Upgrade my DSO... or not?

[IDEngineer]

...if its me, i will use it to its last breath and drop of blood. once its dead then buying (replacing to) another scope it will be.

I'm usually that way, but modern DSO's have become mostly computers with an acquisition deck on the front end. So their feature set changes much more rapidly than the analog scopes of yesteryear. BTW, I still have my original Philips PM3214 dual trace delayed timebase 25MHz pure analog scope sitting here on the bench. And I still use it, there are still things where an analog scope is preferable over a DSO. Thus I totally understand keeping equipment for a long (!) time, but DSO's have changed those rules for me a bit for digital-based tools.

i feel you, thats why i will hold on to my old pc bought around 10yrs ago as long as it can. when the time comes to its last breath, i hope there is still motherboard that i can install WinXP in it, i'm lucky i still have the original WinXP CD working, i've made several backup for it in case the original CD got damaged.

Your biggest risk is that a newer motherboard will have peripherals onboard that won't have WinXP compatible drivers available. That's why I got the new laptop when I did. It hasn't been that long since Win7 was native (or at least a "downgrade" option on Win8 laptops) so I could get a bleeding edge (for ~2 years ago) machine with full Win7 support for all of the integrated peripherals. These days new motherboards and laptops are built with peripherals for which nobody bothers to write drivers that are compatible with older versions of Windows (I suspect there may be some strongarming by Microsoft in that) so unless you are careful to get a platform that has drivers for your preferred version of the OS, you're stuck.

Like analog electronics, I generally get 10+ years out of a laptop so I figure this one will keep me running until the latter half of the 2020's. By then, hopefully Microsoft will have figured out that folks don't like OS's that snitch on their users back to the mothership, don't like corporate overlords constantly resetting the user's preferences, don't like "upgrades" that happen on a whim, etc. Or else Linux will have become prevalent enough that I won't feel like an unpaid beta tester every time I try it (again). Linux was fun as a playtoy back in the mid-90's, and the 2000's, and the early 2010's, but these days I need tools to be reliable appliances, not their own separate hobby!

[David Hess]

I have used higher resolution DSOs before and the advantage was always qualitative and not quantitative.  I was able to detect things which were not present before but not make measurements which were any better because of limitations in linearity and settling time which remained 8-bits.  Any improvement in noise was only in comparison to an excessively noisy 8-bit DSO.  My analog oscilloscopes and even sampling analog oscilloscopes had the same advantage in qualitative measurements.

A better investment might be additional probes (active, low voltage differential, high voltage differential, current, whatever) instead of a similar but marginally better DSO.

[jjoonathan]

Another good investment could be waiting for a deal. If you're happy with what you have, you can afford to wait for lightning to strike.

I have been in a position uncannily similar to OP's for the last few years. I had a hacked Rigol DS4014 that met my needs -- it really is a sweet spot for value -- and I had allotted myself a $4000 budget to "take the next step" but wasn't settled on exactly what form that should take. Better FFTs and annotations good enough to break my sticky note habit were my big requirements, but not so pressing that they forced me to jump at something for sticker price. The RTB2000 launch promotion got my attention so I waited for a RTM3000 launch promotion that never quite happened, but then last month lightning struck. I scored a used 2GHz Rohde & Schwarz beast from Teledyne Lecroy (crazy) that shipped on April  1 (crazier) and was supposed to be half broken but actually wasn't. Best April fools ever. I got my nice FFTs and sticky-note-beating annotations but also considerably more.


https://gph.is/g/am6XlXP https://photos.app.goo.gl/2cNXcP9UBJ6m39we6


You can't win that big every time, of course. Waiting isn't sufficient, but it is necessary, and the fact that my Rigol DS4014 made waiting easy will forever earn it a fond place in my heart, even though it has now moved on to a new owner.

[tautech]

Nice score. 

[Old Printer]

Beautiful looking scope, congrats. You obviously have quite a bit of experience, and equipment, but what gave you the courage to plunk down 4K+ with the item description as it was? Did you have a reasonable expectation the problem was either nonexistent or fixable at an affordable cost? Just curious.

[jjoonathan]

Yes, I did. Previous success in repairing my DS4014 boosted my confidence and drew my attention to the listing's picture showing four active traces with visible nonzero noise-floors. This suggested that everything from the ADCs on back was OK and that the problem was either in the analog front end (I could build a crummy amp/atten and get it limping if needed) or localized to the space between the scope and chair. As it turned out, the problem was not in the analog front end.

[Mechatrommer]

yeah with some willingness for the risk, and some skills under the belt, we can go the "used equipment" path. i purchased 3 GHz DSOs (3 units, not 3GHz) so far for around that $4K combined, sold the 2nd before buying the 3rd and covered the cost of some GHz active and differential probes. but do mind they are boat anchor type, not slim the latest tech type. ymmv.

[rsjsouza]

IDEngineer, thanks for this thread. I own a DS4014 fully unlocked, with upgraded heatsinks and fan and for several times I thought about the same thing: should I freshen my main test gear or should I simply run until its encoders fall off? The ensuing discussion here shows some very nice alternatives and made me stay put for this time. The DS4000 series from Rigol is a solid performer.

As Mecha mentioned, you may go the used route - a used Keysight DSOX3014A (hackeable), for example, can set you back at around $2200 and a new one can be had for $3400 (fleabay US). However, what you gain in UI responsiveness you lose in memory depth and bandwidth if you don't hack it. YMMV.

[2N3055]

IDEngineer, thanks for this thread. I own a DS4014 fully unlocked, with upgraded heatsinks and fan and for several times I thought about the same thing: should I freshen my main test gear or should I simply run until its encoders fall off? The ensuing discussion here shows some very nice alternatives and made me stay put for this time. The DS4000 series from Rigol is a solid performer.

As Mecha mentioned, you may go the used route - a used Keysight DSOX3014A (hackeable), for example, can set you back at around $2200 and a new one can be had for $3400 (fleabay US). However, what you gain in UI responsiveness you lose in memory depth and bandwidth if you don't hack it. YMMV.

Make note that DSOX3000A exist in 100/200MHz, 350/500MHz and 1GHz hardware board version.
So 100 MHz goes to 200MHz, 350MHz  to 500MHz with software unlock.
If you want to upgrade 100MHz to 500MHz, you need serious soldering on mainboard...
Doable and there are details on EEVBLOG.. But not really just a software hack.

[Martin72]

Mine's not "really really quiet". It's ok, but not that good.
A few bugs so far...

Where´s your review ?? 

[IDEngineer]

IDEngineer, thanks for this thread. I own a DS4014 fully unlocked, with upgraded heatsinks and fan and for several times I thought about the same thing: should I freshen my main test gear or should I simply run until its encoders fall off? The ensuing discussion here shows some very nice alternatives and made me stay put for this time. The DS4000 series from Rigol is a solid performer.

Yep, same conclusion I'm reaching. The DS4000 series is hard to beat if 500MHz and 4GSa/sec is enough for your purposes. And honestly, I bet that covers 90% of what most EE's and hobbyists actually work on most of the time. This thread has quenched my curiosity into a new scope, and renewed my faith in the DS4000, to the point that it will be my default recommendation for most people unless they need to stay under $1K or truly need performance over 500MHz and 4GSA/sec. The good news for Rigol is there are a LOT of folks in that middle ground for whom the DS4000 is an absolutely killer deal. Unless I get my hands on something else in the midrange, I'm directing people toward the DS4000 and that will probably make Rigol a lot of money, since I'm sure they've amortized their R&D costs on that line long ago. It's got to be a pure cash cow now.

[IDEngineer]

A better investment might be additional probes (active, low voltage differential, high voltage differential, current, whatever) instead of a similar but marginally better DSO.

True. In fact I already have a low voltage differential probe (Pintek DP-60HS) that is my second most important piece of test equipment after the scope. So much so that I've been tempted to buy 1-3 more just so I can have more channels of fully differential input. I work on a lot of isolated circuits and the diffprobe has made some impossible measurements possible.

One thing that would be nice to have is one or more fully isolated low voltage diffprobes. Like seemingly everyone else on here, I have some ideas for a novel architecture, in my case incorporating some of the tricks I've learned designing ultra-high-isolation circuitry. Maybe someday,right? 

I've also promised myself I'm going to get one of those old Tek DC/AC clip-on current probes with their little TM5xx amplifier. I used one back in the 80's and really liked it. Would be nice to have one again.

[rsjsouza]

IDEngineer, thanks for this thread. I own a DS4014 fully unlocked, with upgraded heatsinks and fan and for several times I thought about the same thing: should I freshen my main test gear or should I simply run until its encoders fall off? The ensuing discussion here shows some very nice alternatives and made me stay put for this time. The DS4000 series from Rigol is a solid performer.

Yep, same conclusion I'm reaching. The DS4000 series is hard to beat if 500MHz and 4GSa/sec is enough for your purposes. And honestly, I bet that covers 90% of what most EE's and hobbyists actually work on most of the time. This thread has quenched my curiosity into a new scope, and renewed my faith in the DS4000, to the point that it will be my default recommendation for most people unless they need to stay under $1K or truly need performance over 500MHz and 4GSA/sec. The good news for Rigol is there are a LOT of folks in that middle ground for whom the DS4000 is an absolutely killer deal. Unless I get my hands on something else in the midrange, I'm directing people toward the DS4000 and that will probably make Rigol a lot of money, since I'm sure they've amortized their R&D costs on that line long ago. It's got to be a pure cash cow now.

I usually recommend the DS4000 series if the price is right, given the more modern architectures from them (5000 and 7000 series) seem to bring more to the table at a somewhat lower cost - although they are a new platform still plagued by several bugs. I got mine at a real bargain from their clearance store.

It seems Rigol has been somewhat snappier in fixing those when compared to the DS4000 series, which was very frustrating for several users of this forum that intercepted this model at an earlier stage. Mine was a later purchase (2015) and the severe issues were already fixed, that's probably why I still have a good impression of it.

[David Hess]

One thing that would be nice to have is one or more fully isolated low voltage diffprobes. Like seemingly everyone else on here, I have some ideas for a novel architecture, in my case incorporating some of the tricks I've learned designing ultra-high-isolation circuitry. Maybe someday,right? 

Differential and isolated operation usually conflict so I am not aware of any probes like that.  Most differential probes will not operate correctly with an isolated oscilloscope channel and better ones have a warning to this effect.

Isolated single ended probes exist and are very useful.  They have better high frequency common mode rejection than a differential probe but require a low impedance common connection to the circuit under test which is usually not a problem.

I've also promised myself I'm going to get one of those old Tek DC/AC clip-on current probes with their little TM5xx amplifier. I used one back in the 80's and really liked it. Would be nice to have one again.

They are very nice and there are some modern equivalents.  I get by with just a 50 MHz Tektronix AC only current probe.

[IDEngineer]

Differential and isolated operation usually conflict so I am not aware of any probes like that.

That's why I said I have some ideas for the architecture. 

Isolated single ended probes exist and are very useful.  They have better high frequency common mode rejection than a differential probe but require a low impedance common connection to the circuit under test which is usually not a problem.

Agreed, and if I ever get the chance to pursue this isolated probe idea I'll start with single ended to prove things out, then tackle the differential aspect. Even if the latter isn't successful I'll end up with a useful tool.

They are very nice and there are some modern equivalents.  I get by with just a 50 MHz Tektronix AC only current probe.

Yep, AC only makes it easier. But I do enough low frequency stuff that DC is important to me.

[David Hess]

Differential and isolated operation usually conflict so I am not aware of any probes like that.

That's why I said I have some ideas for the architecture. 

It is not difficult to fix.  The "ground" for the differential probe needs to be connected to "ground" of the circuit under test.  The problem is that common high voltage differential probes do not provide this connection point (for good reason, people might use it) although you could use the output "ground".

Isolated single ended probes exist and are very useful.  They have better high frequency common mode rejection than a differential probe but require a low impedance common connection to the circuit under test which is usually not a problem.

Agreed, and if I ever get the chance to pursue this isolated probe idea I'll start with single ended to prove things out, then tackle the differential aspect. Even if the latter isn't successful I'll end up with a useful tool.

These days the isolation for each channel is usually built into the oscilloscope itself but in the past other methods were used.

They are very nice and there are some modern equivalents.  I get by with just a 50 MHz Tektronix AC only current probe.

Yep, AC only makes it easier. But I do enough low frequency stuff that DC is important to me.

The only substitute for an AC/DC probe is to make a different measurement.  They are invaluable in applications which demand them.

[IDEngineer]

It is not difficult to fix.  The "ground" for the differential probe needs to be connected to "ground" of the circuit under test.

Yep, a path for the bias current. Can be a problem, unless the measuring circuit is truly isolated from the rest of the world (in which case it can be safely connected to the circuit under test). BTW, I'm not targeting high voltage measurements, my interest is in small signal environments.

[tggzzz]

It is not difficult to fix.  The "ground" for the differential probe needs to be connected to "ground" of the circuit under test.

Yep, a path for the bias current. Can be a problem, unless the measuring circuit is truly isolated from the rest of the world (in which case it can be safely connected to the circuit under test). BTW, I'm not targeting high voltage measurements, my interest is in small signal environments.

"Bias current" is incorrect in this context; it isn't an op-amp input! With probes placed incorrectly, high currents can flow, where high implies destruction of electronics, wires (as in they disappear), and potentially people.

Note that David Hess puts quote marks around the word "ground". That too is a misleading term; in industrial contexts there can be many volts between points that are all nominally "ground".

[IDEngineer]

Agreed, I was generalizing my terminology. A bit of laziness showing through. {grin}

Your point about different "ground" potentials is spot-on. My point was that, if the portion of the isolated probe that is doing the measuring is truly isolated, its "ground" could be raised to the nonzero potential of the "ground" in the DUT without ill effects as long as common mode limits aren't exceeded. As to the latter, you can provide protection to some degree but ultimately it's the user that must close that loop and make sure limits aren't exceeded. That's all the most true for highly sensitive equipment meant to measure the smallest of signals.

[David Hess]

Yep, a path for the bias current. Can be a problem, unless the measuring circuit is truly isolated from the rest of the world (in which case it can be safely connected to the circuit under test). BTW, I'm not targeting high voltage measurements, my interest is in small signal environments.

As far as I am aware, all low-voltage differential probes provide the "ground" at the probe tips as a third connection in one way or another and there is an expectation that it will be used.  This is one thing which distinguishes low-voltage and high-voltage differential probes.

"Bias current" is incorrect in this context; it isn't an op-amp input! With probes placed incorrectly, high currents can flow, where high implies destruction of electronics, wires (as in they disappear), and potentially people.

That cannot happen with a high voltage differential probe, at least between the differential inputs, because the impedance (and voltage rating) between the differential inputs is high.

The problem comes about when a high voltage differential probe is connected to an isolated input.  Now the common mode voltage cannot be attenuated by the input dividers because they have no connection except to each other.  With the common mode voltage unattenuated, the high-voltage differential probe "ground" can charge to the input common mode voltage presenting a safety hazard and eventually causing breakdown at the isolation barrier which should not have been used.

So the common mode rating of a high-voltage differential probe, which can be thousands of volts, does not apply if its common terminal is not connected.  Typically the common terminal is connected through the oscilloscope input ultimately to earth ground but this does not happen on an oscilloscope with isolated inputs.

A similar problem occurs with common x100 passive probes when AC input coupling is used.

Note that David Hess puts quote marks around the word "ground". That too is a misleading term; in industrial contexts there can be many volts between points that are all nominally "ground".

I used the quotes because it is more properly called "common".  The differential input voltages are referenced to the probe's common terminal which may or may not be attached to actual ground.  This is one of those cases where using the correct term can just lead to further confusion.

[IDEngineer]

"Bias current" is incorrect in this context; it isn't an op-amp input!

WRT this: There are quite a few diffprobe designs floating around that use opamps configured as differential amplifiers. Quite a few on this website, in fact.

[2N3055]

"Bias current" is incorrect in this context; it isn't an op-amp input!

WRT this: There are quite a few diffprobe designs floating around that use opamps configured as differential amplifiers. Quite a few on this website, in fact.

To be more accurate, diff probe is actually nothing more than instrumentation amplifier.  A special version that is either really fast or has really robust inputs with large common voltage range.
And I guess you could say they have input current that you could call bias, but it is not called so. With probes, it is simply input impedance, both differential (between the inputs) and single ended (towards "ground").

And they all have third connection for common reference potential.

High speed low voltage probes come with "ground" terminal because of low common mode range and to maximise CMRR and minimise various spurs in signal...
Connection to the ground will be short and local to DUT.

For a high voltage probe that is not so critical, so their third terminal ("ground", I actually like this quoted term for common reference potential) is directly connected to scope grounding.
Any common voltage on inputs gets suppressed by CMRR of the probe.  That is why they don't usually have third connection, except some do, and a warning it has to be connected to protective ground or some common reference plane in SUT if your scope input is floating (is isolated from common frame of reference). That would happen if you (not recommended) float your scope, or if you have proper isolated channel scope, in which case you mostly don't need diff probe. Unless you want to use diff probe anyways, because it is more robust, or you want to repeat previous measurement so you want to compare with the same circuit loading (same probe).

One not very much discussed point in having a diff probe is mostly overlooked. People speak about reference planes and grounds, dangers and such.
To me, best explanation why you need diff probes, and their most useful characteristics is this:

With classic grounded (passive or active) probe, you take GND wire from the probe and connect it to GND of your circuit. And then you probe various SINGLE points in circuits. So you are observing a signal at certain net in a circuit.  So if you imagine stack of batteries, you can measure voltage on first battery, and then voltage on second battery (that will be in series with first one) etc.

If you wanted check a voltage of every individual cell (you want to check if cells are good), you cannot measure voltage ACROSS every cell, like you can with multimeter. Connecting grounded probe across third cell will connect scope ground in such a way that it will short first two cells and kill your scope if there is enough current in them.

Diff probes will allow you to do just that. With them you can measure across the components. Between two arbitrary points, not only potential on that net in regards to ground.
Like you do with multimeter, and that makes them very useful when trying to understand what is going on in a circuit.
Of course, with usual disclaimer that they have input impedance (complex one) that is usually lower that other probes. So they can interfere with high impedance circuits.
Like any probe. connecting them to circuit can change circuit behaviour.. So you have to understand both probe spec and circuit you're measuring.
Probe is a part of the circuit when connected.

* CMRR  is common mode rejection ratio, for those that don't know.
   SUT- system under test

[Old Printer]

Yes, I did. Previous success in repairing my DS4014 boosted my confidence and drew my attention to the listing's picture showing four active traces with visible nonzero noise-floors. This suggested that everything from the ADCs on back was OK and that the problem was either in the analog front end (I could build a crummy amp/atten and get it limping if needed) or localized to the space between the scope and chair. As it turned out, the problem was not in the analog front end.

The Keysight person between the scope and chair left a lot of money on the table for Keysight to give away. I am happy for you and your good deal, and have to wonder if someone higher up at Keysight might trip across this thread and track down the missing link

[jjoonathan]

It would have been a LeCroy person, unless there was a Keysight spy infiltrating a LeCroy facility to sell me a discounted Rohde and Schwarz scope, which is probably one level of silliness too far.

In any case, I've seen these sales go down from the other side. If I were in a supervisory position my biggest fear would not be that the person in charge of liquidating the old equipment rack left an extra 20%-50% on the table because they weren't familiar with the quirks of an instrument we were selling 1 of. A 100x larger fear would be that in chasing down the "missed opportunity" and berating the seller I would inadvertently encourage the corporate path of least resistance, which is to let a $200k rack of used equipment sit in a corner and gather dust until it has lost 90% of its value and then sell it to the nearest liquidator for 10% of the remaining 10% -- with employees barred from bidding, because *that* would be a conflict of interest. 

Recovering 50-80% of the value in an unused asset really isn't bad at all by the corporate standards I've seen.