Topics

475A Chop Blanking and a Theory Question


 

I've been going through the performance check process with my recently repaired 475A, partly to get familiar with what will be needed to do a calibration, but also to verify that the device is in fact fixed, and I've run into a few snags. The one I'm concentrating on at the moment is the chop blanking: when I did the chop performance check I was clearly able to see the traces joining the channel 1 and channel 2 signals, so there's still a problem in the beam intensity system. I know that the problem isn't the beam intensity amplifier, because I don't see the horizontal retrace anymore, and that makes me suspect the chop blanking signal that feeds into the beam intensity amplifier.

So I'm looking at the vertical channel switching schematic, and I find this chain of inverters (U340) that feed the emitter of a PNP transistor (Q348) whose collector then feeds into the beam intensity amplifier, but whose base is connected to ground.

This is called a common base configuration?

It looks to me as if the signal coming out of U340 is forcing Q348 to sink current to ground from the beam intensity amp, but the diagram seems to indicate that the voltage on the collector side is -2.5 V, so I'm confused.

Maybe this isn't a problem for diagnosing or fixing this circuit, but I really don't understand how this is supposed to work, and that bothers me.

(DISCLAIMER: as I've said before, I'm almost a complete amateur in electronics, and transistors are still in the realm of myth and magic as far as my understanding of them goes)

-- Jeff Dutky


Dave Peterson
 

I'm a bit rusty on my discrete xtor education, but I believe Q348 should be passing the +5v output of the inverters.

The forward biasing of the Base-Emitter junction of a Bipolar Junction Transistor results in a Emitter to Collector current path, not Collector to Base.

Dave

On Thursday, December 17, 2020, 07:37:05 PM PST, Jeff Dutky <jeff.dutky@gmail.com> wrote:

I've been going through the performance check process with my recently repaired 475A, partly to get familiar with what will be needed to do a calibration, but also to verify that the device is in fact fixed, and I've run into a few snags. The one I'm concentrating on at the moment is the chop blanking: when I did the chop performance check I was clearly able to see the traces joining the channel 1 and channel 2 signals, so there's still a problem in the beam intensity system. I know that the problem isn't the beam intensity amplifier, because I don't see the horizontal retrace anymore, and that makes me suspect the chop blanking signal that feeds into the beam intensity amplifier.

So I'm looking at the vertical channel switching schematic, and I find this chain of inverters (U340) that feed the emitter of a PNP transistor (Q348) whose collector then feeds into the beam intensity amplifier, but whose base is connected to ground.

This is called a common base configuration?

It looks to me as if the signal coming out of U340 is forcing Q348 to sink current to ground from the beam intensity amp, but the diagram seems to indicate that the voltage on the collector side is -2.5 V, so I'm confused.

Maybe this isn't a problem for diagnosing or fixing this circuit, but I really don't understand how this is supposed to work, and that bothers me.

(DISCLAIMER: as I've said before, I'm almost a complete amateur in electronics, and transistors are still in the realm of myth and magic as far as my understanding of them goes)

-- Jeff Dutky


 

Dave,

I understand every word you said, individually.

The symbol for the transistor shows an arrow pointing to the base, that's a PNP transistor, right? And the emitter is the connection with the arrow, right?

Well, the connection with the arrow is connected to the output of the chain of inverters (U340, a 7400, elements A, B, and C), and the base is connected to ground. When the inverter signal is positive we have forward biased the emitter base junction, and most of the current will flow through from the emitter to the collector, right?

When the output of the inverter chain is low then we will (I hope) have the voltage on the emitter close enough to ground (it's TTL stuff, so it only really needs to drop below something like 1.5 V, I think) that the transistor will be in "cut off", right?

When the transistor is in "cut off" it's basically an open circuit, so it's not pulling the collector side in either direction: the collector side will do whatever it was already trying to do. This raises a further question: what is the purpose of Q348? It's not amplifying anything. Is it just there to provide a high impedance input to the Z-axis amplifier system? (I used a big word, "impedance" that I don't think I fully understand. I hope I used it right. I meant that Q348 seems to be there to make it very hard for the collector side to have any affect on the emitter side)

I'm also a little fuzzy on how the beam intensity signal works, but looking at the schematic, the beam intensity pot drives between -8V and +15V. I don't recall if -8V is high or low beam intensity.

My ability to read a schematic falls far short of what is required to reason about what is going on in the Z-axis amplifier, and how it will affect the mesh voltage in the CRT, but I think a high mesh voltage should suppress the beam.

I'm going to set up the 475A for investigation over the weekend. I figure that whatever I don't understand it theory will become clear in practice (I may not be able to read a schematic, but I can read a multimeter value or a scope trace just fine). This also gives me a chance to use my just-fixed 2236 for it's intended purpose.

-- Jeff Dutky


Dave Peterson
 

Yes, it's a PNP.

I think I see better now. The description for Z-axis amplifier for the 475 Service Manual, page 3-19 says:

"The current signals from the various control sources are connected to the emitter of Q1338 and the algebraic sum of the signals determines the collector conduction level."

One of those signals coming in is Ice (that's collector/emitter current) of Q348 in the chop blanking circuit.

I suspect the -2.5v in the schematics is a nominal voltage level, but the injection of current from Q348 will raise the voltage at the emitter of Q1338 probably cutting off the forward bias of it's base/emitter junction.

The wiki page for Bipolar Junction Transistors has a nice image of the current amplification that goes on in a BJT. The current from the forward biasing of the base/emitter junction triggers a large current between the collector and emitter. I think the purpose of Q348 is to provide current drive to the Z-axis amplifier Q1338. The output of U340 is likely not able to provide that current drive.

All this examination of these scope circuits is rekindling my fundamental circuits classes, but I know how much I don't remember. So I don't mean to speak as an expert. I'm sure there are plenty on this forum who can explain this better. I hope my explanations help.

Dave


Carsten Bormann
 

On 2020-12-18, at 05:40, Jeff Dutky <jeff.dutky@gmail.com> wrote:

what is the purpose of Q348
I didn’t look at the schematics, but your description sounds like it is a level shifter from the positive-to-ground TTL signal to a negative-to-ground input to the Z-axis amplifier.

»The current gain of a common-base amplifier is always less than 1. The voltage gain is a function of input and output resistances, and also the internal resistance of the emitter-base junction, which is subject to change with variations in DC bias voltage.«
https://www.allaboutcircuits.com/textbook/semiconductors/chpt-4/common-base-amplifier/

Grüße, Carsten


 

Well, what I can't solve by intellectual effort (a humbling experience for me), I have managed to solve by brute force (so to speak).

I checked Q348 with my multimeter's diode function and it appeared to be open between all pins (not what I expected) so I figured I had another blown transistor in the Z-axis circuit (which wouldn't have been surprising after what I found in the Z-axis amp). I pulled Q348 and tried it in my component tester, but it tested fine (identified as a PNP transistor, Hfe = 205, Uf = 678 mV), so I put it back. Then I buzzed out the circuit backwards from the emitter of Q348 toward U340 (the 4700). I found connectivity to R348, and R348 measured as the correct value (~330 ohms), but I could not find connectivity from R348 to any pin on U340 (R348 is supposed to be connected to pin #8 of U340).

Further inspection revealed that there was connectivity to the pad under pin #8 on U340, so I pulled U340, cleaned its pins with IPA (they were a bit dirty, but not terribly so), and reinstalled U340 in its socket. Now the CHOP blanking is working as expected.

On one level I am very pleased with my diagnostic skill, but on another level this feels too easy. Even if I'm not quite up to speed with the underlying circuit theory, I feel like I need more of a challenge. At the same time, I'm pretty sure that I'm not quite up to diagnosing some of the less well defined problems I have with my father's scopes (e.g. a "drifty" channel on the 2213 that seems to clear up if I keep the scope running for any period of time, or the odd wiggles in channel 2 on the 475 that were visible in the CHOP test).

I guess I'll just keep soldiering through the performance checks on the 475A and see what comes.

-- Jeff Dutky


Harvey White
 

Let's take a single inverter, put a resistor in series with the output (current limiting), and then connect it to the emitter of that PNP transistor whose base is grounded.  The collector goes through a resistor to a - voltage.

In this circuit, since the collector current is essentially the same as the emitter current, if we have one ma flowing through the emitter resistor, we have one ma flowing through the collector resistor.  The emitter base junction is essentially a diode, so it's easy to calculate the current.  Assuming that you have 0.7 volts VBE in the transistor, 5.0 volts at the inverter output (we're assuming "perfect" TTL, it will be lower), that's a difference of 4.3 volts.  With a 4.3 K resistor, you'll get 1 ma emitter current.  Run the collector supply to, say, -12 volts. Make the collector resistor 2K, and you get 2 volts/ma collector current.  In this example, we'd expect the collector voltage to be -10 volts (12-2).

So the first thing is that the circuit takes a positive voltage and converts it into a negative voltage.  The output current is the same as the input, and the voltage ratio between the input and output is the same as the emitter resistor's ratio to the collector resistor.  Now the fun can start.  Since you're driving essentially a diode to ground, it makes it very easy to figure the emitter current.  Suppose we add a second inverter with a different resistor, say 4.3K/2 or 2.15K.  That second inverter contributes exactly twice the current to the emitter/base junction as the first.  Since you're driving a diode, the two inverter outputs don't really interfere with each other.  So you can have

inv 1          inv 2           current              output voltage

off             off               0 ma                  -12 volts (transistor off)

on             off               1 ma                  -10 volts

off            on                2 ma                  -8 volts

on            on                3 ma                  -6 volts

The signals add without interfering with each other.  If I made the collector resistor a 1K, I'd be getting -12, -11, -10, and -9 volts, differences of 0,1,2,3 instead of differences of 0,2,4,6.

So this circuit could take any number of positive signals, and add them together, getting a negative going negative signal, which you could then treat whatever way you want.

It's a crude digital to analog converter.  It's accuracy is determined by the resistor precision, but most especially the voltages feeding each emitter resistor.  If you were going to make an accurate D/A converter, you'd want a different kind of gate family driving the emitter resistors.  For quick and dirty, TTL will work, though.

So the circuit is summing the output of all those inverters (or signals), contribution of each signal in proportion to the emitter resistors, and gives you an output scaled by the collector resistor and the total current.

There's not a major limit on how many signals can be summed this way.  This kind of circuit shows up in the horizontal stages in the 4xx series scope (A sweep and B sweep to horizontal amp), and in some video circuits for adding blanking, video, burst, and sync signals together.

Harvey

On 12/17/2020 11:52 PM, Dave Peterson via groups.io wrote:
Yes, it's a PNP.

I think I see better now. The description for Z-axis amplifier for the 475 Service Manual, page 3-19 says:

"The current signals from the various control sources are connected to the emitter of Q1338 and the algebraic sum of the signals determines the collector conduction level."

One of those signals coming in is Ice (that's collector/emitter current) of Q348 in the chop blanking circuit.

I suspect the -2.5v in the schematics is a nominal voltage level, but the injection of current from Q348 will raise the voltage at the emitter of Q1338 probably cutting off the forward bias of it's base/emitter junction.

The wiki page for Bipolar Junction Transistors has a nice image of the current amplification that goes on in a BJT. The current from the forward biasing of the base/emitter junction triggers a large current between the collector and emitter. I think the purpose of Q348 is to provide current drive to the Z-axis amplifier Q1338. The output of U340 is likely not able to provide that current drive.

All this examination of these scope circuits is rekindling my fundamental circuits classes, but I know how much I don't remember. So I don't mean to speak as an expert. I'm sure there are plenty on this forum who can explain this better. I hope my explanations help.

Dave





 

Because everything looks like a bandwidth measurement when all you have is a fast pulse generator I had been measuring the rise time of every scope I have to see if their bandwidth matched their specs, and before I fixed the CHOP blanking problem the bandwidth of the 475A was measuring as something like 175 MHz (rise time of about 2 ns), which seemed very wrong. After fixing the CHOP blanking problem, however, I went back and measured the rise time again, and this time I got a rise time of between 1.5 ns and 1.3 ns, which gives the bandwidth as 233-269 MHz, which seems about right for a 250 MHz scope.

I am at a loss to explain how fixing the CHOP feature could affect the bandwidth of the scope (especially since I didn't have CHOP engaged when I was measuring the rise time), and it's entirely possible that some amount of operator error may be involved, but, as I also measured the rise time of my father's 475 when I got the low bandwidth measurement for the 475A, and found the 475 to be spot on at 200 MHz, I suspect I made both measurements correctly (my notes, sadly, do not tell me enough about what how I set up the scopes to be sure).

I suppose that it's possible that another NAND gate in U340, which contributes to another part of the vertical amplifier system, also had a dirty pin that was causing a bad connection, but I don't see anything like that in the schematic. I'm much more inclined to believe that my boneheadedness (like leaving the 100MHz bandwidth limit pulled out), rather than a dirty pin on U340 was the culprit, but I'd be happy to be proven wrong.

-- Jeff Dutky


Dave Peterson
 

Sorry all, that last response was intended to be direct to Jeff, not the whole group. Disregard.
Dave


Tom Lee
 

Hi Dave,

RE: 465, 475 and 485

The 485 actually predates the other two, and is a very different animal. There was very little cross-pollination between the 485 team and the folks who did the other two. There are several nice features of the 485 that I wish were more commonly offered (e.g., a built-in fast pulse gen, and two levels of input protection that make it hard to blow up the front end). Every time the red light goes on, I know that I owe John Addis another beer, 'cuz he's just saved me from a blown scope for the nth time.

The 465 and 475 had some personnel in common, and a lot of informal collaboration. Those two scopes are similar enough that a good understanding of either of them will take you pretty far in understanding the other. Lots of stuff in common, too, making organ transplants feasible in more than a couple of cases.

-- 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 12/18/2020 16:09, Dave Peterson via groups.io wrote:
There are too many dimensions to all of this!
A) Ha-ha - ya got me! You don't have to get the first edition. It's just the sight of the red cover conjures night sweats.
B) I was looking at the Bodnar FPG. For the $$ I just have too many fundamental things to get first. Like appropriate 10x probes. Send me off into a whole other tread: vertical system calibration and circuit function. I know what I need to do, but I need two proper 10x probes to measure the preamp output. And just to have an appropriate high impedance probe just for poking around. Alligator clips and banana plugs are not appropriate tools for circuit analysis. I could spend several oscilloscopes worth of $$ on just basic bench equipment. I'm trying to prioritize and pace myself. I honestly am considering building my own FPG. Reading Leo's origin thread on EEVblog gives me ideas. Might be fun.
C) Bandwidth, math, transistors, amplifiers and filters: I want to help, but you're quite capable on your own. I also remember wanting to really grok xtor theory, and after getting into it in school I recall the mental rungs on the ladder of understanding. Again, layers on layers: discrete component topologies as applied in 1970 are not synchronous with deep sub-micron CMOS circuits. I'm not as fluent in Tektronix topologies, but I also do recognize a lot of basic BJT configurations. But then there's Tektronix's weird schematics - relative to my experience. The experienced guys on this forum will give better answers, but I'm re-learning myself and remember "the mysterious black box" that was a transistor. The BJT wiki has some really good descriptions and pictures that jive with those mental rungs. Anyway, there's BJT and FET physics, bias topologies, small signal models and analysis, frequency domain analysis, transfer functions, feedback, op-amps, ... There are a lot of facets to the things being done in these scopes. Compartmentalization and experience. Layers and layers.
D) Sorry, after playing electronic tech for 4 years in the Army: 99% of if is mechanical stupidity. But divide and conquer is the methodology, and I'd say you've got it pretty much down. Not as sexy as it seems from the outside. BTW, I just killed and resurrected my cheap Chinese function generator. I just stopped working a couple hours ago. I walked away in a bit of disgust, but after 5 minuets I went back and unscrewed the cover. After poking around online I got the courage to turn it back on and start checking some basic things - power etc. When I noticed a connector partially in it's socket. That's all it took. None of the voltage probing had anything to do with it. Just that it brought my focus to the innards of the box helping me spot the loose connector. Back in business.
E) I'm just beginning to dip my toes into the Trigger and Sweep circuit descriptions and calibration procedures. One thing I'm observing is a difference between my two scopes' triggering of chop signal observation. I want to figure out why they're behaving differently. The "working" scope has a very stable trigger, the "parts" scope is being a bit finicky. Then it occurred to me it would be helpful to see the chop signal without the blanking. Hmm. Wonder how I could do that?! Thanks to some guy on the TekScopes group I know just the transistor to pull to make that happen. Interesting.
I recall my struggle to understand BJT function, and I sense you have a mix of understanding and uncertainty. I'm re-examining the fundamentals I've gotten away from since being a CMOS jockey. If there are facets of xtor theory and operation that are frustrating you, and you can articulate them, I enjoy helping. I might say stupid things along the way, but I enjoy blundering my way into understanding. Let me know if there's something tripping you up.
F) Here's another one: 465, 475, 485. These are all concurrent designs from Tektronix. Yet they're so different in so many ways, while still being the same in a lot of ways. Did they have some mix of independent, common, and cooperative departments working on these? E.g. Your 475 ALT/CHOP circuit is so different than the 465. Why? Doesn't seem that significant a circuit to have such different designs while being produced at the same time? What's the R&D rationale behind that? Similar things with the 465 and 485 - though I haven't had the chance to delve too deep into it. Just rhetorical thoughts to share. Yet another thread/dimension. I am having fun tho!
Hope you're having fun. I got to have the day off to play today, and hope you're getting to enjoy some time off too. Thanks for sharing your thoughts and experiences. It's helping me as well.
Dave

On Friday, December 18, 2020, 03:11:18 PM PST, Jeff Dutky <jeff.dutky@gmail.com> wrote:
Because everything looks like a bandwidth measurement when all you have is a fast pulse generator I had been measuring the rise time of every scope I have to see if their bandwidth matched their specs, and before I fixed the CHOP blanking problem the bandwidth of the 475A was measuring as something like 175 MHz (rise time of about 2 ns), which seemed very wrong. After fixing the CHOP blanking problem, however, I went back and measured the rise time again, and this time I got a rise time of between 1.5 ns and 1.3 ns, which gives the bandwidth as 233-269 MHz, which seems about right for a 250 MHz scope.

I am at a loss to explain how fixing the CHOP feature could affect the bandwidth of the scope (especially since I didn't have CHOP engaged when I was measuring the rise time), and it's entirely possible that some amount of operator error may be involved, but, as I also measured the rise time of my father's 475 when I got the low bandwidth measurement for the 475A, and found the 475 to be spot on at 200 MHz, I suspect I made both measurements correctly (my notes, sadly, do not tell me enough about what how I set up the scopes to be sure).

I suppose that it's possible that another NAND gate in U340, which contributes to another part of the vertical amplifier system, also had a dirty pin that was causing a bad connection, but I don't see anything like that in the schematic. I'm much more inclined to believe that my boneheadedness (like leaving the 100MHz bandwidth limit pulled out), rather than a dirty pin on U340 was the culprit, but I'd be happy to be proven wrong.

-- Jeff Dutky








Dave Peterson
 

What got me about the 485 - I was perusing the 4-series scopes on TekWiki - was that it was released before the 465/475s, but had the separate B-sweep trace. But the Wiki mentions that this feature was on the 465B. I'd forgotten about that, and was probably one of the reasons I preferred the 465B when working with them back in the Army. It made me realize that there's this weird mix-n-match of components and features between them. I realize the 485 is a different animal, and I suspect the separate B-sweep trace is implemented in different ways. The 485 is a full dual-trace system.

After getting into the guts of the 465 directly, and the 475 indirectly, and now the 485 just via the TekWiki description, why is the 465 so extensively implemented in discrete components and lesser bandwidth than the earlier 485 (350MHz) and concurrent 475 (200MHz)? Cost? Size? Weight? All of the above? Sure, I'm sure Tektronix made a market analysis and product development plan. And that these solutions addressed expected markets. Anyone have sales numbers? I bet the 465 cost lest, sold more, and probably made Tek more money.

One of the things that I've realized about being a circuit designer, vs. a system designer, is a lack of market awareness and knowledge. It's fine, I'm not a marketing type, and I don't come at the engineering profession as a product solutions person. I admire people who have the inclination and insight to find and implement market solutions. My interest lies closer to the physics of things. But I do find the product development decisions fascinating. Nobody really builds this stuff for fun. They build it to make money!

Steering it back to the original thread, why are the 465 and 475 chop blanking circuits so different, yet so similar? Seems a product development optimization that didn't happen? But I well know being on the inside of product development for the past 30 years, "you go to war with the army you have, not the one you'd like to have". The reality is likely a mix of planning, accident, and circumstance. It fascinates me to consider I was such a kid pedaling around Beaverton with my friends who's dads were engineers in Tektronix struggling with all these developments and the associated stresses and occupations. Appreciating now the realities of their experiences I was ignorant to as a kid, looking forward to unwrapping my presents under the tree. Some things just never change!

Dave


Harvey White
 

I certainly can't give you an answer for that, but perhaps a bit of insight into how things might be done.  While I was only a "senior design engineer (i.e: I says do it like this and they does.....) in a few instances, in which cases I was the "ONLY" design engineer (more software than hardware, but still.....)

1) design guidelines come (in theory) from the senior hardware or software engineer.  The overall design concept, adherence to what the customer wanted, and compliance to their specifications) was the responsibility of the "senior systems engineer".  He made the decisions, and others followed them.  In theory, the software and hardware "lead" engineers had to show him that their designs matched the specifications.  Note that this was in a multi-layer military/government contract, so documentation, compliance, and at the very last, functionality, were important.  (yes, cynical I am, been there, saw that, tried to design around it).

2) design teams tend to use approaches that they know work, and hope that they fit the current situation.  I've seen examples of 1) it worked and we're fine and 2) well, that kinda does it....

I suspect that Tektronix had the same kind of outcome.  From what I've heard of HP designs, I think they did the same.

So it's likely that the different scopes did designs based on their specifications, what the designs were capable of, and only innovated when a design didn't meet the spec.  Tektronix may have had different limitations here, but there's likely to be an element of the "we do it like that" built into every product.

I've seen that the horizontal amplifier designs of the 468 (which I have) and the 465 seem to be quite similar.  Same design team? Possible.  I will say that I will reuse a design until it fails, then I rework it.  How much that applies to anyone else, and how much "they" are allowed to do that, well.....

Harvey

On 12/18/2020 8:59 PM, Dave Peterson via groups.io wrote:
What got me about the 485 - I was perusing the 4-series scopes on TekWiki - was that it was released before the 465/475s, but had the separate B-sweep trace. But the Wiki mentions that this feature was on the 465B. I'd forgotten about that, and was probably one of the reasons I preferred the 465B when working with them back in the Army. It made me realize that there's this weird mix-n-match of components and features between them. I realize the 485 is a different animal, and I suspect the separate B-sweep trace is implemented in different ways. The 485 is a full dual-trace system.

After getting into the guts of the 465 directly, and the 475 indirectly, and now the 485 just via the TekWiki description, why is the 465 so extensively implemented in discrete components and lesser bandwidth than the earlier 485 (350MHz) and concurrent 475 (200MHz)? Cost? Size? Weight? All of the above? Sure, I'm sure Tektronix made a market analysis and product development plan. And that these solutions addressed expected markets. Anyone have sales numbers? I bet the 465 cost lest, sold more, and probably made Tek more money.

One of the things that I've realized about being a circuit designer, vs. a system designer, is a lack of market awareness and knowledge. It's fine, I'm not a marketing type, and I don't come at the engineering profession as a product solutions person. I admire people who have the inclination and insight to find and implement market solutions. My interest lies closer to the physics of things. But I do find the product development decisions fascinating. Nobody really builds this stuff for fun. They build it to make money!

Steering it back to the original thread, why are the 465 and 475 chop blanking circuits so different, yet so similar? Seems a product development optimization that didn't happen? But I well know being on the inside of product development for the past 30 years, "you go to war with the army you have, not the one you'd like to have". The reality is likely a mix of planning, accident, and circumstance. It fascinates me to consider I was such a kid pedaling around Beaverton with my friends who's dads were engineers in Tektronix struggling with all these developments and the associated stresses and occupations. Appreciating now the realities of their experiences I was ignorant to as a kid, looking forward to unwrapping my presents under the tree. Some things just never change!

Dave





 

Tom Lee wrote:

There are several nice features of the 485 that I wish were more commonly offered (e.g., a built-in fast pulse gen,
and two levels of input protection that make it hard to blow up the front end). Every time the red light goes on,
I know that I owe John Addis another beer, 'cuz he's just saved me from a blown scope for the nth time.
Heaven knows that I don't need a 485, but you're making me really WANT one.

-- Jeff Dutky


Mlynch001
 

The 485 is a marvelous instrument. I was very fortunate to find a near pristine and mostly working example in Dallas in 2019. With a bit of restoration, it works like new.

--
Michael Lynch
Dardanelle, AR


 

Dave and Harvey,

I've been wondering about this specific difference between the 400 series scopes myself. I specifically bought two 2000 series scopes that had the ALT horizontal mode because I liked it better than the 475/475A's MIX mode, and wanted a scope (or two) that had that feature. The more that I looked at the 400 series models, however, the more it became obvious that, unlike the 2000 series, the 400 series scopes were largely independent designs executed over many years, often without seemingly any thought the economics of manufacture (reuse of knobs, for instance, or a common internal components). It also almost looks like I can discern some kind of logic behind which scopes implemented MIX and which implemented ALT, but I can't quite put it into words.

Tek changed their engineering culture with the 2000 series, which have almost all their components (frame, power supply, main board, sweep and vertical amp boards) in common and are designed specifically to cut manufacturing costs as much as possible. For the most part this resulted in better scopes (lighter, more reliable, easier to check during service) but there were some drawbacks (the press-on knobs were not durable, and having everything soldered down made some repairs harder). Still, when you look at the 2000 series models they all seem to be of a piece, even the ones that were produced years later (e.g. compare the 2213 to the 2236 and you can see that one just added stuff onto the other, for the most part) while the 400 series scopes look almost completely different (even scopes that are very similar have glaring differences: the 465 and 475 are nearly identical, but the V/DIV ranges are different, and the attenuator boards are quite different for little obvious benefit: did an extra 100 MHz in bandwidth really require the special PCB substrate?).

Anyhow, I don't actually have any knowledge about how or why the 400 scopes were built the way they were, but I've been reading a lot about them and had these observations.

-- Jeff Dutky


teamlarryohio
 

Tom, I always thought of the 485 as a 7904 / 7A19 / 7B92A shoehorned into a portable cabinet. It is truly a thing of beauty. One of the tweaks for setting up the front corner was to adjust how far the preamp chips were plugged into the sockets :-)

-ls-


Dave Peterson
 

Something that struck me about your U340 problems, and the oscilloscope development arc:

The U340 connectivity/continuity issue you had was evidently due to the plug-in nature of the component. As opposed to soldered. The Apollo Guidance Computer used welded connections. I'm sure other applications used various more robust component installation methods.

But mix that with the context of the times: uncertainty regarding reliability of silicon devices, the culture of reparability, the very rapid evolution of electronics as integrated circuits were ramping exponentially, mass production of greater and greater levels of integration. Not to mention competition. Wasn't HP horning in on Tek's market share around this time too? I was trying to allude to this somewhat previously. I can't imagine how frightening it was to work at Tek in the 70's and 80's. I feel for those guys.

I find the 400 series scopes a fascinating glimpse into the world of the early 70's. I think that's why I like them so much. Those were my "wonder years". I wonder how stressful they were for the men and families around me. I too am casting lustful glances at the 485. And the 465B. I'm intrigued too by the 2000's, and the 5k and 7k mainframes.

Dave


 

Dave,

Don't forget that semiconductor devices can be damaged by too much heat. I think it's pretty likely that the wave soldering methods in use in the early 70s might have used higher temperatures than these parts could tolerate, so socketing was absolutely mandatory. But, man, that must have made the 400 series horribly expensive to manufacture (all those tiny components being inserted on each unit by hand). When it came time to design the 2000 series the manufacturing techniques had likely improved to the point that soldering semiconductor components directly to the PCB was doable without worry of overheating.

The 2000 series are pretty primitive in their early incarnations (the 2213 and 2215) and the 2200s never really move too far beyond that in some ways (they never get the rear panel outputs or the external B trigger input, for example), but they are generally nice scopes, especially if you have to lug the things around: they weigh half of what the 400 series weigh. The only real shortcoming of the 2200 series is that they never broke 100 MHz; that was left for the 2400 series scopes, which had a lot more "black magic" in their construction. Of course that same shortcoming of the 2200 series also has advantages: few, if any custom ICs to worry about.

If you're looking for a nice 2000 series scope I would suggest the 2236/2236A, which has a nice DMM/CT module whose readout is a separate vacuum fluorescent display. I got one pretty cheap off eBay, and it was pretty easy to diagnose and fix. The CT module uses a 10 MHz OCXO (at least mine does, I think that's called "option 14"), and it seems to still be spot on. Otherwise it's a lot like the 465/475 with DM44: same dual delayed timebase, some physical rotary switches, same vertical sensitivity range as the 475 (2 mV - 50 V) and the same timebase range as the 465 (.5 s - 0.05 us). The DMM/CT module is based on a Motorola 6802; a processor family for which I have a sweet spot.

If you get a 2236 you'll want a probe with the 10x readout pin to use for CH 1 which feeds the DMM. Without the readout pin there's no way to tell the DMM that you're using a 10x probe, and the voltage readings will be wrong. I splashed out for a P6121 on eBay, which was the probe designed for this scope, but that's because I'm some kind of nostalgic fool who wants everything to look "original". It was cheaper than a modern probe with readout pin, so I can't really complain.

The 2220/2221/2230/2232 are interesting, but I almost feel like getting a digital storage scope of this age is pointless. The sample rate is nowhere near the analog bandwidth of the scope, and the memory depth is a joke compared to modern DSOs (that won't stop me from getting a 468 to play around with, but it's stopped me from pulling the trigger on 2230 or 2232). I figure that whatever I don't spend on a 2230/2232 is money I can put toward a nice, modern DSO, which will round out my bench quite nicely.

The 7000 series mainframes are fascinating, as are the 11000 series scopes, but, again, I think that the 475 is well in excess of what I need, and the 7000 series still fetch real money, if the eBay prices are any guide. I've seen some 11000 series scopes on eBay for very reasonable prices, considering what those scopes are capable of, but a man should know his limits. I'm almost tempted to get a 11403 just to be able to say that I have a 1 GHz scope, but, again, I know my limits.

Also, like you, my interest lies firmly in the scopes of the seventies and early 80s, before everything went to microcomputers, rotary encoders, soft buttons and on-screen menus (and, heaven forbid, Windows! *shudder*). I'll put a modern DOS on my bench at some point, but I won't enjoy it in the same way that I enjoy the 475.

-- Jeff Dutky


Tom Lee
 

The 465 and 485 targeted different markets. The 485 was aimed at the "gotta have the highest bandwidth in a portable, price is no object" market. I once knew the prices at introduction, but I don't any longer. It was certainly quite a bit more expensive than the 465 (maybe a factor of e? I dunno). To get the 350MHz bandwidth, they used two different paths in the preamp, each optimized for the 1M and 50 ohm impedances, rather than having just one high-Z path and terminating the input.

There is also, as you would expect, an extensive use of bridged T-coils. Even with that magic, John Addis and Wink Gross realized that an all-discrete implementation would not get there; you could not get the parasitic inductance low enough. Addis designed a custom IC (the M84) to overcome that problem, and even exploited bondwire-to-bondwire coupling to implement part of the T-coil internally. There are also lots of external adjustments to make the vertical amp clean over such a large bandwidth. Until the 2465 showed up, the 485 had the highest bandwidth of any portable scope. Even then, the bandwidth edged up only another 50MHz, even after 15 years of Moore's law working its magic. To save space, the delay was implemented with helical lines, wrapped around the crt's neck. Because of the peculiar nature of such lines (high frequencies can take a short cut across turns, so are delayed less than the lower-frequency components), a special equalization network --
implemented with more T-coils -- is needed to flatten the delay curve. The sub-1ns rise time source and input protection circuits are conveniences that you appreciate even more when you have to go back to a scope that doesn't have these features. I not infrequently use the fast source to do quick checks of slower scopes -- it's not only fast, it's very clean. And the switching supply saves enough weight to allow the jam-packed 485 to remain portable. The 485 really was ahead of its time.

The 465 aimed for more modest performance at a more modest price. It hit a major sweet spot, marketing-wise, and until the 2465, it was the best selling portable (both in total units sold, and in total revenue generated). It's sort of the VW Bug of scopes. Because it didn't target bleeding-edge performance, it could use many more off-the-shelf components, including a conventional delay line (bulkier, but didn't need a delay equalizer and its attendant adjustments), and a conventional power supply. The relative absence of unobtainium custom parts makes the 465 eminently repairable.

I love both models (and the 475). I've learned a tremendous amount from studying them. There's much there that you would be hard-pressed to find in textbooks.

--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 12/18/2020 17:59, Dave Peterson via groups.io wrote:
What got me about the 485 - I was perusing the 4-series scopes on TekWiki - was that it was released before the 465/475s, but had the separate B-sweep trace. But the Wiki mentions that this feature was on the 465B. I'd forgotten about that, and was probably one of the reasons I preferred the 465B when working with them back in the Army. It made me realize that there's this weird mix-n-match of components and features between them. I realize the 485 is a different animal, and I suspect the separate B-sweep trace is implemented in different ways. The 485 is a full dual-trace system.

After getting into the guts of the 465 directly, and the 475 indirectly, and now the 485 just via the TekWiki description, why is the 465 so extensively implemented in discrete components and lesser bandwidth than the earlier 485 (350MHz) and concurrent 475 (200MHz)? Cost? Size? Weight? All of the above? Sure, I'm sure Tektronix made a market analysis and product development plan. And that these solutions addressed expected markets. Anyone have sales numbers? I bet the 465 cost lest, sold more, and probably made Tek more money.

One of the things that I've realized about being a circuit designer, vs. a system designer, is a lack of market awareness and knowledge. It's fine, I'm not a marketing type, and I don't come at the engineering profession as a product solutions person. I admire people who have the inclination and insight to find and implement market solutions. My interest lies closer to the physics of things. But I do find the product development decisions fascinating. Nobody really builds this stuff for fun. They build it to make money!

Steering it back to the original thread, why are the 465 and 475 chop blanking circuits so different, yet so similar? Seems a product development optimization that didn't happen? But I well know being on the inside of product development for the past 30 years, "you go to war with the army you have, not the one you'd like to have". The reality is likely a mix of planning, accident, and circumstance. It fascinates me to consider I was such a kid pedaling around Beaverton with my friends who's dads were engineers in Tektronix struggling with all these developments and the associated stresses and occupations. Appreciating now the realities of their experiences I was ignorant to as a kid, looking forward to unwrapping my presents under the tree. Some things just never change!

Dave




Dave Peterson
 

That's awesome Tom. The TekWiki has a history section, but I don't think it gets into this kind of product history. Another thing I haven't had the opportunity to dig into. Thanks for this little bit of extra insight.
Dave

On Friday, December 18, 2020, 11:12:11 PM PST, Tom Lee <tomlee@ee.stanford.edu> wrote:

The 465 and 485 targeted different markets. The 485 was aimed at the
"gotta have the highest bandwidth in a portable, price is no object"
market. I once knew the prices at introduction, but I don't any longer.
It was certainly quite a bit more expensive than the 465 (maybe a factor
of e? I dunno). To get the 350MHz bandwidth, they used two different
paths in the preamp, each optimized for the 1M and 50 ohm impedances,
rather than having just one high-Z path and terminating the input.

There is also, as you would expect, an extensive use of bridged T-coils.
Even with that magic, John Addis and Wink Gross realized that an
all-discrete implementation would not get there; you could not get the
parasitic inductance low enough. Addis designed a custom IC (the M84) to
overcome that problem, and even exploited bondwire-to-bondwire coupling
to implement part of the T-coil internally. There are also lots of
external adjustments to make the vertical amp clean over such a large
bandwidth. Until the 2465 showed up, the 485 had the highest bandwidth
of any portable scope. Even then, the bandwidth edged up only another
50MHz, even after 15 years of Moore's law working its magic. To save
space, the delay was implemented with helical lines, wrapped around the
crt's neck. Because of the peculiar nature of such lines (high
frequencies can take a short cut across turns, so are delayed less than
the lower-frequency components), a special equalization network --
implemented with more T-coils -- is needed to flatten the delay curve.
The sub-1ns rise time source and input protection circuits are
conveniences that you appreciate even more when you have to go back to a
scope that doesn't have these features. I not infrequently use the fast
source to do quick checks of slower scopes -- it's not only fast, it's
very clean. And the switching supply saves enough weight to allow the
jam-packed 485 to remain portable. The 485 really was ahead of its time.

The 465 aimed for more modest performance at a more modest price. It hit
a major sweet spot, marketing-wise, and until the 2465, it was the best
selling portable (both in total units sold, and in total revenue
generated). It's sort of the VW Bug of scopes. Because it didn't target
bleeding-edge performance, it could use many more off-the-shelf
components, including a conventional delay line (bulkier, but didn't
need a delay equalizer and its attendant adjustments), and a
conventional power supply. The relative absence of unobtainium custom
parts makes the 465 eminently repairable.

I love both models (and the 475). I've learned a tremendous amount from
studying them. There's much there that you would be hard-pressed to find
in textbooks.

--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 12/18/2020 17:59, Dave Peterson via groups.io wrote:
What got me about the 485 - I was perusing the 4-series scopes on TekWiki - was that it was released before the 465/475s, but had the separate B-sweep trace. But the Wiki mentions that this feature was on the 465B. I'd forgotten about that, and was probably one of the reasons I preferred the 465B when working with them back in the Army. It made me realize that there's this weird mix-n-match of components and features between them. I realize the 485 is a different animal, and I suspect the separate B-sweep trace is implemented in different ways. The 485 is a full dual-trace system.

After getting into the guts of the 465 directly, and the 475 indirectly, and now the 485 just via the TekWiki description, why is the 465 so extensively implemented in discrete components and lesser bandwidth than the earlier 485 (350MHz) and concurrent 475 (200MHz)? Cost? Size? Weight? All of the above? Sure, I'm sure Tektronix made a market analysis and product development plan. And that these solutions addressed expected markets. Anyone have sales numbers? I bet the 465 cost lest, sold more, and probably made Tek more money.

One of the things that I've realized about being a circuit designer, vs. a system designer, is a lack of market awareness and knowledge. It's fine, I'm not a marketing type, and I don't come at the engineering profession as a product solutions person. I admire people who have the inclination and insight to find and implement market solutions. My interest lies closer to the physics of things. But I do find the product development decisions fascinating. Nobody really builds this stuff for fun. They build it to make money!

Steering it back to the original thread, why are the 465 and 475 chop blanking circuits so different, yet so similar? Seems a product development optimization that didn't happen? But I well know being on the inside of product development for the past 30 years, "you go to war with the army you have, not the one you'd like to have". The reality is likely a mix of planning, accident, and circumstance. It fascinates me to consider I was such a kid pedaling around Beaverton with my friends who's dads were engineers in Tektronix struggling with all these developments and the associated stresses and occupations. Appreciating now the realities of their experiences I was ignorant to as a kid, looking forward to unwrapping my presents under the tree. Some things just never change!

Dave