Older high-end vs new mid-range scopes.

[cyr]

I have been looking for a while at getting a decent DSO (upgrading from a DS1102E), my budget will just about stretch to something like the Rigol DS4024 (~3000 USD).

I wonder though if there are any decent used options at the same price range (or below) worth considering. Quickly looking around ebay I found Agilent Infiniiums, some LeCroys, TDS784, TDS5000. "Floppy-era" 4ch stuff, 500-1000MHz, 2-8Gs/s, 1-16Mpts.

The main advantage with the older scopes seems to be that I could get around 1GHz of bandwidth instead of 200M. There are some obvious disadvantages, like the size/weight and fan noise of these old beasts and of course the fact that they are used and without any warranty.

I do play around a bit with some pretty high speed logic (FPGA interface to various stuff) so I think the higher bandwidth might be useful sometimes, along with features like graduated persistance display and histograms which I believe most of those scopes have but the Rigol is missing. I have no experience with any of those scopes though, I have no idea how good they are in real use. Would I be crazy to go for one of the old ones?

[EEVblog]

1GHz bandwidth is great, but pretty useless without the right high speed FET probes, and they can cost more than the scope!
500MHz is usually more than enough.

Check out Agilent Certiprime ebay store from time to time:
http://stores.ebay.com.au/Agilent-Remarketing-Solutions-Store/Oscilloscopes-/_i.html?_fsub=3420071011&_sid=869664151&_trksid=p4634.c0.m322
Fully calibrated and warranted by Agilent.

A good Agilent 6000 series unit would be a great buy, or any of the Infiniiums.

Dave.

[AndyC_772]

I agree with Dave on the bandwidth, 500 MHz is about the upper limit for a passive probe, and for that reason I'd regard it as a sensible upper limit for a general purpose bench scope. If you need anything faster then you'll already have a specific application in mind and you'll know how much bandwidth that application requires. Expect to pay a lot for active probes.

I have a Tek TDS754D which is 4 chans, 500 MHz and has a proper intensity graded display - albeit an early implementation which you have to enable manually and in which all channels become the same colour. The UI is a little bit sluggish, and if you want to capture waveforms from it then you'll need a floppy disc drive, but it's still a perfectly usable and capable piece of equipment. It would certainly be a major step up from a 'toy' scope like the 1102.

The Tek TDS5xx/6xx/7xx series are plentiful enough on the s/h market and you can get them from dealers with warranties, though I'm inclined to feel they're overpriced in many cases. Earlier models (prior to the 'D' suffix variants) contain electrolytic capacitors which leak corrosive electrolyte and cause all sorts of problems, so I'd suggest getting either a 'D' or an earlier one which has been refurbished, had the caps replaced and any damage repaired. Avoid anything that fails the start-up self test; many of them have faulty acquisition boards where the caps have leaked.

[nctnico]

When it comes to the TDS500, 600 and 700 series the capacitor problems depend on the age of the scope. The early models from around 1992 have leaking capacitors. Older models >1995 have better capacitors or use ceramics. One or two errors mean the capacitors are leaking but there problably is no real damage yet. There are a couple of electrolytics on the  PLL board which fails due to the capacitors going out of spec. I have restored a TDS644A and a TDS544A without much problems. My TDS510A and TDS744A which are from around 1995 show no signs of leaking capacitors.

[Wuerstchenhund]

I have been looking for a while at getting a decent DSO (upgrading from a DS1102E), my budget will just about stretch to something like the Rigol DS4024 (~3000 USD).

I wonder though if there are any decent used options at the same price range (or below) worth considering. Quickly looking around ebay I found Agilent Infiniiums, some LeCroys, TDS784, TDS5000. "Floppy-era" 4ch stuff, 500-1000MHz, 2-8Gs/s, 1-16Mpts.

Good idea. I personally would take an older scope from HP/Agilent, LeCroy or Tektronix any day.

The main advantage with the older scopes seems to be that I could get around 1GHz of bandwidth instead of 200M.

As others said, forget 1GHz unless you really need the bandwidth and are prepared to shell out for an appropriate active probe. More than 500MHz is a waste for general purpose use.

There are some obvious disadvantages, like the size/weight and fan noise of these old beasts and of course the fact that they are used and without any warranty.

Right, but then if you buy from a Chinese seller your warranty may be worth nothing (especially a problem with Rigol), and the warranty doesn't necessarily cover firmware or hardware bugs.

I do play around a bit with some pretty high speed logic (FPGA interface to various stuff) so I think the higher bandwidth might be useful sometimes, along with features like graduated persistance display and histograms which I believe most of those scopes have but the Rigol is missing. I have no experience with any of those scopes though, I have no idea how good they are in real use. Would I be crazy to go for one of the old ones?

No, but you have to be realistic about what you want (see the point about the bandwidth). Many older scopes come with CRT and may show some burn-in marks. They also lack USB or Ethernet.

On the other side, a 2nd hand scope from the big names gives you a tool for professional use.

On what scope to get, have a look at the LeCroy Waverunner 2 series (LTxxx). They come with a wide range of signal analyzing tools (especially if you find one which has the additional software options) which you won't find in other scopes. A LT374 (500MHz 4Ch 4GSa/s) might be worth a look, you should be able to find one in good condition for your budget. If you can do with less bandwidth/sample rate, look at the LT264 (350MHz 4Ch 1GSa/s) as well.

As to the Agilent Infiniiums, I'be be careful with them. Early models run Windows95 and are prone to crashing in the worst possible moments. The later variants running Windows 2000 or XP are much more reliable, but also much more expensive.

And quite frankly, I'd also avoid the Tek TDS500/600/700 Series, unless it has to be really cheap. They were good scopes in their days but the screen is small and the UI slow, and there's not too much in terms of signal analysis. The TDS5000 is a much better choice.

[nctnico]

1GHz bandwidth is great, but pretty useless without the right high speed FET probes, and they can cost more than the scope!
500MHz is usually more than enough.

Actually there are passive divider probes which are useful to 1GHz. You can even make them yourself. The only problem with those is that the attenuation is quite high so for small signals you'll need a FET probe.

[digsys]

+100% with Wuerstchenhund
My Favourite LeCroy is the 9354AL 500MHz, 4ch Find one with all options and you're in heaven.

[mikeselectricstuff]

As others said, forget 1GHz unless you really need the bandwidth and are prepared to shell out for an appropriate active probe. More than 500MHz is a waste for general purpose use.

Also beware old gear with 50 ohm-only input as you will need expensive probes.

You also need to do a bit of research as to which models are useful - some early digital scopes (e.g. HP) were pretty ropey to use.
Things like lack of interfacing (or GPIB only) aren't much of an issue, as mostly all you want is a screenshot, and digital cameras are now good enough to do that.

[nctnico]

+100% with Wuerstchenhund
My Favourite LeCroy is the 9354AL 500MHz, 4ch Find one with all options and you're in heaven.

Sure about that? At some place I saw a brand new Lecroy scope from same series IIRC somewhere in a corner. I soon learned it was a complete waste of money because it didn't have peak detect and they showed me why. Without peak detect you can't see narrow pulses. If a scope doesn't have peak detect just walk away.

[Wuerstchenhund]

+100% with Wuerstchenhund
My Favourite LeCroy is the 9354AL 500MHz, 4ch Find one with all options and you're in heaven.

It's a very good scope but after nearly 20 years the LeCroy 93xx series is becoming a bit long in the tooth, unless you find one cheap enough. The screen is a monochrome CRT, and finding one which doesn't have heavy burn-in will be nearly impossible. And most 93xx scopes on ebay seem to be beaten up units that I wouldn't touch for the asking price.

Considering that 93xx scopes are still pretty expensive (especially the models with higher sampling rates) I'd rather invest a bit more and get a Waverunner which even in standard config does all what a 93xx with WP01, WP02 and WP03 does, and much more with the various Waverunner options installed. The color LCD helps a lot over the monochrome CRT when working with multiple signals or the analysing functions, and processing is also much faster on the Waverunner than it is on most 93xx scopes.

[Wuerstchenhund]

+100% with Wuerstchenhund
My Favourite LeCroy is the 9354AL 500MHz, 4ch Find one with all options and you're in heaven.

Sure about that? At some place I saw a brand new Lecroy scope from same series IIRC somewhere in a corner. I soon learned it was a complete waste of money because it didn't have peak detect and they showed me why. Without peak detect you can't see narrow pulses. If a scope doesn't have peak detect just walk away.

The 93xx Series does have Peak Detect. But the user interface is a bit different to other scopes, and any feature is useless if the user doesn't know how to use it or can't be bothered to read the manual.

[JoeyP]

...I soon learned it was a complete waste of money because it didn't have peak detect and they showed me why. Without peak detect you can't see narrow pulses. If a scope doesn't have peak detect just walk away.

Wow, a *complete* waste of money because you thought it didn't have one single feature which you consider important? Not everyone uses scopes the same way or for the same tasks.

For at least the past decade, every digital scope I've used had/has peak detect, but I very rarely turn it on. It's only useful if you've zoomed out to a point that the display span exceeds buffer depth at max sample rate. If there is plenty of memory then peak detect is rarely needed. I generally zoom in and trigger on the signal of interest when looking at fast signals, in which case peak detect does absolutely nothing. If I've zoomed out beyond the max-rate buffer span, its usually because I'm looking at something slow, and don't care about (in fact may want to exclude) faster signals/noise, so I leave peak-detect off.

[nctnico]

...I soon learned it was a complete waste of money because it didn't have peak detect and they showed me why. Without peak detect you can't see narrow pulses. If a scope doesn't have peak detect just walk away.

Wow, a *complete* waste of money because you thought it didn't have one single feature which you consider important? Not everyone uses scopes the same way or for the same tasks.

Those are not my words, it was the person who bought it. Read more carefully.

For at least the past decade, every digital scope I've used had/has peak detect, but I very rarely turn it on. It's only useful if you've zoomed out to a point that the display span exceeds buffer depth at max sample rate. If there is plenty of memory then peak detect is rarely needed.

The problem is that long memory often means slower update rates. And even the longest memory will have gaps at low sampling rates. Peak detect usually catches glitches which are a few ns wide. I often look at fast signals at low sampling rates to look at the overall amplitude so I have peak detect on most of the time.

[JoeyP]

...I soon learned it was a complete waste of money because it didn't have peak detect and they showed me why. Without peak detect you can't see narrow pulses. If a scope doesn't have peak detect just walk away.

Those are not my words, it was the person who bought it. Read more carefully.

That last sentence doesn't sound like a quote from someone else. You wrote the post. You made the hyperbolic assertions.

The problem is that long memory often means slower update rates.

Peak detect does nothing to change this. If you're zoomed out far enough for peak detect to be active, then the time required to sample the chosen data span will dictate the update rate, not the memory depth or processing time.

[cyr]

Thanks everyone.

The series I've been looking at mostly is the Infiniium 5483xB/D. Seems early ones are 98 based and later ones are XP. I wonder if it's possible to upgrade the motherboard and O/S, it's just a micro-ATX board with a couple of PCI cards (acquisition interface, GPIB, display controller) as I understand...

600MHz or 1GHz probably wouldn't make much difference for me, although 1G is a very nice round number. 

I was thinking I could get away without expensive active probes by using DIY resistive probes soldered into the circuit on the few occasions I need the bandwidth.

[JoeyP]

The series I've been looking at mostly is the Infiniium 5483xB/D. Seems early ones are 98 based and later ones are XP. I wonder if it's possible to upgrade the motherboard and O/S, it's just a micro-ATX board with a couple of PCI cards (acquisition interface, GPIB, display controller) as I understand...

I saw someone selling these on ebay recently, who claimed to not only upgrade the OS, but also converted it to SSD.

600MHz or 1GHz probably wouldn't make much difference for me, although 1G is a very nice round number. 

I was thinking I could get away without expensive active probes by using DIY resistive probes soldered into the circuit on the few occasions I need the bandwidth.

I used the HP 10020A resistive divider probe for a long time before I finally got an active probe. Once I got the active probe I was kicking myself for not doing it a lot sooner. You can find them pretty reasonably priced on ebay. I think the 1152A 2.5GHz active probe is compatible with the 5483x series, and I just saw one of them sell for under $400 on ebay (with a 30-day warranty).

[Wuerstchenhund]

The series I've been looking at mostly is the Infiniium 5483xB/D. Seems early ones are 98 based

Windows 95 (we had one of these), but they could be upgraded to Windows 98 easily. And shortly after launch of the Infiniiums the Win95 OS was replaced with Win98 anyways.

and later ones are XP. I wonder if it's possible to upgrade the motherboard and O/S, it's just a micro-ATX board with a couple of PCI cards (acquisition interface, GPIB, display controller) as I understand...

You may be able to upgrade the OS if you can get hold of the XP recovery media, and maybe a 548xxD mainboard. It's unlikely any other ATX mobo will work as if I remember right the recovery media checks for the BIOS signature. Considering the price you'll very likely pay because the mobo has an HP part number and a HP BIOS ID I don't think it's sensible unless you already have access to all the parts. And I won't even touch the legality of running an essentially unlicensed copy of XP.

In addition, while Agilent makes very good scopes, the 548xx Series was not known for their high reliability.

600MHz or 1GHz probably wouldn't make much difference for me, although 1G is a very nice round number. 

I was thinking I could get away without expensive active probes by using DIY resistive probes soldered into the circuit on the few occasions I need the bandwidth.

If you want 1GHz+ then have a look at the LeCroy LC Series (which are successors of the 93xx scopes). LT5xx have color CRTs, LT6xx have LCD screens. The same scopes were also sold with the Disk Drive Analyzer package as 'DDA'.

There are quite a few LT584 and LT684 (and DDA-120/125) on ebay sitting there for over a year, so with a bit of luck you may be able to get a 1+GHz scope with 8GSa/s for a low price.

And they are much more reliable than the Infiniiums.

[nctnico]

Thanks everyone.

The series I've been looking at mostly is the Infiniium 5483xB/D. Seems early ones are 98 based and later ones are XP. I wonder if it's possible to upgrade the motherboard and O/S, it's just a micro-ATX board with a couple of PCI cards (acquisition interface, GPIB, display controller) as I understand...

You probably can unless they use a specially crafted BIOS. Be carefull with the amount of power required and choose a high quality motherboard.

[alm]

I agree with Dave on the bandwidth, 500 MHz is about the upper limit for a passive probe, and for that reason I'd regard it as a sensible upper limit for a general purpose bench scope. If you need anything faster then you'll already have a specific application in mind and you'll know how much bandwidth that application requires. Expect to pay a lot for active probes.

I would say 500 MHz is already stretching it for high-Z passive probes. I know the manufacturers claim 500 MHz, but this is from a terminated 50 Ohm signal generator having a 25 Ohm output impedance, so the probe might be 50 Ohm or so at 500 MHz (check the input impedance graph in the datasheet). Is the circuit node you're probing that low impedance? The already mentioned resistive divider probe is much better in this regard.

Used active probes aren't that expensive. A passive 500 MHz probe will cost about $100 on eBay. A used active probe might cost $200-300 or so for the low bandwidth (<= 1 GHz) ones.

[nctnico]

Most high-speed circuits don't like to be probed with the capacitance of any high impedance probe. My rule of thumb is to use a passive divider HF probe or an active probe for measuring anything over 100MHz.