Basic scope requirements except need to see ~10 ns pulses

[stephencox]

I know there are a ton of posts seeking purchasing advice, but I've read through a bunch of them and I don't think my situation has been covered. I am an academic (scientist, not an EE) and I am shopping for a new oscilloscope after changing labs and research focus. I previously worked more directly with RF and had access to much better test equipment. Now my needs are pretty basic for the most part and would be satisfied by a $2000 scope (plus some expensive high voltage probes) with one major exception: I work with electron multipliers and need to be able to look at 10–1000 mV pulses with a rise time of 1.5–3 ns and a total pulse width of ~20 ns. It can be very useful to trace these signals down the amplification chain from the source to troubleshoot both electronics problems and external sources of noise. I don't need to be able to produce a textbook image of the pulse, but I need to be able to see it well enough to see that it's really a single pulse, for example. I recently tried to troubleshoot one of these with a cheap 40 Mhz scope that was already in the lab and could not even get it to reliably trigger, let alone visualize the pulses.

Most recently I used a Tektronix MDO3034 for this purpose, and the bandwidth was sufficient for my purposes. The direct equivalent is now ~$15000 from Tektronix. I do still have some RF equipment but I'm unlikely to be troubleshooting it without an engineer beyond trying to isolate things to reduce noise, so I'm wondering whether I should consider a cheaper option. For example, I am wondering if I should consider buying a slower Tektronix scope for daily/high voltage use and something less nice to use but with the necessary bandwidth for this one specific job. I'd be much happier spending around $5000 if I can get away with it.

[nctnico]

You'd probably need a scope with at least 300MHz of bandwidth. If it is just for looking at pulses, almost anything is useable so it should be possible to stay under the $5k mark if you look at other brands.

[TimFox]

Are you concerned only with pulse-pair resolution (how close the pulses can be while still resolved as separate) or do you need other information about pulse width, pulse height, or rise time?

[tggzzz]

... and need to be able to look at 10–1000 mV pulses with a rise time of 1.5–3 ns ...

You should be a bit more explicit about what you mean by "look at". In particular, what aspects of the pulses will you be measuring and what are the fidelity requirements. You will be able to see a pulse is present and its relative time of occurance even with a 50MHz scope

Having said that, the key points are indeed the risetime and amplitude.

The rule of thumb is that t_r=0.35/BW, where the risetime and bandwidth are the combination of the scope plus probe plus signal. Hence 1.5ns=>250MHz.

The rule of thumb is that the risetimes increase as root-sum-of-squares, so the combination of two components with a risetime of 1.5ns is 1.414*1.5ns.

That should give you sufficient information to enable you to determine the required scope+probe bandwidth.

Given that bandwidth and the impedances, you should be able to calculate the noise. The noise won't be trivial, but may well be OK - depending on your fidelity requirements.

[stephencox]

Thank you all for the replies so far. It sounds like I'm not missing some magic solution, so I'm looking at either a ~$15000 Tektronix scope or maybe a combo of a $2000 Tektronix scope for something nice and usable for daily use and then a still pretty expensive scope from a budget manufacturer that can do 350+ MHz for the one specific task.

Are you concerned only with pulse-pair resolution (how close the pulses can be while still resolved as separate) or do you need other information about pulse width, pulse height, or rise time?

... and need to be able to look at 10–1000 mV pulses with a rise time of 1.5–3 ns ...

You should be a bit more explicit about what you mean by "look at". In particular, what aspects of the pulses will you be measuring and what are the fidelity requirements. You will be able to see a pulse is present and its relative time of occurance even with a 50MHz scope

Having said that, the key points are indeed the risetime and amplitude.

The rule of thumb is that t_r=0.35/BW, where the risetime and bandwidth are the combination of the scope plus probe plus signal. Hence 1.5ns=>250MHz.

The rule of thumb is that the risetimes increase as root-sum-of-squares, so the combination of two components with a risetime of 1.5ns is 1.414*1.5ns.

That should give you sufficient information to enable you to determine the required scope+probe bandwidth.

Given that bandwidth and the impedances, you should be able to calculate the noise. The noise won't be trivial, but may well be OK - depending on your fidelity requirements.

It's a little tough because I am mostly using this for troubleshooting and so I probably can't even imagine all of the potential problems that I might see in a sufficiently resolved pulse shape. That said, I can definitely be more clear. In addition to pulse-pair resolution, which only requires ~20 ns resolution, it is useful to be able to see stacked pulses that the pulse-counting electronics also can't resolve, and to do that I need to be able to resolve ~the rise time of the pulse rather than the width. That's how I end up at the requirement of around 350 MHz as you outlined. And I can say from experience with the MDO3034 that I could see the pulses relatively well and was able to resolve both dead time and ringing/double counting issues this way in the past.

[bdunham7]

I work with electron multipliers and need to be able to look at 10–1000 mV pulses with a rise time of 1.5–3 ns and a total pulse width of ~20 ns.

 For example, I am wondering if I should consider buying a slower Tektronix scope for daily/high voltage use and something less nice to use but with the necessary bandwidth for this one specific job. I'd be much happier spending around $5000 if I can get away with it.

You need something with low front end noise so you can use the 5mV/div setting with a 10X probe.  If your source is low-impedance and you can use a 10X or 20X low-impedance probe with a 50 ohm scope input, that will work even better.  You also need 350MHz+ bandwidth.  I have a Siglent SDS2354X+ and a Tektronix P6156 10X probe (among others) and I would easily be able to get a very clear picture of a signal like the one you're describing.  As far as being "nice to use", I think it is probably more of becoming familiar with it more than it actually having a bad UI.  If you can get used to a touchscreen it's actually fairly nice to operate.

As for whether you also need a Tektronix scope for other uses, if you already have or intend to use specific Tektronix probes that connect via their proprietary interface then yes you might consider one of their lower-end models that at least has those features.  Otherwise the Siglent is likely quite sufficient for all your needs, although you'd have to say exactly what your "high voltage" application and requirements are to make sure the appropriate probes are available. 

[stephencox]

You need something with low front end noise so you can use the 5mV/div setting with a 10X probe.  If your source is low-impedance and you can use a 10X or 20X low-impedance probe with a 50 ohm scope input, that will work even better.  You also need 350MHz+ bandwidth.  I have a Siglent SDS2354X+ and a Tektronix P6156 10X probe (among others) and I would easily be able to get a very clear picture of a signal like the one you're describing.  As far as being "nice to use", I think it is probably more of becoming familiar with it more than it actually having a bad UI.

Thanks, I will look at the Siglent options. I have never used their equipment but saw them mentioned here quite a bit. The prices are certainly attractive.

Edited to add: I am not sure how to think about the impedance of an unterminated electron multiplier signal output, but I guess very high. However, these are always terminated into 50 Ohms using either a separate RF terminator or the input impedance of the preamplifier. I'm not sure there would be any need or benefit to trying to measure without that, although my ignorance might be showing.

If you can get used to a touchscreen it's actually fairly nice to operate.

That's definitely something I'm not excited about, but it seems to be the way everything is going. Unless you buy the remaining stock of old Tektronix units it looks like that's what you get from them, too. Aside from the touchscreen, I can adapt to anything that's relatively user-friendly as long as it has a decent standalone interface (definitely don't want to have to mess with something like the PC scopes when I just need to grab something for a quick measurement.

although you'd have to say exactly what your "high voltage" application and requirements are to make sure the appropriate probes are available. 

I no longer intend to measure high voltage RF, so I only expect to have the need to measure DC high voltage up to 20 kV to look at flutter and drift on pedestrian timescales (like kHz or slower). I think anyone will have a single-ended high voltage probe that I could get away with for this.

[bdunham7]

That's definitely something I'm not excited about,

I no longer intend to measure high voltage RF, so I only expect to have the need to measure DC high voltage up to 20 kV to look at flutter and drift on pedestrian timescales (like kHz or slower). I think anyone will have a single-ended high voltage probe that I could get away with for this.

I wasn't either, but there are so many features and functions that you need nested menus and touchscreen buttons for them.  The regular buttons for basic functions are still there, but once you adapt to the touchscreen it actually makes older models seem like flip-phones.  The Siglent SDS5000 series retains individual channel controls if that's important. 

A 20kV probe actually is a tricky deal.  One one hand, there is the very economical 40kV version that is commonly used with DMMs (Fluke 80k-40, BK Precision et al. PRT-28A) that works to 40kV but only for DC and power-line frequency AC.  It's input impedance is 1G-ohm, so good for DC with low loading (20µA at 20kV), but it has a lot of parasitic capacitance and is unusable beyond 0.1kHz or so.  Then there's the expensive Tektronix P6015A (or the discontinued P6015 like I have) that is good to 75MHz and has only 3pF input capacitance, but has a 100M resistance so 10X the loading and 0.2mA at 20kV.   And then, AFAIK, there's not much in between.  Maybe this:

https://www.mouser.com/datasheet/2/701/CT4028_datasheet_20171016-1483614.pdf

[tautech]

39kV 1000:1 2pF 50 MHz 900M HVP-39pro fits the bill.

[stephencox]

39kV 1000:1 2pF 50 MHz 900M HVP-39pro fits the bill.

The price is right, for sure! I had a P6015 (I think without the A--it looked pretty old) hat I inherited in my previous setup and I didn't realize it was quite such a pricey piece of equipment. I guess I was thinking it was less of an upgrade from the Fluke 80k-40s that we use all the time with the multimeters.

[shapirus]

FWIW, as an example, here's what I am able to capture with a Rigol DHO804 tricked into thinking that it's a DHO924. Cost $342.16 at the time of purchase, but I did use some Aliexpress sale coupons active at that time.


1. A ~10 nanosecond pulse:




2. The shortest pulse that it can capture and detect its true amplitude (of course including the overshoot), meaning there's at least one sample point present at the peak:




3. About the shortest pulse that it is able to trigger on; here the displayed amplitude is below the true value (vertical scale is the same, but I raised voltage of the signal source to output ~5 V):

(I am not sure whether I hit the limitations of the scope or of the pulse generator here, but naturally I would expect that it's the scope, because it's clearly about the spot where 1.25 Gsa/s becomes the limiting factor)




As you can see, it has no issue triggering on a 1 kHz repetition rate. As a matter of fact, it happily triggers on a 0.1 Hz signal:




All of the above was captured using a 100:1 probe, so ~25 mV on the scope input. It can trigger on lower values just fine, but I haven't tested how low it can go when the pulses are very short.

I also haven't tested it on capturing two consecutive short pulses, but really there's no reason for it to be able to capture one, but not two.

A caveat: this was captured with a single channel being active. When you switch on another one, sampling rate is halved, and the results become accordingly worse.

Now, I'm not saying that it's necessarily a good idea to get, for scientific work, a scope like this, for which the observations you want to make are not very much above the limits of its capabilities, but I do want to point out that a multi-thousand dollar scope is not necessarily a strict requirement. I think it's the probes that you'll need to spend most of the money on, at least as soon as it comes to high voltage and/or differential probing.

[tggzzz]

39kV 1000:1 2pF 50 MHz 900M HVP-39pro fits the bill.

Make sure the voltage vs frequency derating curve is acceptable

[tautech]

39kV 1000:1 2pF 50 MHz 900M HVP-39pro fits the bill.

Make sure the voltage vs frequency derating curve is acceptable

OP states:
like kHz or slower

[blackdog]

Hi,

Good pulse response is not possible with passive probes.
I have several 500MHz probes and none of them have as few abberations as a good coax connection.

Coax type
If possible, then always use good 50 Ohm coax, normal RG58 is not good enough for correct pulse response.

Bandwidth
I am thinking more of a 500MHz model scope with at least 2GHz sample frequency.

My 2 cents :-)

Kind regards,
Bram

[tggzzz]

39kV 1000:1 2pF 50 MHz 900M HVP-39pro fits the bill.

Make sure the voltage vs frequency derating curve is acceptable

OP states:
like kHz or slower

Yes. So?

[stephencox]

39kV 1000:1 2pF 50 MHz 900M HVP-39pro fits the bill.

Make sure the voltage vs frequency derating curve is acceptable

OP states:
like kHz or slower

Yes, this would definitely be fine. Anything I have that is over 3 kV at least *should* be a constant voltage.

FWIW, as an example, here's what I am able to capture with a Rigol DHO804 tricked into thinking that it's a DHO924. Cost $342.16 at the time of purchase, but I did use some Aliexpress sale coupons active at that time.

Thanks, this is interesting . . . I am not sure what to make of the budget vs. quality brand debate for this particular purchase. Depending on what you read, brands like Rigol are either a great value and pretty decent to use for anything that doesn't require critical calibration/certification, or are lousy, buggy, and failure-prone pieces of junk with terrible UI. Seems like there's a missing middle for my part of the market, which I know is true for lots of sectors (microscopes come to mind).

Hi,

Good pulse response is not possible with passive probes.
I have several 500MHz probes and none of them have as few abberations as a good coax connection.

Coax type
If possible, then always use good 50 Ohm coax, normal RG58 is not good enough for correct pulse response.

Bandwidth
I am thinking more of a 500MHz model scope with at least 2GHz sample frequency.

My 2 cents :-)

Kind regards,
Bram

The multiplier pulses I would always be measuring with a coax connection directly to the scope, but maybe I should be thinking more carefully about how I split the signal when I need to do this with the other electronics still attached.

[tautech]

39kV 1000:1 2pF 50 MHz 900M HVP-39pro fits the bill.

Make sure the voltage vs frequency derating curve is acceptable

OP states:
like kHz or slower

Yes. So?

No derating @1kHz. 

From the manual....easier than copying the graph:
39KV:1KHz /20KV:10KHz /5KV:100Hz /300V:100MHz

[tggzzz]

Hi,

Good pulse response is not possible with passive probes.
I have several 500MHz probes and none of them have as few abberations as a good coax connection.

Coax type
If possible, then always use good 50 Ohm coax, normal RG58 is not good enough for correct pulse response.

Bandwidth
I am thinking more of a 500MHz model scope with at least 2GHz sample frequency.

My 2 cents :-)

Kind regards,
Bram

Have you used a resistive divider probe, a.k.a. a Z0 probe. Passive, multi GHz.

If the source impedance is 450ohm, just connect the source directly to a 50ohm input and you have a  *10 probe (950ohm *20 probe, etc).

With low end scopes, make sure it is a real 50ohm input. Just putting a 50ohm resistor in parallel with a 15pF capacitor is inadequate. Use a pad if necessary.

[bdunham7]

Depending on what you read, brands like Rigol are either a great value and pretty decent to use for anything that doesn't require critical calibration/certification, or are lousy, buggy, and failure-prone pieces of junk with terrible UI. Seems like there's a missing middle for my part of the market, which I know is true for lots of sectors (microscopes come to mind).

Both can be true even for a particular model depending on what the user needs and how they use it.  The 'missing middle' is actually there in the mid-to-higher end offerings of the B-brands (Rigol, Siglent, GW Instek) and if  you just skip the entry-level stuff trying to compete in the $500 price space you'll avoid most of the commonly complained about issues.  If you want to spend less than $500 for what is often a fairly capable instrument, you'll just have to put up with some issues.

The multiplier pulses I would always be measuring with a coax connection directly to the scope, but maybe I should be thinking more carefully about how I split the signal when I need to do this with the other electronics still attached.

Then be aware that a lot of the lower priced scopes do not have 50-ohm inputs, which is really a necessity for what you are describing with or without probes being used. 

[tautech]

I work with electron multipliers and need to be able to look at 10–1000 mV pulses with a rise time of 1.5–3 ns and a total pulse width of ~20 ns.

Hardly challenging with a modern DSO......
1kHz rate 20mV 20ns pulse with 2ns edges ex SDG6022X AWG.
100 MHz DSO with BNC connection.....

[Fungus]

A sub-$500 oscilloscope will easily see those pulses. A 1.5ns rise time corresponds to about 200MHz bandwidth.

If all you need is to see those pulses then a Rigol DHO812 will do it.

(Yeah, we know it says "100Mhz" on the front but lots of people here have measured them and they have about 200Mhz real bandwidth...)

If you're prepared to spend half an hour fiddling you can buy a Rigol DHO804 and trick it into thinking it's a DHO924 which has 280Mhz measured bandwidth.

They're great little 'scopes.

The big question is: Will you be using it for other things and what sort of probes do you need? A sub-$500 oscilloscope will do this job but your $2000 budget would easily let you buy a Siglent 2000 series which is a much more powerful everyday oscilloscope.

[Aldo22]

A sub-$500 oscilloscope will easily see those pulses.

That's true, it's really nothing special.
A toyscope like Zeeweii can show a 10 or 20ns pulse (attachment, unfortunately I don't have a better pulse generator).
Rise time, noise, sensitivity etc. are a different question.

Of course I do not recommend the Zeeweii here. This is for illustration purposes only. 

[David Hess]

When I bought my first 100 MHz 100 MS/s Tektronix 2232 it was for that sort of application, and I tested it.  It very reliably captured 10 nanosecond pulses, as given in its specifications, at faster sweep speeds or when glitch detection was active at any sweep speed.  The slower Tektronix 2230 exactly met its specification of 100 nanoseconds.

So almost any modern DSO will at least capture the pulses, but if you want to get an accurate representation of a pulse shape with 1.5 nanosecond edges, higher bandwidth will be required.  As usual, bandwidth is king here.

Given your other requirement, I would look for a low noise DSO, however what manufacturer's commonly call "low noise" is more about marketing than reality.

[2N3055]

Thanks, this is interesting . . . I am not sure what to make of the budget vs. quality brand debate for this particular purchase. Depending on what you read, brands like Rigol are either a great value and pretty decent to use for anything that doesn't require critical calibration/certification, or are lousy, buggy, and failure-prone pieces of junk with terrible UI. Seems like there's a missing middle for my part of the market, which I know is true for lots of sectors (microscopes come to mind).

Rob is not discussing Rigol here. They are what they are.

But Siglent IS exactly this middle market you are talking about. And more than that actually. Their lower end scopes are shaming that segment from A brands. Their mid range is shaming A brands..
In this segment they are the leading brand.

A brands only have edge in higher end, where they have 20+GHz scopes and likes of LeCroy who have miles and miles of all kinds of advanced analysis nobody else has (not even other A brands).

Age of total supremacy of A brands is gone. Now they have edge only in super sophisticated products.
And that is not because legendary A brands couldn't keep the step. It is because their MBA managements only want to manufacture products where they have no competition. That way you can have predatory pricing and numbers  look impressive.

[tggzzz]

A brands only have edge in higher end, where they have 20+GHz scopes and likes of LeCroy who have miles and miles of all kinds of advanced analysis nobody else has (not even other A brands).

Age of total supremacy of A brands is gone. Now they have edge only in super sophisticated products.
And that is not because legendary A brands couldn't keep the step. It is because their MBA managements only want to manufacture products where they have no competition. That way you can have predatory pricing and numbers  look impressive.

Let's put that in perspective.

HP chose to go for markets where they are demonstrably a technology leader, and can offer reliable products and good service for products that customers need. Consequently HP has never been cheap

David Packard once said that if he ever found anybody suggesting market share was a valid business objective, he would fire them.

HP still exists, except it is now called Agilent Keysight.

I doubt the OP gives a hoot about these points; quite right too.

[2N3055]

I doubt the OP gives a hoot about these points; quite right too.

If you read a bit more carefully to OP quote I included, you would see that it was in direct response to his sentence...

And it is relevant.

Because there is large portion of T&M market that is not covered by A brands anymore.
And they became irrelevant in that segment.

More than decade ago, A brands started outsourcing these segments and simply rebranding manufacturers we know today as Rigol, Siglent etc...

To the point that Keysight makes no modern scope in that segment anymore (all their scopes are 10 years old designs, repainted black), R&S has no scope for less than 2000+€ except 15+ year old HMO1000 Hameg rebranded R&S. LeCroy entry level is rebranded Siglent.

So you go to brands that actually serve that segment.

[tggzzz]

I doubt the OP gives a hoot about these points; quite right too.

If you read a bit more carefully to OP quote I included, you would see that it was in direct response to his sentence...

And it is relevant.

Because there is large portion of T&M market that is not covered by A brands anymore.
And they became irrelevant in that segment.

More than decade ago, A brands started outsourcing these segments and simply rebranding manufacturers we know today as Rigol, Siglent etc...

To the point that Keysight makes no modern scope in that segment anymore (all their scopes are 10 years old designs, repainted black), R&S has no scope for less than 2000+€ except 15+ year old HMO1000 Hameg rebranded R&S. LeCroy entry level is rebranded Siglent.

So you go to brands that actually serve that segment.

I was referring to my points.

[Performa01]

I would not stumble from one extreme to the other – any sub 500 $ DSO isn’t the right choice to replace a 15 k$ Tek.

My recommendation is the Siglent SDS2354X HD, which is a 12 Bit / 350 MHz / 4-channel MSO with deep 100 Mpts/Ch. Sample memory and extended analysis capabilities, switchable true 50-ohm inputs, providing a VSWR <1.5:1 up to >1 GHz. The actual -3 dB bandwidth is more like 460 MHz.

This scope is reasonably responsive and does not pass out for several seconds as soon as it has anything more demanding to do.

Attached is an example, tailored towards your use case.


SDS2504X_HD_PR_Pair_W10ns_RT1.5ns_d4ns

The signal consists of 10 ns wide double-pulses with 1.5 ns rise time and 14.28 ns period (~4.28 ns spacing) between the individual pulses; the burst period is 100 µs.

The main window shows 4 such double-pulses to illustrate the before mentioned 100 µs period.

The Zoom window shows a detailed view on the pulses and the automatic measurements are pretty spot on with regard to pulse width and rise/fall time. The Rising edge Overshoot measurement ROV reads ~3.2% on average and is not entirely stable, as can be seen at a glance in the measurement statistics (high peak to peak variance and standard deviation) as well as the Histogram Icon (which could be enlarged into a separate window for closer inspection.

The instrument provides an always visible 7-digit trigger frequency counter and the time base has 1 ppm accuracy.

[2N3055]

I doubt the OP gives a hoot about these points; quite right too.

If you read a bit more carefully to OP quote I included, you would see that it was in direct response to his sentence...

And it is relevant.



Because there is large portion of T&M market that is not covered by A brands anymore.
And they became irrelevant in that segment.

More than decade ago, A brands started outsourcing these segments and simply rebranding manufacturers we know today as Rigol, Siglent etc...

To the point that Keysight makes no modern scope in that segment anymore (all their scopes are 10 years old designs, repainted black), R&S has no scope for less than 2000+€ except 15+ year old HMO1000 Hameg rebranded R&S. LeCroy entry level is rebranded Siglent.

So you go to brands that actually serve that segment.

I was referring to my points.

Ahh, my apologies for misunderstanding. Mea culpa.
Best,

[tggzzz]

Ahh, my apologies for misunderstanding. Mea culpa.

Fortunately I have never ever made such a mistake.
...
...
...
Oh.
 

[2N3055]

Good advice.

I would like to add fact that scope also have segmented /history mode that you can use to capture many individual events and than analyse them one by one.
You also have search to automatically find certain signal parameters.

[stephencox]

Thanks, everyone. I am going to chat with some former collaborators before making a decision, but I am now leaning toward the Siglent option.

The topic of the different focuses of legacy and budget or upstart brands is definitely an interesting one! I live in different market segments for different types of equipment, so it comes up a lot. For example, I think my needs are in the mid-range for lasers, electronics that aren't part of the mass spectrometers, and microscopes. Unfortunately for microscopes I don't think that market segment exists, so we're wasting money because we can't tolerate the budget equipment. Also there are some egregious monopoly practices with proprietary software and camera compatibility, etc., but that's a digression . . .

[Fungus]

I would not stumble from one extreme to the other – any sub 500 $ DSO isn’t the right choice to replace a 15 k$ Tek.

If all the Tek was doing was look at 10ns pulses then why not?

[2N3055]

I would not stumble from one extreme to the other – any sub 500 $ DSO isn’t the right choice to replace a 15 k$ Tek.

If all the Tek was doing was look at 10ns pulses then why not?

Because it was 500MHz Tek with proper 50Ω inputs with good SWR etc.
Simple in processing as those were, they had serious signal integrity, those old scopes.
They were not toys, but serious machines.

So for a modern replacement you need to replace with same quality front end.
That excludes lower end. For instance, although in processing functions SDS800xHD is way more powerful than OP needs, we would need to go to minimum SDS2000x+ class.
Rigols first level would be DS7000 or DHO4000.
It is shame they don't have anything in between like Siglent's 2000 class (8/12bit).