Trigger Isolator

Hi, Joerg,

I've done pretty fast stuff with a high-mu core and separated windings. The tradeoff here is more capacitive coupling (with multifilar) versus slower risetimes (separated windings.) Since the op needs tens of ns response, pig slow, I'd think that separated would work OK. Either way, it's easy to try.

And/Or drive it bridge-wise on the trigger side for twice the bang.

The laser gadget is something like $180, and the receiver is maybe $350. I can never remember the prices of anything. Even if you never sell stuff like this to anybody else, it makes press releases.

John

Hi Y'all :

I need to simultaneously (within a few 10's of ns) trigger a number (4) battery powered 'scopes from a single trigger event.

The problem is that each 'scope is at a different reference ground. I have high dV/dt's between the "grounds" so isolation transformers have too much C.

Obviously, I could design myself a circuit with a battery powered trigger comparator feeding the LED's of 4 high speed opto's with four battery powered secondary circuits giving the trigger signals to the 'scopes.

Is there a better way ? Does anybody make a suitable signal isolator ?

Thanks Gary

Hello Gary,

Assuming this is 10 or 20 nsec and not a 10th of a nsec. That makes it easy.

I use #43 ferrite toroids for that. Four to five bifilar turns (twisted), one winding to the trigger input and the other to wherever trigger is coming from. You'll need a C in series with the primary if your trigger source doesn't like a DC short. The number of turns depends on how sluggish the signal slope is. If rise times exceed 100nsec you may need more than five turns bifilar.

If you take wire with a fairly thick insulation on a 1/2" OD core the capcitance is just a few pF. If higher voltage spikes could be present between primary and secondary use 20mil Teflon insulated wire or something similar. Make sure the core has no burrs or carefully sand these down before winding.

Don't use ferrite material that was intended for switche mode power supplies (#77 and the like). It won't be any efficient in the nsec range.

If it's lots of scopes and the trigger source can't stomach all these loads you may have to follow with CMOS buffers there (HC series bus drivers or Schmitt inverters if slopes are slow). One for each toroid. They don't like DC shorts so you'll need the caps.

Regards, Joerg

Hello John,

You need #43 material or similar and the coupling might get a bit too loose. If it has to be all on one toroid I'd twist them all gently together and then wind. Or I should say, wrestle it around the core. Let's see, how would that be called? Quintufilar or something like that?

Big old bus drivers should do. When in the same chip they can often be paralleled for more oompf.

Nice housings. Looks upscale and expensive.

Regards, Joerg

Have you ever looked at high frequency pulse transformers - not the one's intended for triggering triacs, but those aimed at digital and data processing?

IIRR Newport offered a bunch of parts that did the job. I last looked at using their pars for this sort of job 15 years ago, but it looks as if their 766 series - some of which are available from Farnell (order codes 179-984 to 179-987) - might be worth looking at.

When I looked at them, it seemed to me that they had to be bifilar wound as transmission line transformers, from the inter-winding capacitance, and that is precisely what you and John Larkin are recommending, as I would have done if I'd posted a response last night.

Because they are transmission line transformers, high permeability manganese-zinc ferrite cores seem to give the same sort of rise-times as nickel-zince ferrites, even though MnZn is lossier at high frequencies.

----------- Bill Sloman, Nijmegen

Hello John,

Yes, if he can tolerate a tiny deviation between the individual triggers it should indeed be ok here.

Another option for those who hate winding toroids or don't have cores handy is Ethernet. Often these cards contain a set of four transformers (Murata etc.) in a block. Probably they should be good if the common mode spikes don't exceed 500V. The stunner came when I needed one of these quickly because the client didn't have cores and shipping took too long. It was cheaper to buy an Ethernet card than buying the isolation xfmrs as a part. I guess a cheap port switch would have been even more of a bargian because it should contain lots of them.

Do you find it easy to get press releases published in the EE media? I have never done it there, just in the medical world. There it often helps if a "luminary" uses the gear or at least comments on it.

Regards, Joerg

If you have an interesting product and a nice picture, and you blanket

20-50 mags with press releases, odds are pretty good a few will pick it up. If you're lucky, you'll get into the online editions too, and get a link, and that improves your Google status.

The problem with a small business making niche products is that it's so hard to connect with your potential customers scattered around the world. So press releases and links and stuff are necessary to gradually build up visibility and momentum.

John

Hello Bill,

I did, sometimes. But the minute I saw the price tags I decided it's better for the client to roll their own or contract that out. Much cheaper. Lots of times you can also find a ready-to-go "catalog transformer" from companies such as Murata. They just aren't advertised as pulse transformer but more for their target markets, such as modems or LAN ports.

Almost 7 Euros. A lot of money, unless it's for a prototype. For production runs that usually won't fly.

Usually I just look what's there from the mainstream manufacturers and then design the xfmr around that. #43 works great in this application and is very cheap.

Regards, Joerg

Hello John,

Thanks, that sound like a good strategy.

That is true. I always wonder why there isn't any kind of exchange for technology services. A place where managers could come to when they are at the end of the rope and need someone they can't find in-house.

Regards, Joerg

Hey look, it worked!

Of course, this is buried pretty deep in an obscure list of new product releases under the heading "interconnects", so it's not real likely anybody will find this starting at the E-E home page. But press releases do create more search-engine targets, and the link on this page does boost us in Google's scoring system.

The advertising concept of "building a public presence" never made sense for small businesses, but has new meaning in the age of search engines.

John

Hello John,

Yes, that's great! Interesting, you went metric in the text. Kilometers. The only other places I see that in the US are IEEE publications.

Yes, it sure does. However, there still is a huge vacuum when it comes to finding the right talent for a project of limited scope where people don't need to hire someone permanent. When they are just looking for someone or some company that can fix a dilemma or take over a nasty part of a design.

Regards, Joerg

Well, if you had a 50 ohm resistor, and behind it you had a 5-volt source, then, well, there you'd be. And a BNC is a sort of round shiny gadget used to help connect coaxial cables from some things to other things.

I like SMBs. I like MCXs even more, but they're not as common in my user base, and don't detent very hard so are more likely to work loose. No complaints so far, plus we get to sell SMB-BNC cables as accessories. No way I could have fitted BNCs into that tiny box, and SMAs are just too much a pain to screw/unscrew all the time. Don't even mention LEMOS to me.

No fiber optic connector is well-suited to use by lumberjacks standing in mud. But you might note that this message to you is no doubt being transmitted over fiber optics that has been pulled through pipes buried in ditches, clear across the country, by regular worker-guys.

Lots of fiber is being installed in cities and buildings lately, mostly actually blow-in and not literally "pulled." The fiber is so thin and light and flexible, you just clip a sort of little parachute to it, stick it in a conduit, and air-blast it all the way, around corners and everything.

What's astounding about fiber is the quality of fast signals over distance; multimode fiber has a bit of dispersion, maybe a ns per km, but nothing like the ghastly skin-effect losses of copper. Single-mode is even better.

STs are pretty good, easy to mate/unmate and quite reliable; ST cables are available off the shelf at at Fry's, and are used for audio and ethernet links.

Yeah. People are surprised at how cheap they are. We're gonna have to raise the price eventually. I actually did this for a guy at Arnold AFB who does EMP testing of military gear, at field strengths of

50,000 v/m. I just did the fiber stuff to get him to buy the serious timing hardware.

John

What does "+5 volt nominal behind 50 ohms, BNC connector" mean exactly for that J720 "INPUT"? - wish you did not use that crummy SMB for the J730 digital output...And how rugged is that ST-ST fiber patch cord anyway? Is it some dainty lab type thing or can it withstand being manhandled and pulled through conduits not so gently and repeatedly? And this product runs roughly $500 for an xmt/rcv pair?

I thought that terminology was fairly common: an open-circuit voltage source of 5 volts, with a 50-ohm source impedance. Not to be confused with a 50-ohm source that puts 5 volts onto 50 ohms, which would be "ten volts behind 50 ohms". If you just say "5 volts at 50 ohms" it can be taken either way.

Sorry. Don't know how else to explain "BNC" connector.

John

That still makes no sense- so skip it.

If that's your attitude then keep them.