Can someone explain a bit about function generators? They seem pricey!?

[rwgast_lowlevellogicdesin]

So I have a PC scope with an add on card which has a DAC and has options to do Sine, Square, Triangle, NTSC Signal, and Arbitrary. It does DC, and BiPolar wave forms up to 4.7v. It seems do operate decently to about 2mhz but over 1mhz I see a bit of reflection(maybe output distortion, just random spikes in some waves). The PC Scope and Generator are 200 bucks with probes.

Now that I have a better scope i have been looking at better function generators. I was surprised to see how expensive they are! Even companies like rigol, siglent, atten are pretty expensive for a 20mhz function generator, and by expensive I guess I mean more than there scopes most of the time! Im having a hard time figuring out why? Is it the type of hardware that goes in to them? I wouldnt have thought that, DDS chips arent all that expensive, and I wouldn't think its much harder to lay out a 100mhz scope then it is to lay out a 100mhz DDS based generator all the high frequency and transmission line rules apply to both.

Im also having a difficult time figuring out the cost to usefulness ratio. What do people use a function gen for? I mostly see it as a nice tool to test other tools and to inject signals in to analog circuits to verify they work correctly. Obviously if your going to inject digital signal in to a circuit you could just use a micro. Sweeping seems nice for verifying a circuit works well across a wide range of frequency's, but i think its more of a gimmick than anything else unless your doing RF. Im sorry if that statement may have come off as ignorant, I understand there are probably some special cases where im dead wrong about sweeping, so i guess im just talking about more of an in general use than a niche thing.

So whats the difference between something like this
http://www.ebay.com/itm/10Mhz-DDS-Function-Arbitrary-Waveform-Signal-Generator-Waveform-Editor-Softwar-/111624982011?hash=item19fd5e0dfb

and this
http://www.ebay.com/itm/Hantek-HDG1012A-Arbitrary-Function-Waveform-Generator-40uHz-10MHz-/150982364714?hash=item2327400e2a

I mean how much more are you getting for that extra 200 hundred dollars? I guess in my mind I would think a 50-100mhz gen, that comes from a reputable company would cost about 300 to 400 bucks, not a 10-20mhz version!

Im kind of asking this becuase I always though it would be cool to shove a DDS chip in a fat dry erase marker, with a conductive tip. It would be neat to have a pen style probe that you could inject signals with. I always thought it would be pretty cheap and easy to build one up to about 30mhz looking at the DDS chips out there. Im starting to think im wrong and there is more to all this than I thought. I was under the impression a TCXO a DDS chip a DAC and a Micro were all it really took, im not sure how arbitrary waves or generated but that also wasnt on the list for this project although sweeping was.

[Noise Floor]

Well, they can be pricey because generating high frequency signals reliably is not something that has yet to make its way into cheap parts.  Don't get me wrong, there are cheap ways to generate signals, but its all about the tolerance you can have for signal quality. 

Are you a hobbyist or working in an industry?  The answer may affect what you can use. 
You want to see expensive, look at RF signal generators.

[German_EE]

I share your curiosity when it comes to this particular piece of test equipment. Couple a decent microprocessor or FPGA to a DAC and add a bit of memory to store the software and you should be 75% of the way there, all that's left is the front panel and the output amplifier. If my programming skills were any good I'd attempt the project myself .

[rwgast_lowlevellogicdesin]

Well at the moment I am not working, Im an industrial sparky but Im switching my field in to CELL base stations, if I cant pick up a few technician jobs I applyed for but in either case my job would supply the tools I need. I do consultation on the side some times but I have never needed a function gen better than the one I have for that.

Im more of a hobbyist in to building test equipment and Software Defined Radio, along with a multi year long robot project whos point is to do slam and chase things around . If i take this job doing LTE I will be flying around the country for 6 weeks straight with 10 days off. My plan is to take a portable lab and use all that hotel time to develop a Kit and hopefully bootstrap it on tindie. So you could say im aspiring to work for myself in some ways. Right now all my kit plans are for precision type instruments and dont really need to much frequency generation, except for some impeadence testing. One thing I want to do is make a decent Z/LCR/ESR meter down the road, Im not real sure how high frequency I will need but considering id like to be able to measure cable impeadence including coax I would think I would need at least 100mhz before coax started resisting frequency? Im not sure its these kinds of things I find myself needing a better (f)gen for, things require a bit of experimentation though. Kind of a catch 22 since I cant experiment without a better function gen. So for my 2mhz unit has served me well in the mostly digital or simple analog realm.

[eas]

I don't understand why people think that all products are or should be commodities and priced as such, with a small margin on top of the cost of materials and assembly, particularly for something fairly specialized and industrial like test equipment.

At the very least, there are fixed costs for design and validation that have to be recovered, and overhead costs like sales, marketing and distribution. Those costs are spread over the total volume sold, and the smaller the market, the more those costs contribute to the selling price. Don't forget the profit margin too. If the payoff on initial and ongoing capital investment isn't sufficient, the company kills the product, or sells it off, or never lets it past feasibility studies in the first place.

Of course, what really matters, or should, is how valuable the product is to the customer. In a business context, people will happily pay more for things that save them time, or money, or reduce risk. If the fully loaded value of your time is, say, $150/hr, then how much sense does it make spending 20+ hours selecting a DSS module and supporting components, integrating it with an MCU, PSU and display, writing some control software, testing and validating it for your own use?

So, in the case of function generators, I think the first thing to consider is that you aren't the only person who isn't sure why they need one. I'm not expert, but from what I can tell, the market for function generators is smaller than that for oscilloscopes.

Absolutely nothing wrong with diving into projects in order to learn something new though. I've got a bunch of used test equipment that I plan to sell off when I'm done working on it. I'm definitely going to be loosing money on it if I value my time at all, but the truth is, even I sell the stuff for less than what I paid for it, even after fixing it up, I won't feel bad, because I'm learning a lot.

[tautech]

Now that I have a better scope i have been looking at better function generators. I was surprised to see how expensive they are! Even companies like rigol, siglent, atten are pretty expensive for a 20mhz function generator, and by expensive I guess I mean more than there scopes most of the time! Im having a hard time figuring out why?

Garanteed Specs
Cal certs
Risetimes
Warranty

They all cost real money to produce.

If your needs are basic, then stick with basic gear.

Have you checked out the SDG800 series?

[daqq]

Im also having a difficult time figuring out the cost to usefulness ratio. What do people use a function gen for? I mostly see it as a nice tool to test other tools and to inject signals in to analog circuits to verify they work correctly. Obviously if your going to inject digital signal in to a circuit you could just use a micro. Sweeping seems nice for verifying a circuit works well across a wide range of frequency's, but i think its more of a gimmick than anything else unless your doing RF. Im sorry if that statement may have come off as ignorant, I understand there are probably some special cases where im dead wrong about sweeping, so i guess im just talking about more of an in general use than a niche thing.

RF signal generators are their own special category, so lets ignore them.

As with all test equipment a signal generator is nothing that you couldn't do without, but more slowly, less practically etc. My most recent use of a signal generator (yesterday) was playing around with a MOSFET in a nasty switching application - I needed to try out various gate times, switching frequencies and other stuff. I could have slapped together something like a small micro, add to that a gate driver, add to that some protections (nasty kickback), add to that some filtering, add to that some control... and after a few hours/days I would have ended up with a dumb single purpose square wave signal generator, hacked together, barely holding shape. Thanks to a signal generator, I just fiddled with a few knobs, found the right pulse width, frequency and then my transistor exploded.

Arbitrary function generators are even more useful. Imagine you have recorded a one time event from a sensor. To test your electronics you want to feed the signal into it, preferably repeatedly. You have to recreate it. You can either do this with the actual sensor (which might get difficult) or with a function generator and a recording, which is comfortable and infinitely repeatable.

As to using a micro to inject digital signals - depends. You can spend some time coding a 10MHz generator, or you can press a button and be done with it. At 20MHz, well, not all GPIOs can cut it. 100MHz?

I was under the impression a TCXO a DDS chip a DAC and a Micro were all it really took, im not sure how arbitrary waves or generated but that also wasnt on the list for this project although sweeping was.

You also need: A reasonable power supply, output filtering, output buffering, output protection (you assume that test equipment will be abused to a degree), a fair bit of analog stuff, user interface (display, buttons), a PCB

I mean how much more are you getting for that extra 200 hundred dollars? I guess in my mind I would think a 50-100mhz gen, that comes from a reputable company would cost about 300 to 400 bucks, not a 10-20mhz version!

At a very quick glance: Sturdy box, two channels, double the bandwidth, proper, fast, friendly interface, a display not meant for ants, more than double the sample rate, a frequency coutner. There's probably more.

Is it the type of hardware that goes in to them?

Not at all* - it is the fact that a lot of engineers sat behind a desk to make a nice tool and wish to feed their kids, with the profits. It's called a free market. If you think you can make the same/better for less, go for it.

* - in the more high end stuff you really do get exotic stuff.

[Matje]

I share your curiosity when it comes to this particular piece of test equipment. Couple a decent microprocessor or FPGA to a DAC and add a bit of memory to store the software and you should be 75% of the way there, all that's left is the front panel and the output amplifier. If my programming skills were any good I'd attempt the project myself .

One would wish for this to be so. It is not.

After the digital part (which is reasonably simple, but e.g. may need to employ special algorithms to avoid spurs in the output as much as is practical) comes a quite difficult analog part.

First you need a decent low-pass filter, degree 7 or 9 or so, or you will suffer bad crap on the signal (Nyquist). Then the stuff necessary to add a configurable DC offset. Then the output amplifier, not really trivial for the usual amplitude requirements, with configurable amplification, low noise, low distortion, the works.

And all this analog stuff has to be controllable from the processor while being reasonably stable against temperature changes, aging and so on. Doesn't sound that easy.

[nctnico]

In another thread the Hantek is not a recommended piece of equipment. I think you can't go wrong with a cheap DDS signal generator (not the one from the link above but there are others which are just function generators) and for most measurements it should do. Most low end AWG function generators have such a short memory that their use as an AWG is severely limited. These kind of generators also need more sophisticated software so the cheap ones are likely to have issues (the Hantek ones being particulary bad and the Siglent ones reasonable to OK).

@Matje: even then the hardware of an AWG is the easy part. Writing the firmware will be a much larger project than the hardware.

[rwgast_lowlevellogicdesin]

The recommended siglent looks to be about $100 per 10mhz. That is comparable to all the other big Chinese manufactures out there, so yes its nice but seems quite expensive for the bandwidth to me.

I wasnt complaining so much as wondering if there was more to a Signal Gen than some digital control and a DDS, along with some filtering. I was mostly amazed at the price becuase when compared to scopes from the same manufactures you get much less bandwidth for the price which is what threw me off. But I guess there are more people wanting/needing scopes than there are people who need a sig gen and supply and demand is probably the answer to my question. It seems to me all the basics are the same They both have big screens nice cases a clean psu etc... so the only hardware difference is the scope has a sampling front end while a Signal Generator has an a front end meant to output clean stable wave forms, in my mind the cost of the scopes analog pieces would probably be much more than a signal generators analog front end.

[rwgast_lowlevellogicdesin]

BTW if I were to buy a cheap one it would probably be this

http://www.ebay.com/itm/24Mhz-Dual-ch-DDS-Function-Arbitrary-Waveform-Signal-Generator-sweep-Software-/121478483178?hash=item1c48ae8cea

They have some issues but some guys have already taken these apart and shined them up so they put out quality stable signals, the reason I linked to the one I did is because it had a nice display like the nice ones from big companies

[Matje]

@Matje: even then the hardware of an AWG is the easy part. Writing the firmware will be a much larger project than the hardware.

I don't think so. But then I write software all day long, it's what I get paid for. ;-)

[Matje]

The recommended siglent looks to be about $100 per 10mhz. That is comparable to all the other big Chinese manufactures out there, so yes its nice but seems quite expensive for the bandwidth to me.

I wasnt complaining so much as wondering if there was more to a Signal Gen than some digital control and a DDS, along with some filtering. I was mostly amazed at the price becuase when compared to scopes from the same manufactures you get much less bandwidth for the price which is what threw me off. But I guess there are more people wanting/needing scopes than there are people who need a sig gen and supply and demand is probably the answer to my question.

Hmm, maybe.

It seems to me all the basics are the same They both have big screens nice cases a clean psu etc... so the only hardware difference is the scope has a sampling front end while a Signal Generator has an a front end meant to output clean stable wave forms, in my mind the cost of the scopes analog pieces would probably be much more than a signal generators analog front end.

I think the difference is in the control one has about what is being connected.

A scope has an analog input, 1 Meg || some pF, whatever voltage appears there needs to be processed/displayed. IOW anything behind the input BNC (in terms of signal flow) is controlled by the device manufacturer and thus relatively easy.

For a siggen there is no telling what may be connected to the output BNC. Yet the siggen must do it's best to produce a reasonable signal, must not be destroyed by extremes, must not be disturbed overly much by unusual stuff on it's output - especially for multi-channel siggens, any coupling between channels would be bad. And it must be able to output a little bit of power, maybe several 100 mW into 50 Ohms. Some people call that a power amplifier/output stage and demand a license for suitably high frequencies, so don't connect an antenna to it...

Think of the difference between a DMM and a lab power supply. The PSU has more conditions to fulfill, so tends to be more expensive for a similar level of "goodness" (yes, I know, not objectively comparable really).

[nbritton]

I was wondering why they were so expensive myself. You would think you could wrap a USB interface around a DAC and call it done, but apparently not. I think one reason they are so expensive is not a lot of people want them, it's an issue with economies of scales.

[Bud]

... so the only hardware difference is the scope has a sampling front end while a Signal Generator has an a front end meant to output clean stable wave forms, in my mind the cost of the scopes analog pieces would probably be much more than a signal generators analog front end.

For a scope application you can get away with a sampling frequency x3 of the scope bandwidth, i.e. 1GSa/S for a 300MHz bandwidth. For a waveform generator you need many more samples -> much higher sampling frequency. In other words, for a given sampling frequency a waveform generator will have much lower bandwidth. You cannot create a quality arbitrary waveform with just 3 samples, can you? You can do a sinewave or square wave (sinewave squared) but not a triangle let alone an arbitrary one. If you check any arbitrary waveform gen specification, chances are you will find they are spec'ed to generate sinewave/squarewave of much higher frequency than triange/arbitrary waveform. Even for a 10MHz triangle you would need a GHz DDS. You can check a DDS datasheet such as AD9910 to understand why.

[Joule Thief]

To your original question,

I use audio frequency range function generators in the repair and calibration of automotive tachometers (square wave) and in hobby level design of stereo speaker crossover networks (sine wave). Neither application requires top shelf, high end equipment.

in addition, I just re-coned a pair of JBL woofers and "broke in" the new suspensions by running low frequency sine waves thru the stereo amplifier into the woofers. Large excursions of the cone at low frequency will check for any voice coil / suspension interference on the finished product.

[Muxr]

I have one of those HMS-5200 cheap ebay sig gens, and I can tell you even though it says it can generate 6Mhz square wave, it can only really generate an ok looking squarewave at 4Mhz. It has all sorts of other issues, the UI is pretty bad, but it's good enough for injecting signals for troubleshooting circuits, if for whatever reason it gets fried it's not a big loss and it might be easier to repair.

Devtty0 (user on here I think) has some videos describing some issues with it and correcting some of its design shortcomings:



My SDG 1025 is in a different league. And even though it's not exactly a pro level sig gen it definitely offers UI and performance worth a premium over the HMS-5200. You definitely get what you pay for, at least in my experience.

[Wuerstchenhund]

I think the difference is in the control one has about what is being connected.

A scope has an analog input, 1 Meg || some pF, whatever voltage appears there needs to be processed/displayed. IOW anything behind the input BNC (in terms of signal flow) is controlled by the device manufacturer and thus relatively easy.

For a siggen there is no telling what may be connected to the output BNC. Yet the siggen must do it's best to produce a reasonable signal, must not be destroyed by extremes, must not be disturbed overly much by unusual stuff on it's output - especially for multi-channel siggens, any coupling between channels would be bad. And it must be able to output a little bit of power, maybe several 100 mW into 50 Ohms.

Not really. There's not a much difference between a simple digital scope and an AWG in terms of complexity, really. And it's not that an AWG takes what happens past its BNC into consideration, all it does is to produce an output calibrated at 50ohms or 1Mohms and that's it. It doesn't adjust for impedance mismatch, or any other disturbance that might occur, and if the output is shorted then the AWG might get damaged as easily as if the output is exposed to excessive voltages, unless the AWG has protection in place (not all do).

For a scope application you can get away with a sampling frequency x3 of the scope bandwidth, i.e. 1GSa/S for a 300MHz bandwidth. For a waveform generator you need many more samples -> much higher sampling frequency.

No, you don't. You make quite a few misconceptions here. First, the best thing about the sampling theorem is that it works exactly the same both ways, i.e. the amount of samples/s a scope needs to accurately reproduce a waveform is exactly the same as for an AWG to produce the same waveform with the same accuracy.

Second, the rated analog bandwidth of a scope refers to a pure sine wave. That means a 300MHz scope can still "see" a 300MHz sine wave (with the displayed amplitude 3dB lower than the actual amplitude). If you feed the same scope a 300MHz square wave all you'll see is crap as most signal contents that make out the 300MHz square wave will be invisible to the scope (and the lower harmonics will be at visible at various stages of attenuation by the scope's front end). So that means a 300MHz scope which is spec'd for 300MHz for sine waves only and which comes with a sample rate of 1GSa/s is perfectly fine, as 3 points per cycle are enough to replicate a sine wave.

In other words, for a given sampling frequency a waveform generator will have much lower bandwidth.

Not necessarily, there are many AWGs (mostly low-end) which have the same or higher bandwidth to sample rate ratio than a scope, for example Rigol DG1062z (60MHz 200MSa/s = 0.3/Sa), Siglent SDG1050 (50MHz 125Msa/s = 0.4/Sa), Siglent SDG2122X (120MHz 1.2GSa/s which really is just upscaled 300MSa/s = 0.1/Sa datasheet or 0.4/Sa real). Compare this to scopes like the Rigol DS1104z (100MHz 1Gsa/s = 0.1/Sa), Siglent SDS1152CML (150MHz 1GSa/s = 0.15/Sa), Siglent SDS1302X (300MHz 1GSa/s = 0.3/Sa), Siglent SDS2304 and Rigol DS2302A (300Mhz 2GSa/s = 0.15/Sa).

You cannot create a quality arbitrary waveform with just 3 samples, can you?

That depends on the waveform, but in general, no, 3 samples per cycle would be a bit low.

You can do a sinewave or square wave (sinewave squared) but not a triangle let alone an arbitrary one. If you check any arbitrary waveform gen specification, chances are you will find they are spec'ed to generate sinewave/squarewave of much higher frequency than triange/arbitrary waveform.

Yes, especially in the low end class. But the main reason for that is that, as with scopes, the AWGs bandwidth is usually defined for sine waves only (which again comes back to what I said above that the sample rate requirements for scope and AWG are the same). All composite waveforms (i.e. waveforms that consist of two or multiple sine waves, like square, pulse, triangle, whatever) naturally follow this limitation, which means that if your sine wave limit is say 100MHz then the max fundamental of any composite waveform must be much lower to produce an accurate waveform. But again, that is the same with scopes.

Some AWGs (especially low end ones) offer lower sample rates in arbitrary mode than in any of the fixed modes, but that is merely a limitation of their specific design. On AWGs which offer the same sample rate as in fixed modes the same bandwidth limit as for the fixed modes applies.

Back to the original question why function generators are expensive, well, they aren't. Function generators (which many people mix up with AWGs) are pretty cheap these days but aside from the entry level have mostly died out in favor of Arbitrary Waveform Generators (AWGs). Low end AWGs from the typical B-brands (Siglent, Rigol) are roughly as expensive as their scope, which is understandable considering the similar complexity. However, big brand AWGs like the ones from Keysight are indeed pretty expensive, and (aside from the benefits of a big brand like quality, actual support and such things) is mostly due to their proprietary technology (i.e. Keysight's Trueform) that makes composite signals like square waves look less like crap as they do on the cheap AWGs which employ pure DDS.

[-jeffB]

For a scope application you can get away with a sampling frequency x3 of the scope bandwidth, i.e. 1GSa/S for a 300MHz bandwidth. For a waveform generator you need many more samples -> much higher sampling frequency. In other words, for a given sampling frequency a waveform generator will have much lower bandwidth.

It sounds like you're confounding fundamental frequency and bandwidth. Generating or displaying a convincing 300MHz triangle wave will take a good bit more than 1GHz bandwidth.

With a 1GHz sample rate, sure, you could generate a 300MHz signal -- but square waves, sawtooths, and triangles would all look like more or less wobbly sine waves. (Or, if unfiltered, like various Lego pieces.)

[Bud]

There was no misconception or confounding, it was your interpretation of what I wrote. At the end of the day we all saying same thing that such generators (talking DDS based ) produce sinewave of higher frequency than other waveforms. This is a no brainer, sinewave is what DDS is for, with a good output filter one can get a usable sinewave of 0.45 f_clock. This is 450MHz for a 1GSPS DDS. But to get a triangle wave you need to build it using incremental steps and ,say , for AD9910 a step size in this mode is f_clock/4, so it takes you down from 1GSPS to 0.25GSPS right  to start with, for 1 time step (or sample if you wish). Now, lets say we use 20 such steps to synthesize a triangle wave, 10 to ramp in and 10 to ramp out. that gets us 250/20=12.5MHz of the synthesized fundamental frequency, So there it is, 12.5MHz of a triangle wave is all we can get from a 1GSPS DDS compare to 450MHz sinewave.

For a squarewave we just take the filtered sinewave and put it through a high speed comparator or squarer, and we have squarewave of the same high fundamental as the sinewave (in general no need to route the resulting squarewave through a filter). Looking at what is available on the low end  market it appears they even do not bother doing this as their squarewave is often spec'ed for much lower frequency limit than sinewave. Perhaps they do it by bruteforce synthesis same way as other arbitrary waveforms and not by squaring the sinewave. But there may be other limitations in the technology used, i.e. incapable output buffers, simplified architecture etc.

[rf-loop]

For a squarewave we just take the filtered sinewave and put it through a high speed comparator or squarer, and we have squarewave of the same high fundamental as the sinewave (in general no need to route the resulting squarewave through a filter). Looking at what is available on the low end  market it appears they even do not bother doing this as their squarewave is often spec'ed for much lower frequency limit than sinewave. Perhaps they do it by bruteforce synthesis same way as other arbitrary waveforms and not by squaring the sinewave. But there may be other limitations in the technology used, i.e. incapable output buffers, simplified architecture etc.

Example in Siglent SDG1000 series use comparator method for sguare/rectangle wave.  Also it can do it using pulse function but more rejected frequency.

This serie is specified up to 50MHz sinewave and up to 25MHz square wave (derived from internal sinewave)

Comparator method have its advantages but also disadvantages what can understand easy if know even Basic Fundamentals. Specially with low frequencies, timing jitter is this method one disadvantage.

You wonder why square or other not sinewaves are frequency rejected.

Please before start wondering Basic things take care you know what are waveforms (other than sine) and how they are sum of  different frequencies and levels of sinewaves.

I try very simplified...

If you think example 10MHz pure square. It is sum of 10, 30, 50, 70, 90 etc sinewaves depending its risetime...  every harmonic have its own level what is less than first harmonic what give some help for amplifier BW. With not infinite rise and fall time it make game more easy for end amplifier but still, it need much more than sinewave.

Amplifier what is good for "up to" 50MHz sinewave  is not good at all for 50MHz square depending what risetime we want, not even for 25MHz, not even foor 10MHz if we want fast risetime.  Think risetime and harmonics and then wonder agen why Square (and other waveforms) are mostly specified under max sinewave.

If we want 0.1Hz (!)  square and if we want 1ns risetime can you imagine that it is bit difficult to make cheap and easy amplifier for it. Even if we want <10ns risetime it still need linear amplifier from DC to over 50MHz example for 10 or 20Vpp. Yes, for this 0.1Hz square wave you need least DC to 50MHz linear amplifier. (of course these are not difficult but there need also think if this circuit  can cost under 1 dollar or very expensive 5 dollar.  With high money all is simple, diffulties start when price need reduce to as near zero as possible and still keep specs.)

[Ben321]

It appears to have a lot to do with what country they are manufactured in. If they come from the USA, Europe, or Japan, a decent one will cost a few hundred dollars, and an arbitrary waveform function generator will cost over $1000. However, there's this generic (no company name displayed anywhere on the case of the device) function generator that's made in China, that's less than $100 on eBay, and it appears to have a significant number of features, including multiple waveforms, and frequencies that go up to 25MHz for sinewave (and 6MHz for all other waveforms). It also supports completely arbitrary waveform generation if you have a file with the data to generate a particular waveform (it has a USB port on the back to allow it to be connected to a PC for uploading waveform data). It uses the DDS (direct digital synthesis) technique to actually generate the signal. Basically it has a microcontroller with pre-programmed waveforms (and ability to add new ones by uploading them from a PC), and a DAC that converts the digital output from the microcontroller to an actual analog signal. This makes it much easier to output a perfect sinewave (something that's very difficult with a purely analog function generator, and probably accounts for 90% of the cost of most analog function generators). It features 2 output BNC connectors, and also one input BNC connector for signal modulation (AM or FM). If this was manufactured in any country other than China, I'd expect it to probably cost about $1500.

If you want to check it out, here's the link to it http://www.ebay.com/itm/121478483178

[tautech]

If you want to check it out, here's the link to it http://www.ebay.com/itm/121478483178

Rise and fall time of square wave ?100ns

[Wuerstchenhund]

It appears to have a lot to do with what country they are manufactured in. If they come from the USA, Europe, or Japan, a decent one will cost a few hundred dollars, and an arbitrary waveform function generator will cost over $1000. However, there's this generic (no company name displayed anywhere on the case of the device) function generator that's made in China, that's less than $100 on eBay, and it appears to have a significant number of features, including multiple waveforms, and frequencies that go up to 25MHz for sinewave (and 6MHz for all other waveforms). It also supports completely arbitrary waveform generation if you have a file with the data to generate a particular waveform (it has a USB port on the back to allow it to be connected to a PC for uploading waveform data).

If you really believe that this $100 thing is as good as a big brand AWG manufactured in the US, Europe or Japan and only so cheap because it lacks the brand label and is made in China then you're taken for a fool.

The simple reason why instruments produced in US, Europe or Japan are more expensive is because they usually are professional lab-grade instruments and not crap toys which barely deliver on the alöready low promises they make. That $100 AWG on ebay you referred to has truly horrible specs (i.e. 10ppm frequency "accuracy", 2048pts sample memory  ), which in reality in doesn't even comply with. The sample memory is so small that it's useless, and most likely the only reason for its existence is that this thing can be sold as "Arbitrary Waveform Generator".  It's probably OK-ish if all you need is a device to create some LF sine waves but as an AWG it's a pretty poor joke.

The other thing you're ignoring completely is that not everything coming out of China is actually cheap. The often recommended AWGs made by Siglent and Rigol are noticably more expensive than your poor ebay example, but then they offer vastly better specs which make them pretty decent entry level AWGs. But again, the reason why these are still cheaper than big brand AWGs is simply because the latter have better signal quality, tighter specs, much better software tools (the software of all Chinese B-brand AWGs I've seen is shit, plain and simple), much more stable firmware, support that actually deserves that name, better quality control, longer support cycles and so on.

The country of manufacture is actually pretty irrelevant, even big brands produce some stuff in China and it's still pretty expensive, but then the level of quality of a product has actually much more to do with its design, the manufacturing processes and the presence of appropriate quality controls and a lot less with the place of manufacture.

This makes it much easier to output a perfect sinewave (something that's very difficult with a purely analog function generator, and probably accounts for 90% of the cost of most analog function generators).

Whatever this thing outputs will be far away from a perfect sine wave because a perfect sine wave is a mathematical construct that doesn't occur in nature. A good signal generator comes very close to a perfect sine wave, but to do so a lot of effort needs to go in the analog front end. Your $100 toy will probably produce something that resembles a sine wave close enough to be recognizable on a scope as sine, though.

[Ben321]

It appears to have a lot to do with what country they are manufactured in. If they come from the USA, Europe, or Japan, a decent one will cost a few hundred dollars, and an arbitrary waveform function generator will cost over $1000. However, there's this generic (no company name displayed anywhere on the case of the device) function generator that's made in China, that's less than $100 on eBay, and it appears to have a significant number of features, including multiple waveforms, and frequencies that go up to 25MHz for sinewave (and 6MHz for all other waveforms). It also supports completely arbitrary waveform generation if you have a file with the data to generate a particular waveform (it has a USB port on the back to allow it to be connected to a PC for uploading waveform data).

If you really believe that this $100 thing is as good as a big brand AWG manufactured in the US, Europe or Japan and only so cheap because it lacks the brand label and is made in China then you're taken for a fool.

The simple reason why instruments produced in US, Europe or Japan are more expensive is because they usually are professional lab-grade instruments and not crap toys which barely deliver on the alöready low promises they make. That $100 AWG on ebay you referred to has truly horrible specs (i.e. 10ppm frequency "accuracy", 2048pts sample memory  ), which in reality in doesn't even comply with. The sample memory is so small that it's useless, and most likely the only reason for its existence is that this thing can be sold as "Arbitrary Waveform Generator".  It's probably OK-ish if all you need is a device to create some LF sine waves but as an AWG it's a pretty poor joke.

The other thing you're ignoring completely is that not everything coming out of China is actually cheap. The often recommended AWGs made by Siglent and Rigol are noticably more expensive than your poor ebay example, but then they offer vastly better specs which make them pretty decent entry level AWGs. But again, the reason why these are still cheaper than big brand AWGs is simply because the latter have better signal quality, tighter specs, much better software tools (the software of all Chinese B-brand AWGs I've seen is shit, plain and simple), much more stable firmware, support that actually deserves that name, better quality control, longer support cycles and so on.

The country of manufacture is actually pretty irrelevant, even big brands produce some stuff in China and it's still pretty expensive, but then the level of quality of a product has actually much more to do with its design, the manufacturing processes and the presence of appropriate quality controls and a lot less with the place of manufacture.

This makes it much easier to output a perfect sinewave (something that's very difficult with a purely analog function generator, and probably accounts for 90% of the cost of most analog function generators).

Whatever this thing outputs will be far away from a perfect sine wave because a perfect sine wave is a mathematical construct that doesn't occur in nature. A good signal generator comes very close to a perfect sine wave, but to do so a lot of effort needs to go in the analog front end. Your $100 toy will probably produce something that resembles a sine wave close enough to be recognizable on a scope as sine, though.

These things are made in (I assume) the stereotypical "Chinese sweatshops" and the employees get VERY low wages. That cuts down on the vast majority of the price that goes into manufacturing these items. I'm sure the total price of the components in an AWG is not $500 worth, and that the high price of most of these brand-name devices comes from the labor used to make them (in the "parts and labor" calculation of the price). Items made entirely in China (because of their very low wages) cost only a tiny fraction of those that are either made entirely outside of China, or items that are made mostly outside of China (but with a few parts being made in China).

A product that is sold is priced according to 3 factors. These are parts, labor, and profit. Lets say the price of all the parts in an AWG is $50, and the price that it costs employees to make it (at a reasonable wage) is $500, and the company is trying to make a $100 profit on each item. They will sell it for $650 ($50+$500+$100). But now in China, they probably are going to only charge the employees $10 for each device manufactured, and be willing to be content with a lower profit, of maybe only $10 per unit. That means that for a similar product made in China it's going to be sold for only $70 ($50+$10+$10). That's why I think stuff made in China is so cheap.


And by the way, what do you consider to be reasonable frequency accuracy, if 10ppm isn't? And what do you consider to be reasonable sample memory, if 2048samples for a waveform isn't reasonable?