Want faster and newer oscilloscope

[lordvader88]

My best scope is a Tektronix 2430A with 200MHz bw, 5ns/div, and usually only 2mV/div (it has 200uV/div but I have to read the book to get that working). It's fine for most everything I do, but it's hybrid chips run super hot, die a lot, and the fan is super loud.

I plan to get a Rigol 1054Z for it's lightweight and 4 channels/ but it's only 50MHz.

So I'd like something faster, with more voltage zoom, and a lot newer. But I don't want to spend $2,000. Looking around at 500Mhz used scopes, they are still mostly 1980s scopes at about $500-1000

Doesn't anyone make a fast 2-ch scope for lower budget people ?

[Paul Moir]

No, because you need high budget probes to go with a high bandwidth oscilloscope.
(Also around the east coast of Canada, it is pretty slim pickings when it comes to used 'scopes.  I had to put up with a "5MHz" heathkit for way too long before I found a 465.)

Why do you need MHz but not memory? 

(EDIT:  now enjoying a Siglent 1202X-E which has a nice 500uV range)

[tautech]

Yes, the SDS1202X-E with 200 MHz BW and 500uV/div (@ full BW) is the best choice currently.
If your budget can stretch further, a 4ch SDS1204X-E could be a good choice.

[Mechatrommer]

I plan to get a Rigol 1054Z for it's lightweight and 4 channels/ but it's only 50MHz.

correction. Rigol 1054Z is a 100MHz scope after you put serial number from riglol. at 200MHz it only attenuates at few dB down.

[Fungus]

I plan to get a Rigol 1054Z for it's lightweight and 4 channels/ but it's only 50MHz.

They're easily unlockable to 100MHz (plus all other options).

The "100Mhz" is quite nominal, many people have measured theirs much closer to 200Mhz in reality.

nb. With 4 channels on the sample rate drops to 250Mhz which isn't really enough for that much bandwidth. Advice: Switch to 1 or 2 channels for best readings on the limit.

If you're 100% sure you can live with 2 channels there's also a Siglent for a little bit more money. 4 channels is good though.

[nctnico]

My best scope is a Tektronix 2430A with 200MHz bw, 5ns/div, and usually only 2mV/div (it has 200uV/div but I have to read the book to get that working). It's fine for most everything I do, but it's hybrid chips run super hot, die a lot, and the fan is super loud.

I plan to get a Rigol 1054Z for it's lightweight and 4 channels/ but it's only 50MHz.

So I'd like something faster, with more voltage zoom, and a lot newer. But I don't want to spend $2,000. Looking around at 500Mhz used scopes, they are still mostly 1980s scopes at about $500-1000

Doesn't anyone make a fast 2-ch scope for lower budget people ?

You could look at a GW Instek GDS2072E or GDS2074E (these can be uphacked to 300MHz). Another option would be to look at the Tektronix TDS500 or TDS700 series as these are from the 90's. A little newer are the HP/Agilent 54800 series Infiniium scopes which are also nice but are more geared towards working with high speed signals. If you want to get your hands dirty then a 'for parts' oscilloscope can be brought back to life with a little bit of effort. Most of the issues and the fixes are well known. I have owned a few Tektronix TDS500 and TDS700 series scopes but I must say I like the HP/Agilent 54800 series better because these can calibrate themselves (unlike the Tektronix scopes).

[sn4k3]

+1 for SDS1202X-E and SDS1204X-E
They are new (2018) as you like

[macboy]

I spent slightly over $1000 USD on a used 4 channel 2 GHz scope, Lecroy Wavepro 960. I got lucky... it worked great and was optioned up with everything (advanced math, jitter analysis, LAN card, etc.). It's big and deep, but has an LCD (10") not CRT so isn't quite a boat anchor. Virtually identical hardware is available as the DDA-260 (Disk Drive Analyzer) or JTA-260 (Jitter/Timing Analyzer). Those have only slight differences in the front panel quick-access buttons but are functionally equivalent otherwise, for a given set of options*.  Mine was built 2002 IIRC, so these are not quite ancient, but are also not quite modern. They land somewhere in the sweet spot for price/performance/age IMHO. They come up frequently enough on ebay.

The main problem with a > 500 MHz scope is finding probes to use the extra bandwidth. Passive 10 MOhm probes don't go over 500 MHz, with very few, expensive exceptions. You can use any passive probes, of course, within their own limitations. I really lucked out on four Tek P6158 "Low Z" probes (one broken) for next to nothing at a local recycler. Low-Z probes are also easy to make yourself, though DIY probes come with DIY response curves, which means you need to characterize it yourself if you want to know how well it works.

* p.s. There is a thread here on "option recovery" on Lecroys in case you find a low-spec one. I can't comment on effectiveness as I have no need for that.

[nctnico]

Yeah, the older Lecroys are also worth looking at but many of them don't have peak-detect.

[DaJMasta]

Yeah, the older Lecroys are also worth looking at but many of them don't have peak-detect.

It seems like this comment comes up every time a LeCroy scope is mentioned.... isn't the glitch-finding functionality it offers the same as a persistence mode?  You're not going to be catching any glitches with peak detect that you won't detect with persistence on because you're limited by the waveforms per second update rate of the acquisition system, so why would a peak detect acquisition mode matter?


I think the unlocked Rigol or the Siglent options are good choices for the OP.  500MHz or more on an older 2000s era scope should be quite doable within the budget, and while you need expensive probes for GHz plus speed stuff.... if you're not so concerned with that, 500MHz passive probes do reasonably and don't cost an arm and a leg.  Back that down to 350MHz and you can get them very cheaply, and they'd still work fine on the front of your 1GHz or more scope, you'd just have a ton of extra sample rate and the option to measure a 50 ohm termination signal directly.

[nctnico]

Yeah, the older Lecroys are also worth looking at but many of them don't have peak-detect.

It seems like this comment comes up every time a LeCroy scope is mentioned.... isn't the glitch-finding functionality it offers the same as a persistence mode?  You're not going to be catching any glitches with peak detect that you won't detect with persistence on because you're limited by the waveforms per second update rate of the acquisition system, so why would a peak detect acquisition mode matter?

It has nothing to do with glitches but with aliasing and masking due to undersampling. There are two problems which are solved with peak-detect:

1) Every DSO I know of isn't adjusting the bandwidth of the front-end filter when sampling with a lower sampling rate. This means that you can get aliasing artefacts or get signals which aren't there. This can be confusing and/or make you draw the wrong conclusion.

2) If a signal contains narrow pulses (think about a vertical synchronisation signal for a display) then the pulses will dissapear or become random when the sampling rate drops (masking).

IMHO peak-detect is a must have on a general purpose DSO due to the problems I described above. The reason I mention it often is that it isn't very obvious at first sight and 99% of the non-Lecroy DSOs have peak-detect so people may assume the Lecroy DSOs also have peak-detect.

What peak-detect does is let the ADC run at the full samplerate and record the minimum and maximum values during the sampling interval (not the ADC sampling rate but the sampling rate dictated by memory length and time/div settings).

[lordvader88]

Yes, the SDS1202X-E with 200 MHz BW and 500uV/div (@ full BW) is the best choice currently.
If your budget can stretch further, a 4ch SDS1204X-E could be a good choice.

Yeah that wouldn't take too much more saving than a 1054Z.

I forget the hacking of the 1054Z, IDK if I would tho, depends on if I get a warranty I guess.

Those late 80s-90s scopes could be great, I wish they didn't weigh 30lbs

[Fungus]

I forget the hacking of the 1054Z, IDK if I would tho, depends on if I get a warranty I guess.

a) You can lock it up again just as easily as unlocking it, and without leaving a trace.
b) There's been no reports of Rigol not honoring a warranty.

(actually, there's not many reports of people needing the warranty, those things are very reliable)

[macboy]

Yeah, the older Lecroys are also worth looking at but many of them don't have peak-detect.

The mentioned Wavepro 960 has:
- analog persistence with variable saturation threshold
- history to go along with the analog persistence so you can "rewind" back through a many acquired waveforms to find the one which had caused a glitch seen on the persistence display (including timestamps).
- advanced smart triggering to either start or stop acquisition based on height or duration of pulses, missed pulses, extra pulses, patterns, etc.
- extrema function which captures the envelope, a.k.a. floor and roof, a.k.a min/max, over many acquisitions.
... and so on.
The types of processing available on the captured waveforms is mind-boggling. Then manual is over 340 pages, and there is another entire manual to explain all the timing and jitter analysis functions (if so equipped).

[nctnico]

The Lecroy scopes seem to have superior waveform analysis features. Sadly enough I never owned one.

[David Hess]

I plan to get a Rigol 1054Z for it's lightweight and 4 channels/ but it's only 50MHz.

They're easily unlockable to 100MHz (plus all other options).

The "100Mhz" is quite nominal, many people have measured theirs much closer to 200Mhz in reality.

Bandwidth on the unlocked Rigol DS1000Z series apparently varies with vertical sensitivity and signal level.  For most applications this does not matter but if you are running up against the bandwidth limitation of your 2430A, this may be an issue.

Yeah, the older Lecroys are also worth looking at but many of them don't have peak-detect.

It seems like this comment comes up every time a LeCroy scope is mentioned.... isn't the glitch-finding functionality it offers the same as a persistence mode?  You're not going to be catching any glitches with peak detect that you won't detect with persistence on because you're limited by the waveforms per second update rate of the acquisition system, so why would a peak detect acquisition mode matter?

Persistence relies on multiple triggered acquisitions to catch a glitch.  Peak detection applies the maximum sample rate so it catches glitches on every acquisition including single shot acquisitions.  "DPO" type modes also do this but capture more than the minimum and maximum producing a 2-D histogram.

[bson]

What peak-detect does is let the ADC run at the full samplerate and record the minimum and maximum values during the sampling interval (not the ADC sampling rate but the sampling rate dictated by memory length and time/div settings).

The longest time div it can run the ADC at maximum rate depends on the memory depth.   For the WaveSurfer 3000 with 10M per channel and 2GS/s this comes out to 50µs/div to capture a 500ps "pulse" (more like noise really).  In real use you're just not going to miss anything significant if you work with jellybean LVC logic or other modestly fast CMOS, even at 1M samples or much less, simply because the timebases used will have you running at full sample rate with much smaller memory sizes.  I don't know what you do, but to me it's a total non issue.

[Fungus]

Bandwidth on the unlocked Rigol DS1000Z series apparently varies with vertical sensitivity and signal level.  For most applications this does not matter but if you are running up against the bandwidth limitation of your 2430A, this may be an issue.

Only when you overdrive the input amplifier by switching below 500mV (relay clicks over and you get 200mV/div) then selecting 'fine control' and manually cranking the volts/div up over 400mV/div.

ie. Almost never.

PS: What do you want for $350?

[David Hess]

The longest time div it can run the ADC at maximum rate depends on the memory depth.   For the WaveSurfer 3000 with 10M per channel and 2GS/s this comes out to 50µs/div to capture a 500ps "pulse" (more like noise really).  In real use you're just not going to miss anything significant if you work with jellybean LVC logic or other modestly fast CMOS, even at 1M samples or much less, simply because the timebases used will have you running at full sample rate with much smaller memory sizes.  I don't know what you do, but to me it's a total non issue.

But how much slower does the WaveSurfer operate when such a long record length is used?

It depends on the application, but I would not consider wanting to see every minimum length glitch even at sweep speeds slower than 50us/div unusual.

Bandwidth on the unlocked Rigol DS1000Z series apparently varies with vertical sensitivity and signal level.  For most applications this does not matter but if you are running up against the bandwidth limitation of your 2430A, this may be an issue.

Only when you overdrive the input amplifier by switching below 500mV (relay clicks over and you get 200mV/div) then selecting 'fine control' and manually cranking the volts/div up over 400mV/div.

ie. Almost never.

PS: What do you want for $350?

We do not know all of the conditions where this occurs yet.  The measurements which show different bandwidths at different vertical sensitivities without variable adjustment show that it is not an edge case.

The overdrive situation reveals that bandwidth also depends on signal level independent of vertical sensitivity.  It should not be possible to overdrive an oscilloscope input while remaining inside of its input signal range for a given vertical sensitivity.

PS: I want not to be lied to in both function and documentation at any cost.

[Fungus]

It should not be possible to overdrive an oscilloscope input while remaining inside of its input signal range for a given vertical sensitivity.

I dunno. I think I might simply use the 500mV rage if I was looking at that sort of signal.

Going into 200mV range, selecting fine adjust, cranking it up close to 500mV seems like a lot of work.

PS: I want not to be lied to in both function and documentation at any cost.

Good luck with that.

[bson]

But how much slower does the WaveSurfer operate when such a long record length is used?

A lot slower, but if I want to look into the noise at that level I'll turn up the record  length.  Then when no longer needed I'll turn it down again. 

Just like I'd turn on peak detect mode when needed, then turn it off when no longer needed.  It's a tool to turn on and off as required.

[nctnico]

What peak-detect does is let the ADC run at the full samplerate and record the minimum and maximum values during the sampling interval (not the ADC sampling rate but the sampling rate dictated by memory length and time/div settings).

The longest time div it can run the ADC at maximum rate depends on the memory depth.   For the WaveSurfer 3000 with 10M per channel and 2GS/s this comes out to 50µs/div to capture a 500ps "pulse" (more like noise really).  In real use you're just not going to miss anything significant if you work with jellybean LVC logic or other modestly fast CMOS, even at 1M samples or much less, simply because the timebases used will have you running at full sample rate with much smaller memory sizes.  I don't know what you do, but to me it's a total non issue.

But the older Lecroy scopes which are in the price range of the OP usually have much less memory. More likely 500kpts per channel with all channels enabled.

[macboy]

What peak-detect does is let the ADC run at the full samplerate and record the minimum and maximum values during the sampling interval (not the ADC sampling rate but the sampling rate dictated by memory length and time/div settings).

The longest time div it can run the ADC at maximum rate depends on the memory depth.   For the WaveSurfer 3000 with 10M per channel and 2GS/s this comes out to 50µs/div to capture a 500ps "pulse" (more like noise really).  In real use you're just not going to miss anything significant if you work with jellybean LVC logic or other modestly fast CMOS, even at 1M samples or much less, simply because the timebases used will have you running at full sample rate with much smaller memory sizes.  I don't know what you do, but to me it's a total non issue.

But the older Lecroy scopes which are in the price range of the OP usually have much less memory. More likely 500kpts per channel with all channels enabled.

The ones I had specifically mentioned are usually equipped with 16 MS or 32 MS but can be up to 64 MS. Divide by number of channels in use.

My Wavepro is much older than the Wavesurfer, but it can still be set up to capture oddities over a very long interval. You can set up the trigger for any of many types of glitches - extra/missing pulses, edge slope too fast or slow, waveform exceeds envelope or min/max levels, etc. - then you segment the memory (up to thousands of segments), start acquisition, and the ADC will capture the input continuously at the configured rate up to 16 GS/s, writing data to the next memory segment whenever a trigger event occurs. Once all segments are written (which could be less than a second, or many days), you can do interesting things with the captured data including animated playback, math, statistical analysis, etc.

[Mr Nutts]

But the older Lecroy scopes which are in the price range of the OP usually have much less memory. More likely 500kpts per channel with all channels enabled.

I think that's only for very old ones and very basic models. Even the ones I saw that had a yellow crt had 4M or 8M.

My old Waveruner lt374m has 4M but could be equipped with up to 8M. And similar to macboys wavepro the functionality is just insane

The last time I used peak detect was in college but that was with some truly awful Tek scopes that couldn't do more thanthe basics (and sucked even at that).