PC based DIY Oscilloscopes, logical analyzers, spectrum analyzers, etc...

Probe -> ADC -> FPGA/CPLD -> Computer-I/O

Computer-I/O can be ethernet, firewire, usb, pci, pic-e etc.. depending on your choice for complexity, speed, and latency.

USB has an isochronous mode that will transmit frames without retransmitting any faulty ones. That mode should fit this purpose. USB have serious overhead problems thoe.

I hope you know your groundplanes, shielding, impedance matching etc.. ;)

Hello Jon,

If you come across a USB spectrum analyzer for EMI pre-compliance work (150kHz through 1GHz) let me know. There is a huge market for this stuff but companies that make spectrum analyzers do not seem to understand. Even the small ones don't.

--
Regards, Joerg

http://www.analogconsultants.com

Joerg;

Are you looking for "fewer features" just to do pre-EMI? Lower cost?

Can you elaborate?

Dave

Patience. Working on it :-)

--
Reply to nico@nctdevpuntnl (punt=.)
Bedrijven en winkels vindt U op www.adresboekje.nl

seen this

? It even plays MP3s while you are working, and picks up blooteeth viri as well

martin

Hello Martin,

Yes, I have seen those. A company in Germany (Aaronia) makes a whole range of them.

I am looking for something cheaper and most of all smaller. No display, no battery, no keyboard. A laptop or PDA that you already have to schlepp along anyway can do all that. Or, pretty soon I guess, a fancy cell phone.

--
Regards, Joerg

http://www.analogconsultants.com

Hello Dave,

Yes. You don't need narrow IF filters. In fact, you don't need a whole lot of parameter settings at all. Experienced users know what's going to leak, they just need to know how bad it is.

Typically the situation is: Client got black eye at the compliance lab. Sends you the plots, you look at them, choke a little, cuss a little, and book a flight. All you really need to know is i.e. that 360MHz is the third of the processor clock and it need to be muffled by another

15dB plus a 10dB "angst" margin.

Yes, but most of all size. It would be nice to have something the size of a pack of cigarettes or two that plugs into USB.

Pretty much what Martin suggested but without display, battery and keyboard. And smaller.

--
Regards, Joerg

http://www.analogconsultants.com

Hello Nico,

Let us know when it matures.

--
Regards, Joerg

http://www.analogconsultants.com

How fast shall the swep be done ..?

I've seen 50 MHz USB2 "scopes" already on the market (VERY small). Maybe such could be programmed to do your spectrum measurements? (frequency might be the akilles heel thoe I suspect :)

Math = for sampling and reconstructing a sinusoidal wave you need _at least_ two samples per cycle, usually you use more, because you want to see distortions and so on. So your ADC has to be able to pass the processed info in digital word to some storage before the next sampling occurs. The rest is trivial as your display can reconstruct this info at its own pace. How simple it is you will find out!

Good luck

Stanislaw Slack user from Ulladulla.

Abstract Dissonance wrote

We have the logic analyzer stuff at DIYish prices. See

It doesn't matter much, a few seconds would be ok.

Not really. You need a mixer and a sweep oscillator. Similar to what a scanner radio does. Also, it has to have enough dynamic range so it isn't thrown off the rocker by a strong local TV signal or a cell tower, or by someone in the building using a cell phone.

--
Regards, Joerg

http://www.analogconsultants.com

Joerg, here is a wwebsite that has a design for a spectrum analyzer, and a complete kit if parts, that looks as if it would work for your requirements.

And here is the website that goes into the technical details.... Actually, this fellow did the original design and the other one liked it so much he produced a set of PC boards and the components.

Personally, I am impressed and seriously thinking of building one.

Jim Pennell

--

18:40 Pacific Time Zone
Jun 14 2006

International Time
01:40 UTC
15.06.2006

This reminds me of ed, whose only error is '?', about which it has been said: The experienced user will know what is wrong.

--
Ben Jackson

http://www.ben.com/

Forget analog bandwidth for a minute, what sample rate do you need?

100MS/s is about the top you get in low priced units. That gives you an effective single shot bandwidth in the order of 10MHz. If you want 100MHz single shot analog bandwidth you need at least 1GS/s, you can't do that yourself. You can push it to a few hundred MS/s is you rolled your own, but it's not easy. If you don't know what "single shot bandwidth", "effective bandwidth", and "repetitive sampling" are, you should learn before you embark on such a project. Repetitive sampling sucks, you don't want to go there, although if you really need 100MHz analog bandwidth you will have no choice. The hardest part is the analog front end. You need an input buffer and a selectable gain stage. If you want it for a single input voltage range it makes your life easier, but I would go to all that trouble and not include a proper CRO front end attenuator.

USB is the key, but you need sample buffering. You would not stream it via USB in real time, no one does that. Everyone buffers it inside an FPGA these days if you only need a few KB of sample memory (which is not very good), or using external SDRAM for larger sample buffers.

Dave :)

Hello Jim,

Thanks. Interesting, but why on earth do most of these contain a bazillion little circuit boards that must be interconnected. Makes no sense to me. The best approach would be one board, one BNC connector and one USB plug.

If I was retired I'd do that as well. But I'd rather buy something right now.

--
Regards, Joerg

http://www.analogconsultants.com

You'r not the first one. :-)

Btw, should anyone else care to lookup more projects in this area, pls post the links..

Consider if you want storage or realtime. If storage is your way then it's no big deal with transfers as it can happen in any speed. However for realtime your pipe(s) better keep up with capacity and latency.

Also consider what interface you want to use to get the data into the pc.

There's other gotchas.. like if you do 3Gsps 8bit. Then PCI-e bus be 79% busy with sampleing data. And you can't save anything because your bus is simple stuffed. Unless there's two of them. Be happy you don't need 1TByte/s like cern.ch ;) For 3GByte/s you would need something like Raid-0 with 40 discs. For 1TB/s you would need to divide signal and then use raid systems.

Joerg skrev:

Would a 2Gs ADC and an FFT on the PC be any good? the ADCs at that speed doesn't provide too many bits, so the dynamic range would be limited. maybe with some selectable input filters and bandpass sampling?

-Lasse

Hello Lasse,

Not really. You need well north of 80dB of true dynamic range because there is so much pollution nowadays. Cell phones everywhere, plus wireless this, that and the other thing.

The other issue is power. Such an ADC would take one big slurp and your laptop battery is gone.

Best would be something like a scanner. One analyzer I built back at the university used a ladder of simple varicap-tuned oscillators and one mixer. This created a "zero IF" scheme where you'd always get a double peak but for this kind of application it might be ok. Unfortunately I built if for the RF institute so I had to leave it there. Sigh. Well, it didn't quite go to a GHz anyway.

Later I built something like that around a Plessey SL6440 but again not for myself. Man, those were incredible mixers. History, AFAIK. Somehow much of the good stuff doesn't seem to make it. But I still do have some of those Plessey mixers. They boast a dynamic range from here to the Klondike.

--
Regards, Joerg

http://www.analogconsultants.com