How about a Function Generator + Oscilloscope 2-in-1 device GW INSTEK GOS-620FG?

[matrixofdynamism]

I am looking up for oscilloscope and function generators that I shall be buying in the near future.
I came across this "GW INSTEK GOS-620FG (20 MHz) ANALOG OSCILLOSCOPE+FUNCTION GENERATOR". Apparently this device is both: An Oscilloscope & a function generator at the same time. The function generator is limited to 1MHz though. It costs about 300£.

What do you think about it? How does this compare with a buying a separate oscilloscope and function generator?

[alm]

My first impression: mediocre specs: a 20 MHz analog scope is the bare minimum. When people like Dave recommend people to start with a 20 MHz scope, this is because they can often be found for $0-50 on the used market, not to buy them retail for close to the price of a decent 50 MHz digital scope.

The 1 MHz function generator is probably analog, which is old technology, inferior to DDS. Not necessarily a bad thing if you can get it cheap, but again, ~$500 s not cheap for a scope and function generator. At tequipment.net, you could get a Rigol DS1052E plus a GW-Instek SFG-1003 3 MHz DDS function generator for $520. I realize UK prices are higher than US prices. If getting a separate scope and function gen is outside your budget, consider starting with just a scope and getting the function gen later. A scope is more important than a function gen, in my opinion.

I don't usually like combined equipment. Inevitably one of them will die or be obsolete sooner. For example, you might ditch the scope for a better one in the future, then you would be without a function generator. Either that or you're stuck with one big function generator. For the same reason, it would be harder to sell. A GW-Instek scope is not in the same league as the old Tek and HP analog scopes, so that will bring the expected resale value.

[Mechatrommer]

The function generator is limited to 1MHz? though. It costs about 300£

they put a toy func gen to make sale on 20MHz DSO. wtf! i dont know what their engineers are loafting around? f*ck them, they only think about money!

[free_electron]

ask yourself this question  :
what are you going to use a function generator for ?
what are you going to use the scope for ?

you will very quickly come to the conclusion that a function generator is something that is used only rarely , while a scope is something you need frequently to look at a signal ...

@alm: be carefull with statements as 'The 1 MHz function generator is probably analog, which is old technology, inferior to DDS'
Depends... you'll need a really good DDS to beat a simply wien bridge sinewave oscillator when it comes to THD and spectral purity. You can make a wien bridge oscillator with an opamp and a few passives ... but you will need a damn good dds with a 14 bit or more dac followed by a really good high-order filter .... it aint gonna be a 100$ DDS function generator....
If the reference clock of that DDS drifts all over theplace it aint gonna be frequency stable either... and we havent talked about jitter yet ...

DDS is one of these terms that is being used as a 'marketing ploy'. Just because something is DDS doesn't mean it is good.
The DDS engine is just a means of performing FREQUENCY control. it doesn't do anything else. It has nothing to do with the quality of the produced signal. Making a DDS engine is simple. All you need is a big accumulator ( which can be constructed out of an ADDER and a register ).
under control of a reference clock you simply keep adding a set value to the accumulator. the accumulator MSB is used to drive the clock of a toggle flipflop. the output of the flipflop gives you a weighted ratio of reference clock and the value you set. that's all there is to it. you can make it with a bunch of TTL ic's , or cram it in a CPLD if you want.

The problems start to make the analog waveform.... the linearity and number of bits in the DAC have the biggest impact. the precision of the lookup table is also important. especially at low frequencies... it'll better have a lot of numbers or you will see the steps in the table at the output. and then there is the output filter and the output amplifier. these things have bandwidth constraints. it's pretty hard to keep output voltage constant over frequency , even when the dac delivers correct voltages.

[alm]

I was not applying that DDS is superior to any analog signal source. Generating square and triangular waves, which are standard on any standard function gen, is quite hard with a Wien bridge oscillator. So is sweeping (no clue if this FG has sweep capability, probably not). Any general-purpose DDS function generator I've seen is superior in both frequency stability and spectral purity compared to any general purpose analog function gen, which work by generating a triangular wave and running it through a sine shaper to make it something that resembles a sine wave on an analog scope CRT. I guess you could build a 4-bit DDS generator with an RC oscillator as frequency reference, but I've never seen one. The standard AD DDS IC's work quite well. It wouldn't surprise me to see them for sale on eBay in the foreseeable future, though, once someone designs a low-end DDS IC.

[free_electron]

I was not applying that DDS is superior to any analog signal source. Generating square and triangular waves, which are standard on any standard function gen, is quite hard with a Wien bridge oscillator.

how so ? add a comparator driven by the sine wave to make the square .... want pwm ? make the reference voltage for the comparator adjustable with a potmeter ...
want a triangle ? inject square wave in an integrater built with 1 opamp , 1 resistor and 1 capacitor. if you employ the pwm mechanism on the square wave  you can even control the slope of the triangles now....
All you need is 3 or 4 opamp and a few resistors and capacitors.

So is sweeping

/
replace one of the resistors in the wien bridge with a j-fet. apply another varying signal there and you can sweep ...

which work by generating a triangular wave and running it through a sine shaper to make it something that resembles a sine wave on an analog scope CRT.

those are CRAP !

I guess you could build a 4-bit DDS generator

you don't seem to understand how dds works.... you need something in the order of 28 bits or more to even begin working with this mechanism...

It wouldn't surprise me to see them for sale on eBay in the foreseeable future, though, once someone designs a low-end DDS IC.

not sure what you mean here ... low end dds chips a plenty... AD9843 in 8 pin package costs a few dollars. but that chip in itself does not make a function generator....

[alm]

they put a toy func gen to make sale on 20MHz DSO. wtf! i dont know what their engineers are loafting around? f*ck them, they only think about money!

I don't think 1 MHz for a basic function generator is that bad. It easily covers the audio range, and 1 MHz can also serve as a clock signal for many micros. Perfectly fine for a beginner to do basic experiments like capacitor charging/discharging. People don't have to worry much about signal integrity at these frequencies (since the square wave are usually also quite lazy). I wouldn't complain if it were available separately for $100 or so. That scope isn't worth an extra $400, however.

how so ? add a comparator driven by the sine wave to make the square .... want pwm ? make the reference voltage for the comparator adjustable with a potmeter ...
want a triangle ? inject square wave in an integrater built with 1 opamp , 1 resistor and 1 capacitor. if you employ the pwm mechanism on the square wave  you can even control the slope of the triangles now....
All you need is 3 or 4 opamp and a few resistors and capacitors.
[...]
replace one of the resistors in the wien bridge with a j-fet. apply another varying signal there and you can sweep ...

Would this be practical across the usual 1 Hz to 1 MHz or so range of a function gen? Wouldn't you have to switch the integration capacitor/resistor? Do you have an example of a general-purpose function generator (square/triangle/sine, 1 Hz to 1 MHz, < 1V to 20V amplitude, DC offset, usually symmetry/duty cycle adjustment) that used this principle? How many of them were affordable? Would the sweep rate and amplitude stability across the frequency range be similar to common function generators?

which work by generating a triangular wave and running it through a sine shaper to make it something that resembles a sine wave on an analog scope CRT.

those are CRAP !

Do you disagree that the function gen in this scope most likely uses this principle? My impression that essentially all of the general-purpose analog function generators (including HP and Wavetek) used this principle, but I wouldn't mind to be proven wrong. Some implementations were of course less crappy than others.

not sure what you mean here ... low end dds chips a plenty... AD9843 in 8 pin package costs a few dollars. but that chip in itself does not make a function generator....

Many of these (eg. AD9834) provide sine, square and triangular waves. Somewhat similar to the MAX038. Add an output stage for amplitude and DC offset control, a micro and some controls and you essentially have a function generator. Many of the cheap ones do this.

[free_electron]

Would this be practical across the usual 1 Hz to 1 MHz or so range of a function gen? Wouldn't you have to switch the integration capacitor/resistor?

yes, so ? rotary switch to switch a resistor bank around. nothing wrong with that. they have been doing this for year ( or reed relays to switch them around )

Do you have an example of a general-purpose function generator (square/triangle/sine, 1 Hz to 1 MHz, < 1V to 20V amplitude, DC offset, usually symmetry/duty cycle adjustment) that used this principle ?

the older hp's like the 332x series (3324 / 3325) did that. i believe also the 8116, although that one may have used a sine shaper. Bk precision did it ( sold as tektronix generators)

Many of these (eg. AD9834) provide sine, square and triangular waves. Somewhat similar to the MAX038. Add an output stage for amplitude and DC offset control, a micro and some controls and you essentially have a function generator. Many of the cheap ones do this.

the problem is this filter, gainstage and output stage ... also the dc offset control is problematic. but you still can't sweep, have no slope control or anything. most of these designs come no further than the micro and AD9834 chip.. that's easy. its just digital and software... the problems start when they have to cross in the analog domain...
The price goes up very quickly at this point.

If you want to get an idea of what is involved : download the service manuals of something like an Agilent 33120 or 33250 and take a look at the filter , dc offset and output driver stage... it is -VERY- complex.. None of those fleabay generators will dare tread in that area...

As for value : you can pick up a used hp 3325 sometimes as low as 100US $ on ebay. DDS and al... 0 .. 60 MHz 40vPP with sweeping / modulation and plenty of other stuff. It even has a marker output. Ideal to use if you want to do bode plots. Hook up generator output to scope channel 1 and input of system under test. Output of system under test goes to scope channel 2. marker output goes to scope channel 3. Set scope to substract cannhel 2 from 1. and show trace 3 + math trace. the marker output is a signal that goes low when the sweep begins and goes from low to high when a preset frequency is set. so you create a 'cursor' this way. you can program this marker frequency. so now you can read gain versus frequency. the generator tells you exactly where you are looking in the sweep. Such machines are incredibly useful. If you trust the output to be constant you can get away with a dual channel scope simply look at the output of the system under test and the marker. put scope to external trigger and let the geenrator trigger the scope when the sweep starts.

I bought mine for 150... with all options installed (3325B). it'll run circles around those fleabay things.

And this 'driving microcontrollers with it ? why would you want to do that ? slap a crystal on them and off you go. Besides most micros run at 16MHz or faster these days ...

Anyway, i would not buy a scope with signal generator in it. Invest a bit more and buy a better scope. the times you need a signal generator are very limited ...

[alm]

Do you have an example of a general-purpose function generator (square/triangle/sine, 1 Hz to 1 MHz, < 1V to 20V amplitude, DC offset, usually symmetry/duty cycle adjustment) that used this principle ?

the older hp's like the 332x series (3324 / 3325) did that. i believe also the 8116, although that one may have used a sine shaper. Bk precision did it ( sold as tektronix generators)

Frac N synthesis is a Wien-bridge oscillator? They do start with a sine wave and shape this into a square and triangular wave, I stand corrected on this part, although the triangular shaper works somewhat different from an integrator, avoiding the need for switching integrating resistors. It would be yet another thing to switch, and many-pole switches or reed relays are somewhat expensive.

I bought mine for 150... with all options installed (3325B). it'll run circles around those fleabay things.

No argument here.

And this 'driving microcontrollers with it ? why would you want to do that ? slap a crystal on them and off you go. Besides most micros run at 16MHz or faster these days ...

Many people like the Arduino. Those smart engineers at Atmel thought that it was a good idea to require a working clock source for changing the clock fuses. Many beginners set these fuses to 'external clock' instead of 'external crystal'. This can be fixed by feeding it an external ~1MHz square wave. 1 MHz is plenty fast for this.

[free_electron]

Frac N synthesis is a Wien-bridge oscillator?

no it is not, but you asked for a generator that uses the sine-> square -> triangle system.
you can do the same with a wien bridge based machine. i think the older full analog machines from agilent worked that way like the 3310. i don't know, i'd have to check the schematics.
The 3325 is the 'oldest' Agilent i have in my posession. Everything else i have is newer than that ( a couple of  8116's, a 33120, a 8904 and a 8657B)

the 3325 is a very advanced machine ( even thoug it is old ) i use it mainly to do that bode plot thing. the 8904 is very useful to test differential systems as it has a floating symmetrical output.

Frac-N is the same underpinning technology used in DDS. Old technology...

Many people like the Arduino.

sigh ...

Those smart engineers at Atmel

double sigh ...
Any CPU that can be messed up that way should be recalled ...

[Kilroy]

...you asked for a generator that uses the sine-> square -> triangle system.
you can do the same with a wien bridge based machine. i think the older full analog machines from agilent worked that way like the 3310.

Yup. I have a 3310B model. Wein bridge...with a Hewlett Packard spin on it, of course.

Nice generator...really nice waves, *very* stable. Amazing really. And brilliantly executed design and build, as you would expect from that generation.

[Mark]

Take a look at the Hantek DSO8060 "5 in 1 " handheld oscilloscope. 
http://www.hantek.com.cn/english/produce_list.asp?unid=80

Five-in-one mobile laboratories: oscilloscope/arbitrary source/frequency spectrum analysis/counter/multimeter

I have the DSO1202B handheld oscilloscope, I have found the built-in DMM to be very useful.  Having a AWG built in would save me carrying another box around when I travel and reduce desk clutter. 

[codeboy2k]

The problems start to make the analog waveform.....[....]  and then there is the output filter and the output amplifier. these things have bandwidth constraints. it's pretty hard to keep output voltage constant over frequency , even when the dac delivers correct voltages.

I once thought about making a DDS function gen myself, but I never got around to actually building it, I just got some of the design started. As you said, it's really not that hard. 

I gave the output section the most thought, and did some breadboarding and simulations for it. As you said, it was extremely difficult to keep the output level constant over a wide range of frequencies, and to keep the harmonics down. One filter could not cover the entire output range, and I ended up compromising in my design, using several filter groups that would need to be switched in for different frequency bands.  And this is why the good DDS's aren't cheap.

As for maintaining consistent output levels (because of filter rolloffs, or filters which have gains in the passband, or other non-linearities) I thought about using a true-rms chip at the output and feeding that back to a gain stage before the filter. The goal was to try to maintain a programmed true-rms level at the output.  I never breadboarded this section, so I don't know if it would work, but it seems plausible.  Do you have any comments on that idea? If I get back to it, I might try it out.

[Mechatrommer]

todays opamp can maintain flatness easy up to few tens MHz. hundreds or GHzs if you want to. just slap one of those on your dac and off you go.
edit: and then there's another word... software black magic tricks.

[Kremmen]

Those smart engineers at Atmel

double sigh ...
Any CPU that can be messed up that way should be recalled ...

Without going into whether Atmel engineers are smart or not, and without wishing to start another religious war - what do you think is so broken in Atmel processors? And is it all of them or just some specific series or models?

Yes i use Atmel but then i use STM, TI and Xilinx as well. I don't need Atmel to "be the best", just curious because i seem to have missed this one.

P.S. sorry for the OT...

[free_electron]

The default mode should have been : if i am in programming mode i fall back to my internal calibrated oscillator ... so at least they can load me new code...
That is a bit of an oversight ... if you wrongly set the fuse you can't load new code... very annoying. and then you need to kludge around with feeding it an external clock ...

another problem is that the fuses are not saved in the hex file... you cannot simply compile and spit out a single hex file... you need a separate file with fuse settings. more room for mistakes... when developing you change the hexfile but forget the fusechange you needed and then spend half an hour wondering how come it doesn't work ...
Also for mass production deployment this is more cumbersome. The fusefields should have been mapped in the rom image.

[nctnico]

And Atmel lies about the operating voltages of their controllers. When they spec a part to run from 1.8V it means the brown out detector trips at 25mV below 1.8V and the internal EEPROM doesn't work reliable at that voltage as well! No more Atmel for me!
edit: to keep things clear: it was an inherited design for which someone else choose the Atmel controller.