Topics

3T77 tunnel diodes (again)


 

Down the rabbit hole again... of unobtainium tunnel diodes, that is.

I couldn't get my "new" 3T77A to trigger at all. I'm actually not sure if the 3S2/3T77A/561A combo allows internal triggering or not. Manual says it can trigger any 3T timebase and any 561 scope, but there didn't seem to be any signal getting to the 1st stage trigger amp (Q14). The internal slide switch on the 3S2 is in the correct (Sampling timebase) position - and the 3S2 does work with a 3B4 timebase in real-time mode.

Anyhow. I hooked up the trigger-out from the 3S2 to the ext. trigger input of the 3T77A. That provides a nice big signal to Q14. But... no tunnel diode action. No free-run with the Trig Sens all the way clockwise. I'll do a bit more measuring to make sure the bias currents are reasonable but I suspect I need at least one 1N3129 20 mA Ge tunnel diode, which sadly no longer exist. They were rare as hen's teeth 10 years ago.

Looking for recommendations to resurrect this timebase... trying to alter bias resistors so it'll work with Russian surplus 20 mA GaAs? Or 10 mA Ge? I searched the archives and some posters have had success with both methods but they didn't share the details.

Or maybe just redoing the trigger with a more modern (and simpler) ultrafast comparator on an outrigger board?


 

After adventures in (mis)using my homemade extender cable (different thread), I FINALLY got back to working on the plug-in. D22 (unobtainable 1N3129 20 ma TD) just would not trigger no matter what, just sat there around .025 volts and wouldn't get much higher despite adjusting its bias (R21) all the way up. At one point the other TD (D25) fell out of its clip. The factory solder joints hadn't penetrated through the vias far enough and the clip had worked its way loose. Resoldered that. D25 then showed .057v and by tweaking its pot, could be seen to snap from low to high state... I swapped D22 for D42, the trigger regeneratorand it started merrily free-running, stops when I reduce the Trig Sens as intended, and triggers on my 114 pulse generator :)

So now I need to figure out a mod to use some other TD in place of D42. Seems easier than modifying the two-TD trigger circuit.


 

I did some more digging in my junkbox and discovered a Tek-numbered bag also labeled "10 ma tunnel diode" :) so I won't have to wait two months for the Russian TDs. It has wire leads (the tiny "top hat" case), but that will certainly work for now - I can make them quite short for a permanent install later. I tested it and it switches at 8.9 ma.

Then I played with the resistor values in the D42 trigger regen circuit. Doubling R43 (300 ohm) didn't give me enough range to get to 8-9 ma on R44 (200 ohm pot). Ended up with 150 ohms in series with R43. I next increased R41 from 5.1 ohm to 10 ohm, adjusted R44 and a nice trigger output appeared at the front panel jack! Good pulses also coming out of Q94 which provides the staircase advance pulse and sampling drive to the 3S2.

I next removed the BSM connector (trigger out) from the 3S2 front panel and replaced it with a solderable SMA. I ran a cable from there to the ext trig input on the 3T77A. At this point I had sampling dots appearing on the screen - but no sweep (i.e. a messy vertical line). It moves horizontally with the position control and adjustments so the output stages are working. Tracked the pulse train down through Q124 ok, but it disappears at the base of Q135 completely, and of course nothing on Q135 collector. It's not shorted to a simple diode-function check B-E or B-C on my Fluke 73, nor is D134 (reverse-bias protection from B-E). But when I pull Q135 from its socket, the signal comes back there...

Now I need to dig out a PNP Ge switching transistor that can withstand at least 25 volts Vce to substitute for this 2N1516... I tried a handy 2N414 but it didn't work. Turns out its Vce of 15v is lower than the -20v power supply... oops. Also it's slower (7 MHz vs. 35 which may not be needed). I think a silicon PNP will also work there, which greatly increases the choices - I will just have to experiment.

This thing is absolutely stuffed with tunnel diodes... sure hope the rest are ok.


Tom Lee
 

That 2N1516 is almost certainly afflicted with the tin whisker problem, as are basically all devices in a TO-7 (and others whose cans are substantially of tin). The case is likely showing a short to one or more of the terminals. It's often temporarily fixable with a capacitive discharge, but just for grins and giggles, not as a repair.

From a quick glance at the circuit, it looks like a garden-variety silicon PNP should get the job done. Try a 2N3906 if you have one handy.

Look forward to hearing how that works out.

--Cheers,
Tom

--
Prof. Thomas H. Lee
Allen Ctr., Rm. 205
350 Jane Stanford Way
Stanford University
Stanford, CA 94305-4070
http://www-smirc.stanford.edu

On 2/20/2021 16:56, Charles wrote:
I did some more digging in my junkbox and discovered a Tek-numbered bag also labeled "10 ma tunnel diode" :) so I won't have to wait two months for the Russian TDs. It has wire leads (the tiny "top hat" case), but that will certainly work for now - I can make them quite short for a permanent install later. I tested it and it switches at 8.9 ma.

Then I played with the resistor values in the D42 trigger regen circuit. Doubling R43 (300 ohm) didn't give me enough range to get to 8-9 ma on R44 (200 ohm pot). Ended up with 150 ohms in series with R43. I next increased R41 from 5.1 ohm to 10 ohm, adjusted R44 and a nice trigger output appeared at the front panel jack! Good pulses also coming out of Q94 which provides the staircase advance pulse and sampling drive to the 3S2.

I next removed the BSM connector (trigger out) from the 3S2 front panel and replaced it with a solderable SMA. I ran a cable from there to the ext trig input on the 3T77A. At this point I had sampling dots appearing on the screen - but no sweep (i.e. a messy vertical line). It moves horizontally with the position control and adjustments so the output stages are working. Tracked the pulse train down through Q124 ok, but it disappears at the base of Q135 completely, and of course nothing on Q135 collector. It's not shorted to a simple diode-function check B-E or B-C on my Fluke 73, nor is D134 (reverse-bias protection from B-E). But when I pull Q135 from its socket, the signal comes back there...

Now I need to dig out a PNP Ge switching transistor that can withstand at least 25 volts Vce to substitute for this 2N1516... I tried a handy 2N414 but it didn't work. Turns out its Vce of 15v is lower than the -20v power supply... oops. Also it's slower (7 MHz vs. 35 which may not be needed). I think a silicon PNP will also work there, which greatly increases the choices - I will just have to experiment.

This thing is absolutely stuffed with tunnel diodes... sure hope the rest are ok.




 

thanks, I was thinking of that (and possibly swapping Q145 for a 2N3904 too?). Hopefully it won't oscillate since those have a 250 MHz Ft, a lot faster than the original Ge device. Could always put ferrite beads on the leads, I suppose.

Q64 in the Staircase Inverter is also a 2N1516, so a tin whisker suspect as well. It's a DC coupled stage and has a zero-level pot on the base, so replacing with a Si PNP shouldn't present any problem either.


Gary Robert Bosworth
 

We used to design tunnel diodes into the trigger circuits of pulse
generators at Datapulse back in 1970. Even back then, we were aware of the
drift in tunnel diode characteriatics with age. Some circuit operational
characteristics would change radically over the years. It was due to the
heavy doping of the semiconductor materials. It is a shame that newly
manufactured tunnel diodes are no longer in production for replacement
purposes.

Gary

On Sat, Feb 20, 2021, 17:53 Charles <charlesmorris800@centurytel.net> wrote:

thanks, I was thinking of that (and possibly swapping Q145 for a 2N3904
too?). Hopefully it won't oscillate since those have a 250 MHz Ft, a lot
faster than the original Ge device. Could always put ferrite beads on the
leads, I suppose.

Q64 in the Staircase Inverter is also a 2N1516, so a tin whisker suspect
as well. It's a DC coupled stage and has a zero-level pot on the base, so
replacing with a Si PNP shouldn't present any problem either.






Tom Lee
 

In speaking with some of my colleagues whose professional careers extend back to the days when Ge technology was mainstream, it seems that a lot of the drift problems were due to contaminants. There was a prevailing belief that the very heavy doping needed to make TDs implied that you didn't need cleanrooms -- you could make them in dirtyrooms (take a look at early die photos of ICs -- you can see all sorts of crud; apparently, the industry then relied on cigarette ash as a dopant). It really wasn't until CMOS started to become important that the industry was willing to pay for the requisite processing environments. By that time, TDs were in decline, so they probably continued to be processed more or less the old way.

-- Tom

--
Prof. Thomas H. Lee
Allen Ctr., Rm. 205
350 Jane Stanford Way
Stanford University
Stanford, CA 94305-4070
http://www-smirc.stanford.edu

On 2/20/2021 18:03, Gary Robert Bosworth wrote:
We used to design tunnel diodes into the trigger circuits of pulse
generators at Datapulse back in 1970. Even back then, we were aware of the
drift in tunnel diode characteriatics with age. Some circuit operational
characteristics would change radically over the years. It was due to the
heavy doping of the semiconductor materials. It is a shame that newly
manufactured tunnel diodes are no longer in production for replacement
purposes.

Gary


On Sat, Feb 20, 2021, 17:53 Charles <charlesmorris800@centurytel.net> wrote:

thanks, I was thinking of that (and possibly swapping Q145 for a 2N3904
too?). Hopefully it won't oscillate since those have a 250 MHz Ft, a lot
faster than the original Ge device. Could always put ferrite beads on the
leads, I suppose.

Q64 in the Staircase Inverter is also a 2N1516, so a tin whisker suspect
as well. It's a DC coupled stage and has a zero-level pot on the base, so
replacing with a Si PNP shouldn't present any problem either.







Jim Ford
 

Oh, that's funny about cigarette ash getting into IC dice! Lots more people smoked back then too, and indoors. Fortunately for us nonsmokers, these days it's basically illegal to smoke indoors in California. At the beginning of the year in 1990, I was working at the Lockheed Martin Skunk Works, and Lockheed Corp declared there would be no smoking indoors in any Lockheed building. Very nice, and we all got used to it. Later that year, in September, I started taking graduate microwave engineering courses at California State University, Northridge (several years before the big earthquake). I was appalled to see students smoking in the hallways in the engineering building! Gross!
Hard to imagine such gross (no pun intended) contaminants in ICs these days, with the cleanrooms and bunny suits and all.

Jim Ford

------ Original Message ------
From: "Tom Lee" <tomlee@ee.stanford.edu>
To: TekScopes@groups.io
Sent: 2/20/2021 6:25:25 PM
Subject: Re: [TekScopes] 3T77 tunnel diodes (again)

In speaking with some of my colleagues whose professional careers extend back to the days when Ge technology was mainstream, it seems that a lot of the drift problems were due to contaminants. There was a prevailing belief that the very heavy doping needed to make TDs implied that you didn't need cleanrooms -- you could make them in dirtyrooms (take a look at early die photos of ICs -- you can see all sorts of crud; apparently, the industry then relied on cigarette ash as a dopant). It really wasn't until CMOS started to become important that the industry was willing to pay for the requisite processing environments. By that time, TDs were in decline, so they probably continued to be processed more or less the old way.

-- Tom

-- Prof. Thomas H. Lee
Allen Ctr., Rm. 205
350 Jane Stanford Way
Stanford University
Stanford, CA 94305-4070
http://www-smirc.stanford.edu

On 2/20/2021 18:03, Gary Robert Bosworth wrote:
We used to design tunnel diodes into the trigger circuits of pulse
generators at Datapulse back in 1970. Even back then, we were aware of the
drift in tunnel diode characteriatics with age. Some circuit operational
characteristics would change radically over the years. It was due to the
heavy doping of the semiconductor materials. It is a shame that newly
manufactured tunnel diodes are no longer in production for replacement
purposes.

Gary


On Sat, Feb 20, 2021, 17:53 Charles <charlesmorris800@centurytel.net> wrote:

thanks, I was thinking of that (and possibly swapping Q145 for a 2N3904
too?). Hopefully it won't oscillate since those have a 250 MHz Ft, a lot
faster than the original Ge device. Could always put ferrite beads on the
leads, I suppose.

Q64 in the Staircase Inverter is also a 2N1516, so a tin whisker suspect
as well. It's a DC coupled stage and has a zero-level pot on the base, so
replacing with a Si PNP shouldn't present any problem either.












Tom Lee
 

There's a famous photo of the first IC from Fairchild -- a four-transistor flip-flop, in a round die! One of the contact pads is obviously overetched and barely there, and the chip is full of debris. Some years ago, Gordon Moore was at a conference (ISSCC) and I asked him about the photo. I learned that he was the one who took the photo, and also the one who overetched the pad. He said the only reason the photo was taken was to perform failure analysis. They didn't think that what they were doing was so historic as to merit photographic preservation, so that's the only photo taken of their first real IC. He said he's embarrassed that this photo is now in the Smithsonian's collection, forever enshrining his inattention to etch times.

I then asked him about the debris. He smiled and said, "Well, Noyce was a chain-smoker."

-- Cheers,
Tom

--
Prof. Thomas H. Lee
Allen Ctr., Rm. 205
350 Jane Stanford Way
Stanford University
Stanford, CA 94305-4070
http://www-smirc.stanford.edu

On 2/20/2021 19:38, Jim Ford wrote:
Oh, that's funny about cigarette ash getting into IC dice!  Lots more people smoked back then too, and indoors.  Fortunately for us nonsmokers, these days it's basically illegal to smoke indoors in California.  At the beginning of the year in 1990, I was working at the Lockheed Martin Skunk Works, and Lockheed Corp declared there would be no smoking indoors in any Lockheed building.  Very nice, and we all got used to it.  Later that year, in September, I started taking graduate microwave engineering courses at California State University, Northridge (several years before the big earthquake).  I was appalled to see students smoking in the hallways in the engineering building!  Gross!
Hard to imagine such gross (no pun intended) contaminants in ICs these days, with the cleanrooms and bunny suits and all.

Jim Ford

------ Original Message ------
From: "Tom Lee" <tomlee@ee.stanford.edu>
To: TekScopes@groups.io
Sent: 2/20/2021 6:25:25 PM
Subject: Re: [TekScopes] 3T77 tunnel diodes (again)

In speaking with some of my colleagues whose professional careers extend back to the days when Ge technology was mainstream, it seems that a lot of the drift problems were due to contaminants. There was a prevailing belief that the very heavy doping needed to make TDs implied that you didn't need cleanrooms -- you could make them in dirtyrooms (take a look at early die photos of ICs -- you can see all sorts of crud; apparently, the industry then relied on cigarette ash as a dopant). It really wasn't until CMOS started to become important that the industry was willing to pay for the requisite processing environments. By that time, TDs were in decline, so they probably continued to be processed more or less the old way.

-- Tom

-- Prof. Thomas H. Lee
Allen Ctr., Rm. 205
350 Jane Stanford Way
Stanford University
Stanford, CA 94305-4070
http://www-smirc.stanford.edu

On 2/20/2021 18:03, Gary Robert Bosworth wrote:
We used to design tunnel diodes into the trigger circuits of pulse
generators at Datapulse back in 1970. Even back then, we were aware of the
drift in tunnel diode characteriatics with age. Some circuit operational
characteristics would change radically over the years. It was due to the
heavy doping of the semiconductor materials. It is a shame that newly
manufactured tunnel diodes are no longer in production for replacement
purposes.

Gary


On Sat, Feb 20, 2021, 17:53 Charles <charlesmorris800@centurytel.net> wrote:

thanks, I was thinking of that (and possibly swapping Q145 for a 2N3904
too?). Hopefully it won't oscillate since those have a 250 MHz Ft, a lot
faster than the original Ge device. Could always put ferrite beads on the
leads, I suppose.

Q64 in the Staircase Inverter is also a 2N1516, so a tin whisker suspect
as well. It's a DC coupled stage and has a zero-level pot on the base, so
replacing with a Si PNP shouldn't present any problem either.















Tom Lee
 

There's a famous photo of the first IC from Fairchild -- a four-transistor flip-flop, in a round die! One of the contact pads is obviously overetched and barely there, and the chip is full of debris. Some years ago, Gordon Moore was at a conference (ISSCC) and I asked him about the photo. I learned that he was the one who took the photo, and also the one who overetched the pad. He said the only reason the photo was taken was to perform failure analysis. They didn't think that what they were doing was so historic as to merit photographic preservation, so that's the only photo taken of their first real IC. He said he's embarrassed that this photo is now in the Smithsonian's collection, forever enshrining his inattention to etch times.

I then asked him about the debris. He smiled and said, "Well, Noyce was a chain-smoker."

-- Cheers,
Tom

--
Prof. Thomas H. Lee
Allen Ctr., Rm. 205
350 Jane Stanford Way
Stanford University
Stanford, CA 94305-4070
http://www-smirc.stanford.edu

On 2/20/2021 19:38, Jim Ford wrote:
Oh, that's funny about cigarette ash getting into IC dice!  Lots more people smoked back then too, and indoors.  Fortunately for us nonsmokers, these days it's basically illegal to smoke indoors in California.  At the beginning of the year in 1990, I was working at the Lockheed Martin Skunk Works, and Lockheed Corp declared there would be no smoking indoors in any Lockheed building.  Very nice, and we all got used to it.  Later that year, in September, I started taking graduate microwave engineering courses at California State University, Northridge (several years before the big earthquake).  I was appalled to see students smoking in the hallways in the engineering building!  Gross!
Hard to imagine such gross (no pun intended) contaminants in ICs these days, with the cleanrooms and bunny suits and all.

Jim Ford

------ Original Message ------
From: "Tom Lee" <tomlee@ee.stanford.edu>
To: TekScopes@groups.io
Sent: 2/20/2021 6:25:25 PM
Subject: Re: [TekScopes] 3T77 tunnel diodes (again)

In speaking with some of my colleagues whose professional careers extend back to the days when Ge technology was mainstream, it seems that a lot of the drift problems were due to contaminants. There was a prevailing belief that the very heavy doping needed to make TDs implied that you didn't need cleanrooms -- you could make them in dirtyrooms (take a look at early die photos of ICs -- you can see all sorts of crud; apparently, the industry then relied on cigarette ash as a dopant). It really wasn't until CMOS started to become important that the industry was willing to pay for the requisite processing environments. By that time, TDs were in decline, so they probably continued to be processed more or less the old way.

-- Tom

-- Prof. Thomas H. Lee
Allen Ctr., Rm. 205
350 Jane Stanford Way
Stanford University
Stanford, CA 94305-4070
http://www-smirc.stanford.edu

On 2/20/2021 18:03, Gary Robert Bosworth wrote:
We used to design tunnel diodes into the trigger circuits of pulse
generators at Datapulse back in 1970. Even back then, we were aware of the
drift in tunnel diode characteriatics with age. Some circuit operational
characteristics would change radically over the years. It was due to the
heavy doping of the semiconductor materials. It is a shame that newly
manufactured tunnel diodes are no longer in production for replacement
purposes.

Gary


On Sat, Feb 20, 2021, 17:53 Charles <charlesmorris800@centurytel.net> wrote:

thanks, I was thinking of that (and possibly swapping Q145 for a 2N3904
too?). Hopefully it won't oscillate since those have a 250 MHz Ft, a lot
faster than the original Ge device. Could always put ferrite beads on the
leads, I suppose.

Q64 in the Staircase Inverter is also a 2N1516, so a tin whisker suspect
as well. It's a DC coupled stage and has a zero-level pot on the base, so
replacing with a Si PNP shouldn't present any problem either.















Tom Lee
 

There's a famous photo of the first IC from Fairchild -- a four-transistor flip-flop, in a round die! One of the contact pads is obviously overetched and barely there, and the chip is full of debris. Some years ago, Gordon Moore was at a conference (ISSCC) and I asked him about the photo. I learned that he was the one who took the photo, and also the one who overetched the pad. He said the only reason the photo was taken was to perform failure analysis. They didn't think that what they were doing was so historic as to merit photographic preservation, so that's the only photo taken of their first real IC. He said he's embarrassed that this photo is now in the Smithsonian's collection, forever enshrining his inattention to etch times.

I then asked him about the debris. He smiled and said, "Well, Noyce was a chain-smoker."

-- Cheers,
Tom

--
Prof. Thomas H. Lee
Allen Ctr., Rm. 205
350 Jane Stanford Way
Stanford University
Stanford, CA 94305-4070
http://www-smirc.stanford.edu

On 2/20/2021 19:38, Jim Ford wrote:
Oh, that's funny about cigarette ash getting into IC dice!  Lots more people smoked back then too, and indoors.  Fortunately for us nonsmokers, these days it's basically illegal to smoke indoors in California.  At the beginning of the year in 1990, I was working at the Lockheed Martin Skunk Works, and Lockheed Corp declared there would be no smoking indoors in any Lockheed building.  Very nice, and we all got used to it.  Later that year, in September, I started taking graduate microwave engineering courses at California State University, Northridge (several years before the big earthquake).  I was appalled to see students smoking in the hallways in the engineering building!  Gross!
Hard to imagine such gross (no pun intended) contaminants in ICs these days, with the cleanrooms and bunny suits and all.

Jim Ford


 

That's an amusing anecdote, but did you have to post it three times in a row? :)

Meanwhile, transistor changes didn't help. The staircase never starts (which is what resets the Q135/Q145 latch through R145 Swp Length), and the plate of the Miller tube is sitting at the bottom of the ramp, about 74 volts. There are a lot of carbon comp resistors that may have drifted high in value. Also several switching diodes that could have failed.
I'm tired and don't feel like working around circuits with 300 volt supplies... Will have another crack at it tomorrow!


Tom Lee
 

Very sorry for the multiple posts -- I kept on getting an error message, but apparently the messages got through, despite them.

Yes, it's best not to press onward with VT circuits when tired. Rest up and re-engage in battle tomorrow.

Tom

--
Prof. Thomas H. Lee
Allen Ctr., Rm. 205
350 Jane Stanford Way
Stanford University
Stanford, CA 94305-4070
http://www-smirc.stanford.edu

On 2/20/2021 20:29, Charles wrote:
That's an amusing anecdote, but did you have to post it three times in a row? :)

Meanwhile, transistor changes didn't help. The staircase never starts (which is what resets the Q135/Q145 latch through R145 Swp Length), and the plate of the Miller tube is sitting at the bottom of the ramp, about 74 volts. There are a lot of carbon comp resistors that may have drifted high in value. Also several switching diodes that could have failed.
I'm tired and don't feel like working around circuits with 300 volt supplies... Will have another crack at it tomorrow!




 

Having rested and studied the circuit description carefully, it appears that the pulses *should* go away at Q135 because it (and Q145) latch on during each staircase! So Q135 B-E junction is effectively a short.
The latch are reset when the staircase voltage rises sufficiently high, as I noted previously, which it's not doing.

I have also found that the Miller-stepping blocking oscillator is functioning (despite using a 2N2207 in a tin can). Pulses are available at the charge transfer caps (C157 or C158 depending on whether 10 or 100 dots/div selected).

Now my attention shifts to the Miller integrator section. Two of the charging diodes D161, D162 are 1N4152. The common 1N4148 actually has higher voltage/current ratings, but max 4 pf capacitance vs. 2 pf, probably why Tek speced them. But the most important one that actually passes the charge to increment the staircase (D160) is only listed as "152-0219-00 GaAs, Tek made". Sure hope it's not failed...

Another possibility is the disconnect diodes D152, D153, "Tek spec selected from 1N3579". I can't find a full data sheet (the Google entries lead to blind alleys) but on one site: "1N3579 SI-D 275V 0.15A". The reverse voltage doesn't get near that high, so not sure why (unless it's a transient during power-up).

Finally the 6688 Miller tube and its associated 6DJ8 could be at fault. The 6688 is replaceable with a Russian 6J9P which are relatively cheap. But I don't have one in the tube box.

So I will just have to work my way through each part of these closed loops and eventually find the failure. I hope.


 

One step forward, one back.
I quickly found that D152 had turned into a resistive lump of silicon (in both directions). Tacked in a 1N4148 and the sweep started up :)

I watched the voltage across it during warm-up and it does go to 160v or more, hence the rating. Surprised it didn't break down (4148's are rated for 100v). Anyway there is now a nice linear ramp coming out of the horizontal amp.
But I will have to use a better diode. Tempted to try a UF4007 which is fast enough but probably has more capacitance (17 pf) which I'm not sure I could adjust out. Or maybe a hack with two 1N4148 in series... Anyone got a 1N3579 handy? :)

After a few seconds, the trace bloomed and disappeared. Sure enough, the HV supply just quit on the 561A. Aaaaaarrrrrgh!!!

Power supplies are all good otherwise. Tubes are lit. Pulled the 12BH7A error amp, no HV still. The 27K 2W screen resistor to the oscillator tube appears to have zero volts on both ends... strange, now where is the 5.6K that is supposed to be in series with it to +300?

I hate having to fix my tools while I'm repairing something else!!


 

The 5.6k (carbon comp) resistor was indeed completely open. It looked perfect, no cracks or swelling, other than some whitish oxidation on the leads. Replaced it with a modern resistor and resumed work on the 3T77A.

The display is garbled - can't get it to trigger properly on a nice 6 MHz square wave from my 114 pulse generator. The "pull to sync" mode won't give me the 9 pulses before holdoff, either. It's either 2-3 pulses or it free runs with no holdoff.

Meanwhile, the trace decided to move vertically offscreen (down) and can't be brought back on. Pulling the invert switch on the 3S2 sends it offscreen (up). This is the behavior I first saw when the staircase & sweep weren't working at all. But they are fine, and the 3T77A is sending the proper (sampling strobe) pulses back to the 3S2. Whatever the latest problem, it doesn't appear to be in the time base plug-in (although I will still have an interesting time with the triggering before it will be useful). Could be the 3S2... or the S-2 head...

That's it for now - I have to go out of town for a week and I'm tired of messing with things that break while you're looking at them.


 

I took the S-1 sampling head apart again and discovered one reason it didn't work was that I'd gotten one of the 2N5769 B-E diodes backwards. Turned it around, still very noisy with no input. Decided not to waste any more time on transistor substitutes, since the MA4E1339 Schottky diodes had arrived from Digikey while I was traveling.

That was an interesting experience, working with packages not much larger than a pinhead. I finally cut small pieces of copper-clad board, cut the foil on both sides and with my smallest iron and tweezers, tacked the diodes (SOT-323 package) across the cut. Then tacked the diode + board to the sampler board, and used tiny pieces of copper foil to make short wide electrical connections. As a test, with 2.5 vdc through 1K, there was .690 or so volts across each diode pair. No reverse current. So far so good - no shorts, no opens.

Put it all together and stuck it in the 3S2. Of course the triggering is still screwed up in the 3T77A so I will eventually get back to work on that. In the meantime, the manual does say that a non-sampling timebase can be used at .1 ms/div or slower, so I flipped the internal slide switch on the 3S2 and stuck a 3B3 in the right-hand slot.

The S-1 head will indeed display the output of my 114 pulse gen (approx 17 ns risetime) through a 20 db pad. I stuck a tee at the input and ran a cable to my bench scope to verify the signal and it's correct. But the 561A/3S2/S-1 is showing a VERY slow rise and fall time, with about 8 or 9 dots at the edges, which is an indicated time of close to 100 us, not 17 ns! Something ain't right but I am not sure what it is. The manual says I should be able to use the 3S2 with full bandwidth of the sampling head, even though I can't display anything really fast.

Note that I have not yet tweaked the other adjustments in the S-1 (since I can't use my extender cable that's missing the trigger pickoff connectors. I'm still waiting for those push-on coax connectors to arrive in the mail. Canada to US is *really* slow these days). But I can't believe that the adjustments for bridge voltage, avalanche voltage, etc. on the S-1 could make THAT much difference. Especially since it's displaying the square wave from the pulse generator.

I'll upload a couple of pics to "Sampling with 3S2" album. Any thoughts?


 

I also found (in the 3S2 manual) one line that says, "The TRIG OUT signal is not useful for externally triggering a real-time time-base unit". Hmm. But it does not explain *why*. It certainly seems to be doing exactly that... and the 3S2 does not appear to be capable of internal triggering a sampling time-base like the 3T77A so the external line is definitely needed there.


Albert Otten
 

Hi Charles,

This is probably because the purpose usually is to display/measure rise time. A real-time time base needs a pre-trigger for this. If you have a stable repetitive signal then a sampling time base unit can display details of the rising edge of the next pulse after the trigger event.
For varying or jittery pulse periods you would in both cases be forced to view the edge of the trigger event itself, again requiring a pre-trigger. In for instance the 3S76 this is accomplished by delaying the signal after the trigger pick-off circuit.

Albert

On Sun, Feb 28, 2021 at 06:04 AM, Charles wrote:


I also found (in the 3S2 manual) one line that says, "The TRIG OUT signal is
not useful for externally triggering a real-time time-base unit". Hmm. But it
does not explain *why*. It certainly seems to be doing exactly that... and the
3S2 does not appear to be capable of internal triggering a sampling time-base
like the 3T77A so the external line is definitely needed there.


Albert Otten
 

Hi Charles,

I can easily imitate the "staircase" effect, see the picture Real-time.jpg which I added to your album https://groups.io/g/TekScopes/album?id=261250 .
Very likely your loop gain is far below unity, I estimate it will be about 0.1 . You can vary the loop gain with the front panel DOT RESPONSE knob, but probably not far enough to correct it all the way. It's probably due to a very low sampling efficiency of the S-1 with the new "diodes".
In real time mode the 3S2 samples at 10 us intervals. That's very fine compared with your sweep rate and the pulse top and bottom lengths. Hence you see the successive sampling results in sequence from left to right in the display. Because of the low dot response several samples are needed in order to "jump" from pulse top to pulse bottom and vice versa.

Albert

On Sun, Feb 28, 2021 at 04:34 AM, Charles wrote:

---
Put it all together and stuck it in the 3S2. Of course the triggering is still
screwed up in the 3T77A so I will eventually get back to work on that. In the
meantime, the manual does say that a non-sampling timebase can be used at .1
ms/div or slower, so I flipped the internal slide switch on the 3S2 and stuck
a 3B3 in the right-hand slot.

The S-1 head will indeed display the output of my 114 pulse gen (approx 17 ns
risetime) through a 20 db pad. I stuck a tee at the input and ran a cable to
my bench scope to verify the signal and it's correct. But the 561A/3S2/S-1 is
showing a VERY slow rise and fall time, with about 8 or 9 dots at the edges,
which is an indicated time of close to 100 us, not 17 ns! Something ain't
right but I am not sure what it is. The manual says I should be able to use
the 3S2 with full bandwidth of the sampling head, even though I can't display
anything really fast.