New to DSO's, specification advice needed. (I've been out of the game a while!)

[Harrow]

I've always worked with analog Tektronix CROs. I now need to get a DSO to work on SMPS in the low kW range. I really have no idea what DSO specifications are required for accurately capturing the bridge transistor ringing and switching transients. My budget is relatively low but potentially flexible, so I'd be happy to understand minimum requirements, but also 'nice to haves'.

Any advice would be greatly appreciated.  Thank you!

[David Hess]

I think any "modern" DSO will be suitable, with modern including DSOs going back 40 years.

I would be more concerned with the probes, and having a high voltage differential probe to make floating measurements, or a DSO which has galvanically isolated inputs so it can make these measurements directly.

[Harrow]

Thanks, I'm familiar with measuring floating bridge voltages as I used to do work in this area some 30 years ago, so no problem there. It's really the digital sampling rate and bandwidth figures that I was wondering about since all my experience is with analogue equipment. For example, I wasn't sure if 100 MHz BW and 1GS/s shared between channels would be sufficient for capturing the transients and ringing accurately, or would I need something better?

Here are my rough thoughts on the subject, but I have no idea if my logic is correct or way off. To be working with rise times in the order of 10ns, that's around 100 MHz, so I figure a 200 MHz -3dB point would give me reasonable measurements there, and working on the basis of sampling at 10x the 10 ns to get accurate transient waveforms would mean 1GS/s, but since that gets shared between channels, then ideally that would be 4GS/s when shared between 4 channels? That sounds like maybe I could get away with something like a RIGOL 900 and just use 2 channels when it matters?

[ebastler]

There are a lot of differences between digital and analog scopes, but input bandwidth works pretty much the same on both. So if you were happy with 100 MHz on your analog CRO, it should work just as well on a DSO.

Sampling rates of 3x bandwidth are considered adequate today, since DSOs can do proper sin(x)/x reconstruction of traces. 5x bandwidth puts you on the safe side. So 100 MHz with a shared 1 GSa/s is perfectly fine for 2 channels, and a bit marginal if the sampling rate is divided over 4 channels.

The Siglent SDS814X HD is about to be launched in Western markets. For a rumored price of $499 it has a 12-bit A/D converter and 2 GSa/s shared across up to 4 channels, at 100 MHz bandwidth. (There is also a 200 MHz "824" version, noticeably more expensive at $699.) And it will offer a "power analysis" software option at extra cost, which is unusual in this entry-level segment. The following threads provide more detail about this new series than you ever wanted to know: 

https://www.eevblog.com/forum/testgear/sds800x-hd-review-demonstration-thread/
https://www.eevblog.com/forum/testgear/siglent-sds800x-hd-12-bit-dsos-coming/

If you prefer a 10" screen, the companion SDS1104X HD will be launched at the same time, at a significantly higher price ($1099?).

[tszaboo]

DSOs are nice for such work, but you should really budget in some money for these two things:
AC/DC current clamp
Differential high voltage probe

[Harrow]

Thanks eblaster.  I had the SDS824X in my sights, but am wondering how long I can wait. The 2GS/s seems to give it the edge over the DHO924 for what I want.

[ebastler]

Not that long a wait: release date should be tomorrow.

Several distributors in Europe are already shuffling their feet and some have (somewhat) hidden web pages ready in their online shops. One user here has even reported that he managed to place an order and has received tracking details for the shipment. Tomorrow we should know more!

[David Hess]

The oscilloscope I regularly use is a Tektronix 2232, which is 100 MHz and a combined 100 MS/s, but supports equivalent time sampling to a combined 2 GS/s.  (1)  Between sin(x)/x reconstruction and averaging, a modern budget DSO will provide similar performance so it is not something I would worry about.

If I use my Tektronix 2440, then the sample rate increases to 500 MS/s per channel, and I have to admit that is nice, but for usability it does not matter much for switching power supply design.  It just makes screen regeneration faster, and if I used my Tektronix 2230 with a combined 20 MS/s, things just take longer.

I would consider a DSO with galvanically isolated inputs a major improvement for switching power supply design, but these are few and far between.  They allow high fidelity high side measurements which a high voltage differential probe will have difficulty with.

(1) The limitation with the 2232 is that sample memory is shared between channels.  Each channel has a completely separate digitizer.

[2N3055]

DSOs are nice for such work, but you should really budget in some money for these two things:
AC/DC current clamp
Differential high voltage probe

This!!

[bdunham7]

Here are my rough thoughts on the subject, but I have no idea if my logic is correct or way off. To be working with rise times in the order of 10ns, that's around 100 MHz, so I figure a 200 MHz -3dB point would give me reasonable measurements there, and working on the basis of sampling at 10x the 10 ns to get accurate transient waveforms would mean 1GS/s, but since that gets shared between channels, then ideally that would be 4GS/s when shared between 4 channels? That sounds like maybe I could get away with something like a RIGOL 900 and just use 2 channels when it matters?

A 100MHz scope has an imputed rise time of 3.5ns and a typical 100MHz DSO with 250MSa/s is probably adequate to look at signals with 10ns rise times.  A 200MHz BW and 500MSa/s or more would be ideal.  There are plenty of low-cost scopes that would meet the latter spec, so there's probably no reason to cheap out unless your budget is very tight. 

Since you're transitioning from a CRO, one thing that might trip you up is overload recovery.  If you are used to using any measurement technique where part of the signal is off the screen vertically in order to magnify some other part, it may not work as well on some modern DSOs.  Hopefully the new models being discussed will have a vertical zoom function that can help with this issue.

If your budget can be stretched, a better scope with a larger screen may be very helpful.  In the Siglent line, the 2204X-HD would seem to me to be ideal, although the less-expensive 2104X-Plus (with hacked options) would also be very good.

[Harrow]

I'll need to wait for them to trickle to Australia, lol. :-)

[David Hess]

Since you're transitioning from a CRO, one thing that might trip you up is overload recovery.  If you are used to using any measurement technique where part of the signal is off the screen vertically in order to magnify some other part, it may not work as well on some modern DSOs.

With some exceptions, analog oscilloscopes usually do not handle overload well either.

The exceptions include oscilloscopes specifically designed to handle overload gracefully, or those which are simply immune to overload.  An example of the former is any Tektronix 7000 mainframe using a 7A13 vertical amplifier which includes a feedback circuit to enforce clamping early in the vertical signal chain.  Sampling oscilloscopes are an example of the later as they are completely immune to overload short of that which will cause damage.

Otherwise the results of overload are unpredictable and vary between designs.  Except for the case of the 7A13 mentioned above, I do not remember having seen it included in specifications.

[bdunham7]

With some exceptions, analog oscilloscopes usually do not handle overload well either.

The exceptions include oscilloscopes specifically designed to handle overload gracefully, or those which are simply immune to overload.  An example of the former is any Tektronix 7000 mainframe using a 7A13 vertical amplifier which includes a feedback circuit to enforce clamping early in the vertical signal chain.  Sampling oscilloscopes are an example of the later as they are completely immune to overload short of that which will cause damage.

Otherwise the results of overload are unpredictable and vary between designs.  Except for the case of the 7A13 mentioned above, I do not remember having seen it included in specifications.

I think the official statement from Tektronix is that they typically won't specify it and you'd need to characterize your scope's overload recovery time yourself.  I've never undertaken that task in a systematic way.  As I detailed in a thread a few years back, I found that all of my older scopes--including 3 older DSOs--had fast enough overload recovery to not be an issue for my application while some newer models were unusable.  The recovery times were orders of magnitude larger.   I wouldn't have mentioned all this except that the OP is planning on using it in a way where he might have this issue.

[David Hess]

I should have included that the 7A13 was designed with overload recovery because of its intended application.  It can operate as a differential comparator, meaning that it effectively has a vertical position control with 20,000 divisions of range for monitoring small DC changes at large DC offsets.  The modern equivalent is an "offset probe".

[2N3055]

I should have included that the 7A13 was designed with overload recovery because of its intended application.  It can operate as a differential comparator, meaning that it effectively has a vertical position control with 20,000 divisions of range for monitoring small DC changes at large DC offsets.  The modern equivalent is an "offset probe".

Also called power rail probe...