wanted for itself). Did your guys use it?

Someday I hope to run into someone who's used both Analog Device's VisualDSP++ and TI's DSP/BIOS so as to do a little comparison and contrast...

---Joel

Bluetooth stack certainly asks for a RTOS formalism.

There are some old farts that perfected for many years their art of solving problems using nothing but ropes and sealing wax. It is very frustrating for them to see a code monkey flashing LEDs using a RTOS.

Of course you realize that the problem is not because of DSP BIOS.

DSP BIOS is used quite often; mainly because TI DSP periferals are horribly complex.

Me tried VDK, me tried DSP BIOS. Despised both, developed RTOS of my own.

Vladimir Vassilevsky DSP and Mixed Signal Design Consultant

Interesting...

I've actually never used an RTOS myself... although I have written various little schedulers for cooperative threads and, by measuring execution times and creative use of interrupts, convinced myself that there'd never be a thread starved for attention.

But these things weren't as complex as a Bluetooth stack.

Not directly, yes.

I figure they probably know to work around some of the hardware bugs as well. :-)

Keith has mentioned that his guys found that sometimes the DMA engine just doesn't work quite right, and we've found that trying to used the McBSP in interleaved mode doesn't work the way the databook says (and probably doesn't work at all -- so far no one at TI has actually even claimed that it *does* work, only that it *should*!).

I'm getting the impression that ADI DSPs tend to be not as bad in this regard -- but other than playing with an eval kit for perhaps a day some years ago, I've never touched one, so it's possible I'm just suffering from "greener grass" syndrome. But you use mostly ADI DSPs, don't you? Have any opinion on how Blackfin compares to the C55x parts?

That's great!

Would you write me a fixed-point MP3 encoder for pizza and beer money? :-) (As a hobby project I keep trying to find one of those... LAME is excellent but of course assumes a high-power desktop PC-type CPU with an FPU, and I'm told that the one commonly referenced fixed-point encoder -- SHINE -- performs pretty poorly.)

---Joel

No. I am greedy.

IIRC both AD and TI offer their DSP-optimized mp3 codecs, as well as jpeg and mpeg. You can download binary libraries for free.

Vladimir Vassilevsky DSP and Mixed Signal Design Consultant

Darn. :-)

They do, but they're provided as object files meant to work within VisualDSP++ or DSP/BIOS. ...and I'd rather not pony up the ~$3k to go that route for a hobby project; I need that kind of dough to buy spectrum anazlyers!

Hey, how was the comp.dsp 2010 conference?

---Joel

If you can recommend my services... let's see what else we can do.

Yes, it was:

It is such pleasure to talk with the brilliant people who understand what you are doing. Perhaps we need to do more of the informal gatherings, may be, every two years or so.

Vladimir Vassilevsky DSP and Mixed Signal Design Consultant

I'm not sure I follow this, but I like the sound of it.

Cool, nice photos! ...Eric Jacobson looks a little crazed during his presentation... ;-)

---Joel

Like I said, look for the black and blue trunk on the sidelines. You won't need to zoom in so far. ;-)

Ok, I thought that's what you meant. I thought I might have missed a good reason to use an RTOS. ;-)

I don't think so (I've told them I don't know how to spell 'C'). AFAIK, the only TI code used is the first stage bootloader. The chip was some PITA to use. The I/O and DMA are totally braindead (broken by design). If I were to do the hardware today I'd bypass the DMA almost entirely.

AIUI, ADI's Blackfin was the loser in the DSP-off. TI has left a really bad taste in the owner's mouth and it wouldn't take much convincing to go that way were we to do it again.

One of the big faults in the DMA engine is that there is no framing between the peripherals and DMA. In particular, there is no way to guarantee that data from the I2S peripherals ends up aligned in memory. If something gets dropped (another bug in the DMA engine) the DMA data *won't* be aligned and won't fix itself. The work-around was a feedback loop through an unused Codec to detect alignment and force a DMA and Codec reset when unexpected data is found. Messy!

The McBSP is just impossibly broken. You found one bug. There are also "magic" bits needed in undocumented registers to get the thing to work. No one can explain why they work, but they must be set.

That's where our firmware folks are too. The product works, so it's not likely to be redone anytime soon.

I find it amazing that if a customer has gotten to the point where they can clearly demonstrate the part doesn't work per the dats sheet specs and you're now giving them undocumented register bits to write to, that customer hasn't been put in contact with the guy who actually designed the peripheral (i.e., wrote the VHDL or whatever) in questoin who can just look at the raw design and explain what's really going on.

Well, at least if anyone can. Maybe TI has a lot of turnover and "that guy" no longer works there...

---Joel

Yes, but you added an H-bridge. That's a) a lot more complicated than a resistor and b) doesn't make a clean sinewave magnetic field. The resistor is dirt simple, and it works.

Well that's true and I thank you for choosing a better one. I of course knew it wasn't the right resistor--you could see that from the FFT--but it was 2a.m. on the road and I had other work to do, before rising for an early morning that same day. IOW I was busy, but when SED calls one must harken. (SED is a guilty pleasure.)

So, you pitched in and worked the details out--thanks.

a

Right. The OP, long gone in the tradition of OPs, is just the pretext ;-)

-- Cheers, James Arthur

Not only that, but the FAE could only find some settings by experimentation; "well, there are only eight possibilities, try 'em all".

They promise that "things are better now". They probably are, at least somewhat. The McBSP no longer exists. ;-)

The H-bridge may be more complicated than the resistor, but the resistor implies a bunch of op amps to filter the output of the resistor and to buffer the coil that is actually being driven. You chose the complication that you felt happiest with - or rather the one that John Fields felt happiest with, since your circuit is a modification of his. Op amps and active filters have their own complications.

Some of them. I didn't check out your circuit with enough capacitance to make it look like an integrator - you can't buy better than 1% capacitors and with an integrator the resistor and capacitor values set the output amplitude. You can buy 0.1% - and better - resistors, with temperature stabilities that make it worth paying for the tight tolerance.

d a

Too true. He did show up once, to point out that his relay coils weren't driving relays, so presumably he did do some lurking. My guess would be that he ran a few of the ideas past the boss, and the boss told him why one idea was much better than all the others ...

-- Bill Sloman, Nijmegen

[snip]

"Flaky" just died :-(

So what's best/reliable? ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |
             
I love to cook with wine.     Sometimes I even put it in the food.