Scoping Tiny Signals ~250uV

The problem is that the dominant signal may be within spec while you're trying to look at another part of the spectrum where another weaker signal is NOT in spec.

I haven't done any hands-on EMC qualification testing in two decades, but I can't imagine it's gotten any easier.

Not clear what you're doing. If you're doing certification testing, you've got a lot of issues. If you're doing pre-qualification engineering tests, you still have many issues.

Bottom line: BUY A SPECTRUM ANALYZER

Why? Because you don't want to fail the tests, re-engineer the product and retest. Because you don't want to build in enough safety margin to account for all the additional uncertainties in your test setup.

A scope has HORRIBLE dynamic range. It has a wide band front end with wide band noise. It digitizes that signal plus noise with a wide band low resolution A/D. Only then do you get to do the digital magic. You have to set the gain so the system doesn't saturate for ALL signals from DC to Light. There ain't enough SNR to do what you want for small signals. Do the actual math on the problem using the actual noise and quantization numbers from your system.

A spectrum analyzer mixes the input down through relatively narrow band amplifiers at MUCH lower frequencies. The amps are typically log amps. You get a LOT more dynamic range and effective SNR with a spectrum analyzer.

Bottom line: Beg, borrow, rent or BUY A SPECTRUM ANALYZER Compare the cost to all the screwing around you're gonna do trying to make it work...sorta. A VERY old used spectrum analyzer is gonna be VERY much better than what you're gonna cobble together. YMMV

Reply to
mike
Loading thread data ...

Usual problem is picking it out of the line crossing or input rectifier harmonics.

You might consider a current transformer, instead. A simple multiplication turns uA into uV, if your LISN is terminated. Pick the right transformer and you'll get better attenuation of the low frequency stuff.

RL

Reply to
legg

Yikes.. I feel like the skipper just hit me. :P (Gilligan's Island reference.) A very convincing response against messing around with a booster amp on a LISN to a DSO for conducted EMI compliance analysis.

Yeah, I'm doing pre-qualification. iows.. My smps has to pass conducted EMI limits on my bench so I can learn to manage conducted EMI. I'll shop around for a spectrum analyzer. My motto for test equipment is: If it doesn't have USB...it's too old.

Thanks Mike

Reply to
D from BC

The LISN design is 50ohm terminated. I'll think about popping in a transformer.

Reply to
D from BC

Gee... Buying a spectrum analyzer is sounding better and better. :)

Reply to
D from BC

Bah. USB is great, but there's a lot better value for the buck with some of the older (used) spectrum analyzers... that generally had GPIB and sometimes a serial port, but not USB. I mean, I really think you're better off buying a really good older instrument along with an inexpensive USB to GPIB converter than instead only being able to afford a relatively low-end modern instrument that has USB built-in.

Reply to
Joel Koltner

Ah big picture thinking. :)

Reply to
D from BC

In your case it sure is. However, there is the occasional need to monitor a tiny one-time event on the scope. In my case that's mostly ultrasound echoes. For that I use a HP8447A dual amp. You can cascade the amps and the result is a total of 40dB of unfettered gain from

100kHz to 400MHz.
--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.
Reply to
Joerg

ter

Oh? There is such a thing? And it comes with drivers and software that'll be useful?

I'm looking for a way to talk to a TDS2024 scope with GPIB with the atrocious TekVISA/Openchoice software.

Which USB->GPIB adapter will work with the big boy's software?

I bought a cheap (40$) USB->GPIB adapter on Ebay, all it does is it shows up as a printer port. I could have done that myself.

Reply to
a7yvm109gf5d1

Whatever you get, don't scrimp on a good attenuator/suppressor for the front end.

RL

Reply to
legg

Well, the Prologix adapter ($150 --

formatting link
is supported by John Miles' excellent (and free!) GPIB toolkit
formatting link
I'd call that useful...

...but it is true that old software that expects a National Instruments or Agilent adapter won't "just work" with a Prologix adapter, unfortunately.

There is a reference on the Prologix site to a "wrapper" library so that C code written to the NI API standard will work with the Prologix adapters... so if you have source code, it's not too bad.

One of the guys here on SED works for Alciom, who also makes a cheaper-than-NI/Agilent-but-more-expensive-than-Prologix adapter:

formatting link
. Seems more like a European thing, though -- I don't think they even have a U.S. distribtor.

If it's so atrocious, why do you want to use it anyway? :-)

The ones made by the big boys? :-) But yeah, they're not cheap.

Note that the big guys are known to eventually drop support as well: National Instruments doesn't support their original USB to GPIB adapter on anything past Windows 98, I think it is... nor do their support ISA cards on newer OSes either. Pretty crappy given the premium prices they charge, you know?

It should actually show up as a serial port, I believe...

---Joel

Reply to
Joel Koltner

John

..

Good pointers, thanks. I was looking at the Prologix adapter on eBay. Perhaps after the holidays.

I don't "want", I *have* to. The scope is from before my time. If it had been up to me, we'd have a Rigol scope. No more Tek for me! They ain't what they used to be.

No kidding.

onal

g

OSes

It costs money to keep people around that can program to older OSes.

I don't remember for sure, but I'm not finding out. The last time I installed the "driver" as it was, it caused blue screens on my laptop when I undocked...

I think I'll dig up my trusty commodore 64 and its GPIB interface.

Reply to
a7yvm109gf5d1

Miles' excellent

call that useful...

I use this to save scope captures from my Tek TDS 640A with a Prologix GPIB adaptor, it's very useful, much faster than messing about with floppy disks etc.

Agilent adapter won't

code written to the NI

code, it's not too bad.

It does.

I did some experiments trying to drive an Agilent 34970A data aquisition box from VBA in Excel via the com port.

This was a while ago but think I got this to the point where I was talking to the aquisition box, reading values back and inserting them in the Excel worksheet.

I'm no software engineer but if anyone wants to look at my amateurish hacking you can get the spreadsheet here.

formatting link

Alt-F11 when you open the spreadsheet to see the VBA editor. Open up Forms and Modules to get at the code.

I hope this helps,

Nial

Reply to
Nial Stewart

I have some thoughts on getting gpib on the cheap.

There was a usb/gpib converter sold on ebay by Softmark. Stay away from that one. It worked, sometimes with their demo code on some instruments, some of the time. Was not able to get ANY of my code to read reliably. Most serious problem was that the hardware can't even read DAV. Long delays to make sure the data was valid before reading it didn't help much. Attempts to modify the gpib interface so the internal processor was attached in a way that could read DAV failed cause the function to directly read the Softmark chip registers didn't work. They always answered emails promptly and ACTED very helpful, emphases on the "acted". When presented with the fact that their product couldn't work because it didn't even read DAV, they ACTED very helpful. they'd look into it, they'd send me a firmware update, yada, yada, yada... I even told them how to fix their hardware. But they actually delivered NOTHING, no info, no code, no nothing. I've never encountered anyone who was so at ease with promising things they wouldn't deliver. Others with the same problem had the same observation. Never found anyone who made one work. If you find one of these for cheap, like I did, STAY AWAY. There's a reason it's cheap. It don't work. If I'd bought it new, I'd have sent it back for a refund.

The national instruments stuff works. If you're doing production work, just buy it and be done with it. Unless you're trying to build a hundred stations, it's just not worth the hassle to look elsewhere.

If you're cheap like me, Some of the older NI cards can be had for cheap. Some of the old 8-bit ISA cards (I said old) work great. There's a matrix on the NI site that says which cards work with which drivers on which operating systems. You have to pay close attention, cause some cards with very similar names don't work with newer systems. I decided that spending a dollar on an old NI card and another dollar on a PC with an isa slot and win2K was a LOT cheaper than trying to buy a PCI card or USB/GPIB device.

The NI PCI cards are coming down in price. I found one for $10 on craigslist...but that's a rare occurrence. Was glad to get rid of the old ISA computer.

For NONAME GPIB interfaces, watch out for proprietary stuff. I have one noname card that swizzled the data lines. I had to unswizzle the bits before sending data to the card. No idea why they did that except to require you to buy their drivers.

For hobby work, it's not too hard to roll your own. GPIB is EXCEEDINGLY complex. Good news is that if you're programming only one instrument with ascii commands,

99% of the GPIB hardware/software/protocol is unnecessary. You can just pass ASCII from serial to/from GPIB a byte at a time. I built an interface around a PIC16F627 (only 'cuz I had a tube of 'em). Slow, but many applications don't need fast.

People have used the parallel port hardware. I chose serial because I wanted to use a Palm III as my controller. For the PC, a USB/Serial converter can drive the GPIB thru the PIC. Another fun thing to do is to use a serial/bluetooth converter on each end for wireless gpib control. That way, you can use a bluetooth-enabled PDA as a wireless controller. "Excuse me while I phone my spectrum analyzer for an update..."

I wouldn't recommend rolling your own for anything mission critical. You don't want your milling machine to misfire.

Whatever you do, put a hardware abstraction layer between your code and the GPIB driver. If you need to switch GPIB hardware, it's just a matter of changing the HAL. You may not think you need it, but you'll thank me later when you find a better gpib interface or want to change to a different protocol altogether. Changing my hardware from NI interface to homebrew serial-gpib interface is as simple as clicking a checkbox.

Bottom line: GPIB can be fun to play with and is a great time waster. If you're doing production work, go by National Instruments hardware and get back to what you do best, generating profit for your company. I don't know what your burdened labor rate is, but I expect you can buy a brand new NI card for less money than you've spent in time reading this thread...probably less than I've spent writing this page.

as always, YMMV.

Reply to
mike

I'll second that. With some older analyzers one can clearly say that they don't make'em like they used to no more. I mean, where can you buy the true performance of a HP8566 these days?

I have the Prologix adapter and it works great. With John Miles' software you can get the HPGL plot from the analyzer onto a PC and then store it in a more contemporary format.

AFAIR there is also an Ethernet version of the adapter.

[...]
--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.
Reply to
Joerg

I initially had problems with the Prologix adapter. However, the designer was very cooperative and he and I figured out what it was: The logic threshold on the adapter was too high and the old TTL chips in vintage gear just couldn't get up there. Or barely could. I asked him whether the uC he used had programmable port resistors. Lo and behold, it did. So he changed the firmware and, voila. If anyone has an older Prologix it may be good to spool the latest firmware onto it.

Who knows, maybe other adapters have that problem as well? Terminated LS-TTL just won't get up to 3V.

[...]
--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.
Reply to
Joerg

And never, ever, turn the DUT on or off at the mains switch while the coax is connected to your lab gear. Else ... phssst ... *PHUT*

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.
Reply to
Joerg

I have the Ethernet Prologix. Works great, but costs $250. Next spring I'll put a wireless bridge on it, so I can run my spectrum analyzer from the hammock. ;)

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal
ElectroOptical Innovations
55 Orchard Rd
Briarcliff Manor NY 10510
845-480-2058

email: hobbs (atsign) electrooptical (period) net
http://electrooptical.net
Reply to
Phil Hobbs

Why? What happens if you do that?

-ek

Reply to
E

You get to buy a new diode mixer for your spectrum analyzer.

Reply to
mike

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.