Least expensive but reliable used-market scope with at least 1 GHz?

[Howardlong]

It sounds like you have a similar intrigue that I do into insanely tiny time.

When I first built computers in the 70s, 450ns was the typically static RAM access time, and at the time we were amazed at that.

Although I do plenty in RF mostly in S band and below (although I have been known to go up to K band experimentally), pretty much everything is measured in the frequency domain.

Much more recently, I have become more interested in looking at stuff like transmission lines in the time domain as well (as well as the frequency domain with VNAs as I traditionally have done), and it gives an interesting, and different, perspective.

What I would say is that if you did go to a 1GHz bandwidth scope from a 500MHz one and are trying to measure ~300ps rise times such as with the JW pulse generator, there is a difference, but it's nowhere near as much as you'll get with a sampling scope in the 10s of GHz. What I would recommend, if you go for a real time scope in the low GHz is to get one that works in equivalent time too. Neither interpolation nor fixed, limited period sampling give a faithful reproduction of what's happening, there will inevitably be some bastardisation, but with ETS you can avoid the interpolation and sampling period limitations, and will generally get a fair bit closer on repetitive signals.

Even with ETS, typically the next problem to deal with is front end analogue artefacts like overshoot and/or ringing that often might only exist inside the scope! One way to show what's happening is to shoot a step from a TDR such as that in a sampling scope set up and look at the response when it hits a 1GHz scope's BNC, it's rarely pretty.

A high speed sampling scope is not the same as a regular day to day scope, almost like a spectrum analyser isn't like a regular scope or a kettle isn't an oven, they are different tools for different jobs. My guess is that if, like me, you invest in a 1GHz scope after using 500MHz units, it's just won't be enough.... you have been warned!

[KE5FX]

I appreciate all the advice here - some of the recommended scopes would add more functionality but this thread helped me realize that I'm sitting on just a few incremental pieces short of what I need for pretty fast risetime viewing with the Tek7000.

Don't know if anyone has suggested the Tektronix TDS 694C.  It's a good way to get all the way to 3 GHz without resorting to sampling.  They can sometimes be had for around $1K, and it's a good idea to buy another one as a parts mule if possible.

They aren't great 3 GHz scopes since they "only" sample at 10 Gs/s, but they get the job done, and they are extremely good at 1 GHz.  They are new enough that they don't suffer from the infamous Tek capacitor plague.

Trigger chip failures are very common, but strangely enough, as I recently learned, the 694C has a couple of spares on the board that are soldered in but unused.  So you can save some bucks by buying one with a triggering problem and swapping the chips. 

True to Tektronix's usual balls-to-the-wall engineering practice, you need to use a fan on the acquisition board when servicing it with the power on.  A lot of people didn't know/care about that, and my personal suspicion is that this is what causes a lot of the triggering failures.     Point being, these scopes are somewhat cheaper than they really ought to be, considering how good they are.

Edit: they do require 50-ohm probes, which is something else you have to budget for.  This is true of any scope once you go above 500 MHz, where high-impedance passive probes become useless.  Tek's active probes, like the P6245, are ideal.

[Electro Fan]

It sounds like you have a similar intrigue that I do into insanely tiny time.

When I first built computers in the 70s, 450ns was the typically static RAM access time, and at the time we were amazed at that.

Although I do plenty in RF mostly in S band and below (although I have been known to go up to K band experimentally), pretty much everything is measured in the frequency domain.

Much more recently, I have become more interested in looking at stuff like transmission lines in the time domain as well (as well as the frequency domain with VNAs as I traditionally have done), and it gives an interesting, and different, perspective.

What I would say is that if you did go to a 1GHz bandwidth scope from a 500MHz one and are trying to measure ~300ps rise times such as with the JW pulse generator, there is a difference, but it's nowhere near as much as you'll get with a sampling scope in the 10s of GHz. What I would recommend, if you go for a real time scope in the low GHz is to get one that works in equivalent time too. Neither interpolation nor fixed, limited period sampling give a faithful reproduction of what's happening, there will inevitably be some bastardisation, but with ETS you can avoid the interpolation and sampling period limitations, and will generally get a fair bit closer on repetitive signals.

Even with ETS, typically the next problem to deal with is front end analogue artefacts like overshoot and/or ringing that often might only exist inside the scope! One way to show what's happening is to shoot a step from a TDR such as that in a sampling scope set up and look at the response when it hits a 1GHz scope's BNC, it's rarely pretty.

A high speed sampling scope is not the same as a regular day to day scope, almost like a spectrum analyser isn't like a regular scope or a kettle isn't an oven, they are different tools for different jobs. My guess is that if, like me, you invest in a 1GHz scope after using 500MHz units, it's just won't be enough.... you have been warned!

Howard, if you or others here would be up for an educational post on the fundamentals of ETS vs. the alternative(s) and anything significantly related - either here or in a fresh thread - that would be good.  I think there is an entire layer of distinctions I'm not really seeing much less understanding.  Thanks, EF

[Electro Fan]

I appreciate all the advice here - some of the recommended scopes would add more functionality but this thread helped me realize that I'm sitting on just a few incremental pieces short of what I need for pretty fast risetime viewing with the Tek7000.

Don't know if anyone has suggested the Tektronix TDS 694C.  It's a good way to get all the way to 3 GHz without resorting to sampling.  They can sometimes be had for around $1K, and it's a good idea to buy another one as a parts mule if possible.

They aren't great 3 GHz scopes since they "only" sample at 10 Gs/s, but they get the job done, and they are extremely good at 1 GHz.  They are new enough that they don't suffer from the infamous Tek capacitor plague.

Trigger chip failures are very common, but strangely enough, as I recently learned, the 694C has a couple of spares on the board that are soldered in but unused.  So you can save some bucks by buying one with a triggering problem and swapping the chips. 

True to Tektronix's usual balls-to-the-wall engineering practice, you need to use a fan on the acquisition board when servicing it with the power on.  A lot of people didn't know/care about that, and my personal suspicion is that this is what causes a lot of the triggering failures.     Point being, these scopes are somewhat cheaper than they really ought to be, considering how good they are.

Edit: they do require 50-ohm probes, which is something else you have to budget for.  This is true of any scope once you go above 500 MHz, where high-impedance passive probes become useless.  Tek's active probes, like the P6245, are ideal.

I'm sure I'm comparing apples to watermelons, but what is the significance of the Tek S-4 spec that says it has a bandwidth of 14.5 GHz?  I realize the sampling head is just part of the Tek 7000 architecture but what will that 14.5 GHz do or not do vs. the  3 GHz and 10Gs/s of the TDS694C?  Thanks

[Electro Fan]

The plot thickens.  I noticed that the standard Tek configuration refers to the use of a pulse generator with a trigger out.  In my case I just have my two fairly simple JW pulse gens - neither of which has a trigger out.

I'm trying to digest the Tek manual (it seems to offer some alternative setups) but if anyone has any suggestions on the easiest way to connect and measure a JW pulser (without a trigger out) to the Tek 7000/Tek 7S11 with Tek S-X/Tek 7T11A please let me know.  Thanks

http://exodus.poly.edu/~kurt/manuals/uploads/TEK%207T11A.pdf
- see discussion on 2-6 under "Observing a Fast-Rise Pulse"

[KE5FX]

I'm sure I'm comparing apples to watermelons, but what is the significance of the Tek S-4 spec that says it has a bandwidth of 14.5 GHz?  I realize the sampling head is just part of the Tek 7000 architecture but what will that 14.5 GHz do or not do vs. the  3 GHz and 10Gs/s of the TDS694C?  Thanks

If you know you're always going to be viewing repetitive signals that will look the same at every trigger, then a sampling scope can be a good way to go.  But they are finicky to set up and use, and there are some things they either can't do at all or aren't very good at.  There's a deck here that covers the major differences.  The 7000-series samplers are very finicky to use, lacking any built-in intelligence for setup and triggering.

The 694C is a real-time scope that can capture up to 100K points (IIRC) at 10 GS/s in one single-shot acquisition.  So while you could use a sampling scope to look at an eye diagram, for instance, you could use a 694C to look at specific glitches in a data stream.  It's an ordinary scope just like any other, except that its 3 dB bandwidth is 3 GHz.  Kind of a jittery display, but it's a wonder they could do it at all in the early 1990s.

[Electro Fan]

There's a deck here that covers the major differences.

That was a very good deck - very informative - I hadn't previously been able to understand how the relatively slower sampling rate could sample the faster signal.  Both approaches seem to have good merit - just depends on the functionality required and the affordable budget.  Thanks for the link.

[Electro Fan]

Just re-posting in case anyone has any thoughts on how to deal with triggering for this.... Thx

The plot thickens.  I noticed that the standard Tek configuration refers to the use of a pulse generator with a trigger out.  In my case I just have my two fairly simple JW pulse gens - neither of which has a trigger out.

I'm trying to digest the Tek manual (it seems to offer some alternative setups) but if anyone has any suggestions on the easiest way to connect and measure a JW pulser (without a trigger out) to the Tek 7000/Tek 7S11 with Tek S-X/Tek 7T11A please let me know.  Thanks

http://exodus.poly.edu/~kurt/manuals/uploads/TEK%207T11A.pdf
- see discussion on 2-6 under "Observing a Fast-Rise Pulse"

[edavid]

Just re-posting in case anyone has any thoughts on how to deal with triggering for this.... Thx

You misread that, you don't need a pretrigger to measure a pulse generator's risetime.

(In some situations, it's handy to have a DL11 or 7M11 though.)

[Electro Fan]

Just re-posting in case anyone has any thoughts on how to deal with triggering for this.... Thx

You misread that, you don't need a pretrigger to measure a pulse generator's risetime.

(In some situations, it's handy to have a DL11 or 7M11 though.)

Thanks, so the diagram is just showing an optional triggering capability?  I guess the 7T11 and 7S11 talk to each other via some channel on their adjacent location in the mainframe and triggering happens manually with 7T11 trig level knob?  Thx

[EV]

Thanks, so the diagram is just showing an optional triggering capability?  I guess the 7T11 and 7S11 talk to each other via some channel on their adjacent location in the mainframe and triggering happens manually with 7T11 trig level knob?  Thx

7T11 and 7S11 must be sibe by side. They have connectors there and are connected to each other.

As I told earlier it is easyer to use if you have 70 ns pretriggered pulse connected to ext-trigger. I think in your picture is Tek 284 pulse generator. It has this kind pretriggered pulse.

If you are going to see JW pulse with these plugins you are in trouble. The pulse must be attenuated at least 20 dB to be seen on the tube screen. You must use internal trigger because you dont have pretriggered pulse. The JW pulse is so narrow that there can be difficulties to get triggering work. It will not be easy to see the pulse. Anyway it is better to start practice with something like 1 MHz square wave signal with 250 mV amplitude.

[Electro Fan]

Thanks, so the diagram is just showing an optional triggering capability?  I guess the 7T11 and 7S11 talk to each other via some channel on their adjacent location in the mainframe and triggering happens manually with 7T11 trig level knob?  Thx

7T11 and 7S11 must be sibe by side. They have connectors there and are connected to each other.

As I told earlier it is easyer to use if you have 70 ns pretriggered pulse connected to ext-trigger. I think in your picture is Tek 284 pulse generator. It has this kind pretriggered pulse.

If you are going to see JW pulse with these plugins you are in trouble. The pulse must be attenuated at least 20 dB to be seen on the tube screen. You must use internal trigger because you dont have pretriggered pulse. The JW pulse is so narrow that there can be difficulties to get triggering work. It will not be easy to see the pulse. Anyway it is better to start practice with something like 1 MHz square wave signal with 250 mV amplitude.

The 284 looks like a very nice piece of equipment but it is out of my justifiable budget.  So, the question is, can a fairly rapid pulse with a rise time of less than 500ps be captured by the 7S11/7T11 combo with one of the Tek S-X sampling heads?  According to the specs even the S-1 (at 350ps) should be able to do it. 

I realize that some JW pulse gens produce a very narrow "peaky" pulse - I have one of those (shown in the first photo) and I'd like to figure out how fast it's rise time really is - which is what started me on this mission.  I also have a pulse generator (on a board made by Tektronix) that has a flatter top on the pulse - it is shown in the second and third photos.  It appears to produce about a 20ns wide pulse, so presumably the Tek 7000 sampling system should be able to pick this up in terms of the pulse duration.  Supposedly, the Tek Pulse Gen board is capable of a rise time of <200ps - so this is what I'd like to see with the 7S11/7T11.

What I am a little nervous about is exceeding the voltage capacity of the sampling system.  My peaky pulse gen does seem to produce a pulse that is on the order of 7-8 volts Pk-PK (I'm not sure why or if it is operating properly).  But the Tektronix pulse gen seems to produce a pulse that is well under 1 Volt Pk-Pk (less than 350mV), so I'm thinking that should work.

See any reason why the pulse shown in the second and third photos won't be displayable on the Tek 7000 (S711/S-1/7T11) sampling system?  The only reason I see that it might not work is if there is an issue with the triggering.  If I understood edavid's post he seems to think it will trigger ok.

Thanks for the help/advice.

[Howardlong]

EF: I did a video showing the 54120B/54121A sampling scope combo and its TDR step generator for use in measuring the rise times of assorted scopes, including switching between real time and ETS on scopes that supported it.

[Berni]

I built one of those Jim Williams pulse gens. The transistor i used in it seamed to not want to reverse breakdown anywhere near where the datasheet said it would so i ended up running the circuit from over 100V DC and that made 30V output pulses that i had to run trough an attenuator to get in to my scope. I got 160ps rise out of mine in the screenshot. It appears that this is about as fast as JW pulse gens go. Enugh to test the bandwidth of scopes up to 1GHz but past that not so much.

The pulses are indeed very short unless you use some trick to extend them (I seen a coax cable transmission line used for that). I would guess a sampling scope would have a lot of trouble triggering off this.

[Electro Fan]

EF: I did a video showing the 54120B/54121A sampling scope combo and its TDR step generator for use in measuring the rise times of assorted scopes, including switching between real time and ETS on scopes that supported it.

Howard, that is an EXCELLENT video!  The star of the show is a toss-up but I think the 54120B/54121A comes in second and you come in first!    (The performance of the 54120B/54121A is really something wonderful.) 

Dave, if you happen to catch Howard's video it might be worth featuring.  Between the various scopes/test equipment, the few tens of picosecond performance, and the education..... it's delightful! 

EF

[Howardlong]

EF: many thanks.

I am pretty new to this, so there may be some corrections that come out. I am having second thoughts on the overshoot on the two Agilents, it may be largely due to a designed filter response rather than limitations on the analogue side or impedance bumps.

[Electro Fan]

I built one of those Jim Williams pulse gens. The transistor i used in it seamed to not want to reverse breakdown anywhere near where the datasheet said it would so i ended up running the circuit from over 100V DC and that made 30V output pulses that i had to run trough an attenuator to get in to my scope. I got 160ps rise out of mine in the screenshot. It appears that this is about as fast as JW pulse gens go. Enugh to test the bandwidth of scopes up to 1GHz but past that not so much.

The pulses are indeed very short unless you use some trick to extend them (I seen a coax cable transmission line used for that). I would guess a sampling scope would have a lot of trouble triggering off this.

Yes, I think the JW pulsers with the peaky pulse are challenging - both because of the short/peaky duration and also possibly due to the fairly high voltage - although it looks like you got it captured on your scope.  It's my understanding that Jim Williams also designed another version of the pulse generator that had a wider pulse.

For what I have in mind I'm inclined to use the pulse gen built into a Tektronix demonstration board; think the >20ns wide pulse will ease the measurement process.

[Howardlong]

I built one of those Jim Williams pulse gens. The transistor i used in it seamed to not want to reverse breakdown anywhere near where the datasheet said it would so i ended up running the circuit from over 100V DC and that made 30V output pulses that i had to run trough an attenuator to get in to my scope. I got 160ps rise out of mine in the screenshot. It appears that this is about as fast as JW pulse gens go. Enugh to test the bandwidth of scopes up to 1GHz but past that not so much.

The pulses are indeed very short unless you use some trick to extend them (I seen a coax cable transmission line used for that). I would guess a sampling scope would have a lot of trouble triggering off this.

It's not only the trigger pulse width, it's also that the gap between pulses is very long and not at all stable, so you'll probably need to see the trigger pulse itself. As well as probably needing to extend the pulse (or get a 54118A) you'd need a delay line such as HP 54008A or the Tek equivalent to allow you to see the trigger. Note that by their very nature, delay lines will affect your signal to varying degrees.

[dom0]

Am I the only one who thinks that you can't really measure a meaningful rise time with a signal that never reaches a reference level (i.e. flat top)? You'd have to measure with something that's guaranteed to be fast enough the pulse height first...

[Electro Fan]

Am I the only one who thinks that you can't really measure a meaningful rise time with a signal that never reaches a reference level (i.e. flat top)? You'd have to measure with something that's guaranteed to be fast enough the pulse height first...

I don't think you are the only one.  I agree with what you are saying.

Unless we are sure the measuring system is capable of performing better/faster than the DUT I don't see how we can depend on the measurement.   I'm not sure that I know how wide a pulse needs to be for a scope to determine where the top is, but I believe that even if a pulse is wide enough, if the pulse's rise time is faster than the scope's rise time then what scope is going to show is largely it's own limiting rise time.  I think we saw this in Howard's video. 

So there are two issues here:  One issue is whether the scope's rise time is fast enough to reveal the DUT's rise time.  The second issue, which is related to the first, is whether the pulse from the DUT is wide enough (long enough in duration) for the scope to effectively find the top of the pulse.

I could be wrong about this, but that's the way I understand it so far.

[EV]

Here is the best I could do with my Free-Electrons JW pulser. I used 20 dB attenuator, internal triggering and random sampling. Scope is Tek R7103 with 7T11 and 7S11 (S-4) pluins.

[Berni]

I agree it should be a proper step rather than a short pulse for testing rise time, but it turns out its not all that easy to create a 10s of picosecond rise time pulse out of things you might have laying around the lab.

In my case the scope had a specified rise time of 85ps so the pulse generator was the limiting factor here. Being a modern real time scope it has no trouble triggering on that.

Given the difficulty of properly triggering these sampling scopes are they useful for anything else than looking at quick rise times or perhaps looking at the signal integrity of a stable clock signal?

[EV]

Here is the best I could do with my Free-Electrons JW pulser. I used 20 dB attenuator, internal triggering and random sampling. Scope is Tek R7103 with 7T11 and 7S11 (S-4) pluins.

Here is my picture again and picture By Free-Electron with 33 GHz scope.

[JoeO]

"The 284 looks like a very nice piece of equipment but it is out of my justifiable budget.  "

So what is your budget and what country are you in?

[Howardlong]

I agree it should be a proper step rather than a short pulse for testing rise time, but it turns out its not all that easy to create a 10s of picosecond rise time pulse out of things you might have laying around the lab.

In my case the scope had a specified rise time of 85ps so the pulse generator was the limiting factor here. Being a modern real time scope it has no trouble triggering on that.

Given the difficulty of properly triggering these sampling scopes are they useful for anything else than looking at quick rise times or perhaps looking at the signal integrity of a stable clock signal?

As long as you have a frequent, regular, repetitive signal they are fine, but they do indeed have limitations. For checking signal integrity and eye diagrams, generally those features go hand in hand, and that's the point of a sampling scope. They're useless for single shot or infrequent or irregular signals. There are some similarities to CROs in that respect, especially in being unable to see pre-trigger (or even immediately post trigger) without a delay line. I guess as I was brought up on CROs, I can live with it! Imagine the awe I had when I had my first DSO and could see pre-trigger, it was like magic, some kind of time machine!