EEVblog 1632 - WHY sell 2CH Oscilloscopes? + Test Engineering

[EEVblog]

Why do companies still sell 2CH oscilloscopes? Plus a massive tangent into test and production engineering. Also showing one of my old test engineering projects.



LCR meter and Transmission Line Tester design:
Relay Matrix IR Measurement:
PC104 Embedded DOS PC:
Linus Tech Tips Test Rack:
Test rack Part 2:

[pdenisowski]

In my experience there are many scope users who are unaware of how scopes are used in manufacturing / production, and this video is a great explanation - thanks!

[EEVblog]

In my experience there are many scope users who are unaware of how scopes are used in manufacturing / production, and this video is a great explanation - thanks!

I got hammered in the comments about cheap 2CH scopes!   

[BrianHG]

In my experience there are many scope users who are unaware of how scopes are used in manufacturing / production, and this video is a great explanation - thanks!

I got hammered in the comments about cheap 2CH scopes!   

Putting aside that there are valid markets for 2 channel scopes, boy how spoiled we have all become demanding 4 channels.  When I grew up, any 2 channel scope was a luxury.

[Gyro]

Just because cheap 4 channel scopes are on the market, there's no need to question the use of 2 channel scopes, we're not all doing the same stuff. Why stop at 4? Why not insist on an 8 channel scope.

- Fundamentally a scope is there to examine the timing relationships and integrity of signals. The minimum you need to do that is... 2 (ok you can look at the integrity with 1). It's an analogue domain tool.

- The number of channels needed to do an X-Y phase relationship is... 2.

- Everyone knows that any decent 2 channel scope has an external trigger input anyway, so that gives you the timing relationship of 3 signals with intelligent use, and no loss of sample rate.

- Screen resolution on DSOs (or minimum analogue spot size, lest we open up that argument!) is very low - typically 480 pixels vertical on an 8 bit and a whopping 600 pixels on a 12 bit! In the grand scheme of things, that's pathetic!   Trying to squeeze 4 traces on a a screen and see any reasonable vertical resolution is a joke.

- Just because manufacturers can cheaply stuff in another couple of front-ends is a nice sales feature doesn't make them good. On a typical cheap scope, enable all four channels and the sample rate and buffer size drop through the floor. For example, going back a bit, what was the point of hacking a DS1054Z to 100MHz when the sample rate was only 250Msps with 4 channels enabled.

- Protocol decoding on scopes is a cheap (free) S/W gimmick to sell more scopes. Half the time they get it wrong and the display format is typically very basic. If you are working on digital stuff, after verifying signal integrity in the analogue domain, move to a tool suited to the digital domain... a logic analyzer. Singing the praises of the minimal  decode facilities of a 4 channel scope is just discouraging the manufacturers from putting decent and affordable digital domain tools on the market.

- I wonder how long it will be before you put out a video saying 'geez why would you buy a 4 channel 12 bit scope when you can get a cheap 6 or 8 channel 16 bit'. I bet the vertical resolution will be up to, oh, 1024 pixels by then.


Edit: Reading that back after a couple of hours, I sound a bit irate, sorry about that.

[nctnico]

- Protocol decoding on scopes is a cheap (free) S/W gimmick to sell more scopes. Half the time they get it wrong and the display format is typically very basic. If you are working on digital stuff, after verifying signal integrity in the analogue domain, move to a tool suited to the digital domain... a logic analyzer. Singing the praises of the minimal  decode facilities of a 4 channel scope is just discouraging the manufacturers from putting decent and affordable digital domain tools on the market.

I guess you never had a DSO with proper protocol decoding in your hands then  . R&S pretty much nailed it. And I disagree about getting a logic analyser as in many cases protocol problems stem from problems in the analog domain. If the analog domain works properly then protocol decoding on a large scale using a logic analyser won't do you much good. Higher level tools embedded in the firmware are way more useful as these can produce statistics like number of messages, amount of data, bad messages, etc collected over days, months or even years of runtime. A USB protocol decoder won't do that as it doesn't know the context of the messages it is looking at (for starters). All in all a USB logic analyser is more like an in-between instrument in case you have a DSO with crappy protocol decoding.

[thm_w]

Just because cheap 4 channel scopes are on the market, there's no need to question the use of 2 channel scopes, we're not all doing the same stuff. Why stop at 4? Why not insist on an 8 channel scope.

I'm not seeing where the logic is going here. More channels adds more cost, while increasing the number of tasks you can complete. Probably 95% of common EE work can be done with 4 channels and adding 4 on top of that adds enough cost to not be worth capturing the final 5%.

- Screen resolution on DSOs (or minimum analogue spot size, lest we open up that argument!) is very low - typically 480 pixels vertical on an 8 bit and a whopping 600 pixels on a 12 bit! In the grand scheme of things, that's pathetic!   Trying to squeeze 4 traces on a a screen and see any reasonable vertical resolution is a joke.

hdo1000 is 1280x800 which looks ok with 4 channels.
You can always overlay traces too, its not like they have to each take up 1/4 of the screen.

- Just because manufacturers can cheaply stuff in another couple of front-ends is a nice sales feature doesn't make them good. On a typical cheap scope, enable all four channels and the sample rate and buffer size drop through the floor. For example, going back a bit, what was the point of hacking a DS1054Z to 100MHz when the sample rate was only 250Msps with 4 channels enabled.

Low sample rate is useful to me a lot of the time, eg. monitoring 3V, 5V, 12V at the same time while looking at some input power signal.
I'd gladly take a few extra channels that are limited to only 1 or 10MHz bandwidth for rail monitoring.

Some of the old Tek stuff had simpler secondary channels, capped at 5V, to reduce the knob space/cost and front end cost, presumably.

[twospoons]

Sometimes budget constraints combine with use cases to make 2-ch the favorable option. When I bought my first scope I could get a 4 ch low bandwidth or a 2ch higher bandwidth scope for about the same money. I chose 2 ch with more bandwidth, based on my expected use cases.

[EEVblog]

Just because cheap 4 channel scopes are on the market, there's no need to question the use of 2 channel scopes,

Copied from my response to the pinned comment on the video:
Sure, don't pay for something you don't need. Where did I say a hobbyist should pay for something they know they won't need? Why are people taking this so personally? As always Your Millage May Vary. In this video I was clearly talking about how things typically work ** in the industry **, it was not a treatise on hobby scope selection.
As I have discussed in the Rigol 1054Z and DHO800 videos, 4CH scopes have now become very affordable, and they offer great flexibility in usage scenarios and sample rate split, and as ** a general rule ** I advise getting a 4CH scope even if you think you may not need it. But if you want to save every cent, obviously taking my advice would be dumb, just buy a 2CH scope and be happy.

- I wonder how long it will be before you put out a video saying 'geez why would you buy a 4 channel 12 bit scope when you can get a cheap 6 or 8 channel 16 bit'. I bet the vertical resolution will be up to, oh, 1024 pixels by then.

If 6 or 8 channel scopes become low cost then I'll say that. It won't happen, but that's not the point. The point is that like it or not 4CH has become the new entry level standard at that traditional entry level price.
If you want to say every cent, sure buy that 2CH channel scope, it will serve you well.
As a leading voice in this community I woud be laughed at if my general advice to people these days was to buy a 2CH scope. It is very clear that the best "bang per buck" these days is with 4CH scopes. Has been since the Rigol DS1054Z came out.
As always, YMMV.

[Bud]

A 4-channel scope in general is a solution looking for a problem. I've been doing RF for 25 years and I needed 4 channels may be 2 or 3 times. People doing digital may need it but there are better tools for that and they likely have such tools already.

[EEVblog]

A 4-channel scope in general is a solution looking for a problem. I've been doing RF for 25 years and I needed 4 channels may be 2 or 3 times. People doing digital may need it but there are better tools for that and they likely have such tools already.

The market seems to disagree, as practically every entry level scope maker has affordable 4CH scopes now.
Sure, in general, most of the time you only need one or two channels, but with the low cost ASIC front ends it's now trivial for makers to include 4 channels at the traditional 2CH price point.

[Bud]

The market these days puts RGB lighting in everything including toilet seats. I am yet to see any useful utility it provides.

[pdenisowski]

in many cases protocol problems stem from problems in the analog domain.

In my experience, it's "most"  

[pdenisowski]

Why stop at 4? Why not insist on an 8 channel scope.

Speaking as someone whose company just released its first 8 channel scopes (MXO5 and MXO5C):  there are a lot of cases, e.g. power sequencing (i.e. timing between power rails), when 8 channels is useful / necessary. 

Believe me, we wouldn't have developed (two models of) an eight-channel oscilloscope if we didn't have a really strong business case

And speaking of power:  one thing that people often overlook when it comes to channel counts is power measurements: generally speaking, if you want to measure power, you're going to need two channels: one for voltage and one for current.  A basic power efficiency measurement (Pout/Pin) requires 4 channels all by itself.

But speaking as a hobbyist (spending my own money), I think it depends entirely on one's needs / budget, so IMO it's good to have a choice between two and four channels - there was a time when that wasn't really a choice (see attached)

[EEVblog]

in many cases protocol problems stem from problems in the analog domain.

In my experience, it's "most"

And that's the problem with logic analysers, they can't show that. They only work on electrically robust systems. The first thing I do when troubleshooting a digital system is to use a scope to eliminate any signal integrity issues.

[EEVblog]

But speaking as a hobbyist (spending my own money), I think it depends entirely on one's needs / budget, so IMO it's good to have a choice between two and four channels - there was a time when that wasn't really a choice (see attached)

And that's the point I'm making. In general, scope choice has always been about the best "bang-per-buck", and 4CH has never offered that in the past, so it usually wasn't a viable choice. Now it's clear that best bang-per-buck option is the 4Ch domain.
My first >2CH scope was a used Kikusui COS6100M, and even then I didn't have 4 full channels.

[Peabody]

Your original question was not whether someone should buy a two-channel scope, but why manufacturers even bother to offer them anymore, the implication being that they shouldn't, which would be bad news for hobbyists.  And haven't we been down this road with you before?  "Is an $800 scope overkill for a hobbyist?"  What do you think electronics hobbyists do?

One of my other hobbies is photography.  Ask any professional photographer what kind of tripod you should buy, and he will answer a carbon-fiber model costing about $400.  It will be very light, amazingly sturdy, and will last a lifetime.  And that's the right answer for a professional, but clearly the wrong answer for an amateur.  If you buy a $50 aluminum tripod instead, or even an aluminium tripod, the remaining $350 will get you two complete flash light stands, including the stands, the flashes, the umbrella adapters, the umbrellas, and the radio triggers.  And the tripod will work well enough.  When it comes to buying gear, the professional can be the amateur's worst enemy.  The needs and requirements are often just completely different.  And you know, people do have budgets, and the amateur may have no revenue from his hobby to pay for all this fancy stuff.

Well, I guess for an EE a 4-channel scope is the right choice.  But if there is any difference in price, a 2-channel scope might well be the better choice for a hobbyist.  I could even make a case for getting a two-channel, battery-powered, hand-held "toy" scope of the type you've summarily dismissed.  They are pretty good now, and cost $100 or less.  That's what I use (but very rarely both channels).

[Kleinstein]

With more channels there is a chance that one runs in the ground loop problems, at least for small signals.

With 2 channels one often has an extra trigger input, as a very limited 3rd input. So the step from 2 to 3 is not that large. Granted that a real input channel makes triggering easier with adjustable attenuation / gain.

With a dedicated 2 channel design (not just cut back from 4) one is more likely to have seprate knobs for the vertical range and thus a more intuitive UI - at least for those who where used to analog.

The power sequencing case is indeed one where many channels really help and even more than 4 may be ideal. One can however still use a master for the trigger and than do it with a few channels at a time.
Of cause it gets nasty when one is looking for glitches in the power on sequence (e.g. depending on mains phase).

Quite a few of the features of a scope are rarely used, but it is still nice to have them. The 3rd / 4th channel can also act as spares - in case one gets damaged.
With some scopes the step from 2 to 4 channels also comes with a 2nd ADC and thus more sampling rate when using 2 channels.

If a $800 scope is overkill for hobby use really depends on the income situation. If you have the money this may be no issue at all - in countries like India or China this may very well be an issue. DSOs got really affordable - they were pretty expensive only some 30 years ago.

[EEVblog]

Your original question was not whether someone should buy a two-channel scope, but why manufacturers even bother to offer them anymore, the implication being that they shouldn't, which would be bad news for hobbyists.

That was not my intention at all. It was to explain why 2CH is still needed and requested by the industry. As discussed in a recent forum thread on the Keysight scope which provided the inspriration, I thought a lot of people my not know about the educational market and the industrial test markets.

And haven't we been down this road with you before?  "Is an $800 scope overkill for a hobbyist?"  What do you think electronics hobbyists do?

*snip*
Well, I guess for an EE a 4-channel scope is the right choice.  But if there is any difference in price, a 2-channel scope might well be the better choice for a hobbyist.  I could even make a case for getting a two-channel, battery-powered, hand-held "toy" scope of the type you've summarily dismissed.  They are pretty good now, and cost $100 or less.  That's what I use (but very rarely both channels).

I will repeat a response I put on the video comments:
I was talking about how some people seemingly can't understand what *general advice* is. As if this video was specific advice *for them* and *their situtation*. I literally said it's best to get a 4CH "unless you need to save every last cent". As a leading voice in this field I would be a laughing stock if my *general advice* was to get a 2CH scope when a 4CH scope today is *clearly* better bang-per-buck. As always, YMMV. As a creator I can't possibly make a video that takes everyone's personal financial and requirements situations into account.

It is literally impossible for me to make an advice video that cater to every need an requirement, that's just silly. Not that this was an "advice" video anyway, it was nothing of the sort!

As for "summarily dismissed", rubbish, have you actually watched some of my pocket scope reviews? Unless it's absolute garbage, my general advice has been along the lines of "this is ok and can be useful for certain requirements".
I don't know how many times I have to say it, YMMV.

[EEVblog]

With a dedicated 2 channel design (not just cut back from 4) one is more likely to have seprate knobs for the vertical range and thus a more intuitive UI - at least for those who where used to analog.

Yes, handy.
But 2CH scope usually has only one ADC, so you get half the sample rate for two cahnnel.
A 4CH scope usually has two ADC's and you can get the full sample rate on two channel by using 1/3 or 2/4.
Handy benefit.

[EEVblog]

If a $800 scope is overkill for hobby use really depends on the income situation. If you have the money this may be no issue at all - in countries like India or China this may very well be an issue. DSOs got really affordable - they were pretty expensive only some 30 years ago.

IME, the $800 price point has always been about what you pay for a decent entry level scope.
Back in my day it was a 20MHz dual channel analog.
Then you'd get progressively better bang-per-buck over the years.
It wasn't until the DS1052E did you get a digital for that $800 price point.
Then that price actually started to drop where "entry level" is around the $400-$500 price point now.

[Gyro]

in many cases protocol problems stem from problems in the analog domain.

In my experience, it's "most"

And that's the problem with logic analysers, they can't show that. They only work on electrically robust systems. The first thing I do when troubleshooting a digital system is to use a scope to eliminate any signal integrity issues.

This is where there is a disconnect. A scope is fine for the 'basic' slow interfaces - I2C, SPI, serial etc. (although as I said the decoders can be buggy, for instance the Siglent SDS1202X-E doesn't get the stop bit settings right on simple async serial - unless they've fixed that with an update by now). To be fair, these are the ones that are more likely to be 'casually' implemented and so suffer more signal integrity issues.

There's a difference between these types of interfaces and 'big boy' interfaces like high speed parallel video busses, DDR3 etc. where you have to pretty much guarantee signal integrity by design. It's one thing to 'coo' over the cute length matched traces, it's quite another to actually have to design that stuff. Both ends disappear under BGA packages, and you can't afford the stubs caused by test headers, you have to align traces, leave a rectangular solder mask window and use Mictor probes. My personal logic analyser is a 34 channel Agilent, 250MHz timing, with 100k 1.5pF probes - and that's nowhere near current crop, it's around 25 years old. Try putting your x10 passive scope probes on those signals and see what happens - at very best, you'll be looking at the signal integrity that you just trashed. Of course you can buy 4 very expensive active probes, but they will cost more than the scope. With luck, you might get away with custom Z0 probes (effectively what the LA probes are). The best you can do in signal integrity terms (apart from designing it right in the first place) is to wind down the thresholds to below the IC worst case input specs and look for reflections.

Anyway, leaving aside logic analysers, it's the effect of "WHY sell 2CH Oscilloscopes?" type videos that worries me. The very obvious answer is that there is a market for them. Looking around, I can get a Rigol DS1102Z-E for reasonable £250 in the UK, a pretty good deal compared to a short buffer buggy Hantek. A DS1054Z on the other hand can be had for about £350 - now that's not just £100 difference, it's also a 40% more. You can go on more or less indefinitely, adding £50-100 more to get this or that model with a 'must have' fancy feature. It's a honey trap for beginners [Edit: "best bang-per-buck" doesn't cover it, it should be 'how many bucks do they need to spend'].

It's probably fair to say that the viewership of your videos is heavily slewed towards beginners (not your fault). It worries me when, particularly over recent months, I've seen a number of beginner scope threads where the OP's 'entry level' shopping list consists of various 2Gsample and/or 12 bit (and yes, 4 channel) models, because that's all they see you getting excited about in recent videos. I wonder how many of them will spend £500 on a scope, play around unsuccessfully with some 555s and opamps on a breadboard, get disillusioned and stuff it in a cupboard, together with their fancy AWG and other stuff. Maybe the 20MHz bandwidth filters should be enabled by factory default.

Bottom line, maybe it's time to start putting out some videos on bargain scopes and equipment that will suit beginners, get them through the first couple of years of learning, playing with basic stuff, Arduinos etc. (maybe a lot of them will never need to go past that stage), and see if they're actually going to stick with the hobby. Either that, or start putting a 'parental warning' saying that beginners really don't need this level of stuff, save your money. I think there's a danger of disconnect with your actual (versus target) audience.

[madires]

When repairing stuff I very rarely need more than two channels. And dedicated knobs for the second channel are quite handy.

[nctnico]

A 4-channel scope in general is a solution looking for a problem. I've been doing RF for 25 years and I needed 4 channels may be 2 or 3 times. People doing digital may need it but there are better tools for that and they likely have such tools already.

My rule of thumb:
For doing repairs a two channel oscilloscope is typically more than enough. And 100MHz bandwidth is plenty. 4 channels is something you need if you are working on digital microcontroller circuits and more complex circuit development in general.

[David Hess]

But 2CH scope usually has only one ADC, so you get half the sample rate for two channel.

That is implementation dependent.

Increasing integration made digital triggering less expensive than analog triggering, so if digital triggering and all of the built in trigger modes apply to the external trigger input, then the external input has to go through the same digitizing process into the FPGA as the vertical inputs.  What is not required is the sample memory.

So in this case, the only difference between an external trigger input and a vertical input is the lack of sample memory and analog signal conditioning like attenuators.  The digitizer is still shared, or duplicated.

The cost difference could be just sample memory and analog signal conditioning.  I suspect the most important reason to provide two channel DSOs is price discrimination through product versioning.  They can charge more for their 4 channel "premium" DSOs by offering 2 channel "economy" models while the costs are practically identical.