Products > Test Equipment

Oscilloscope specs - which are really important

(1/5) > >>

I'm still (after 6 months) trying to determine which oscilloscope to get and how much to spend. I know the theory more or less (I have a degree in Electronics and many years experience of working with microwave integrated circuits but very little practical experience with oscilloscopes, digital circuits etc.) but I'm finding it hard to balance what is marketing and what is really useful. I'm looking to get a good quality 100 or 200 MHz scope for my own use ranging from hobby circuit building to perhaps physics like experiments, which will hopefully last me many years.

Any comments by owners of scopes on the following would be great!

1.) Memory. This seems to vary from a few k (definitely too small) to 56M (the Rigol) with 1M seeming typical.
500k gives a zoom level of a 1000 times while 50M gives a zoom level of a 100,000 times. Is a 1000 times enough or do people often require higher levels? My gut feeling is that memory levels are being pushed higher as a headline spec but I wonder how much importance should be attached to the difference between say 1M and 10M.

2.) sin x/x interpolation. With 1GS/s and 500 pixels of screen resolution interpolation only comes in at faster than 50 ns/div. Even at 2 ns/div there are a couple of sample points per div so the error introduced by linear interpolation is less probably than the dc gain error. Additionally, as sin x/x interpolation assumes a repetitive waveform the same effect could be obtained with equivalent time sampling?

3.) Hi-res mode. The author of the on-line blog article oscilloscope dreams states he wouldn't buy an oscilloscope without hi-res mode but it seems to me that it offers the illusion of increased accuracy which might even be misleading. The dc gain accuracy of the oscilloscopes I've been looking at range from around 2% to 5% or more. 8 bit convertors will have a least significant bit error of less than 0.5%. I can see the advantage of averaging out white noise but to go to 12 bits of precision when the uncertainty of the values is so much more seems to be a bit optimistic. (I remember being in a taxi in Tokyo once and seeing a plate attached to it giving its height to the nearest mm!)

4.) Built in serial decode options. These seem attractive but they are generally quite pricey and have to share rather restricted screen space on the oscilloscope. Stand alone solutions are of a similar price and can use say laptop display space. The main question is how much should a DSO buying decision be based on possible future options being available which I suspect I might not end up wanting to fork out for.

5. General quality. Comparisons between scopes that I can find seem to only consider features such as memory and fps. General quality, precision of timebase and gain don't seem to be mentioned yet I would have thought that these should be at least as important.

I realise that how useful features are depends on what you need to do. My problem is I don't know my future needs in detail. I don't want to buy features that I can live without but don't want to buy a scope and find I'm frustrated a year or two down the road.

What I'm looking for is people to say that feature x is great because they need it for y, I can then decide if I'm likely to also do y and therefore need x!

If you don't know what your needs are now, how can someone else tell you? I understand your desire to try to make an optimal choice now, but I think there is too little information on what you want to do with the oscilloscope. My advice is to by the cheapest scope that satisfies your minimum needs and see how that works out, after using it for a while you will gain experience and then  you will be better prepared to make a better choice if you feel the need for it. Don't invest to heavily into something uncertain, there is always the possibility to sell and get a better scope. A  Rigol DS1102E maybe?

What is really important?

Bug-free firmware and support that fixes issues in a timely manner.

If you search this forum for the WaveAce224 threads, you will see what you don't want. I once owned this device, which has multiple nasty issues. I documented and reported all of them. Not a single one has been addressed or fixed so far. The latest firmware update for this device is almost 3 years old. In the end I returned the device and got my money back.

Not getting firmware fixes seems to be especially likely for devices that weren't developed in-house and bought from somewhere else, just to bolster up the low-end range of scopes for some brands.

It's really only you that can decide what your requirements are. Long memory depths, for example, are handy if you want to be able to capture a stream of data and then zoom right in to see what the shape of the signal looks like in detail at some point - but then again, if you know what you're looking for, you may well be able to trigger on the event of interest and speed up the time base. The scope I use most often is a fairly old Tektronix with a 10k point memory, and for 99% of the work I do, that's plenty.

Serial decodes are nice to have if you often need to probe a serial bus for some reason, and completely pointless otherwise. I use I2C quite a bit, and on more than a few occasions I've ended up decoding it a bit at a time by hand, so automated decoding would have been nice, and I'll definitely get it on my next scope if the option is available. But if you don't plan on using interfaces like this, don't bother.

I quite like hi-res mode, though I wouldn't regard it as essential. It can give a very clean trace which allows underlying small changes in a signal to be shown which might otherwise be masked by noise. On the other hand, it does limit bandwidth and can hide important signal behaviour too.

Here's an example of how different acquisition modes would show the same signal.

Intensity graded "analogue" display (an essential feature IMHO, standard on any professional grade scope):

Simple digital sampling, typical of an entry level, hobby grade scope:

Peak detect acquisition - tends to exaggerate noise, but doesn't miss important signals like simple sampling can:

Hi-res mode - good for the simple digital signals, not quite so good for the high frequency modulated analogue signal!

1) Granted, long memory is a useful feature sometimes. But.

I think the memory depth is less important for the zoom level, than it is for the possibility to retain high sample rate at low timebase settings. Unfortunately there's no 1-to-1 correspondence betwen depth and sample rate decrease. (Have a look at my rant here if you like.) I've haven't seen any scope datasheet stating at what timebase the sample rate goes down. You have to test that yourself to get an objective figure.

Having to choose between different memory depth is annoying. Automatic depth selection makes life easier.

Long memory may slow down scope response. Annoying. Check if it does.

"Segmented memory" is a very nice feature and in my opinion more useful than 10x more memory. If it doesn't have it, see to that it's an option.

I think at least 1Mpts/ch, but more than 10Mpts/ch no great advantage. There are pretty cheap protocol analyzers that can do better job for SW debug.

2) not necessary if you have 1GSa/s and only 200 MHz BW

3) Useful somtimes but may play tricks on you, especially when you're not so experienced with scope. But anyways, most scope have it, don't they?

4)  Select a scope that already have, or at least you can upgrade to serial trigger/decode. Some scopes come with serial trigger but decode is optional upgrade. You'll likely to end up with a I2C or SPI interface to a DAC, ADC, etc on some design some day.

5) Naturally is quality important, not so much more to say. Meaning "avoid the chepeast crap".

6) At least 4 channels. You can get away with two channels but it's tedious. Startup behaviour and SPI are two examples. You may even find yourself in a situation where two channels just ain't enough. Option for digital channels will be useful for any design involving CPLD/FPGA(external memories etc.

7) General usability and user interface. Separate knobs for each vertical channel is mandatory (for me). Fast response to knobs and buttons. Frequent used features shall not need you to move your hand over the scope like you're waving away a mosquito. Fast saving of data to USB stick or computer.

8 ) Display resolution: the higher the better. However, how data is processed/displayed is probably more important (with reasonable minimum resolution). Anti-aliasing function avoids confusion.

9) Capture/acquisition rate...the higher the better of course. But the datasheet capture rate doesn't say much; test yourself and see what happens with more channels, different timebase settings, etc.

10) Rise time is more important that bandwidth, but of course they are related.  Logic rise/fall times are down around 1 ns nowadays, so go for 200 MHz (or 100 MHz to start with if you can upgrade to 200 MHz).



[0] Message Index

[#] Next page

There was an error while thanking
Go to full version