DSO spec confusion

The net rate must be half the frequency but there is a technique called equivalent time sampling where each a/d sample is delayed in time on the next sweep. At 100msa/s each sample is at 10nsec spacing. If you take ten sweeps with the sample point displaced in time by an additional 1nsec, then you have a 1Gsps equivalent rate. The tradeoff is a cheaper a/d for a slower data acquisition rate. Many years ago we had boxcar integrators which effectively took one data point per sample. By moving the delay, you assemble a waveform. Clearly the digital scopes are better than this.

At 1Gsps you get the waveform sampled with a 1nsec resolution so you can actually see some details. If you only sample at just over 2 samples per cycle, the amplitude will be small and the waveform will look odd.

For what you want, most of the digital scopes from the Chinese imports to the (probably chinese import) Tektronix will do fine. The biggest adjustment you will have is that there are aliasing effects when you adjust the sweep rate.

For audio and lower frequencies you can also do will with a good used Tek 2465 family oscilloscope. Analog scopes are a pain when you are looking at transients but they are good for continuous signals. However, unless you like screen picturesm, the digital are good for documenting results.

I have an assortment of scopes but I find I use my Tek TDS3032 almost exclusively. They are expensive but are great scopes.

My recomandation would be for a Tektronix 2465B (B is the must with auto setup), it's the RR of the analog scopes. Old, built in the 80's but much metter than a low cost digital scope in wich all the person I know who purchased were desapointed.

pf

"Brett Pantalone" a écrit dans le message de news: snipped-for-privacy@corp.supernews.com...

Not really. If you have only one sample per half cycle, you'll be only able to say that there's a signal there.

Say your scope's A/Ds are 8-bits and you want to avoid aliasing anything into that least significant bit. That means you'll want a stop band that's down at least 48 dB at the Nyquist limit where frequencies start "folding back" on you. With a 4th order filter, you'll get that much attenuation in 0.6 decades or at about 240 MHz. That's where your sampling Nyquist limit will need to be, so your sample rate should be at least 480 MHz per channel or 960 MHz for two channels.

For audio freqs, an inexpensive analog scope is fine. You'll probably mostly be looking at repeating waveforms (or at least would rarely want/need to use single-shot).

For digital *data* I'd strongly recommend a logic analyzer rather than a scope. There are some decent, inexpensive PC-based ones. I've been using the LogicPort from for some time now and like it quite a lot.

All that said, there seems to be a consensus around here that Instek scopes are pretty close to today's sweet spot. Similar to the Tektronix models but with a larger data buffer, RS232, & USB. for one example. I'm still using my "ancient" TDS220 but if it ever goes up in smoke ...

Absolutamente! That's about the first thing I suggest to clients who try to diagnose something with a fancy DSO, to go on EBay and snatch a 2465. After it arrives they often feel like somebody turned on the lights. "Hey, we can see!"

But since Brett is a SW-engineer he may have to look at digitals stuff such as SPI sequences a lot and then a DSO wins.

Yep, I just wanted to suggest Instek. I've got the big one here (GDS-2204) and it runs circles around Tektronix. Also very nice USB connectivity. If you don't have a laptop at hand just pop in a USB stick and it stores the images and raw data on there.

Tektronix has inferior sample memory depth and the TDS210/220 series is prone to 40/80kHz noise, probably from its backlight inverter. IMHO a disgrace to Tektronix. There was also some kind of safety recall because of a sub-par ground connection. You might want to check that.

Brett: Get something with at least 500MSPS, better 1GSPS and 100MHz bandwidth or more. If you do stuff with FPGA or fast micro controllers you'll need it. An old rule is that the sample rate should be at least

Mine's a fairly late model TDS220 and was produced well after the range that was affected by the power supply recall -- thanks for the tip, though! It's also quiet with respect to the inverter noise so I don't really have any excuses to ditch it... ;-)

Then it should be ok. Except that the ones I used had no connectivity, I had to snap digicam pictures. I did use them for lots of serial bus debugging though and they performed those jobs nicely.

Thank you to everyone for your helpful advice!

-- Brett