Tek / Tektronix 2225 and/or 2230 adding "trigger out" output

[KhronX]

The idea in the title is the simplest way i could think of, to "sync up" my two scopes, in case i'll ever need more than 2 channels triggered from the same signal.

Now, in the schematics, i think i've found the places where i can pick off the appropriate signal.

2225, page 167, top right (output of U480B): http://bee.mif.pg.gda.pl/ciasteczkowypotwor/Tek/2225%20SM.pdf
2230, page 340, bottom right (output of U460, eventually): http://bee.mif.pg.gda.pl/ciasteczkowypotwor/Tek/2230%20.pdf

The "interesting" thing is, the chips employed in these are ECL (emitter-coupled logic), so i'll need to figure out some clever level-shifting 

I'll gladly welcome critique / ideas, before i delve even deeper into this 

[David Hess]

I am just going to cover the 2230.  You probably want the a-gate signal and not the a-trigger signal.

The a-trigger signal follows the trigger source continuously so it would be suitable if you wanted to feed it into a frequency counter.  On frequency counters this is called the shaped output.  The a-gate signal indicates the start of a sweep and is where you would connect a second oscilloscope's trigger input.

The a-gate signal is shown in the upper right corner of 2230 schematic 5.  There is a positive and negative a-gate signal and you could use either one.  These signals are fast so a buffer driving a coaxial cable should be used.

[KhronX]

Thanks for the input (pun slightly intended) 

Of course, i never considered just hooking a wire to a BNC straight from that I figured using a hex inverter should work just fine - one gate as a buffer, and the other five paralleled to drive the output.

After checking that area of the schematic, i see an added bonus - no ECL-to-TTL level shifting required Although i might need some sort of clamping, so as not to toast the buffer's inputs...

EDIT: Nevermind that last bit - the "A gate" signal ends up being fed into a 74F153 mux, so it oughtta be safe for (other) 5v-logic chips too 

EDIT no.2: Should (only) the "A gate" signal be the "trigger out" i'd need? Or would the output of said "trigger mux" (schematic 18, grid 1D) be it?

[David Hess]

Of course, i never considered just hooking a wire to a BNC straight from that I figured using a hex inverter should work just fine - one gate as a buffer, and the other five paralleled to drive the output.

There actually is a way to get the cleanest possible signal without any buffering or effect on the circuit in this case.  This trick is sometimes used to get high fidelity signals into an oscilloscope or other low impedance test instrument without affecting the circuit using a probe.

The input impedance of a 50 ohm coaxial cable which is shunt terminated to ground at its other end is 50 ohm resistive to ground.  So any of the pull-down resistors to ground, R578, R576, R582, and R580, can be replaced with a resistor 50 ohms less connected to the center conductor of the coaxial cable with the coaxial shield going to the original ground connection.

After checking that area of the schematic, i see an added bonus - no ECL-to-TTL level shifting required Although i might need some sort of clamping, so as not to toast the buffer's inputs...

EDIT: Nevermind that last bit - the "A gate" signal ends up being fed into a 74F153 mux, so it oughtta be safe for (other) 5v-logic chips too 

The PECL (positive emitter coupled logic) signal directly from U506 will have a little bit less delay which might be important in some applications.  If the delay really matters though, then both instruments should be connected to the same trigger signal.

The 74F153 (U4227) is at an easy place to tap the gate signals since it is exposed on the top of the storage board.  The 10H131 (U506) is on the top of the bottom board so access to it is also possible from the solder side.

EDIT no.2: Should (only) the "A gate" signal be the "trigger out" i'd need? Or would the output of said "trigger mux" (schematic 18, grid 1D) be it?

The theory section covers what signals are present at the input to the multiplexer but I think you want to use the A and/or B gate input signals to the multiplexer.  From what I remember, the multiplexer is for selecting between the two gate signals and internally generated calibration signals.

The 2225 has the same gate generation circuit using an 10H131 but no multiplexer because it has no B gate and no calibration signals.

[KhronX]

Thanks for all the information & indications, i'll be looking into it later this week 

That "partial replacing" of a pull-down resistor sounds interesting, but that would sort of "need" a permanent connection / termination in order to maintain signal integrity, wouldn't it?
Plus i'm not sure how would the extra capacitance & inductance of a wire from that point to the output BNC and back (center & ground), affect the operation of the surrounding circuitry.

I must admit, i'm "still" quite tempted to just buffer the output of U506.

Although, would the output BNC need to be isolated from the case, and grounded to near where the source of the signal is? Or can it be a non-isolated, panel-mounted one?

[David Hess]

That "partial replacing" of a pull-down resistor sounds interesting, but that would sort of "need" a permanent connection / termination in order to maintain signal integrity, wouldn't it?

Yes, it would need to be permanently connected and terminated at the end of the transmission line.

Plus i'm not sure how would the extra capacitance & inductance of a wire from that point to the output BNC and back (center & ground), affect the operation of the surrounding circuitry.

That is the beauty of this solution.  The terminated 50 ohm transmission line looks like a 50 ohm resistor at all frequencies.  As far as the circuit is concerned, it is two resistors in series going to ground with the transmission line having a value of 50 ohms.

In a practical installation with a bulkhead BNC, a 50 ohm termination will need to be installed when it is not in use.  Note that you do not

I must admit, i'm "still" quite tempted to just buffer the output of U506.

Although, would the output BNC need to be isolated from the case, and grounded to near where the source of the signal is? Or can it be a non-isolated, panel-mounted one?

The outputs of U506 are PECL levels so not directly compatible with TTL or other common logic families.  They swing between about +4.1 and +3.2 volts which neatly explains the purpose of Q576 and Q578; they level shift the PECL differential output to TTL levels.

With or without a buffer, the coaxial shield needs to be grounded at the circuit.  A bulkhead BNC may or may not be grounded to the chassis; that is up to you.  In practice the instrument the output is connected to will also ground the shield to earth ground.

[KhronX]

A little update

Today i got around to looking into this issue again. I ended up whipping up a little buffer board in Eagle (with a 7404), and then proceeded to recreate it in real life.
A small piece of leftover copper-clad board, a pencil, a SOIC14 chip, and a (relatively chewed-up) box-cutter knife  My favourite method for quick & dirty one-off prototyping 

Didn't use a 7404, but found a few 7414's in my stash (hex Schmitt trigger inverter, vs. plain hex inverter). Added an 0603 100n ceramic next to it, for good measure 

Plopped the board on top of some of the IC's in the area, soldered some "recycled" resistor legs to U4227's pin 10 ("A gate" signal") and pin 8 (ground), and to a neighbouring axial ceramic (+5v). Signal looks just as expected, on the output Mostly high, with a "blip" to 0v upon each trigger event.

Now to figure out where i could place that output BNC on the back panel... 

PS: I'll add a photo of the bodge, at some point 

EDIT: Photo attached Ignore those resistor legs sticking up, those were added just to have what to hook scope probes to - "test points", if you will 

[KhronX]

Being a hoarder / DIY-"fiend" isn't all that bad, all the time 

A couple years ago i grabbed a dead (ie. dead PSU) Behringer DDX3216, with the intention of salvaging the motorfaders and whatnot, for a (future) DAW control surface - that's yet to happen 

But today i remembered that hey, the thing has a wordclock in/out via BNC connectors. And whaddya know, they were in a (re)usable format  Shakeproof washer and everything...
I had a couple old network cards with BNC connectors, but just PCB-mount versions, and lacking the mounting nut. But not the ones in the Berry 

[KhronX]

Hmm... Would a single wire from the buffer to the BNC suffice? Or should i bother with some sort of coax cable? I have some salvaged laptop wifi cables... 

[KhronX]

In the mean time, i took care of the mechanical side of things 

Yeah, i know, could be a bit prettier / neater, but that's all i could do (or rather, all i was willing to do at 9-10pm) with a cordless drill, a Proxxon mini-drill, a box-cutter and a couple of rasp-files 

[David Hess]

Hmm... Would a single wire from the buffer to the BNC suffice? Or should i bother with some sort of coax cable? I have some salvaged laptop wifi cables... 

With the buffer, an unshielded wire may be sufficient depending on how much distortion of the trigger edge you can accept.  One trick you can use is to make a twisted pair with the other wire being ground.

[KhronX]

That sounds like something that's worth a shot Thanks for the tip, once more 

I've got a whole bunch of CAT5 cable that i use for hook-up wires and whatnot. I'll report back soon, hopefully with trace photos as well 

[David Hess]

Using a twisted pair is not as good as a coaxial wire but it provides a direct ground return and minimizes the loop area and that counts for a lot.  If you put two twisted pairs in parallel (4 wires twisted together?), you can even get about 50 ohms.

[KhronX]

I ended up with only one twisted pair, and soldered the ground wire only at the buffer end (the shell of the BNC is firmly connected to the chassis / ground / earth).

For the most part, it works, i'm happy to say The "slave" scope is just really touchy with the trigger level - i have to catch it in juuuuust the right position, otherwise it only triggers on the next transition of the trigger signal.

As probably expected, there may be a slight ground-loop issue (at least with both scopes connected to my sig-gen), but that was only visible on the "slave" scope.
That being said, i might need to crack open that one - at one point today it seemed to have trouble triggering cleanly even on a plain sinewave 

Either way, many thanks again for all the pointers! 

[David Hess]

I ended up with only one twisted pair, and soldered the ground wire only at the buffer end (the shell of the BNC is firmly connected to the chassis / ground / earth).

The ground wire will not do anything useful if it is not connected at both ends.  Connecting it will produce a cleaner edge which may be helpful for reliable triggering.

[KhronX]

During yesterday's testing, i did connect the trigger-signal BNC to one of the normal channels of the "slave" scope, and the signal looked clean as a whistle to me. No overshoot or undershoot or ringing or any sort of anomalies.

Not sure if photos would help, but i'll see about maybe shooting some short video clip to demonstrate, one of these days. But until then, at least as a proof-of-concept, (i'm glad) it works 

EDIT: I take that first part back - there's about 1-1.5v of over/undershoot, oopsie  I'll see about catching the other end of that ground wire under the nut / washer of the output BNC 

[KhronX]

Sorry to dig up such an "old" thread, but i totally forgot to report back with the conclusion of the story.

It's a bit crude / heavy-handed, i know, but i ended up just slapping on a T-splitter with a 50ohm termination on one output, and the cable to the external trigger input on the other output.

Overshoot go bye-bye   

Thanks again for the tips & pointers