Hello, i am thinking about oscilloscope purchase. The main use is for welding machines repair. What of bandwidth is able for this work? I think that 50MHz 2channel is enough. Have anybody experience to suggest?

 

hi Teris,
Welcome to AAC
Sounds good enough for that type of work.
Get some decent probes, ideally * 1 and *10 attenuation.

E

 

Get some decent probes, ideally * 1 and *10 attenuation.

To clarify, I believe Eric is referring to probes that have a switch on the probe bocy for either 1x (no attenuation) or 10x (1/10th attenuation).

The 10x can be used for most measurements, as it has a high input impedance (10megΩ) and low capacitance (≤10pf typically).

The 1x is used if you need more sensitivity at the the most sensitive range. The disadvantage is it has roughly 10 times lower resistance and 10 times higher capacitance, so should generally be used only for low frequency signals (i.e. ≤ a few tens of kHz or so).

 

I always use the probes at x10.
When getting a scope, go for the highest bandwidth you can for the money you want to spend. Like all tools, the best you can afford the better I think. The extra bandwidth may come in handy, and if you don't have it, you cannot add bandwidth. That is unless you get a lower bandwidth one and do the firmware hack that is mentioned for some models

 

Hello, i am thinking about oscilloscope purchase. The main use is for welding machines repair. What of bandwidth is able for this work? I think that 50MHz 2channel is enough. Have anybody experience to suggest?

Aside from bandwidth and probes you may want to look at a few other features. Things like delayed sweep functions and other assorted features. As to your intended application a 50 MHz band width should be plenty adequate.

Ron

 

And the ability to save the waveforms to USB (or something) for later printing could be handy.

 

And the ability to save the waveforms to USB (or something) for later printing could be handy.

That is a nice feature I plan on any future scopes. Image memory.

Ron

 

Thank you for your replay and welcome. I keep the 50MHZ and the ability to save the waves. What any feature do you suggest in order to have on my mind? The used analog scopes have similar features or i must to go for digital?

 

Nice thing about the digital scopes is the various save functions, including screen shots. Most also will save settings so if you repeat a test you can recall settings you used the last time you did the test. Also some of the digital flavors off the boat from China offer some decent prices. The Rigol scopes for example offer a 100 MHz bandwidth scope at around $300. The Rigol DS1102E Oscilloscope for example gets some pretty good reviews. They also mention some of the features for comparison.

Ron

 

The Rigol DS1102E Oscilloscope for example gets some pretty good reviews.

Yes, in 2009 when it came out (your linked "test" is from 2012!).

In 2018 it would be foolish to buy a Rigol DS1102E.

(Also, I really wouldn't rely on a goofy site like www.top5reviewed.com to make any test instrument related decisions).

Forget the Rigol DS1102E, this scope is one of Rigol's very early scope designs and now almost a decade old! Compared to modern offerings it's horribly slow, has an awfully small low-resolution screen, and very limited capabilities.

The DS1102E has long been superseded by Rigol's DS1000z Series (1054z, 1104z), which not only has a bigger screen and more functions (incl optional serial decode), it also comes with four channels. And this at a price comparable to the old DS1102E.

Also, there's the Siglent SDS1000X-E series which is in a similar price range and offers comparable functionality to the DS1000z, at similar prices.

There's really no reason to even think about the DS1102E today.

 

To clarify, I believe Eric is referring to probes that have a switch on the probe bocy for either 1x (no attenuation) or 10x (1/10th attenuation).

The 10x can be used for most measurements, as it has a high input impedance (10megΩ) and low capacitance (≤10pf typically).

The 1x is used if you need more sensitivity at the the most sensitive range. The disadvantage is it has roughly 10 times lower resistance and 10 times higher capacitance, so should generally be used only for low frequency signals (i.e. ≤ a few tens of kHz or so).

I'd be *very* careful with these switchable 1x/10x probes, because with most of them it's pretty easy to unintendedly switch from x10 to x1. And if that happens during probing of a high voltage signal (i.e. 300V in a switch-mode PSU) it can very well result in a killed scope front-end, and maybe even personal injury.

This also is one of the reason most scope manufacturers don't include switchable x1/x10 probes with their digital scopes.

I'd avoid switchable probes like a plague and just get a pair of x10 probes which are fine for 99% of general purpose measurements in the sub-300MHz range. There's not really much use for x1 probes, and if you really need to measure very low voltages then an active probe is the proper way to do it.

 

Yes, I agree. Get fixed x10 probes, not switchable or x1 probes.

 

Yes, in 2009 when it came out (your linked "test" is from 2012!).

In 2018 it would be foolish to buy a Rigol DS1102E.

The DS1102E has long been superseded by Rigol's DS1000z Series (1054z, 1104z), which not only has a bigger screen and more functions (incl optional serial decode), it also comes with four channels. And this at a price comparable to the old DS1102E.

Also, there's the Siglent SDS1000X-E series which is in a similar price range and offers comparable functionality to the DS1000z, at similar prices.

There's really no reason to even think about the DS1102E today.

More expensive oscilloscopes like the Keysight EDUX1002A, have more features and higher quality that vindicate the different price?

 

More expensive oscilloscopes like the Keysight EDUX1002A, have more features and higher quality that vindicate the different price?

Well, the EDUX1002A is a cut-down DSO-X1102A variant which comes with laughable 100kpts of memory (the DSOX1102A comes with 1Mpts which itself isn't great, especially as thanks to its MegaZoom ASIC the real available memory might only be a fraction of that).

The DSOX1102A is a nice basic scope and like all Keysight scopes easy to use, but compared with the current Rigol and Siglent counterparts it lacks functionality and features (like Ethernet), plus the included probes are pretty crap. It's solidly built but so are current Rigol and Siglent offerings.

Then there's the price: the EDUX1002A is currently $450 which is $50 more than the much more capable 4 channel Rigol DS1054z (which can be unlocked to 100MHz and with serial decoders for free). The 70MHz DSX1102A is $650 which is $270 more than the 200MHz Siglent SDS1202X-E which goes for $380. Go figure.

At the moment the Rigol DS1054z and the Siglent SDS1000X-E Series are probably the best scopes you can find in the low-cost market today.

 

I haven't looked at scopes for a long time, so I have no familiarity with current offerings.

50 MHz is probably enough for what you want to do, but consider 100 MHz if the cost increment isn't too much. Chances are for most of your work 10 MHz would suffice. Service work usually is considerably less demanding than design in terms of scope bandwidth. Most fast scopes have a bandwidth-limiting feature (e.g. a 100 MHz scope might limit at 20 MHz). This is sometimes extremely useful for removing the high frequency noise "fur" from the screen. The faster the scope, the more time you're likely to spend trying to distinguish what is "real" from artifacts resulting from connection of the instrument to the circuit under test.

There were some scopes that had isolated inputs, which means that the "ground" for each channel is independent and neither is connected to the chassis or line ground. As I recall, the allowable voltage between the channel grounds and earth ground wasn't terribly high, but none the less useful.
This feature can be hugely useful in some circumstances. Consider something like a resistor with one end at 20 volts and the other end at 19.9 volts, relative to circuit zero volts, and you want to look at a 50 mV waveform across it, while also looking at some digital signal that swings between 0 V and 5 V.
With non-isolated inputs you won't be able to turn the sensitivity for the channel up very far before you run out of ability to bring the trace off of the ceiling and back down onto the screen, so it impairs your ability to see detail well. With the crud that digital scopes apparently add to the waveform (really just digitizing uncertainty sometimes coupled with aliasing due to screen resolution), this can make things quite difficult. With isolated channels, you can put the "ground" for one channel on the 19.9 V end or the resistor and the probe tip on the other end, and turn up the gain to get a full screen of just the 0.1 V difference. AC coupling the scope input often is a very workable solution for this general sort of thing, but not always. (I had a high-school kid who was extremely smart but had, not surprisingly, limited experience working for me years ago. He was trying to fix a video terminal and working on the power supply. "Be careful" I said, "what looks like ground around those capacitors isn't", but I didn't explain well enough. He burned the insulation off the ground lead of the scope probe, but fortunately did no damage to the rest of the probe, which was good because those probes were $300-400 a pop.)

Some scopes have a "bucking" feature that essentially adds or subtracts a DC level to the input signal that accomplishes a similar end result, but I think that is quite rare.

A lot of digital scopes have an "autoset" feature. Except on rare occasions, it's been useless or worse on any scopes I've used (for one thing, it encourages someone else who comes to look at your scope while you're working to push it, messing up your time-consuming set-up).

I used an expensive Tek digital that had a nice big spinner knob, but the idiots who did the firmware decided that to scroll through the LONG memory in the thing you should use a tiny little knob of the grasp-turn-release-grasp-repeat variety. Scrolling was the only function the big knob would have been really advantageous for; most of what it was used for a small knob would have sufficed. And people wonder why I carry a truncheon dangling from my belt when I work with programmers.

I'd really like a scope that would accept voice commands for those occasions then have 3 hands just isn't enough.

Tektronix used to publish some really good stuff on selecting and using scopes. I think there was one called some like The ABCs of Probes and one called The XYZs of Oscilloscopes. Both originally date from the days when there were only a few digital scopes around (and early digital scopes were horrible, horrible things in terms of controls design), but the probe info is still fully valid and the XYZs, even if not update, still has pertinent info. Many people who use scopes have little understanding of probes, and overlook things like a decent 10 meg input attenuator probe will have in input impedance of around 250 ohms at 50 MHz, a cheap one even more. The Tek document explained all of that stuff really well.

Beware the scope probe ground wire! The inductance of a 6 inch ground wire with a passive probe can often make fast signals look far worse than they really are, even with a 50 MHz bandwidth.

Having at least one 100:1 attenuator probe is occasionally handy. The ordinary ones usually allow peak voltage of only a few hundred volts, but that's still enough for lots of things. High voltage, high bandwidth probes are huge and very expensive, even on the used market.

Current probes can be really useful and some scopes have support for connecting one directly to the scope instead of an intervening box, but both the scope that will support this and the current probe are expensive items.

Have a look at the prices for probes for some of the GHz scopes that are around. A probe will cost more than what you will likely spend for your scope, probably by a large margin. Some of those probes are actually designed to be soldered into the circuit under test.

If I bought a Keysight product I'd put some ugly black tape over the name, just because I think it so incredibly stupid. It sounds to me like some outfit that will help you find the optimal location to erect a billboard. Agilent was bad enough.

If there are electronics trade shows in your area and you aren't in a hurry to buy, the shows can be a good place to look at scope vendors' wares. Of course they all tell you why theirs is better than all the rest. If you are considering a name-brand scope that is reasonably expensive, it is worth getting in touch with the local sales rep to see if you can arrange a demo or even a loan for a day or two.

 

With the crud that digital scopes apparently add to the waveform (really just digitizing uncertainty sometimes coupled with aliasing due to screen resolution),

There's no "crud". The thing is that a signal on a DSO looks more noisy than on an analog scope simply because the DSO shows noise that on an analog scope usually gets under in the gloom of its phosphor of the base signal.

As to aliasing, on modern scopes that's really only a problem if you don't know your instrument or don't know what you're doing. Admittedly, aliasing was a big issue back in the old days ('80s) when sample rates were low and sample memory tiny, but a lot has changed since then.

A lot of digital scopes have an "autoset" feature. Except on rare occasions, it's been useless or worse on any scopes I've used (for one thing, it encourages someone else who comes to look at your scope while you're working to push it, messing up your time-consuming set-up).

Actually, on auto-set on modern scopes works very well, even in more complex situations, and even on simpler scopes like the Rigols and Siglents mentioned before.

I used an expensive Tek digital that had a nice big spinner knob, but the idiots who did the firmware decided that to scroll through the LONG memory in the thing you should use a tiny little knob of the grasp-turn-release-grasp-repeat variety. Scrolling was the only function the big knob would have been really advantageous for; most of what it was used for a small knob would have sufficed. And people wonder why I carry a truncheon dangling from my belt when I work with programmers.

I know how you felt, Tek must have (and still does I'm sure) employed UX developers who really hate humans and had them design their DSO UIs. But then, Tek gave up its crown as the No. 1 scope manufacturer when analog scopes died out, its digital scopes were lackluster and with little innovation, and today there are few digital scopes that are worse in UI and performance than those made by Tektronix. Which is why their sales dropped rock-bottom (they mostly seem to sell to the Edu market which is more forgiving).

Today, if you want a high end DSO you go to either LeCroy (who does stuff no-one else does) or Keysight, mid-range scopes usually come from LeCroy, Keysight or Rohde & Schwarz, and entry-level scopes from Keysight, Rohde & Schwarz (former Hameg) or, if money is tight, from Rigol or Siglent.

I'd really like a scope that would accept voice commands for those occasions then have 3 hands just isn't enough.

Keysight had voice input for its Agilent 54800 Series and DSO8000A/DSO80000A, and I think also for some current models. It works, but I can't say it's more than a toy.

Tektronix used to publish some really good stuff on selecting and using scopes. I think there was one called some like The ABCs of Probes and one called The XYZs of Oscilloscopes. Both originally date from the days when there were only a few digital scopes around (and early digital scopes were horrible, horrible things in terms of controls design)

Some were, but certainly not all of them. Actually, there were some surprisingly capable DSOs even back then.

but the probe info is still fully valid and the XYZs, even if not update, still has pertinent info. Many people who use scopes have little understanding of probes, and overlook things like a decent 10 meg input attenuator probe will have in input impedance of around 250 ohms at 50 MHz, a cheap one even more. The Tek document explained all of that stuff really well.

While I agree that some of these old documents have parts that are still valid and to some extend even relevant relevant, there are lots of modern documents that deal with the topic in a suitable manner, all while staying relevant to modern-day scopes. The danger with such very old documents is that besides from the still relevant parts they also contain a lot of things that are no longer relevant or appropriate with digital scopes, and carry the danger that the reader learns outdated or even counter-productive information.

For example, this is a good primer regarding probes:
6 Hints for Better Scope Probing - Agilent

There's plenty of other explanatory stuff from LeCroy, Keysight and Rohde & Schwarz, covering probing and all other aspects of 'oscilloscoping'.

 

What I refer to as "crud" is just digitizing uncertainty for a signal with a baseline right at a value where a tiny bit of noise will determine the digitized value, so the baseline is two counts high. On some of the digital scopes I've seen, you'd have to crank the brightness on a CRT up so high it would strip the phosphor in a few minutes to get anything anywhere near as wide (high). It is very conspicuous at times and a problem that is forever insoluble, but can be made less obvious with higher resolution ADCs. ADCs get better all the time, which is why I've not followed developments in digital scopes. I just regard putting time into keeping up as a complete waste unless it particularly interests you. Just like keeping up with development in PCs and the like. My philosophy is to bring myself up to speed when purchase is imminent.

The aliasing I refer to is between the discrete values from the ADC and the discrete lines of an LCD. I assume most modern digital scopes deal with this better than they used to, doubtless in part because LCD resolution has gone up.

Some of the old digital scopes performed quite well but the design of the controls was an abomination. It came from the notion that everything should be done with hierarchical menus with an absolute minimum of knobs and buttons, so even switching the gain or timebase on some of them required much button poking and time wasting. Some of the old HP's were like that. The minimalist approach to controls certainly made scopes look a lot less intimidating than something like a Tek 7000 series with a couple of time base modules and a couple of 2-channel vertical modules. Lord did those have a lot of knobs & buttons! That problem largely went away years ago and a better balance of physical controls and menus was adopted. The more recent digitals I've used are quite pleasant to use. Having multiple memories that allow more or less instant recall of special setups is very nice. Color is useful. There are things I really like about digital scopes and things I absolutely loathe about them, and unlike many people these days, I've use both analog and digital a lot. Some of what I loathe I expect to be remedied in my lifetime, others not. Of course opening up a high performance analog scope and finding the zillion trimpots, trimmer caps and adjustable inductors for calibration is enough to drive many sensible people to go digital (not that there is a choice anymore - it's been quite some time since the last of the high performance CRTs rolled off the line).

I would certainly hope autoset is better than it used to be. It used to be OK for ordinary things, but dismal otherwise. Of course when the sales reps demo this feature, they always use ordinary things for inputs. I've seen people who always used autoset, got settings worse than they needed to be and never seemed to learn how to actually properly use the scopes' functions. They were the types who would get near my carefully set up scope and poke autoset before I could grab their finger and snap it off. Of course lots of people who use any oscilloscope fail to learn to use its capabilities (I never got very good with the Tek analog storage scope I used.) Again autoset performance is the sort of stuff that needs to be pounded into the firmware developers by people who use scopes day in and day out for a wide variety of things. It's easy to believe what you're doing is adequate if you do basically the same stuff all the time. Still, I have a hard time believing it can divine that I want to look at a 100 ns window 2 ms before the trigger even on channel 4.

Now that you mention it, I have seen the Agilent stuff on probes. Very good, as I recall. Any reasonable document on probes and probing should thoroughly cover all the fundamentals that everyone should know. Not all of them cover the more esoteric types like active probes, high voltage probes, 500 ohm probes, etc. Jim Williams of Linear Tech wrote quite a bit of good stuff on probes and probing in his ap notes on things like SMPS design. He produced a large body of excellent work with his trusty ancient Tek scopes (I believe he occasionally used an analog sampling scope and very rarely a DSO) and a scope camera. You can identify his work in old National Semi ap notes, where his name didn't appear but the characteristic odd dot pattern near the centre of the screen in his scope photos was like an artists signature on paintings.

 

What I refer to as "crud" is just digitizing uncertainty for a signal with a baseline right at a value where a tiny bit of noise will determine the digitized value, so the baseline is two counts high. On some of the digital scopes I've seen, you'd have to crank the brightness on a CRT up so high it would strip the phosphor in a few minutes to get anything anywhere near as wide (high). It is very conspicuous at times and a problem that is forever insoluble, but can be made less obvious with higher resolution ADCs. ADCs get better all the time, which is why I've not followed developments in digital scopes.

The 8bit vertical resolution most standard scopes come with isn't much of a problem for most measurement tasks (a scope isn't meant to be a precision voltmeter), and modern scopes contain facilities (like HiRes/ERES) which can increase the vertical resolution up to 12bit at the cost of some BW. And for those applications that do need more vertical resolution, 10bit and 12bit scopes are available, and have been pretty much since the end of the '80s (i.e. Philips PM3320A).

The aliasing I refer to is between the discrete values from the ADC and the discrete lines of an LCD. I assume most modern digital scopes deal with this better than they used to, doubtless in part because LCD resolution has gone up.

I understand, but this isn't much of an issue, especially not on newer DSOs with a reasonable display resolution (I know back in the old days there were some really awful early DSOs with low-res mono displays where in some instances the trace could get so thick you'd have to guess what the actual waveform was, but these days are long gone except maybe for some of the $20 DSO kits you can find on ebay).

Also, on a DSO, the actual display isn't as critical as it is on an analog scope, where "measurements" are readings taken from a screen graticule (which has its own uncertainty). I always found the term "digital oscilloscope" a bit of a misnomer really because the DSO isn't as screen focused (or better, 'how-does-the-waveform-look-like' focused) as the analog scope is, it's really about the sampled data and what information can be extracted from it so "Waveform Analyzer" (which is what a DSO really is) would have probably been a better term.

On all but the most simple DSOs you get measurements from sample data or display memory by the press of a button, and even the cheap scopes discussed in this thread can give you some basic insight into the frequency domain of your signal. Glitches and other signal issues are not found by staring at the screen until the issue appears like in the old times but by setting up triggers and (on better scopes) analysis tools which not only capture every single event the issue appeared (with time stamp) but (on better scopes) can also provide you with its parameters which tremendously helps finding the cause.

I just regard putting time into keeping up as a complete waste unless it particularly interests you. Just like keeping up with development in PCs and the like. My philosophy is to bring myself up to speed when purchase is imminent.

This, and I have to say unfortunately, is very common among EEs. As someone who buys and evaluates loads of test equipment as part of the job I can tell you that this approach doesn't work well, especially if it's about something that is a bit more complex than the simple bottom-of-the-barrel scopes we discussed in this thread. Quickly learning about the changes in T&M during the last 20+ years (not just in terms of technology, also in terms of application/techniques!) and on top of that the properties, functions and how to use them properly of even only the few scopes made by the top two in the scope market (LeCroy and Keysight) is futile. I'm not saying an EE needs to know every scope inside out but he should know the current stand of technology and how to appropriately use (at least in general terms) a DSO for a specific measurement task, including what probing to use.

Some of the old digital scopes performed quite well but the design of the controls was an abomination. It came from the notion that everything should be done with hierarchical menus with an absolute minimum of knobs and buttons, so even switching the gain or timebase on some of them required much button poking and time wasting. Some of the old HP's were like that.

Yes, at some point HP favored the single knob approach for its scopes, starting with the 54100 Series and the earlier models of the 54500 Series. I actually liked it (I was much quicker with the single knob scopes than with the traditional knob setup) but it certainly required some time and effort to getting used to it and know the menu structure, which made them unfavorable especially amongst engineers who had to jump a lot between scope models.

At the same time, Philips/Fluke (back then a team in the 'T&M Alliance') made digital scopes and combi-scopes that came with conventional controls and were pretty easy to use.

The minimalist approach to controls certainly made scopes look a lot less intimidating than something like a Tek 7000 series with a couple of time base modules and a couple of 2-channel vertical modules. Lord did those have a lot of knobs & buttons!

I remember the 7000 Series, not very positive, though. Back then we had lots of them but the plugins, while flexible, have proven very unreliable.

That problem largely went away years ago and a better balance of physical controls and menus was adopted. The more recent digitals I've used are quite pleasant to use. Having multiple memories that allow more or less instant recall of special setups is very nice. Color is useful.

Yes, modern scopes, including the cheap ones discussed in this thread, are very nice, and have many convenient features. Not having to fiddle with a Polaroid camera to get a screenshot is great

There are things I really like about digital scopes and things I absolutely loathe about them, and unlike many people these days, I've use both analog and digital a lot. Some of what I loathe I expect to be remedied in my lifetime, others not. Of course opening up a high performance analog scope and finding the zillion trimpots, trimmer caps and adjustable inductors for calibration is enough to drive many sensible people to go digital (not that there is a choice anymore - it's been quite some time since the last of the high performance CRTs rolled off the line).

I started with analog scopes back in the early '80s and used them until the very early '90s, although by then less and less frequently. Unlike my peers who usually were pretty wary of digital scopes I familiarized myself with them early on, and then used the decent ones (like the Philips PM3320A, which back then was my main workhorse, or the LeCroy 7000/9000 Series, or the HP 54510A) to my advantage. Having the waveform available as data opened up so many possibilities that even back then I'd never even consider going for an analog scope again, and you'd have to pry my DSO from my cold dead hands

There's also stuff I dislike on some modern DSOs, but in general they have become very powerful tools that can tell you a lot about your signal if used appropriately.

I would certainly hope autoset is better than it used to be. It used to be OK for ordinary things, but dismal otherwise. Of course when the sales reps demo this feature, they always use ordinary things for inputs. I've seen people who always used autoset, got settings worse than they needed to be and never seemed to learn how to actually properly use the scopes' functions.

Autoset, if not direct on point, at least brings you somewhere near where you want to be, in most cases. It's not meant as a setup tool for someone who's no idea how to setup his scope properly, and if used in that function it will disappoint with near certainty.

Of course lots of people who use any oscilloscope fail to learn to use its capabilities (I never got very good with the Tek analog storage scope I used.) Again autoset performance is the sort of stuff that needs to be pounded into the firmware developers by people who use scopes day in and day out for a wide variety of things. It's easy to believe what you're doing is adequate if you do basically the same stuff all the time. Still, I have a hard time believing it can divine that I want to look at a 100 ns window 2 ms before the trigger even on channel 4.

No, but it brings you a close enough that all you'd need to do is press the Zoom button and navigate to the point you want to see.

Now that you mention it, I have seen the Agilent stuff on probes. Very good, as I recall. Any reasonable document on probes and probing should thoroughly cover all the fundamentals that everyone should know. Not all of them cover the more esoteric types like active probes, high voltage probes, 500 ohm probes, etc.

Yes, Keysight, as Agilent and HP back then, had some really good documentation regarding scopes and probing. There's also fundamental stuff in the old HP Journals (later issues also deal with digital scopes).

Jim Williams of Linear Tech wrote quite a bit of good stuff on probes and probing in his ap notes on things like SMPS design. He produced a large body of excellent work with his trusty ancient Tek scopes (I believe he occasionally used an analog sampling scope and very rarely a DSO) and a scope camera. You can identify his work in old National Semi ap notes, where his name didn't appear but the characteristic odd dot pattern near the centre of the screen in his scope photos was like an artists signature on paintings.

Jim Williams was a great engineer and is a legend, but it shows that his world was completely analog (I'm not sure he ever used a digital scope, from what I heard he stuck with his old analog scopes 'til the end). His app notes are still valid, though, although I'd take anything that talks about probing techniques or test & measurement in general with a big grain of salt.