Is Bandwidth/memory depth a waste of money in oscilloscopes?

[Pete F]

Hi, my trusty old Tektronix 2215 oscilloscope just blew up and after poking around inside, it was time to admit it's days were over.

While electronics is actually my trade, that was sadly a long time ago and things have changed a LOT. After a career change I don't do much in this field now, but do occasionally still dabble and don't want to be without an oscilloscope. It's quite likely I'll do more work in the field in future however. Most of my work would be in the low frequency area, audio, power supplies etc. I'd consider it was relatively easy work by oscilloscope standards.

I've been looking at the offerings from Hantek in the 5000 series. I've read mixed reports about the Hantek overall quality but that aside was wondering if the differences in bandwidth and memory depth are worth the extra money for the above applications? The series starts with a DSO5072P (70 MHz/40K mem) @ A$320, through to a DSO5102B (100 MHz/1M mem) in the midrange @ A$455 and extends up to the DSO5202DM (200 MHz/2M mem) @ $550. Other manufacturers have similar models in their own ranges.

My question is; given the work I expect to do, is the significant increase in cost between the manufacturers' offerings likely to provide me any real benefits, or should I just spend the difference on cheap wine and hookers? The DSO5072P is a popular model and the overall consensus from buyers seems to be that they're happy with the purchase, it may well do everything I want, and I really don't need to spend more than that.

Advice and opinions gratefully received.

[Rerouter]

If your working in the audio and power supplies area, your requirements are not that high, bandwidth comes down to what you intend to measure, and memory down to how many slices of that ting do you want saved to reveiw.

at my work we went with a cheap 2 channel siglent and it has more than me our requirements (component level repair of mixed signal boards below 10Mhz)

[Muttley Snickers]

I'm still in the market myself but haven't even bothered to look at the Hantek stuff because I don't believe there are any local distributors so if there were any problems with the unit you might have to send it back to Asia at huge expense and no guarantee, there may well be a local Hantek distributor in Australia but I haven't found any who carry any substantial stock.

Not long ago I was looking at purchasing the Siglent SDS-1072 CML+ which is 70 MHz and has 2 Meg of memory in addition to a 7 inch 800 x 480 screen and 3 year warranty but I hesitated because Siglent have failed to release a new firmware update to address the bugs and issues which member here Stilltrying  has gone to great lengths to document.

I spoke to Wavecom Australia and they did have these listed on special at $338AUD for a long time but have recently bumped them up to $468AUD plus tax so that counts me out, if Siglent had of pulled their head out of their arse in regards to updated firmware earlier then I would now be using one of these instead of my TDS-1002 with 60 MHz and 2.5kpts of memory. For my applications I would much rather use a shitty old thing that works than a fancy new thing which doesn't.

[Pete F]

Some of these intro models of various brands are available in Australia, but at around double the price. My experience with the local interpretation of "warranty" from small distributers hasn't been good I'm afraid, not much better than the Asian interpretation, and I personally wouldn't pay 1 cent extra for it. Maybe the locals will lift their game as more and more people give up and shop online/offshore, but I'm not holding my breath. However that's my experience and opinion, and others' milage may well be different. I do try my best to support our economy, and shopped around locally but I have no intention of paying double. At this price range I'd consider it disposable in any case, a shame, but that's just life and a sign of the times.

I did have another look at the 2215, I just can't bear to see her go. I did get excited when I had beat it into life, but one of the channel switched was intermittent and didn't seem to want to clean up, at least one of the probes is dicky, and then it died again. I'll threaten it with screwdrivers and stern words once again, but I think the old girls is telling me it's time to go.

[tautech]

Pete F, just because it's a scope doesn't make it any more special to fix than anything else and there's some way you can go in attempting to fix it with minimal tools and some brain power.
First you need probes you can trust and even the $10/pair P6100's on eBay are shit loads better than nothing.

Grab a 2215 service manual online and do some serious study.
You say it 'blew' up, just what exactly failed ?
There's a few threads on these and models from the same series in the Repair board so hunt them out and see if you have similar symptoms.

Ideally if you're to upgrade your Tek, 2Mpts is what you should set as a starting point for a DSO in order to utilize its capabilities when Single triggering or examining a Stop'ped waveform. 100 MHz is what many these days consider appropriate for most of the modern things we look at.

[Pete F]

By "blew up" I mean the magic smoke escaped. No idea where from, there was enough that it should have been obvious, but upon opening it up ... it wasn't obvious! I suspect it's actually from a PCB type arrangement within the power supply. I have no idea why it's designed that way and it's not worth trying to explain as I haven't seen anything like that before.

These machines come with a service manual, because that's what they did back then. So I have it, no serious study required, they were pretty simple back then. Pretty much everything can be repaired and I'm sure this oscilloscope is no exception. Indeed it may well be not a particularly difficult repair. However sometimes using that brain power is to realise when it's about not wasting time and resources on flogging a dead horse. Yes I could spend what will inevitably turn out to be a considerable amount of time messing about with the current machine and get it going. I would also need to go out and spend another 20-30 bucks on probes that are crap and no doubt just as unreliable as the ones I have. I'm then left with an unreliable machine, with crap probes that takes up a lot of room and is well over 30 years old. These puppies aren't wine, they don't get better with age! If I had nothing better to do with my time it might be a good exercise, but I do, so it's not I don't feel it's worth pursuing it further, if I can find an obvious fix and if that doesn't work that's it. If anyone else wants to mess about with it they can check the Sydney dumpsters and knock themselves out!

Edit: Out of interest here's a thread from this very forum discussing the above machine. It seems many share my opinion on the power supply https://www.eevblog.com/forum/repair/tektronix-2215-problem/

Which brings me back to the point of the post. For many years I made my living with a dual channel oscilloscope and never felt the need for more channels. I had many available but my own personal one wasn't particularly high bandwidth and I also rarely, if ever, felt the need for more. Back then I was doing a lot of RF work too. Most of the time I found the faults were reasonably binary; the device is either working or it wasn't. If it wasn't working properly it was normally relatively obvious and there wasn't much need to zoom in on specific parts of waveforms etc. The people designing might need to do that, maybe the people doing a lot of digital, but that's not my experience anyway. I guess I'm just after something that can do what I has, but isn't 35 years old

[tautech]

 
I get all that, and see you're quite ready to move to something more modern.
Don't overlook SDS1202X-E. 

[bd139]

If your 2215 actually exploded it’s probably the RIFA caps. Dead simple fix.

Don’t buy a Hantek.

[David Hess]

Bandwidth is bandwidth with either an analog oscilloscope or DSO and nothing makes up for lack of bandwidth but if you do not need it, then you do not need it.  I am sure your applications can get by with 50MHz.

Memory depth in DSOs is important for maintaining high sample rate at slow sweep speeds and acts as a crutch to make up for the lack of a delayed timebase which your 2215 has.  It is one of the cheapest things they can add to a DSO for bragging purposes.

DSOs are very handy for power supply work because their storage capability prevents flicker at slow sweep speeds and they can capture startup and shutdown events.

My go-to oscilloscope is a Tektronix 2230/2232 which is basically a 2215 with more bandwidth and complete digital storage capability.  Since it has delayed timebase capability and peak detection, it does not need a large record length as much as a modern DSO which lacks these features.

[Pete F]

If your 2215 actually exploded it’s probably the RIFA caps. Dead simple fix.

Don’t buy a Hantek.

I'm not sure how you would come to that conclusion. How many 2215s have you found have that issue? Is it a known issue with that machine?

Why would you say "don't buy a Hantek"? From what I can ascertain reading through the people who have bought that brand, the build quality is pretty ordinary. If I was still doing this malarky professionally I would want something decent. My original CRO was a Kenwood (not a brand that comes to mind normally), given to me by my father who was also in the game. It was by far the best oscilloscope I'd ever used and I will always miss that machine (it was stolen). Not high bandwidth and with no bells/whistles, but a rock solid machine that always triggered well and did what it said on the box.

In the research I've done so far I didn't see any obvious trend in terms of reliability or otherwise in regard Hantek. People are generally quick to whinge and complain when something goes wrong, and I've seen most brands pretty well represented on the whingometer. Once I narrow down what it is that I actually need I will refine the brands more. I certainly appreciate the reasons why recommendations are being made so I can learn for myself and make up my own mind, as without that the person at the other end of the keyboard could be sitting there with a lucky 8 ball for all I know.

[nctnico]

Since it has delayed timebase capability and peak detection, it does not need a large record length as much as a modern DSO which lacks these features.

All modern DSOs have delayed time bases! That has been pointed out before. You can set the trigger point outside the screen and there you have it.

To the OP:
In general brands like Hantek, Owon, Rigol, Siglent may have hidden problems in their firmware which may take long (or never) to get fixed. Just be aware that you get what you pay for when buying an oscilloscope. For your purpose you may want to look at the GW Instek 1072B or MicSig tBook to1072. The latter is a (thick) tablet sized scope which can be both battery and mains powered. It is very handy to take with you or as a scope you can grab quickly and do some measurements.

[bd139]

If your 2215 actually exploded it’s probably the RIFA caps. Dead simple fix.

Don’t buy a Hantek.

I'm not sure how you would come to that conclusion. How many 2215s have you found have that issue? Is it a known issue with that machine?

Why would you say "don't buy a Hantek"? From what I can ascertain reading through the people who have bought that brand, the build quality is pretty ordinary. If I was still doing this malarky professionally I would want something decent. My original CRO was a Kenwood (not a brand that comes to mind normally), given to me by my father who was also in the game. It was by far the best oscilloscope I'd ever used and I will always miss that machine (it was stolen). Not high bandwidth and with no bells/whistles, but a rock solid machine that always triggered well and did what it said on the box.

In the research I've done so far I didn't see any obvious trend in terms of reliability or otherwise in regard Hantek. People are generally quick to whinge and complain when something goes wrong, and I've seen most brands pretty well represented on the whingometer. Once I narrow down what it is that I actually need I will refine the brands more. I certainly appreciate the reasons why recommendations are being made so I can learn for myself and make up my own mind, as without that the person at the other end of the keyboard could be sitting there with a lucky 8 ball for all I know.

I’ve repaired numerous scopes (tens of them) including a few 22xx units. The age of the units and the power supply design suggests that as a failure mode. Other failures usually result in power supply ticking. RIFAs are the hot side and the only bit that tends to actually let magic smoke out as they’re the only bit that doesn’t fail safe.

Not hantek because you’re paying to have every corner cut. 8 bit sample size. Bad. Low memory depth. Bad. Lowest bidder. Bad. UI and software are just dire. Most of the reviews are from people who haven’t used anything better.

Personally I’d go for the Rigol DS1054Z. 4 channels, crack it to 100Mhz and 24M sample depth and 12 bit ADC.  Software is a bit laggy in places but it seems to win on value. Software is pretty mature as well and most of the bugs have been worked out. Also why it’s still extremely popular today on here. Siglent are also a perfectly good contender.

However one thing to factor in here. Your 2215 has lasted a long time. Nothing you buy today has any guaranteed longevity past 3y really. 5y if you’re lucky. 7y if you’re really lucky. Factor that into the purchase cost. Hantek maybe 1-2y.

Edit: nctnico mentioned GW as well. Definitely worth a look there.

[Fungus]

In the research I've done so far I didn't see any obvious trend in terms of reliability or otherwise in regard Hantek.

From what I've heard the problems with the Hantek are the firmware, both in limited functionality and bugs.

The favorites at the moment would be:
Rigol DS1054Z (4 channels, easily hacked to 100Mhz plus all options)
Siglent SDS1202X-E (200MHz, two channels)
GW-Instek GDS-1054B (nice to use, more 'analog' than 'digital', eg. good FFT but no serial decoders)

The Rigol is an awful lot of bang per buck but there's a lot of irrational Rigol hatred in these forums. There's also a lot of Siglent sellers so opinion isn't unbiased, watch out for that/them.

At the end of the day it probably comes down to where/how you can buy one and how much emphasis you place on the warranty. The hardware seems very reliable so long-term warranty is maybe not important.

PS: If you're thinking about doing any digital stuff at all then go for 4 channels.

[Fungus]

To answer the real question:

Most of my work would be in the low frequency area, audio, power supplies etc. I'd consider it was relatively easy work by oscilloscope standards.

If you were happy with a 60Mhz Tek then you'll be very happy with a 100MHz DSO.

(once you get past the mind-blowing number of things they can do compared to an analog 'scope)

My question is; given the work I expect to do, is the significant increase in cost between the manufacturers' offerings likely to provide me any real benefits, or should I just spend the difference on cheap wine and hookers?

Advice and opinions gratefully received.

It depends on how much wine and hookers you can get for $150 where you live.

[Pete F]

If your 2215 actually exploded it’s probably the RIFA caps. Dead simple fix.

Don’t buy a Hantek.

I'm not sure how you would come to that conclusion. How many 2215s have you found have that issue? Is it a known issue with that machine?

Why would you say "don't buy a Hantek"? From what I can ascertain reading through the people who have bought that brand, the build quality is pretty ordinary. If I was still doing this malarky professionally I would want something decent. My original CRO was a Kenwood (not a brand that comes to mind normally), given to me by my father who was also in the game. It was by far the best oscilloscope I'd ever used and I will always miss that machine (it was stolen). Not high bandwidth and with no bells/whistles, but a rock solid machine that always triggered well and did what it said on the box.

In the research I've done so far I didn't see any obvious trend in terms of reliability or otherwise in regard Hantek. People are generally quick to whinge and complain when something goes wrong, and I've seen most brands pretty well represented on the whingometer. Once I narrow down what it is that I actually need I will refine the brands more. I certainly appreciate the reasons why recommendations are being made so I can learn for myself and make up my own mind, as without that the person at the other end of the keyboard could be sitting there with a lucky 8 ball for all I know.

I’ve repaired numerous scopes (tens of them) including a few 22xx units. The age of the units and the power supply design suggests that as a failure mode. Other failures usually result in power supply ticking. RIFAs are the hot side and the only bit that tends to actually let magic smoke out as they’re the only bit that doesn’t fail safe.

Not hantek because you’re paying to have every corner cut. 8 bit sample size. Bad. Low memory depth. Bad. Lowest bidder. Bad. UI and software are just dire. Most of the reviews are from people who haven’t used anything better.

Personally I’d go for the Rigol DS1054Z. 4 channels, crack it to 100Mhz and 24M sample depth and 12 bit ADC.  Software is a bit laggy in places but it seems to win on value. Software is pretty mature as well and most of the bugs have been worked out. Also why it’s still extremely popular today on here. Siglent are also a perfectly good contender.

However one thing to factor in here. Your 2215 has lasted a long time. Nothing you buy today has any guaranteed longevity past 3y really. 5y if you’re lucky. 7y if you’re really lucky. Factor that into the purchase cost. Hantek maybe 1-2y.

Edit: nctnico mentioned GW as well. Definitely worth a look there.

I didn't intend the post to be about the 2215, so don't want to be drawn too far in to that area. For the record however it's now blowing fuses (didn't do that before). As I said, I believe the source of the smoke was a small PCB type arrangement on one of the semiconductors. It's a weird arrangement and anyone who has actually fixed one of these would know what I'm referring to, but it's a totally crap arrangement in my opinion. I did manage to get a thermal imager on it, and one of the output caps is overheating. It only ran for a few seconds but it was thermally very obvious. That would be in keeping with the age of the machine in my experience, especially one that's not used very often. That was one of the motivations for not pursuing it as it's not unusual to have a whole pile of faulty caps in machines of this age/lack of use. If it was fired up regularly it would likely last much longer. I will reassemble it and put it to one side until I get a replacement. If I find the time/motivation I may throw a new cap in there and see what happens, but it's barely worth the effort in my opinion. These old power supplies can be an absolute PIA to repair sometimes (unless you're into guessing and weegie boards).

I did however want to address the comments made about the Hantek brand. I'm definitely not trying to hold myself out as any form of instant expert, which is why I joined to forum and will accept anyone's opinion. However I also know how to read a spec sheet!

Most oscilloscopes in the price range I'm looking at are 8 bit, including (I believe, though happy to be corrected) the one you recommended. From what I can determine that isn't considered a limitation for this type of work. As I said in my opening post, the memory depth varies from 40K to 2M in the 5000 series. That's the whole point of my question, to determine what is the most appropriate memory for my application. Hantek is a brand, not a model number. I already said I have no need for a 4 channel oscilloscope, I thought the DS1052Z is very dated and didn't overwhelm me. The DS1054 is 2 more channels I don't need and attracts a significant price premium down here in 'stralia. The Hantek can also be hacked, to 200 MHz in that case, no biggie in my opinion, I would just buy the higher bandwidth oscilloscope if it was important to me. Hacking machines isn't a selling point to me.

I was one of the agents for GW when they first came in to Australia. Don't know anything about them now. They were one of the first "affordable" test instrument companies available here, well before the days of the internet of course.

Sadly you're right and longevity is just a sign of the times. What I'm finding is that even buying what was previously a very good brand is no guarantee that you're getting any real value. My wife was only 30 minutes ago bemoaning that fact with our relatively new Asko dishwasher. We've used that brand for many years and are prepared to pay a premium. However the past two appliances from them have been very disappointing. My Hilux is the same. So it goes on. I do sense there is a definite brand snobbery when it comes to the test instrument area, and it appears Hantek is the bottom of the pile. When I was in the game it was GW. That's fair enough, but even I can see that many of the assertions are being made based on no personal or even close experience. Many of the reviews I've seen have indeed come from extremely experienced people in the field and they acknowledged the generally low attention to build detail with the Hantek brand. Nevertheless they have been reasonably balanced reviews. I've looked carefully at both Hantek and Rigol, and it appears to me the Hantek functionality is better. I don't hold out much hope for either of unwrapping an HP when I open the box!

[nctnico]

The Rigol is an awful lot of bang per buck but there's a lot of irrational Rigol hatred in these forums.

The irrational part is in your brains only. The fact is that Rigol puts oscilloscopes on the market with very buggy firmware and not all people are willing to pay good money for a piece of equipment which doesn't work as specified. Especially with a recently introduced oscilloscopes the bang per buck is only there on paper.

[Fungus]

The Rigol is an awful lot of bang per buck but there's a lot of irrational Rigol hatred in these forums.

The irrational part is in your brains only. The fact is that Rigol puts oscilloscopes on the market with very buggy firmware and not all people are willing to pay good money for a piece of equipment which doesn't work as specified.

Is Siglent remembering to put all the components on the PCB these days or do they still post little bags of capacitors to people so they can solder them in at home?

How's the firmware coming along, it is out of beta yet?

[BillB]

 

[tautech]

Should I mention Yaigol.....nah I won't. 

[Pete F]

Yeah look guys (as it invariably is ) I appreciate the passion. I'm very new here, but have been around forums ... well more than one or two! I know how this all works.

My trade and background was in electronics but I don't really have a passion for it. Just the same I do have a strong interest in it and was literally bought up on it. But forums, by definition, attract those with a real strong interest and a lot of time invested in the topic. That's great and why I'm grateful to be invited on here. What comes with that passion however is a bias, particularly a gravitation to pro-quality and serious users. For that reason I didn't ask for a recommendation on brands or models, and instead I asked a couple of specific questions about DSOs in general.

If anyone is in doubt, pretend this is a sport car forum and I've just come on and asked if I should buy a Kia. My post would likely be torn down by the administrators! But for somebody who isn't a sport car enthusiast and only wants a car to drive to the shops or the train station by themselves, driving a maximum of 2,000 km a year a Kia may be perfectly acceptable and they don't need to go out and buy a Ferrari!

So to bring the topic bang to the original question I'm keen to know how a (relatively) high bandwidth oscilloscope (let's say 100 MHz) would help somebody who generally works in the audio range of frequencies. Maybe clock freq, but that's about it at this stage. I personally can't see how 100 MHz has even the vaguest assistance, but that's why I'm here to ask.

I think I'm getting the idea of the memory, and from my understanding if I have a slow sweep (as I might with low frequencies) then want to store and zoom in on that waveform, I'm going to need memory to do so. So how much memory is "enough" to do that. Am I going to get a 40K oscilloscope and always be wanting more memory every time I use it? What if it's 1 M/2M/25M. Bragging rights carry no weight with me I'm afraid, see paragraph 2 above

I appreciate those taking the time to educate me so I can make a rational decision based on facts.

[nctnico]

It seems 50 to 70MHz is at the low end of the memory bandwidth nowadays and affordable oscilloscopes have between 5Mpoints to 10Mpoints of memory per channel. Aim for that. With a lot of memory you can record a single sweep and zoom in on areas of interest. What I often do is just capture a whole lot of data without triggering on a specific event and just zoom in on the area of interest. That is quicker than setting up specific trigger conditions and re-measuring different parts of the signals over and over. Especially if it takes some effort to generate the signal of interest (an intermittent fault for example).

Regarding your sports car remark: I think you got that wrong. Nobody has been suggesting buying a US $10k scope so far. To get back to the car analogy: it is similar to buying a too cheap car which has wobbly wheels, squeeking wind shield wipers, sloppy steering, etc while spending a little bit more buys you a car which drives excellent. Still not a sports car by far but something which will get you from A to B comfortably.

[Pete F]

Feel free to substitute "Ford Taurus" for "sport car" and "Mazda MX5" for "Ferrari" if it makes you feel better. It was the analogy I was getting at and not a literal translation 

The point was, I wasn't asking for ANY brand recommendation, because I know how that ends up. Which it did!
 
A car with squeaky windshield wipers and sloppy steering will still get me to the train station in precisely the same time, and I didn't ask for a car that was "comfortable" I asked if a car with fuel injection and capable of 160 kph might be important to me in driving to the train station. A "comfortable" car might be important to you, but what makes you think I'm interested in paying more for that?

[nctnico]

A "comfortable" car might be important to you, but what makes you think I'm interested in paying more for that?

Well, feel free to spend money on anything you want. However: you wouldn't be the first to be dissapointed by buying something cheap first and end up buying a different (more expensive) piece of equipment from a better brand. I've been down that road a couple of times myself (and lost quite a bit of money in the process) so I write from my own experience.

[tautech]

Stepping from a CRO to a DSO there's a few simple things to understand.
Traces won't appear as crisp and at first you think WTF, what's all this noise ? Check for Daves vids on why DSO's appear noisy.
Mem depth and sample rate are related in so much as a captured waveform represents a portion of the sample rate related to the timebase setting and therefore the # of data points captured. The more mem you have the more data points at any given timebase setting and the more accurate the reconstruction of the waveform. For ordinary repetitive stuff none of this matters too much if the interpolation is accurate based on the points available.
For non-repetitive or complex waveforms is when you want as many as possible so in pro scopes high sampling rates and large mem depth are there to give the best possible reconstruction as to not miss any glitches or spikes.
Also the modern DSO has a truly digital trigger with which any data point that meets the trigger conditions will permit you to capture the waveform that created it.
All fairly straightforward once you get your head around it. 

[Fungus]

So to bring the topic bang to the original question I'm keen to know how a (relatively) high bandwidth oscilloscope (let's say 100 MHz) would help somebody who generally works in the audio range of frequencies. Maybe clock freq, but that's about it at this stage. I personally can't see how 100 MHz has even the vaguest assistance, but that's why I'm here to ask.

The "MHz" in the bandwidth rating of an oscilloscope usually isn't related to the frequency of the signals you're looking at.

Any signal other then a perfect sine wave has harmonics. These harmonics can be much higher frequency than the base frequency. eg. Square waves have infinite harmonics, to see a perfect square wave on an oscilloscope screen you need infinite bandwidth. This is true even if you're only looking at a 1Hz square wave!

It's also true of zero Hz square waves. A single transition from low to high voltage may have infinite harmonics. You need a lot of bandwidth to see the detailed characteristics of a single rising edge. You may see discussion of "rise time" in forums like this, that's what it refers to. "Rise time" is a better measure of oscilloscope performance than "MHz" but rise time would confuse sales people (is a 4ns rise time twice as good as a 2ns rise time...?  )

So .... it doesn't matter if your clock freq has a base frequency in the audible range, you need a lot more than audio range in your oscilloscope if you want to look at it.