Which oscilloscope do you recommend? Moderators: adafruit_support_bill, adafruit

[pearlpearl]

Hi, I am planning to buy an oscilloscope for various projects. What Bandwidth, Sample rate, and Memory depth should the scope have? Could you please provide a list of three recommended products? I am considering both 2-channel and 4-channel scopes.

[adafruit_support_bill]

It would help if you could give us a rough idea of your budget and describe in more detail what you want to use it for so we can narrow down the field.

Scope capabilities range from basic 2-channel analog scopes to multi-channel mixed-signal scopes with built-in protocol analysis capabilities. Storage and bandwidth vary considerably between the portable USB scopes and the high-end dedicated bench instruments. As you might expect, the price range is as wide as the range of capabilities.

[adafruit_support_mike]

You also need to characterize what you're doing.

For audio frequency signals, anything upwards of about 20MHz will work. If you want to debug digital signals at microcontroller speeds, you'll want something in the 100MHz to 200MHz range, and might want to look at a mixed-signal scope. If you want to do radio, you're looking at something in the gigahertz range.

[pearlpearl]

I plan to use the device for various mechatronics and motor control projects. Maybe later for signal processing projects. Not sure yet. A device that could be used with all (or most if there is a big price different) of your products would be nice. The ability to store the data that I could use to display graphs in EXCEL or Matlab would be a plus. Budget size may be somewhere between $500-1500.

[adafruit_support_bill]

Mike is the 'scope expert, so I'd follow his advice over mine. The Rigol scopes in the store offer a lot of features for the money. https://www.adafruit.com/products/1987
Personally, I use a Rigol DS1102D for most things.

[adafruit_support_mike]

The DS1102E is an excellent scope, and Rigol has done an amazing job of lowering the cost of entry for good quality scopes.

Another option at the same price is the DS1054Z: https://www.adafruit.com/products/2145 Twice the channels, but half the bandwidth (50MHz as opposed to the 1102E's 100MHz).

As a rule of thumb, assume you need a scope whose upper limit is 5 to 10 times the fastest signal you want to measure. That's especially true for digital signals, because the rising and falling edges of square waves are bundles of high-frequency information. A 50MHz scope can't tell the difference between a 50MHz square wave and a 50MHz sine wave. The more headroom you have between your signal and the scope's upper limit, the more useful information you can get about the fastest parts of the signal.

My day-to-day scope is a 100MHz 4-channel Tek, and I really like having four channels.

Most of what I do involves looking at the relationships between signals: Does the output behave as expected for the given input? How much lag is there from this part of a circuit to that part? etc. You can do simple comparisons with a 2-channel scope, but with four channels I can probe multiple points along a signal path.

[pearlpearl]

What is the model number of your 100MHz 4-channel Tek? What sampling rate would be sufficient for 99% of the cases?

[adafruit_support_mike]

It's an MSO2014B 100MHz mixed-signal scope: 4-channel analog, 16-channel logic.

The sampling rate you need 99% of the time depends on what you do 10% of the time. With any measuring tool, 80% of the readings you take are fairly sloppy. You care more about the general behavior than the fine details. The remaining 20% of the time, you're looking for something specific and do care about the details. In about half of those cases, you'll want as much information as you can get.

For those measurements, you'll hit one of two limits: the smallest useful detail of the thing you're measuring, or the smallest detail your tools can resolve.

As a rule of thumb, you want an oscilloscope about 10x faster than the fastest signal you want to measure.

Without getting into the details of Fourier wave theory, you get more detail about signals every time you increase the sampling rate. When you stop getting more detail, just wider and wider copies of the same overall signal, you know you've hit the limits of the circuit under test. If you keep seeing new detail until you reach the scope's fastest sampling rate, you know you've hit the limits of the scope. A 10x change in sampling rate is enough to confirm you aren't seeing anything new.

Most of the signals I work with stay below 10MHz, so a 100MHz scope is suitable for my needs. OTOH, I just built a 0-to-1uS adjustable pulse generator that I couldn't characterize below 15nS because the scope couldn't keep up with the signal. I had to calculate the scope's impulse response and run the math backwards to estimate the circuit's performance down to 1nS.

I don't do enough work where the timing below 10nS really matters to justify the cost of a 1GHz scope, but will probably get there eventually.

I needed the pulse generator to trigger a circuit whose response had to be measured at 100mS per division, so even in one circuit I was measuring at both ends of the speed range.