Scope recommendations

Forget the analog bandwith, much more important is the sample rate. The Bitscope's 25 MS/s is really nothing to write home about, it's barely enough to take a look at 10 MHz digital signals.

You won't have to display the CPU clock on the scope, you probably won't even have access to it because it is generated on-chip with a PLL from a lower external frequency. The interesting things to display are serial communications, e.g. between DSP and DAC, and for those a scope with about 0.5-1 GS/s is adequate (e.g. Tektronix 1002, Agilent 3000 series). Unfortunately that will cost you a bit more than $500.

--
http://en.mikrocontroller.net

Ouch. Thanks, I see I wasnt even looking in the right department.

Are there scopes than can sync to a clock signal, +/- some offset? The idea being that you could forgoe the the nyquist oversampling, using the scope basically as an analyser.

I could concoct some way to justify spending a grand, a grand and a half, perhaps maybe even two grand, but only if it made some kind of long term sense. With Nyquist in effect, even 1 GS/s seems insufficient for todays standards: DDR2 or a 600 mhz blackfin and you're already outmoded and out of luck.

The other big question is tech change, as in the scopes themselves. A/D silicon over the past 5 years has gotten a lot cheaper and a lot better. Have scopes advanced similarly, or are we likely to have rapid advancement soon? Where on the curve are we with scopes, and what kind of advances can we expect in the scopes? There seem to be an army of 1 GS/s scopes on ebay, how long in the future until its an army of 2 GS/s scopes?

For a little bit of background, I graduated computer engineer last year. I'm just slinging code now, but I want to start building boards for myself as a hobby activity, mostly using heavily integrated SoC stuff; currently looking at ARM, Blackfin or ColdFire, plus fpgas. My analog knowledge and experience is pretty weak, there's going to be a lot of bumbling as I'm laying out boards. Costly as crap way to learn; I'll try to start with vaguely-moderate goals. With a little luck I'll build some cool toys along the way.

I'm presuming a badass oscilliscope will probably be required? The Tektronics 1002 looks great, some sell reasonably cheap on ebay,

1GSa/s, but only 60 mhz bandwidth (basically useless if I ever want to do GnuRadio stuff). I'm not sure what the 60MHz bandwidth implies, I'm assuming it would still be useful for debugging problems with board layout & signal behavior.

Thanks, rektide

I don't think so, at least not in the price range we're looking at.

If you want to build DSP or FPGA boards the scope will probably be your smallest problem. Are you prepared to design multilayer PCBs, handle BGAs, that kind of stuff? Unless you want to specialize on board design I don't think it makes sense to do this for a hobby. It is much easier to buy adapter boards with FPGA+RAM or DSP+RAM and use these with your own hardware.

Not necessarily. You won't be able to debug the HF daughterboards (a scope isn't the right tool for this anyway), but it could be useful for baseband/IF signals.

Signal components over 60 MHz are attenuated by 3 dB, so basically everything above 30 MHz looks like a sine.

Yes, but not in the xxx MHz range.

--
http://en.mikrocontroller.net

I don't think so, at least not in the price range we're looking at.

If you want to build DSP or FPGA boards the scope will probably be your smallest problem. Are you prepared to design multilayer PCBs, handle BGAs, that kind of stuff? Unless you want to specialize on board design I don't think it makes sense to do this for a hobby. It is much easier to buy adapter boards with FPGA+RAM or DSP+RAM and use these with your own hardware.

Not necessarily. You won't be able to debug the HF daughterboards (a scope isn't the right tool for this anyway), but it could be useful for baseband/IF signals.

Signals with 60 MHz are attenuated by 3 dB, so basically everything above 30 MHz looks like a sine.

Yes, but not in the xxx MHz range.

--
http://en.mikrocontroller.net

THINK about what signals you are actually going to need to look at and what you will be looking for. Note also that the Tek scopes I've used have tiny memories, they're *quite* irksome to use.

$300 is not enough of a budget to acquire anything "badass" or to ensure that you have all the bases covered. I work on designs clocked at up to 667MHz using primarily my old 1Gs/s 50MHz DSO. I never need to look at that 667MHz clock.and couldn't, even if I wanted to, because it's buried inside the uC. Mostly I'm looking at signals that are

You might just want to purchase a decent logic analyzer if you don't need the scope for any analog signals. There are a number of decent logic analyzer for under $600 with sample resolution on the order or

10nseconds.

I've used the USB Scope 50 from Elan -

- which has a pretty clunky interface but has been enough for me with a little working around. I think it costs about $400 but, being PC based means you're not short of storage. It claims it does an effective 1GS/s but I've found it gets a bit unpredictable on signals over about 40MHz. If you really want to measure faster than that then forget it.

You're definately right. Last time I looked at buying a scope was four years ago or so; I guess I was fantasizing prices would've made strides and bounds.

My goal is to learn how to build modern portable SoC-based computing platforms. With everything moving towards BGA, it's just something I've got to adapt to. Without question, I am most definately not prepared, but I also dont really have a choice. I might be able to avoid the pains of modern board design for a while longer, but I'd be ignoring the future.

I'm committed to learning what needs be done, even if it means accidentally melting a couple IC's in a toaster oven and paying through the nose for. What I'm most concerned about is actually engineering the boards, since my analog experience has been fairly skimpy. I was hoping to find a scope that would aid me in discovering all my very painful mistakes.

rektide

My main concern in bootstrapping myself towards modern SoC is high speed signals, as in, I basically at this stage dont know jack about high speed signal design. After I get a little experience with some pretty vanilla high speed systems, building myself a simple reasonably high sample rate logic analyzer is going to be very high on the todo list. But measuring whats getting sent is, at this stage, less important than knowing if the reciever is hearing the same message.

Thanks, rektide

If it comes to this you should be able to find a company that'll mount BGA parts for you, it's getting run of the mill for PCB assembly houses and it's not that expensive, even for 1 off.

Nial.

True. I got a quote on it a couple of weeks ago, and the fellow said that the process was actually easier to control than reflow of some of the fine-pitch leaded parts.