Is a Rigol MSO5000 overkill for a hobbyist?

[nctnico]

I disagree it is viable strategy...
Fungus keeps repeating this all the time...
I don't know why.
He can explain how he thinks it works.

I'll explain it right after you explain how "noise" magically stops right at the point where Siglent oscilloscopes run out of vertical gain. 

But ... you never explained it last time around, and we've already been warned to stay on topic in this thread.

The only reason I answered is because amplifiers might make an interesting video topic for Dave. Is a $5 Aliexpress amplifier going to make a difference?

Not for the internal noise. What you want from an external amplifier is to amplify in a signal that is below the measurement range of the instrument. In addition the amplifier will need to have a low enough noise floor in order not to drown the signal in noise from the amplifier itself. IOW: an external amplifier is not going to make the noise performance of the MSO5000 better. Once noise is added at some point in the signal chain, it cannot be removed (without removing any of the frequency spectrum from the signal).

[Fungus]

Not for the internal noise ... an external amplifier is not going to make the noise performance of the MSO5000 better.

Of course not, but you'll be able to see the signal.

What an amplifier does is change a thing called the "signal to noise ratio".

https://en.wikipedia.org/wiki/Signal-to-noise_ratio

[nctnico]

Not for the internal noise ... an external amplifier is not going to make the noise performance of the MSO5000 better.

Of course not, but you'll be able to see the signal.

What an amplifier does is change a thing called the "signal to noise ratio".

An amplifier will always add noise and thus make things worse. So yes, the signal to noise ratio will be different but not in a positive sense.

[Fungus]

An amplifier will always add noise

How much?

and thus make things worse

You're saying it will increase the noise more that the signal? 

[egonotto]

Hello,

For example, if the SNR is better at 1 V/div than at 1 mV/div then a good amplifier can shift the small input signal to the range with better SNR and be quite useful.

Best regards
egonotto

[alm]

An amplifier will always add noise and thus make things worse. So yes, the signal to noise ratio will be different but not in a positive sense.

How can you make such a blanket statement? Say you have a 10 mVrms signal and a 10:1 signal to noise ratio (so 1 mVrms noise), a scope with a noise floor of 5 mVrms, and a 10x amplifier with also a 10:1 signal to noise ratio. Without amplifier the scope would display a 10 mVrms signal with 5 mVrms noise (1^2 + 5^2 ~ 5^2), a 2:1 signal to noise ratio. That's not very usable.

With amplifier the output of the amplifier would be a 100 mVrms signal with 14 mVrms noise, or a 7:1 signal to noise ratio, which is indeed worse than the original 10:1 ratio. But if you then feed it into the scope, the scope will display a 100 mVrms signal with 15 mVrms noise, or a signal to noise ratio of 6.7:1, which is a lot better than 2:1. So the signal to noise ratio on the scope screen is a lot lower. This principle is often used in measuring power supply noise, for example see LT AN-124 by Jim Williams.

Treating the noise of a scope front-end as constant is a good first order approximation at high vertical sensitivities. As you decrease the sensitivity it tends towards a constant ratio of the vertical sensitivity.

[nctnico]

Hello,

For example, if the SNR is better at 1 V/div than at 1 mV/div then a good amplifier can shift the small input signal to the range with better SNR and be quite useful.

IMHO you can't use SNR together with visualising a signal. If you amplify a signal, you'll add noise so compared to the original signal, the SNR will be worse. The fact that a higher amplitude signal is easier for an oscilloscope (or any measuring device) to work with doesn't change that.

[thm_w]

Sure we can.

We've been mocking the $2k speaker cable buyers since at least the 1970s.

If a $5k mountain bike doesn't buy much over a $1k bike then we can mock the $5k bike buyers as idiots.

Maybe the time of the $800 oscilloscope is here.

(for the "hobbyist" user - there's always going to be a small percentage of people who need extra...no argument there, but it's not everybody)

Yes but no one is an idiot outright and deserves to be mocked, if a difference exists (audiophiles are out), the person has money, and uses the product. Its simply a matter of diminishing returns.
The same applies to almost any consumer product out there.

https://www.cpubenchmark.net/cpu_value_available.html

[Caliaxy]

Hello,

For example, if the SNR is better at 1 V/div than at 1 mV/div then a good amplifier can shift the small input signal to the range with better SNR and be quite useful.

IMHO you can't use SNR together with visualising a signal. If you amplify a signal, you'll add noise so compared to the original signal, the SNR will be worse. The fact that a higher amplitude signal is easier for an oscilloscope (or any measuring device) to work with doesn't change that.

I can't believe you are serious.

Any amplifier would amplify in equal extent the signal *and* the noise presented to its input, while adding a little bit of noise itself. Of course, it won't improve the SNR of the original signal in any way. However, here the assumption is not that the SNR of the original signal is too low to get a decent representation, but that the sensitivity of the input amplifier of the scope is at its limit (can't discern between signal and noise). So, amplifying the original signal would help presenting to the input stage of the scope a signal with an amplitude high enough so the noise it adds is not that relevant anymore.

[hans]

A SNR is a ratio. It's unitless. You can have a 60dB SNR for 1Hz signals with thermal noise floor of -174dBm. Then your signal to measure is still -114dBm. That's 4e-15 W of power, or 0.44uVrms into 50Ohm. For sure RF receivers can detect that, but those also use techniques like matched filters and other forms of correlation.

(Analog) amplifiers always adds noise, or in other words, each stage has a noise figure which is greater than 1 (0dB). This means that SNR can only go down. The best LNA's may utilize noise-cancellation, but even those still have 1-2dB NF due to imbalances and other noise sources that can't be cancelled.

Quantization noise on the other hand is fixed. It's related to Vref and the number of bits. If your signal amplitude isn't utilizing range, then your signal is too weak. If your A/D has more quantization noise than your signal, you should use a higher resolution A/D. Both conditions can also be true, and then you're probably measuring crap. If only signal range is true but not the noise, then it's of no use to amplify further. You could "trade" A/D resolution for amplification given that the signal is noisy enough, especially now that 12-bit A/Ds or higher have become common ground it's a neat trick..

[2N3055]

I disagree it is viable strategy...
Fungus keeps repeating this all the time...
I don't know why.
He can explain how he thinks it works.

I'll explain it right after you explain how "noise" magically stops right at the point where Siglent oscilloscopes run out of vertical gain. 

But ... you never explained it last time around, and we've already been warned to stay on topic in this thread.

The only reason I answered is because amplifiers might make an interesting video topic for Dave. Is a $5 Aliexpress amplifier going to make a difference?

(...except he already sent back his MSO5000  )

I did explain to the length. It's just you either have no mental capacity to understand (because there was enough repetitions)  or you have an agenda.. go figure that one out..

[gf]

Treating the noise of a scope front-end as constant is a good first order approximation at high vertical sensitivities. As you decrease the sensitivity it tends towards a constant ratio of the vertical sensitivity.

Depends on the way how attenuator and gain stages and their interaction are designed. A 100:1 attenuation followed by 10x gain is certainly noisier than a 10:1 attenuation alone. Some scopes (at least some cheap ones) are not optimized in this sense, but attenuate only in coarse steps (e.g. 100:1, 10:1) and then re-amplify the signal. Then 10V/div, 1V/div and 100mV/div have a similarly good SNR_FS, while 2V/div and 200mV/div are 14dB noisier (for example).

[2N3055]

IMHO the whole discussion about what is overkill or not is entirely moot. There are people spending > $5k on mountainbikes, dirt bikes, audio gear or even a BBQ. Somebody I know, spend like 30k euro on an audio amplifier and a couple of speakers.

Nailed it, 100% !
None of us can quantify another's spend as overkill, none of us !

Sure we can.

We've been mocking the $2k speaker cable buyers since at least the 1970s.

If a $5k mountain bike doesn't buy much over a $1k bike then we can mock the $5k bike buyers as idiots.

Maybe the time of the $800 oscilloscope is here.

(for the "hobbyist" user - there's always going to be a small percentage of people who need extra...no argument there, but it's not everybody)

Of course you can claim that.
Since you routinely take a liberty to make all kinds of decisions for other people including hiding data, recommending stuff you never saw in real life, explaining and advocating stuff you don't understand and such.

It is quite a leap from calling an idiot a person that gives 5000 USD for speaker cables that I can buy for 20 USD, and a person that wants to buy 1500USD scope instead of 800USD scope, or 800USD scope instead of 350 USD scope.
These scope WILL have measurable differences in specifications and will in fact provide additional benefit for increased cost.

Whether that is important to purchaser it is only theirs choice.
For instance, my eyesight is not what it used to be.. I realized that I use Picoscope (with 23" screen) and SDS6000A more than my Keysight MSOX3104T for pretty much anything because it is simply easier to see because of bigger screen. So someone might splurge for MSO5000 even if DS1000Z would have been scope enough for them only for bigger screen. Someone will have to go for SDS200X+ if they want 50 Ω signal path in a scope or 500+ MHz BW.
OR if they want 50 Ω and 2 ch is enough from them they might go for odl DS2000A from Rigol they might get on sale from Rigol for very little money. Or something else ,whatever...
Some will work mostly on small signals and would need a scope that is better there.


What we can and should discuss is our opinions as to what we would and would not buy for ourselves and rationale behind our decisions so other people can have a point of reference and as a discussion.

I know, for instance, that on hobby stuff I have a sort of limit that is based on social environment I live in. I can justify buying MSOX1104T for work, where it paid off itself with time. But for "it's for playing", it is different.
It will be different for each country, and general economic status of each person.
I don't think there is a universal rule that would apply to all.
In some countries 100 USD is 2 months of food and in some it is pocket change.
Not to mention huge socioeconomic spread even in 1st world countries. In USA alone there are tens of millions of people in poverty. I assure you they have different purchasing power than someone earning 100K USD a year..

[Fungus]

These scope WILL have measurable differences in specifications and will in fact provide additional benefit for increased cost.

It's about diminishing returns, not absolute price.

I'm not at the point where I'd call somebody an "idiot" for owning a $1500 Siglent, but I unlike certain forum members wouldn't look down on people who bought something less than that.

As Dave's video points out: We live in a time when an $800 'scope is enough for just about anybody.

Most people would be far better off spending that extra $500-$600 on something else than the Siglent. eg. A decent soldering iron/power supply/etc. That's my opinion and the only thing I've ever tried to express here, ie. look at the bigger picture.

[nctnico]

Hello,

For example, if the SNR is better at 1 V/div than at 1 mV/div then a good amplifier can shift the small input signal to the range with better SNR and be quite useful.

IMHO you can't use SNR together with visualising a signal. If you amplify a signal, you'll add noise so compared to the original signal, the SNR will be worse. The fact that a higher amplitude signal is easier for an oscilloscope (or any measuring device) to work with doesn't change that.

I can't believe you are serious.

Any amplifier would amplify in equal extent the signal *and* the noise presented to its input, while adding a little bit of noise itself. Of course, it won't improve the SNR of the original signal in any way. However, here the assumption is not that the SNR of the original signal is too low to get a decent representation, but that the sensitivity of the input amplifier of the scope is at its limit (can't discern between signal and noise). So, amplifying the original signal would help presenting to the input stage of the scope a signal with an amplitude high enough so the noise it adds is not that relevant anymore.

Still, what you see on screen also contains noise from the amplifier. Real world example: recently I did some measurements on low level RF signals that where below my spectrum analyser's noise floor. I used an amplifier that has 20dB of gain in the frequency range I was measuring. This got me about 10 to 15dB extra room. Not 20dB.

Again: IMHO you need to be really carefull to throw terms like SNR around for purpose of overcoming a limitation in a device to do measurements on a signal. IOW: 'improving' SNR is not the right term. Better call it amplification or magnification.

[Fungus]

I used an amplifier that has 20dB of gain in the frequency range I was measuring. This got me about 10 to 15dB extra room. Not 20dB.

Please explain to us how it "made things worse".

[Fungus]

I'll explain it right after you explain how "noise" magically stops right at the point where Siglent oscilloscopes run out of vertical gain. 

I did explain to the length. It's just you either have no mental capacity to understand

Maybe some sort of link to that so people can judge for themselves...? 

[nctnico]

I used an amplifier that has 20dB of gain in the frequency range I was measuring. This got me about 10 to 15dB extra room. Not 20dB.

Please explain to us how it "made things worse".

Simple: the original signal now has 5dB to 10dB of extra noise on top so SNR got worse.

[Fungus]

I used an amplifier that has 20dB of gain in the frequency range I was measuring. This got me about 10 to 15dB extra room. Not 20dB.

Please explain to us how it "made things worse".

Simple: the original signal now has 5dB to 10dB of extra noise on top so SNR got worse.

Ho do you keep "extra room" and "got worse" in your head simultaneously? I'd love to know.

(and why were you using an amplifier if you've already explained to us why Siglents never need one?)

[nctnico]

I used an amplifier that has 20dB of gain in the frequency range I was measuring. This got me about 10 to 15dB extra room. Not 20dB.

Please explain to us how it "made things worse".

Simple: the original signal now has 5dB to 10dB of extra noise on top so SNR got worse.

Ho do you keep "extra room" and "got worse" in your head simultaneously? I'd love to know.

It should be clear that the original signal is distorted with extra noise after the amplifier. Only the strongest parts of the signal remain. The finer details are gone forever.

(and why were you using an amplifier if you've already explained to us why Siglents never need one?)

I never wrote that!

[2N3055]

These scope WILL have measurable differences in specifications and will in fact provide additional benefit for increased cost.

It's about diminishing returns, not absolute price.

I'm not at the point where I'd call somebody an "idiot" for owning a $1500 Siglent, but I unlike certain forum members wouldn't look down on people who bought something less than that.

As Dave's video points out: We live in a time when an $800 'scope is enough for just about anybody.

Most people would be far better off spending that extra $500-$600 on something else than the Siglent. eg. A decent soldering iron/power supply/etc. That's my opinion and the only thing I've ever tried to express here, ie. look at the bigger picture.

That is exactly what I'm saying. There is a difference. If it matters to particular person it's up to them. I personally recommended on numerous occasions to people to buy DS1000Z or SDS110XX-E instead of more expensive scope, if budget was tight and it was good fit for them. Dropping 80% of all your money on scope when you like tinkering with voltage references is bad decision. It would make more sense to buy 350-400USD decent entry level scope and a good bench multimeter or a two. Also I think people should buy "good enough" equipment and spend their money on projects. Otherwise they are just instrument collectors. Which is also OK, but then all this "rational choice" thing goes out of the window.

I disagree with Dave making such a blanket statement. I would say "many" people not "just about anybody". People that doesn't do any daliy practical work can sometimes lose contact with how it actually is when you do some actual work..

Point of diminishing returns comment needs clarification: it is only point of diminishing returns if you already accomplished your goals and you are needlessly optimizing. 
If there is a need that is not fulfilled sometimes you need to pay twice for a single feature. If that is what you need, you can't say "point of diminishing returns".

[Martin72]

Please explain to us how it "made things worse".

In an ideal world the only character of a amplifier would be the amplification of the signal and nothing else.
In real he got several characters which will be added to the signal.

[2N3055]

I used an amplifier that has 20dB of gain in the frequency range I was measuring. This got me about 10 to 15dB extra room. Not 20dB.

Please explain to us how it "made things worse".

Simple: the original signal now has 5dB to 10dB of extra noise on top so SNR got worse.

Ho do you keep "extra room" and "got worse" in your head simultaneously? I'd love to know.

(and why were you using an amplifier if you've already explained to us why Siglents never need one?)

It is simple.

Scope A has 1mV RMS noise on 1V range (full vertical screen, not per/div to make it simpler) that is dynamic range of 1:1000.
Scope B has 100uV RMS noise on 1V range (full vertical screen) that is dynamic range of 1:10000. That is 10x more.

Preamp will only shift this intrinsic dynamic range to lower voltages (and it will be little worse for preamps own noise).

Preamps will have it's own BW (both lower and upper and linearity of it), phase distortions (group delays), it's input and output impedance that has to be matched to DUT and scope respectively. Also their characteristics have to be characterized and normalized in some way on a scope to make measurements.

Like I said many times before, they are useful for specific jobs, require knowledge of preamp, DUT, scope and all practice  involving it's use. You will usually have dozens of them for different purposes. Just ask RF people how many amp, preamps, LNA, attenuators and all kind of crap they have if they work with just different frequencies (on 50Ω systems that should be well standardized).

Preamps with scopes are specialist equipment that are not replacement/fix  for scope bad characteristics. Scope input is marvel of universality.  One size fits kind of stuff. That is what makes scope such an universal instrument. From 10s of uV to 100s of volts with just a flick of a button.  Fast and easy.

Can there be workarounds for specific use. Of course they can. But it gets old fast.

Stop inventing stupid excuses how something that is worse is not "if you just  do this trick".. It ain't so.
In this pathetic struggle to prove you're right by getting yourself involved into topics that are technically way over your head, you didn't even notice there is consensus being made that:

A. MSO5000 is noisy. No amount of lying, hiding and delusional thinking can change that. That one you lost.
B. For many people it might not be a problem. But they have to be aware of shortcomings and make informed decisions to chose it. In case it is good enough for them it is very good hobby scope. Even good for noncritical pro work.

[Bud]

Scope A has 1mV RMS noise on 1V range (full vertical screen, not per/div to make it simpler) that is dynamic range of 1:1000.
Scope B has 100uV RMS noise on 1V range (full vertical screen) that is dynamic range of 1:10000. That is 10x more.

This is irrelevant when it comes to observing a signal on the screen. A 8-bit scope will only fit roughly 200 ADC levels vertically, and a 10-bit scope will fit 820 levels (scopes have 20% of ADC range reserved for top and bottom display margins so they are not visible when the signal is centered with zero vertical offset). On 1V range a test signal will look the same for both your 1:1000 case, and 1:10000 case.

[2N3055]

Scope A has 1mV RMS noise on 1V range (full vertical screen, not per/div to make it simpler) that is dynamic range of 1:1000.
Scope B has 100uV RMS noise on 1V range (full vertical screen) that is dynamic range of 1:10000. That is 10x more.

This is irrelevant when it comes to observing a signal on the screen. A 8-bit scope will only fit roughly 200 ADC levels vertically, and a 10-bit scope will fit 820 levels (scopes have 20% of ADC range reserved for top and bottom display margins so they are not visible when the signal is centered with zero vertical offset). On 1V range a test signal will look the same for both your 1:1000 case, and 1:10000 case.

Well I have 3x 12 bit scopes and a 16bit one. 
Some of these big screen touch scopes support real vertical zoom where you can stretch signal on the screen..

Also that is just neat numbers to illustrate math.
But you are making a good point. In real life we have comparison of a scope that has less than 100 points vertically (because of noise) to one that has 200..
And when using FFT, because of software gain, much bigger difference..