OT: Oscilloscope Purchase

[J Tiers]

Nyquist says generally you want 4-5x sample rate, anything past that and you start getting to the point of diminishing returns. Plus you start wasting capture memory.

Digital scopes will show you a point cloud if you enable it. And in storage mode you can zoom in and see all the points too.

One thing to watch out for is many scopes divide up the sample rate among the channels. So on a 4 channel scope that’s rated for 1GS/s you may get 500MS/s in two channel mode and 250 in 4. Then there are cheap scopes that say that are 100mhz but they only have a 100MS/s rate which does not work so hot. They do some averaging to get a waveform but there are many potentials for weird things and aliasing. They are really only good for 20-30mhz.

But really I don’t know how many of us here will ever look at signals even close to 20MHz.

What you want for SMPS work is to include harmonics, which means transients, and seeing the edge transition times on gate drives etc.

Remember, on ANY scope, analog or digital, if you get close to the frequency limit with the fundamental, you filter off the harmonics. So, if you look at a 20 MHz signal with a 100 MHz 'scope, you see only the first 4 harmonics, the rest are filtered off.

When you think about it that way, you quickly see that a 100 MHz 'scope is not an extravagance at all. And the sample rate is easily seen to be important. A 1GHz sample rate translates to a cutoff which cannot be higher than 500 MHz, but at that frequency, you get only one sample..... with 100 MHz, you now get 5 samples per half cycle, which allows at least some semblance of detail. In reality, the 1 GHz sample rate provides detail similar to an analog 'scope up to around 30 MHz, maybe less.

Of course, the "sample rate" of the analog scope is related to the cutoff frequency. And it will lie to you by showing you a nice continuous waveform, with a slight bump on it. That "bump" is actually your evidence of a big transient at a frequency significantly above the cutoff frequency.

The good news is that you saw it. And that with the analog 'scope, the cutoff is typically more gradual, so things can still be "seen", although the amplitude cannot be trusted in any way, above cutoff. With the digital 'scope, you probably would not have seen that transient at all.

[nickel-city-fab]

Nyquist says generally you want 4-5x sample rate, anything past that and you start getting to the point of diminishing returns. Plus you start wasting capture memory.

But really I don’t know how many of us here will ever look at signals even close to 20MHz.

Last I went up to 26 mHz was around 30 yrs ago, working on a shortwave receiver with an analog scope. The features of the newer digitals are tempting, but quite frankly I'm not familiar enough with their nomenclature, to be confident in buying. Besides the fact that I'm not very active in electronics tech any more. I think more samples/sec is better, correct? But then I see the prices on those scopes and have to check my underpants.....

[macona]

Nyquist says generally you want 4-5x sample rate, anything past that and you start getting to the point of diminishing returns. Plus you start wasting capture memory.

Digital scopes will show you a point cloud if you enable it. And in storage mode you can zoom in and see all the points too.

One thing to watch out for is many scopes divide up the sample rate among the channels. So on a 4 channel scope that’s rated for 1GS/s you may get 500MS/s in two channel mode and 250 in 4. Then there are cheap scopes that say that are 100mhz but they only have a 100MS/s rate which does not work so hot. They do some averaging to get a waveform but there are many potentials for weird things and aliasing. They are really only good for 20-30mhz.

But really I don’t know how many of us here will ever look at signals even close to 20MHz.

[Paul Alciatore]

Oh, if I haven't said it already, thanks to all of you for helping me with this decision. I don't know if any of the comments here have changed the outcome, but at least I can tell myself that I pushed all the buttons.

So thanks very much to all.

[Paul Alciatore]

I could not agree more about the possibility of being deceived by digital scopes. It is commonly thought that the theory behind digital sampling says that any waveform can be reconstructed if it is sampled at twice the highest frequency component in that wave. But a careful examination of what is really being said leads to something a lot different.

It is a simple and easily demonstrated fact that there are not just two or three or even a dozen or a hundred different waveforms that can produce the exact same samples. There are literally an infinity of them. It is only when you start imposing some restrictions on the original waveform that the number of possible, similarly restricted waveforms is lowered to something manageable. And even then, the one waveform that a scope will display is still reliant on the exact filtering or processing that the output of the digital to analog conversion process. And don't think that these scopes do not convert back to analog. An analog waveform is exactly what is shown on the face of the display. They do not show a scattering of digital points. They interpolate between those points to show a continuous, waveform. Some, if not all of these scopes do allow you to view just such a display of the digital points from the samples. But that is not their normal mode of operation. These deficiencies are most evident when you are talking about frequencies that are near the limit of the scope.

On the other hand, analog scopes also have their shortcomings when you are near their upper limit of the bandwidth. I have been there many times, staring at a lump or wiggle on the displayed line and wondering just what it really is. And what it was doing in the circuit that I was working on.

That common misconception about the reconstruction of a waveform is perhaps supported most by the use of digital sampling in audio signals. The human ear is the final target of recorded sound and one of the things that a human ear can not do very well is distinguish between signals where phase shifting has occurred. So a 20 KHz sound wave which is sampled at only slightly more than 40 KHz can sound to our ears very much like the original sound even if it is not a really accurate reconstruction of the original waveform. A sine wave would sound much like a square wave or a triangular wave at that point. And there are very, very few people, if any at all, who could tell the difference in a true, double blind test.

Unfortunately digital logic circuits can and do react to differences that the human ear will be totally unaware of.

Frankly, and in my HUMBLE opinion, if you really want to know everything about a waveform and you are using a digital scope, then you need to sample it at a rate that is AT LEAST 10 or 20 times the highest frequency of interest. That's my opinion and I am sticking to it. But, of course, when one is buying his own equipment, there are, there must be some practical considerations. And that is where I am today.

This is very true..... You can be fooled easily by digital, especially by "aliasing" at the limits of the frequency response.... transients can actually be far different from what you see on digital. Then you want an old analog "storage scope". With those you will spend 10x the time messing around to get to see what you want, but you will get the correct answer if you really need it, at least as far as the waveform. Measurements etc are tons easier on digital, and trigger modes totally blow analog away. You just have to avoid getting fooled.

[skunkworks]

my first 'nice' scope was a b&k 20mhz? dual channel scope... Dad got it from an auction at a tech school. Didn't work. I watched my dad and uncle work on that thing all afternoon (both really good electronic techs..) (probably mid 80's? I don't really remember)

it ended up being a open 1m ohm resistor that was open.

we still have it - don't use it anymore. The rigol was the first scope I had with auto.. OMG - I didn't know what I was missing!

sam

[macona]

One thing I like about the digital scopes (besides the small size and light weight) is the fact that the higher-end models can replace several separate measurement tools, or entire racks of equipment.

Speaking of high-end, don't forget National Instrument: You can often find their older modules and systems a lot cheaper than the new ones. https://www.ni.com/en-us/shop.html

Yup, the lecroy waverunner I picked up a couple years ago has all sorts of features that are software locked, lucky for me someone figured out how to create the activation codes. There are all sorts of things, serial decoders, logic analyzer options (With the USB breakout), basic spectrum analyzers, canbus stuff, and a whole bunch of more special bus interface stuff and eye diagrams.

The only bad thing about modern scopes is the boot time, sometimes I finds myself just turning on my old Tek 2465 because I don't want to wait for the wave runner. I also have a tiny 7MHz Hitachi battery powered scope with about a 3" CRT, great for stuff where I want complete isolation. It had two 6v lead acids in it, replaced that with a small 3 cell 18650 lithium pack with a boost converter to get the full 12v. Actually runs longer now and significantly lighter.

I did order one of the national instruments usb based scopes last week for a test fixture, mainly for the ease of integration into labview. We also use the picoscope usb scopes at work and they seem to work well.

[Paul Alciatore]

Well, I finally pulled the trigger. I ordered a Hantek DSO2D15 plus two 200 Mhz probes from Amazon. I think the included arbitrary function generator took me over the finish line. Now comes the infinitely long wait for delivery. That's OK because I need to clear some space on one of my electronic benches.

[nickel-city-fab]

One thing I like about the digital scopes (besides the small size and light weight) is the fact that the higher-end models can replace several separate measurement tools, or entire racks of equipment.

Speaking of high-end, don't forget National Instrument: You can often find their older modules and systems a lot cheaper than the new ones. https://www.ni.com/en-us/shop.html

[J Tiers]

You have to buy a lot of digital scope to equal the detail you see with an analog scope. Of course if you need to store waveforms or see a lot of pre-trigger, then digital is it. Even though I have access to some very nice digitals (I work at Keysight), I still use my old 1980B to see fine details.

This is very true..... You can be fooled easily by digital, especially by "aliasing" at the limits of the frequency response.... transients can actually be far different from what you see on digital. Then you want an old analog "storage scope". With those you will spend 10x the time messing around to get to see what you want, but you will get the correct answer if you really need it, at least as far as the waveform. Measurements etc are tons easier on digital, and trigger modes totally blow analog away. You just have to avoid getting fooled.

[Forestgnome]

You have to buy a lot of digital scope to equal the detail you see with an analog scope. Of course if you need to store waveforms or see a lot of pre-trigger, then digital is it. Even though I have access to some very nice digitals (I work at Keysight), I still use my old 1980B to see fine details.

[rcaffin]

I started with Tektronix CROs around 1970. Wonderful machines. Yeah, went through the full range as others have mentioned.
My last Tek was a TDS 220. Over time it needed repairs to the PS and to the front-end BNC connectors. That was OK. Then the LCD died. Um.
I was a bit disappointed with the lack of support from Tek: no cct dia for the PS as it was bought-in from China.
So, search on ebay.

Hantek DSO5102P looks exactly like a TDS-220. In fact, I now suspect that Tek bought the TDS 220 from Hantek as an OEM job, which is why Tek did not have cct diagrams. It is virtually an exact copy. No learning curve. And light enough to prop up on the CNC when it needs some TLC.

So, new DSO5102P. It does the same things as the TDS 220, but in colour (oh wow ...) Sold the TDS 220 to someone technical (with full disclosure) who is now trying to buy a functional replacement LCD screen from ebay. The first one was DOA.

The DSO5102P is presently monitoring the current through a brushless DC motor fan out of the fridge. Every now and then the fan seems to lose one phase, and makes a horrible noise. But right now ... the two last replaced fans are running perfectly. I dunno ... May have to look inside it.

Cheers
Roger

[nickel-city-fab]

In 1960-62 I helped Tektronix make the 545A - after they came off the production line I calibrated them. Learned a lot there.

-js

Got mine at a Ham festival for $15, had some bad grounds in the plug-in. Re-soldered and works fine ever since. You'll be glad to know the calibration still seems to be OK according to Mr Fluke. Mine has Raytheon property tags. If you want, I can get the ser. number off it for you sometime.

[J Tiers]

X2 for the Tek TDS scopes, eBay is covered with them recently. I hear nothing but good about the Fluke scope meters, but they are a tad pricey for me. It's time to update a bit... LOL my Tek 545 was property of Raytheon in 1953!

The older Fluke are fairly cheap. Mine was under $300, and I got some other unrelated stuff with it at that price.

The bad news is that I did NOT get the interface cable nor software for downloading the stored traces. I have wanted that from time to time, although I worked around it the old way by simply taking a pic of the screen.

[nickel-city-fab]

Well, like I said: eBay is awash in newer Tek TDS series scopes. And other ones that would fit in a shoebox. All digital, etc. I'm considering one myself in the medium future. Among other things (HP distortion analyzers oscillators etc. And I miss my old Kepco pwr supply. Disclaimer: last time I was active in electronics, *all* my stuff was rack mounted. As in, a rack next to my desk.)

Perhaps a newer, used Tektronix would be nice.