Is DSP useful?

[JC1]

...

Hi Wizard,

Robert Hoolok was always good to talk to.

He was a real engineer.

Realize that a Digital driven PI detector

or VLF

Is NOT a DSP detector.

And certainly not an improvement over

the "state of the art" that exsist now.

Alot of that is just doing the same thing

in digital, using timers or whatever.

We need to see Big improvement over

"state of the art" as it is now.

So have to do more, more processing,

more number crunching, and do all the

other stuff you have to do such as make

a beeeeep.

[TheWizard]

Hi JC1

Who is "We need to see"... Dream Team? Still ready?

I don`t know Robert Hoolok....

Analog Devices DSP - http://avmaster.bnx.homelinux.net/da...s/adsp21xx.pdf

Peacefully, I want not to work for your team...

[JC1]

...

Hi JC1

Who is "We need to see"... Dream Team? Still ready?

I don`t know Robert Hoolok....

Analog Devices DSP - http://avmaster.bnx.homelinux.net/da...s/adsp21xx.pdf

Peacefully, I want not to work for your team...

That is funny Wizard, I do not have a team.

Well actually I do, but we don't make, nor ever will

make hobby detectors, much more money in

other things.

Too bad you don't want to join, this team makes

alot of cash. I just play with detectors for myself.

The ones I build, and I design everything, and they

are not like the ones you see in the commerical world

things like AD630s for demodulators and the such.

Way too expensive for the toy market.

[JC1]

Two processors

Hi Wizard,

It should be obvious I don't want

to be on any "team". I always end

up doing all the "heavy lifting".

Story of my life. Though it sure has

paid off in later years.

Anyway, here is the Garrett patent

and shows on page one how one would

set this up. A/D converter, DSP processor ,

and then microprocessor for the lights and

bells and beeps.

http://geotech.thunting.com/pages/me.../US5721489.pdf

[Carl-NC]

Are you happy with the features on the PicMicro board, and do you have any comments concerning the MikroElektronica offering?

I like the PicMicro a lot. But the MikroElektronica has more add-ons, I think it's cheaper, and it has the dot-matrix display I'd like to mess with.

- Carl

[Carl-NC]

So what type of Digital Signal Processing are you
doing to the signal with this setup?

Just some lowly averaging, pop reduction, and peak detection. Not "true DSP", just as much digital as I need to do for this particular design.

Now I have NO doubt NO real DSP will be done
by the Collective design team.

That could very well be. I am not insisting that any design we might decide to do involve DSP. Sean's comment was, "One thing is for sure, we will need someone with more than a little understanding of DSP." OK, I happen to agree, that in order to evaluate the merits of DSP, we need someone with a strong background in it. DSP has obviously become a big part of other markets, so it must have some advantages. One of the biggest is in signal processing. Metal detectors involve signal processing. Ergo, can DSP benefit metal detector designs? I refuse to state emphatically it cannot.

And I'm glad Carl's new low power detector
only uses 300 uA of current. Which seems
like the Xmit power is a bit on the low side for
this to be a very exciting detector.

No, just the uC. The overall detector is 2mA. And it's not a hobby detector. The point I was trying to make is that I only used exactly as much digital "stuff" as I needed, and no more. If I came up with a good DSP-based design, and could run the DSP at 40MHz and 100mA instead of 400MHz and 1A, I would do so.

BTW, the latest Teknetics T2 and Fisher F75 are almost all-digital detectors. I opened my T2 tonight, and Dave used a TI MPS430 microcontroller (yes, I know, not true DSP). In front of that is an LTC amplifier, and a 4053 demux. Nothing else I can see. So apparently, past the preamp and demod, all further signal processing is digital. And these detectors are getting rave reviews. So far, my T2 comes closest to detecting my 2-foot-deep 3-pound silver cache.

So, again, I refuse to dismiss DSP until I look, and see for myself.

- Carl

[Muntari]

Hi Carl,

The MikroElectronica gear is well supported and reliable. The C compiler is not for me but there are a number of demo downloads with 2k code limits.
For those on the forum who wish experiment or to get started with Microchip or Atmel chips, these boards are pretty good value.

Back to DSP versus analog etc, I think it's horses for courses really. Most of us know there are always trade offs when it comes to producing the end product.

I happen to think the Microchip range of DSPIC30f and DSPIC33 are well suited to PI metal detectors and while not at the high end of DSP devices they do have some nice peripheral options, low power and are reasonably priced.

At the end of the day each design brief will dictate what is best to use.
I started out using TI DSPs in my designs and finally moved across to DSPic devices as a trade off between lower power & speed.

As to DSP or analog, like you say Carl, keep all options open. There is a long way to go.

Cheers

Brian K

Muntari

[Boattow]

Mikroelektronika

I am using their EasyPIC3 and EasyPIC4 development boards. There is nothing better for the price out there that I've found. The only complaint that I have heard on them is the lack of compatibility with MPLAB. I've used MPLAB. All you have to do is compile your code and then use the program utility supplied with the EasyPIC boards to program it. I don't find it to be a problem. For the first time I have strayed from assembly and bought the C and Basic compilers. They have great support for them on their forums and have an unbelievable library of routines which really speed up development. The only complaint I have is their hi level compilers are very inefficient compared to others on the market and so end up using a lot of RAM in your PIC. I don't believe you will be disappointed with their development boards.
Good luck,
Boattow

[JC1]

So Civil

Wow !!!

You guys sure are civil.

Not sure I deserve that, but hey really good posts there.

Making a lot more sense now.

Yes Carl, I knew you meant only the electronics

and not the xmit, but I was hoping you would say

100 ma xmit and then I could ask why the low power

on the rest. 2 ma you say, I prefer about 2 Amps on PI

but always have notice slight improvements with higher

power. I use 12 volt gel cells I charge up with the

cigar lighter plug in the field and don't mind a bit.

Well Sorry I stirred up the soup here a bit, but

after being in the EE business for real for way too long,

I have become sensitive to buzz words like DSP and

the rest. So now whenever I hear any of them, I am like

OK let us see if you are real. Because you see I do

have experience in these areas and have done similiar

things, in "teams" of technicians, programmers, and

engineers. So I didn't do it all, and not going to pretend

I did, but will say I have seen with my own little eyeballs

some of the real struggles that are gone through by

very qualified people playing these games.

Sorry for the rant, but in a way it does at least maybe,

get folks to be able to both know what the other is talking

about, i.e. real signal processing- methods of improving

the performance above the market now, or digital

implementations of analog detectors. Which if done

right might be lighter, cheaper, low power, low temp drift.

Things like that.

But Carl you are right, like I said, I was smelling must

have DSP, so I was OK just exactly what do you mean

by this? If must have prove it. Because I'm not so sure

digital just automatically wins everytime on every issue.

Now I'm sure by asking in broad terms what big deal

DSP thing is someone doing, that they think I am going

to "steal" their genius ideas. Believe me, I could probably

care less.

Keeping it real.

Well kinda.

[Qiaozhi]

I like the PicMicro a lot. But the MikroElektronica has more add-ons, I think it's cheaper, and it has the dot-matrix display I'd like to mess with.

The MikroElectronica gear is well supported and reliable. The C compiler is not for me but there are a number of demo downloads with 2k code limits.
For those on the forum who wish experiment or to get started with Microchip or Atmel chips, these boards are pretty good value.

I am using their EasyPIC3 and EasyPIC4 development boards. There is nothing better for the price out there that I've found. The only complaint that I have heard on them is the lack of compatibility with MPLAB. I've used MPLAB. All you have to do is compile your code and then use the program utility supplied with the EasyPIC boards to program it. I don't find it to be a problem. For the first time I have strayed from assembly and bought the C and Basic compilers. They have great support for them on their forums and have an unbelievable library of routines which really speed up development. The only complaint I have is their hi level compilers are very inefficient compared to others on the market and so end up using a lot of RAM in your PIC. I don't believe you will be disappointed with their development boards.

Thanks for the feedback. It all looks positive, so I think I'll spend my hard-earned money on one of their EasyStart kits.

[Carl-NC]

Yes Carl, I knew you meant only the electronics
and not the xmit, but I was hoping you would say
100 ma xmit and then I could ask why the low power
on the rest. 2 ma you say, I prefer about 2 Amps on PI
but always have notice slight improvements with higher
power.

2mA time-averaged. The TX pulse is 50mA. Again, it is enough for what I want to do.

But Carl you are right, like I said, I was smelling must
have DSP

Yes, I could tell that's what you were thinking.

The whole reason I opened up the "Design Group" concept for discussion, is that a few people have what might be novel ideas, but can't develop those ideas on their own, and don't want to just give them away. Both Sean and myself have ideas that probably cannot be done without heavy digital processing. Now, it might turn out that those ideas are lousy, or it might turn out they are not selected by a design team. But I think they would be interesting to try. In any case, I would strongly encourage anything that gets us beyond the current state of detector technology, which hasn't changed much in 20 years.

- Carl

[Unregistered]

‘Do you dig everything’

The thing with DSP is, what are you going to do with it. Clean the signal would help, pulling a weak signal out from the background is possible but discrimination? That’s going to be a lot harder. I did a little work a few years back on OCR (optical character recognition). On a fresh laser printed text it’s easy; a dirty photo copy it’s a lot harder. While on the bench metals show distinct characteristics put them under wet clay or add a bit of oxidisation and even soil depth, and it starts to fall apart. Which brings you back to the same old argument ‘Do you dig everything’.

[JC1]

Found another Smartyarsy

DTex DSP ultra sensitive metal detectors.

You guys know anything about this detector?

Is it real? Looks like they spent some money

on a Big Gun.

Almost as Big as Me.

I googled dsp metal detector algorithms

and got this guy.

http://209.85.165.104/search?q=cache...nk&cd=31&gl=us

it says: DTex advanced ultra sensitive metal detector

near bottom of page one.

here is resume.

http://newcustomer.lawsonline.net/Dowling/index.html

Did D-Tex ever make anything?

[JC1]

from above link

Consultant - DTex Inc. - Design of algorithms, DSP processing board, and DSP56L811 DSP software to implement an advanced ultra-sensitive metal detector, February 1997 to present.

I can copy this one.

[Carl-NC]

D-Tex used to be Bill Mahan's company in the 60's. It was resurrected in the 80's, maybe by his son, I don't know. But it disappeared after only a couple of years. Maybe he blew all his funds on a DSP designer!