Tool Selection - Scopes for specific needs

[BenKenobi]

So yes I'm a beginner in the electronics world - well I've been playing for a couple of years now via Arduino etc which has led me here - largely through watching a lot of Dave's content. I'm probably about to ask a question that will have all you experts sighing ..

I have though developed specific interest areas and actively pursue robotics and remote control, I also dabble with telemetry and the collection of situational awareness data via the likes of bluetooth, WiFi, 2.4 Ghz and other frequencies but I need to get more serious, I need to be able to inspect and debug at a serial protocol level, I also want to be able to see what 'sensors' are really doing to determine whether they are the issue or my coding skills, I also want to be able to probe what a sensor is seeing vs what it is encoding,  motor noise, RF interference, harmonics and wire borne interference from connected systems etc etc.

So I decided it is time to buy an oscilloscope, there are so many, most will fulfill 90% of my requirements but so few even discuss serial analytics and protocol decoding, I really don't want to spend hour upon hour tallying peaks and troughs and working out who is saying what - if I must then so be it but I'd rather not.

I'd like to be able to focus on specific devices on say an I2C or SPI interface, but also see if some device is talking out of turn or not handshaking properly to isolate timing issues / line capacitance etc etc.  - aka an oscilloscope. As an example I have a very large helicopter, the RC receiver also transmits telemetry data back to the ground - but the RF interferes with the onboard flight controller quite badly - so somebody isn't shielded, I've confirmed that the RF noise is on the wires but not by anything other than trial and error.

To a noob like myself the sheer number of scopes available is daunting, many don't even discuss what I need and focus mostly on settings aimed at the electronics diagnosis world, to find out what they can and can't do you are faced with reading manual after manual to figure what they can really do. I see so many for sale and yet I'm then compelled to play hunt the manual to figure out if they can do what I need and it is usually negative, this is getting frustrating.

Right now the only one on my list is the Rigol 1054Z, can anyone suggest any others that cost hundreds as opposed to thousands, this is a hobby not a business for me I can't justify thousands on buying gear I use infrequently. I'd rather buy a high quality secondhand than a mediocre new if that makes any sense, the Tektronix DPO's don't seem to appear often (or at all at what I consider sensible money).

[tggzzz]

That's a pleasingly well thought out question

I need to be able to inspect and debug at a serial protocol level, I also want to be able to see what 'sensors' are really doing to determine whether they are the issue or my coding skills, I also want to be able to probe what a sensor is seeing vs what it is encoding,  motor noise, RF interference, harmonics and wire borne interference from connected systems etc etc.

So I decided it is time to buy an oscilloscope, there are so many, most will fulfill 90% of my requirements but so few even discuss serial analytics and protocol decoding, I really don't want to spend hour upon hour tallying peaks and troughs and working out who is saying what - if I must then so be it but I'd rather not.

I'd like to be able to focus on specific devices on say an I2C or SPI interface, but also see if some device is talking out of turn or not handshaking properly to isolate timing issues / line capacitance etc etc.  - aka an oscilloscope. As an example I have a very large helicopter, the RC receiver also transmits telemetry data back to the ground - but the RF interferes with the onboard flight controller quite badly - so somebody isn't shielded, I've confirmed that the RF noise is on the wires but not by anything other than trial and error.

You have recognised the principle problems, but haven't quite turned that into a solid debugging strategy...

The first problem you identify is "signal integrity", i.e. are the voltages and times and transitions within specification such that the receiver will be able to correctly interpret those analogue signals as digits. For that an oscilloscope is necessary, and the minimum necessary bandwidth is dictated by the analogue signal's transition time, not by the baud rate, not by a clock speed. The rule of thumb is BW = ^0.35/_tr.

The second problem you identify is the information carried by those analogue signals, which requires protocol decodes. As you realise, there is no point in decoding a protocol until the signal integrity has been verified. However, once you have valid digits then it is best to debug in the digital domain, not in the analogue domain. Traditional tools are logic analysers and printf statements, plus there are useful refinements that are particularly relevant to finite state machines (FSMs). Triggering and filtering are of prime importance if you are to sort the wheat from the chaff.

Don't bother trying to use a scope to look at an RF signal (use a spectrum analyser or network analyser), but do use a scope to look at the demodulated baseband signal.

I'm not going to recommend specific equipment, because with skill and imagination you will be able to use many different tools to skin the cat in front of you. However I will note that:

• one way or another you are going to pay to gain experience; there is an argument that your initial experiments should be as cheap as possible until you determine what you really need
• signal integrity can be assured with analogue scopes, especially if you can arrange that signals are repetitive. Second hand working analogue scopes can offer better bandwidth at a lower cost than new digitising scopes. For modern logic, 100MHz is the bare minimum. Traditional second hand scopes are Tek 465/475/485/24x5 (and 22x5) HP1740, but there are others
• storage scopes help when looking at intermittent signals. Analogue storage scopes are a pain in the backside; digitising storage scopes are infinitely preferable. Some factions on this forum will say that only the DS1054Z need be considered, others that other manufacturers are better; choose your poison!
• some digitising scopes also claim to decode protocols, but they have limitations. You probably can't trigger/filter on protocol messages, and even triggering on protocol violations can be difficult. Some only decode what's visible on the screen, despite having captured far more
• very cheap and simple logic analysers simply capture the bitstream and decode by software postprocessing. They are often sufficient for simple debugging, but can be problematic with the more complex cases because the triggering is poor
• more complex (and expensive) logic analysers are better at capturing rare events and protocol violations
• your own custom-coded software is best at filtering and capturing rare events and protocol violations

If money is tight and you can get the educational discount, consider the Digilent Analogue Discovery (10MHz scope and AWG, 100Mb/s pattern generator and data capture).

[BenKenobi]

Absolutely fantastic information, and I do understand that logic analysers play a part, I'm not ready to debug RF as such but I want to see where interference is present - i.e. that I have clean digital signals.

If I can see the start of a communication but then observe the stream to see if it is OK and noise free - the rise times and peak to peak etc, that there are no 'pulses' where there shouldn't be kind of thing.

I do apologise for being a bit woolly, I've an idea what I need to be looking at, I realise there is more than one tool to do this but I think that at the moment (and my current status in this intellectually) that a really good scope is the best 'initial' investment, money isn't particularly tight but I don't want to be wasting it. I feel that a scope will allow me to make sure signals are 'clean', that devices are doing what I think they are (or not), that interrupts are triggering for genuine reasons not random noise.

Right now I have two interrupt driven routines, one reads SBUS remote signals, one reads a 9 axis gyro via SPI, they're both working fine in isolation, both use different time bases, both are 'exclusive' so cannot interrupt one another but they flat out won't work when both connected, the code freezes - somebody isn't behaving. I also get servo jitter when I connect a 2.4GHz receiver, this is 100% line induced because if I power the receiver from a separate supply to the processor the jitter disappears although the receiver is still next to the processor.

Really wish I could articulate this stuff appropriately.

[rstofer]

The DS1054Z will decode UART, I2C and SPI with no problem (after you unlock it, details elsewhere).  But it won't decode the entirety of "War and Peace".  You can get a couple of packets on the screen (maybe) and the 1054 only decodes what is on the screen.  Other scopes decode what is in memory.  Huge difference!

So, if what you want to see is in the middle of a huge volume of packets, you really need a logic analyzer.  As mentioned above, printf() is a common debugging technique.  Sometimes you can jigger up a synch signal that triggers the scope at some particular event.  That way you don't capture a bunch of nonsense and the display has only the specific event.

Nothing is more important than bandwidth except channels.  That's why there is so much excitement for the DS1054Z, it has 4 channels and this is ideal for decoding SPI.  There is also the new Siglent 2 channel SDS1202X-E which goes to 200 MHz and a yet to be released SDS1204X-E which has 4 channels.  It has already been released in China.  Both do decoding.

As mentioned above, if your basic need is speed, a used Tektronix 485 is good for 350 MHz but it lacks every modern feature of the DSO.  I bought mine about a dozen years ago for $200.  I didn't get rid of it when I bought the DS1054Z.

You have an apparent need for a scope, no question.  If I knew for a fact when the SDS1204X-E was coming out, I would probably say wait.  But the bigger question is "what if it takes a year, two?".  Is it worth waiting?  I don't have an answer for that but I don't need another scope.  I will buy one but I'm in no hurry.

So, why not just go for the DS1054Z and learn to use a modern DSO.  I'm still learning!  The scope has abilities I could only dream of with my analog scopes.  Use the 1054, decide later whether to upgrade and, if necessary, sell the 1054.  Education is never free so you might as well get started.

Analysis paralysis is a problem.  You can read manuals until you are blue in the face but nothing will compare to actually using the scope.  That's when you find out what all the words really mean.

[rstofer]

Really wish I could articulate this stuff appropriately.

What you have said makes all the sense in the world.  You are not the first to try and troubleshoot this kind of thing.  We have all had similar problems.  Maybe not with an RC Helo but on some other system.

You can't fix what you can't see.  Hence the oscilloscope!

[BenKenobi]

Really appreciate the input, I am leaning very strongly toward the 1054Z I don't need a Rolls just need to get started at the right layer as it were.

I just need to be sure that the biggest limitation in the system is me, that I'm not fighting an inadequacy in the equipment because I don't know enough to know the difference right now.

[tggzzz]

If I can see the start of a communication but then observe the stream to see if it is OK and noise free - the rise times and peak to peak etc, that there are no 'pulses' where there shouldn't be kind of thing.

Any analogue or digital scope with adequate bandwidth can do that. If the data is random then you will see an "eye diagram". If the scope has storage (analogue or digital), then you will be able to record the 1-in-1000 events that violate the norm.

I do apologise for being a bit woolly, I've an idea what I need to be looking at,

No apologies necessary. Answering a question is usually much easier than finding the right question to ask!

I realise there is more than one tool to do this but I think that at the moment (and my current status in this intellectually) that a really good scope is the best 'initial' investment, money isn't particularly tight but I don't want to be wasting it. I feel that a scope will allow me to make sure signals are 'clean', that devices are doing what I think they are (or not), that interrupts are triggering for genuine reasons not random noise.

That's a sane starting point. The trick with that is to find a trigger/filter which allows you to spot the important deviations. Unfortunately until you can visualise the stimulus (voltage, timing,. message, sequence of messages), it is difficult to guess exactly what you need to do that.

Right now I have two interrupt driven routines, one reads SBUS remote signals, one reads a 9 axis gyro via SPI, they're both working fine in isolation, both use different time bases, both are 'exclusive' so cannot interrupt one another but they flat out won't work when both connected, the code freezes - somebody isn't behaving. I also get servo jitter when I connect a 2.4GHz receiver, this is 100% line induced because if I power the receiver from a separate supply to the processor the jitter disappears although the receiver is still next to the processor.

If I was in that situation, I would look at an XMOS xCORE processor, but I doubt you have that luxury

Really wish I could articulate this stuff appropriately.

Do cheap experiments. Observe. Learn. Change. Rinse and repeat. That's what we all do.

[BenKenobi]

Do cheap experiments. Observe. Learn. Change. Rinse and repeat.

Yup that's pretty much what I do. The challenge I have now is moving to the next level, I've an idea what I need to do but no darn clue the best way to go about it, my desire for protocol decode of 'some sort' may be limiting my scope - pardon the pun - even when I get an oscilloscope and I'm pretty much committed to this I need to learn how to do the diagnosis to avoid false conclusions and chasing the rabbit - paths with no end.

I may in fact be better off with an older higher frequency for much of what I need, but then you can't store and review without a scope that costs £$£$£$£$ so that's why the 1054Z is currently top of the list, I may wait for the one suggested above and see how that looks. I suppose that whatever I do buy that resale is always an option but I get attached to stuff, particularly tools.

Like learning to walk I guess - sooner or later after all the crawling you've gotta stand up and fall over a few times.

[tggzzz]

If there is a local Hackspace/Makerspace, visit it and see what equipment they have. They can be a suitable place for discussing ideas and experimenting.

[rstofer]

In my very early view of DSOs, a grand total of 1 year experience, the most important feature is the single shot trigger.  If I can create a trigger, I will capture exactly what happens after the trigger and the image will stay on the screen.  That's a really big deal!  Equally important, I get to view what happened before the trigger.  Nice!

The analog (non-storage) scope can trigger just fine but the image is lost  in milliseconds.  I would find myself spending time creating repetitive triggers so I could get some kind of image on the screen.  But no matter what I did, it isn't even close to the single shot feature of a DSO.

Decoding is nice, math functions are helpful and 4 channels is great but it's the single shot capture that makes the DSO such a popular tool.

You will find that decoding SPI works well when you do a single shot capture using the falling edge of CS'.

[BenKenobi]

It's strange that even looking for one of these has been a learning curve. I started out with the 'Oscilloscope for $50' and started wondering how these scopes were going to fit my needs. Many places said you must have an old analog scope because they're so useful. Now I'm not denying that they are but they fit a particular need and I have indeed come to the conclusion that the digital world isn't one of them.

I don't plan to be diagnosing and fixing Amps and such but I may end up in this world if only for interests sake - that and I have an old KEF PSW3000 that won't power on and I'd love to try and get that up and working so who knows.

But I fear that a DSO is indeed what I need, but do I need 200Ms, 1Gs, 2Gs - obviously I want 5Gs - until I see the price, so it looks like 1Gs possibly 2Gs is in my price range, I don't really want less than 100MHz, I believe .35/bandwidth gives the fastest rise time measurable but I'm currently researching how that rise time impacts the detection of serial communication 'bits' - i.e. at what point does my communication become too fast for the oscilloscope to render.

[tggzzz]

Don't confuse Hz with S/s; the two are nearly orthogonal.

To ram home that point, a while ago I used a top-of-the-range boat anchor with a bandwidth of 1GHz (I could easily see 800ns transitions) but only 25MS/s. Currently my fastest scope, a 40 year old Tek 1502 time domain refloctometer, has >2.5GHz bandwidth but only takes <100kS/s. Those only work with repetitive signals.

The key parameter to concentrate on is MHz; the S/s are largely a marketing tool.

[BenKenobi]

Update accepted and acknowledged.

I'm looking at all kinds of things like Nyquist and recommendations for MHz from various manufacturers, hence the .35/bandwidth rise time comment - the faster rise time being the key I think to analysing the digital world. The thing I'm trying to resolve is just what rise time I can live with vs what would I like, there seems to be a lot of conflicting debate on how you decide what this should be, I've scoured the EEV Vids to see if anything is there but not found anything, I'm hoping 100 / 200MHz will do because funding anything much faster I can't justify unless I find a nice secondhand unit. 

I like the look of the sds1202x-e, it looks pretty impressive, not sure I need 4 channels - but sods law says better looking at than for. The way these things seem to time divide the MHz more channels isn't necessarily more useful - or is it.

Analysis Paralysis ..

[tggzzz]

I'm looking at all kinds of things like Nyquist and recommendations for MHz from various manufacturers, hence the .35/bandwidth rise time comment - the faster rise time being the key I think to analysing the digital world. The thing I'm trying to resolve is just what rise time I can live with vs what would I like, there seems to be a lot of conflicting debate on how you decide what this should be, I've scoured the EEV Vids to see if anything is there but not found anything, I'm hoping 100 / 200MHz will do because funding anything much faster I can't justify unless I find a nice secondhand unit. 

Analysis Paralysis ..

You aren't into paralysis by analysis, yet: your understanding is still evolving.

Unfortunately only a few logic families come with a specification of the transition time, but you can guesstimate the relative times from the propagation delay. 100MHz/3.5ns was adequate 40 years ago for bog-standard TTL and LSTTL. Nowadays even jellybean (74LVC) logic has sub-ns transition times, too fast to see completely on a 350MHz/1ns scope. Some logic has transition times <50ps, but you are unlikely to use them! Of particular interest are MCUs; these tend to have transition times that are controlled to be "not too fast", which often implies similar to TTL and that 100MHz is sufficient.

Provided the digital lines are "short" and there are "plentiful nearby" ground connections (either a ground plane or one ground wre for every two signal wires, you can probably ignore the signal integrity issues.

To understand the distinction between "long" and "short", see http://www.edn.com/electronics-blogs/all-aboard-/4436423/How-long-a-stub-is-too-long--Rule-of-Thumb--18

Be aware that probing high frequency signals is far from trivial. 100MHz is relatively easy, 500MHz is cheap with care, higher frequencies are expensive and fragile.

[rstofer]

Nyquist (Shannon) is only part of the equation because those concepts apply to strictly bandwidth limited signals.  Square waves are not strictly bandwidth limited.  More below...

2.5, that's the number.  If you look around, you will find that a 1 GS/s scope with 4 channels will have 250MS/s per channel which, conveniently, is 2.5 times the stated 100 MHz bandwidth of the scope.  You will see this number occur on MANY scopes.  If you use less channels, the S/s per channel increases.  It is thought that 2.5 times the frequency is adequate since it is higher than the Nyquist Limit but it really doesn't deal with square waves very well.

None of which means very much.  What you really want to know is what is the maximum frequency square wave the scope can reproduce without the trace looking like a sine wave.   A square wave is the sum of the odd harmonics from DC to daylight.  But we'll cut off somewhere around the 7th harmonic.  So, you get a pretty good looking 13 MHz square wave on a 100 MHz scope.  You get a sine wave if you plug in a 100 MHz square wave.  More or less...

Here's an interesting article about harmonics and square waves.

https://www.allaboutcircuits.com/textbook/alternating-current/chpt-7/square-wave-signals/

What is your SPI clock rate?  You wouldn't have any problem displaying a 10 MHz clock on a 100 MHz scope.

I have attached a picture of a 15 MHz square wave.  Some of the reason it isn't square could be attributed to the signal generator (Siglent SDG2082X) and some to the DS1054Z.  My signal generator won't generate square waves beyond 25 MHz...   Still, the signal is workable.  The Siglent SDG2082X has a rise time spec of 9 ns and that's about what the trace is showing.  Not bad!

If I can reproduce your SPI rate with an Arduino or an FPGA, I'll post a picture.  The FPGA would have better rise time but it would take a bit longer to get an output.

I have an SPI master core on another project...

[vk6zgo]

Don't confuse Hz with S/s; the two are nearly orthogonal.

To ram home that point, a while ago I used a top-of-the-range boat anchor with a bandwidth of 1GHz (I could easily see 800ns transitions) but only 25MS/s. Currently my fastest scope, a 40 year old Tek 1502 time domain refloctometer, has >2.5GHz bandwidth but only takes <100kS/s. Those only work with repetitive signals.

The key parameter to concentrate on is MHz; the S/s are largely a marketing tool.

It depends on what you are looking at:-

An analog video signal is "sort of" repetitive, but if you look at it at field rate, DSOs reduce their sampling rate radically to fit the sampled data into their memory.

Early DSOs with small memories reduced the sample rate so far that the high frequency components of the signal were lost in a "forest" of aliasing, rendering the instrument useless.
Those instruments had impressive quoted bandwidths!

[tautech]

Update accepted and acknowledged.

I'm looking at all kinds of things like Nyquist and recommendations for MHz from various manufacturers, hence the .35/bandwidth rise time comment - the faster rise time being the key I think to analysing the digital world. The thing I'm trying to resolve is just what rise time I can live with vs what would I like, there seems to be a lot of conflicting debate on how you decide what this should be, I've scoured the EEV Vids to see if anything is there but not found anything, I'm hoping 100 / 200MHz will do because funding anything much faster I can't justify unless I find a nice secondhand unit. 

I like the look of the sds1202x-e, it looks pretty impressive, not sure I need 4 channels - but sods law says better looking at than for. The way these things seem to time divide the MHz more channels isn't necessarily more useful - or is it.

Analysis Paralysis ..

You've got a good handle on what you need AFAICS along with good sound advice thus far.
100+ MHz is a good starting point for what you'll be doing as it will ensure reasonable signal integrity for the BW of signals you'll be looking at. I started with a CRO and looking back all it offered me was a good grounding in scope usage....predominantly the ability to not need the OSD display measurements to quickly assess signal amplitude and frequency. Some of that must be credited to my then scope mentor....always know what you should expect to see before connecting a probe. Even with a DSO this is good advice.
As it seems this will be your first scope and believe me I know how daunting this can be (most of my customers are hobbyists like you), look for a scope that gives a wealth of info about both setup and signal on the base/std display rather than needing to dig into the UI to find it. Hunt out lots of online DSO display screenshots and see what they convey about the setup....or don't.
DSO's DO have a steeper learning curve than a CRO as there's much more functionality to get your head around plus to drive it to any degree you need learn their many features and where in the UI to find them.

In some clue to availability of SDS1204X-E, presumption is early next year but I wouldn't be surprised to see them earlier. There'll be a 100 MHz version too if the Chinese website is to be believed.

[tggzzz]

Don't confuse Hz with S/s; the two are nearly orthogonal.

To ram home that point, a while ago I used a top-of-the-range boat anchor with a bandwidth of 1GHz (I could easily see 800ns transitions) but only 25MS/s. Currently my fastest scope, a 40 year old Tek 1502 time domain refloctometer, has >2.5GHz bandwidth but only takes <100kS/s. Those only work with repetitive signals.

The key parameter to concentrate on is MHz; the S/s are largely a marketing tool.

It depends on what you are looking at:-

An analog video signal is "sort of" repetitive, but if you look at it at field rate, DSOs reduce their sampling rate radically to fit the sampled data into their memory.

Early DSOs with small memories reduced the sample rate so far that the high frequency components of the signal were lost in a "forest" of aliasing, rendering the instrument useless.
Those instruments had impressive quoted bandwidths!

True, but that doesn't invalidate the general point - it adds a special case based on the peculiarities of some instruments.

Personally I'd avoid any digitising scope which couldn't display the "min/max" in a displayed "bucket", and also those with only a few kS memory. Virtually any analogue scope would be better than those, except when capturing single shot events such as PSU switch on transients.

[tggzzz]

I don't know where you are in the UK, but you might like to look at, of all places, gumtree. There's a nice-looking 100MHz Tek 2235 for £40 in Wimbledon at the moment. That would be a great tool for learning what you do and don't need!

https://www.gumtree.com/p/ham-amatuer-radios/tektronix-2235-100mhz-oscilloscope-/1266921982
http://w140.com/tekwiki/wiki/2235

[RoGeorge]

Right now the only one on my list is the Rigol 1054Z, can anyone suggest any others that cost hundreds as opposed to thousands, this is a hobby not a business for me I can't justify thousands on buying gear I use infrequently. I'd rather buy a high quality secondhand than a mediocre new if that makes any sense, the Tektronix DPO's don't seem to appear often (or at all at what I consider sensible money).

Forget about old oscilloscopes, it doesn't worth going backwards.
All their strong points does not help for the kind of usage you described, but you will feel their lack of modern features every time you power them.

Buy the cheapest Rigol DS1054Z you can find, then unlock all options for free. Batronix is a reliable seller for EU, I bought mine from them.
There is no other competitor for an unlocked DS1054Z, not at 400 euro. Just buy one without any hesitation.

An unlocked DS1054Z is the perfect tool for your needs.

[BenKenobi]

I really appreciate all the input, it's interesting how looking for a tool can have somebody studying harmonics - it is all related and important stuff that you would never even think about without prompting from outside.

I've read (and will re read) some of the information at the links provided, it's all useful in aiding 'is this my circuit or my limitations'. I also watched this 12 Tips vid ( ) - if I took everything in that on board I'd need to spend £10k but it does show the consequences of a device behaviour vs the signal it is fed.

So far I'm convinced I need 100MHz or better and I need 4 channels and an absolute minimum of 1Gs.

The system I'm exploring I've reduced the clocking to 1MHz for debugging although this is way to slow but my problem is still there even at that low speed.

I do appreciate all the information, for me this is a puzzle and I like puzzles.

If the Tektronics were closer I'd probably go and get it just to have something to experiment with, at that money it is a no brainer - alas I'm 200+ miles away ...

[tggzzz]

Right now the only one on my list is the Rigol 1054Z, can anyone suggest any others that cost hundreds as opposed to thousands, this is a hobby not a business for me I can't justify thousands on buying gear I use infrequently. I'd rather buy a high quality secondhand than a mediocre new if that makes any sense, the Tektronix DPO's don't seem to appear often (or at all at what I consider sensible money).

Forget about old oscilloscopes, it doesn't worth going backwards.
All their strong points does not help for the kind of usage you described, but you will feel their lack of modern features every time you power them.

In 40 years and with careful design and debugging, I've never needed a scope to "go backwards". OTOH, it is often useful with logic analysers.

The DS1054Z has, I'm told, significant disadvantages w.r.t. some of the OP's protocol decode requirements. Hence any "just get a X" statements should be carefully evaluated. What is ideal in one situation isn't necessarily sufficient in another.

[BenKenobi]

One of the reasons I have the Siglent on the radar is that apparently it can decode the entire memory - but that depends where you read. I saw Dave's hour long on the siglent sds1202x and was pretty impressed, I've since looked at the siglent sds1202x-e but it seems to be inferior to the one Dave tested, they seem to be totally different beasts - I could be wrong of course.
 
I'm not too hung up on decode, I can if need be move to a logic analyser but I need something more 'general' purpose to tell me if the issue is logical or mechanical in nature, I can use the scope for many things, not so the logic analyser. If I have to manually analyse then so be it but that will be futile if the 'anomalies' aren't even exposed because of some limitation.

I've seen some real nice scopes for circa £800 but that's about £300 more than I really want to pay right now, the longer I look the more scope creep I'm encountering in the budget.

Does anyone have an opinion on Gould DSO's they seem very basic but they're also going pretty cheap, is a DSO with a CRT a good option the things are huge.

[Howardlong]

About eight years ago, due to reduced space constraints at home at the time, I had to choose between keeping either my 400MHz CRO (Tek 2465B) or my 200MHz DSO (Tek TDS2024B), both 4 channel.

I chose to keep the DSO, for the simple reason that it would do single shot. While having the extra bandwidth is nice, for low to medium speed MCU mixed signal applications it really was a no-brainer. Even with a 400MHz CRO, I rarely used it properly as my probing skills and the probes themselves at the time just weren't up to it.

You can certainly get by with a CRO. Back in the days before the DSO, it was common to introduce code at various places to produce repeatable trigger signals. You can still use the technique now, but it can turn in to a time consuming rabbit hole, not to mention the increased Heisenberg uncertainty added to the mix.

If the choice is between a 4 channel 100MHz DSO like a DS1054Z, and a 4 channel 400MHz CRO, bearing in mind the additional features of the DS1054Z and in spite of the reduced bandwidth, I'd definitely go for the DS1054Z for low to medium speed MCU applications.

About serial decode: I was brought up without it, and I before I had a scope with serial decode/trigger I used a USB logic analyser, and before that I decoded it by eye on the scope. I know a lot of people recommend both an LA and a scope. The serial decode on the DS1054Z is certainly limited as others have mentioned, but that is where the serial trigger functionality becomes important. Note that you can decode a reasonably long capture, but it will only decode what's displayed on screen, so you'd have to zoom in and scroll through the captured waveforms.

Note that two channels severely limits you when doing serial decodes such as SPI.

[BenKenobi]

Note that two channels severely limits you when doing serial decodes such as SPI

And that's what I figured - I want at least Clock, MISO and MOSI for that, so thats 2 + 1 ext or 4 channel and I don't know if I really want the mucking about with 2 + 1.

I could of course move away from SPI but then you need to find a device that meets your needs in I2C - I'd rather make the tool fit the needs than tailor my needs around the tool.