Advice on Digital Oscilloscope Design

I thought you said "high speed" ? 10-100MHz is slow for an oscilloscope.

A standard SDRAM module from a PC can keep up with these rates and has the added advantage of being readily available, cheap, large and wide.

The width matters when you try to sample on multiple channels simultaneously.

You will need to make an SDRAM controller in something like an FPGA. It will need small FIFOs to allow for the burstiness of the SDRAM accesses (caused by scheduling in refreshes and read accesses).

Regards, Allan

Hey Jim, this one isn't even from a gmail or hotmail addy.

Graham

Google Tektronix. They have done the development further than you will ever imagine.

--
John G

Wot\'s Your Real Problem?

A demanding hobby project. Too bad, that in order to actually get it running you'll need equipment that is far more advanced than the scope itself.

Rene

*** Free account sponsored by SecureIX.com *** *** Encrypt your Internet usage with a free VPN account from ***

It can be done. Go to Cypress semiconductor web site and find the widest and deepest FIFO you can afford. They have ones that go up to

100MHz data rate on the input.

You'll probably have more trouble with the analog front end. Making a good glitch-free attenuator and preamp is a lot tougher than it looks, plus you need a much better scope than you're making to fine-tune the details!

Yeah its for hobbiest use so $40000 Tektronix scopes are not for me. I need a SDRAM not dynamic for simplicity.

Don't know what Pooh bear's problem is.

JJ

Cypress has generic SRAM down to 12ns (83Ms/s) for 256k x 16bit CY7C1041B or 8ns for 64k x 16 bit CY7C1021CV33. You could use two banks to get 4ns (250Msamples/sec) by interleaving.

You could do it with a $5 video ADC and a single FPGA, with the waveform stored in FPGA internal ram. The whole thing could be done in the kluge area of a cheap FPGA eval board. The nasty part would then be the analog interface to the input and trigger signals.

It would be quite a project overall, but do-able and no doubt educational.

John

Interesting idea. Can you suggest a FPGA? I don't have any experience with them.

JJ

Xilinx Spartan 3. The Web Pack development software is free, and I think there are eval boards for $80 or something like that.

John

The may-get-round-to-it-someday bit box has an ADS830 8 bitter and a CYC199C-15PXC 256K static ram (Farnell 891-1541) in it. Long forgotten the details but ISTR 50MSPS as an aiming point. I do know the bits would have been selected for cheapness and availability. (Lost interest in the idea after having to use a customers HP digital scope for all of 5 minutes :) john

Why?

Were you unsatisfied with the HP scope, or consider it already done very well by them?

Actually you could buy a pretty nice fpga eval board, analog front end, and the hdl code needed to build 100Mhz digital oscilloscope kit from:

for about $150. Of course that spoils all the fun of building one yourself.

Mark

Need more work on my cynicism. Hadn't realised the words read validly both ways!. The HP marketing guys could probably phrase it as that I had a somewhat challenged relationship during my developing a strategic approach towards an enriching measurement experience. Myself, I'd just say their product was a bag-o-s**te :) john

Shame.

HP used to be synonymous for making top quality gear and viewing the price tag as a consequence, not a constraint.

Was it the user interface or the fundamental performance of the thing that let it down?

Plus the summ of the parts are likely more a cheap comparable 100MHz Tek Scope.

Rene

--
Ing.Buero R.Tschaggelar - http://www.ibrtses.com
& commercial newsgroups - http://www.talkto.net

HP made some appalling digital scopes in the '90s.

Design faults include (but are not limited to):

- Slow response to knobs and buttons.

- Poorly designed user interface.

- Low sample frequencies.

- No digital antialiasing.

- Small memory size.

More modern scopes are much better. I wouldn't part with my Agilent

54855 (?) with its 20Gsa/s capture on four channels simultaneously, etc.

Regards, Allan

Yep, you can pick up some of those early digital scopes, 54000 thru

54599 for a song on eBay sometimes. They're barely worth the $100-$300 bucks IMHO. Really rotten interface on many of them, no indication of clipping or aliasing, and nasty green or amber screen CRT's on many of them.

Hi Kryten. Biggest complaint was the user interface. Buttons, menus, sub menus, then more buttons and more sub-sub menus. I got sick of reading screen text. Finally got a stepped, somewhat raggedy, rather pathetic, trace on view. -Knew- I had intermittent (many volts) noise problems riding on the signal I was looking at but despite best efforts couldn't see it. A major problem was the "disconnectnedness" of the display from reality. Like trying to listen to a delayed version of a singer with a 90% faulty microphone. Borrowed someones 'real' cheap bench scope and confirmed the noise at a glance. To me it was obvious that the whole team involved in the design, management and production of this digital scope had never had cause to use a conventional instrument since their leaving school. There's a Windows sheep mentality at work here, like "Enjoy those fine quality printouts, enthuse over the network connectivity and forget about doing any work". In former times a battle scarred manager could have called into their development lab'. Spotted the nascent 'user interface' and quickly nipped it in the bud. Can't happen now as no one knows any better. Despite all this, I've still high regard for much of the other HP (Agilent) kit I've come across. john