5

This question is somewhat related to this other EE.SE question, and some hints are given there, but are not completely satisfactory for me.

I've understood that, besides safety considerations, the biggest obstacle to making a fully differential, i.e. floating, input stage is that it is far more difficult (hence costly) to design a wide band DC-coupled differential amplifier than to design a single-ended amp having the same bandwidth (or the same performance, in general).

The usual answer could be "OK, you need a differential input, so buy a (costly) differential probe like this and be done with that".

On the other hand, oscilloscopes are constantly dropping in price, with companies offering ever more functionalities and performance at lower prices (heck! In the low-end market the venerable Rigol DS1054Z has had 4 channels and lots of features for years and now is getting competition from Siglent and others Chinese manufacturers, too).

So, my question really involves answering to the following points: why embedding the differential probe circuitry in a scope is not deemed a right step to do to gain market share? Wouldn't it be beneficial to the average user to be able to use a scope input like a floating multimeter input? Surely the circuitry of the diff probe would become less expensive once embedded in a bigger device and mass market production dynamics kicks-in. Am I wrong in assuming this? Is there something in a differential probe circuitry inherently so difficult to design/build that makes it not quite amenable to cost reduction by volume production?

Note that I know that there are some rare models that have differential inputs, but my question is really about why, with the continual production cost reduction of oscilloscopes, this has not become the rule among manufacturers.

LorenzoDonati4Ukraine-OnStrike
  • 24,350
  • 5
  • 62
  • 107
  • Don't forget that new functionality is all in displays and computers, where bigger players compete and drop costs. At the same time analog circuitry, fast adcs, remain a niche, so there is not as much competition. –  Dec 19 '17 at 07:36
  • @GregoryKornblum Yep, I understand that the analog front-end and other high-speed sections can be harder to be dropped in costs, but they *do try*. IIRC in the reverse engineering EEVblog video of the DS1054Z front-end, Dave Jones actually showed how they used discrete JFETs to cut costs down relative to their higher level models. Therefore I often felt they don't try hard to add differential inputs just because the market doesn't require that, which seems funny. – LorenzoDonati4Ukraine-OnStrike Dec 19 '17 at 07:57
  • Trust me, uf basic device becomes a little more expensive and you can't sell it to all your customers that don't need the new feature, it's a huge problem. –  Dec 19 '17 at 08:00
  • My guess is that many of the potential customers (for a cheap scope) would not even understand **why** they would need a diff. input. or how to benefit from it. Then the probes / inputs: with a diff. input you cannot use BNC anymore. Which means new probes. What will these look like. The red/black testleads on a diff. probe. do not give me confidence of having a high bandwidth. I mean: a diff. input can be a good idea but there are significant implications/changes to be made to a scope's design as it is now. – Bimpelrekkie Dec 19 '17 at 08:12
  • @Bimpelrekkie I understand that low-end scope users wouldn't understand all the implications, but I cited that example only to show how cheap the market has gone. Even more high level scopes from high-end manufacturer are getting cheaper. That's why I was interested in why companies are not investing in such a "paradigm" switch. – LorenzoDonati4Ukraine-OnStrike Dec 19 '17 at 08:44
  • What do you mean by high-end, and cheaper? On my bench I have a 40k+ Lecroy, and if I need it we have a 5k differential probe. Not something the average Joe would ever buy... – Vladimir Cravero Dec 19 '17 at 08:47
  • @VladimirCravero Of course I'm not talking about the bleeding edge, top of the range scopes here. But definitely highly reputable companies like Keysight, Rhode&Schwarz, etc. are trying to compete on lower-end markets (e.g., see EEVBlog reviews of last two years, for example). It seems that a floating differential input is not a feature they want to invest into. Therefore I'd like to know what are the technical aspects that makes it so difficult. – LorenzoDonati4Ukraine-OnStrike Dec 19 '17 at 09:09
  • To whoever voted to close: I'm not asking for opinions, but for documented technical/industrial reasons (or expertise from people in the field) and explanations of technical difficulties about why what I ask has not been implemented. This has nothing to do with an opinion-based answer (of course I assume there is someone that has the information I seek, but the information is not opinion-based at-all, unless an industrial decision related to electronics design is deemed merely "opinion"). – LorenzoDonati4Ukraine-OnStrike Dec 19 '17 at 09:52
  • Look around how people use their scopes, and you might be able to see that almost nobody really needs a fully differential input. For the few situations where you might not want to be grounded the way you are, isolation transformers can be the answer, or that one differential probe that is shared among 20 scopes in the whole lab (instead of being implicitly paid for in each of the 20 scopes). – PlasmaHH Dec 19 '17 at 11:25
  • "why embedding the differential probe circuitry in a scope is not deemed a right step to do to gain market share? Wouldn't it be beneficial to the average user to be able to use a scope input like a floating multimeter input?" at least these two questions out of the five are in my opinion opinion based, and the obvious opinion of the contenders of the market is "No it is not worth it". If you don't want that to be answered, don't ask these questions. – Arsenal Dec 19 '17 at 12:18
  • @Arsenal "If you don't want that to be answered...", on the contrary, I'd like them to be answered. The point is, I'd like them to be answered with hard data, if possible. Up to now no one has posted an answer where there is an attempt to explain why is it still not feasible/useful. It is self-evident that market didn't go that direction, but I asked *why* it didn't go there (bench multimeters have floating inputs, and everyone is happy with them; why can't scopes have those?) and I suppose there are technical and/or economical reasons, which are not "opinions". – LorenzoDonati4Ukraine-OnStrike Dec 19 '17 at 15:06
  • @PlasmaHH Well, that's the situation as it was until now. If a diff probe costs that much, that is the right usage pattern: buy one and share among many users. But if the production costs went down (and scope hardware *is* getting cheaper), why wouldn't user be happy with having any channel behave like a diff probe? Maybe the answer is "the technology is not that cheap yet", but the fact is that no one has posted an answer with some hard data or a well fleshed out explanation, yet. – LorenzoDonati4Ukraine-OnStrike Dec 19 '17 at 15:13
  • @LorenzoDonati: It should be obvious that full differential will always be more expensive than the classical solution we have now. Even more so if you get to really high bandwidth. Given that lower end scopes are getting cheaper mainly because they get lower in parts counts (my phone looks more densely populated than some of the recent ones) this is likely to be a significant part of the cost. Especially on a budget, would you pay 20% more for a feature you are not likely to use? N(scopes)xM(channels) diff amps instead of one or two differential probes for your lab? Handhelds fill the gap. – PlasmaHH Dec 19 '17 at 15:43

1 Answers1

2

Differential probes usually used to monitor very fast signals with relatively small amplitude. In such case diff amp must be placed as close to the source as possible. Proximity is a key feature.

Also, I'd like to point out that single-ended isolated input usually is not equal to differential input. It's easy to make 1500 V isolation of the whole single-ended circuit, but it's utterly not so easy to create fast diff amp with such allowed common-mode voltage.

Eugene K
  • 918
  • 5
  • 9