[solved!] Function generator, mainly for audio applications (D/EU)

[frozenfrogz]

Dear all,

lately both of my very old function generators died (Hungarian made EMG TR-0463) - due to old caps I guess, but still need to look into this.
I always wanted to get an upgrade, but since the old tools were good enough, I never really looked into that. Now, I am in need of some new gear. I read a bunch of threads on various devices but would love to get some recommendations by you.
Mainly, the FG is used for audio applications, generating input signals for troubleshooting amplifiers and for building guitar effects pedals.

I would like to buy trusty used gear if possible. Chinese (FY6600, UTG-9005C,..) might be an option if there are reliable makes. A rather compact device would be nice but not strictly mandatory. I am not keen on buying a boat anchor though.

Apart from audio applications, I am tinkering with Arduino and Raspberry Pi based stuff. I am also owning a Rigol DS-1054Z that satisfies a lot of my measuring needs.

What should I look for? Hameg, Philips, RS,...? Any specific models that you would recommend?
Also: I guess that somewhere, someone already answered this question, but I could not find it on the forums. In case there is a thread that has all the answers, please let me know.

Thank you and kind regards!

Edit: I would like to stay under 200€ if possible.

TLDR: I went with the Rigol DG812 and am very happy with that choice.

[HighVoltage]

If you look for old school audio generators, look for "LEADER" made in Japan.
They used to have some of the best audio generators and from time to time
they can be found on ebay DE.

[2N3055]

Take a look at FY6800. For 100 USD it will give you more than you need.
For your use more than enough.

[nctnico]

Take a look at FY6800. For 100 USD it will give you more than you need.
For your use more than enough.

I agree. The old RC generators are very unstable compared to the modern DDS based ones.

[radiolistener]

FY6800 and FY6600 unfortunately has too high jitter. That's the real issue for this gen, with no jitter it will be the best gen for the price

[2N3055]

FY6800 and FY6600 unfortunately has too high jitter. That's the real issue for this gen, with no jitter it will be the best gen for the price

For his intended use not an issue.

[0culus]

The function gen I ended up picking up to cover the audio work I want to do was an HP 3325B. Sure, it's older and not fancy but it does linear/log sweeps, DC offset, phase angle, sine to 20 MHz, square wave to 10 MHz as well as sawtooth and ramp to 10 KHz. Mine has Option 001 (OCXO timebase) and Option 002 (high voltage, up to 40 Vpp from 0-1 MHz). It can also accept external modulation. If you get one with high voltage do be mindful of 50 scope frontends with it.

[frozenfrogz]

Thank you for your input so far!

Currently there are a couple of options I compiled from the small ads:

Hameg HM8030-4 90€
Thurlby TG 550 230€
Rohde & Schwarz 336.3019.02 150€
Stanford Research DS345 190€

I am thinking the DS345 might be worth getting, but I am thankful for your opinion.
One of the above or wait and hunt for an HP / Agilent?

[Zenith]

You need to define what you want this for, apart from AF. For instance, if it was for accurate distortion or level measurements I'd go for a low distortion audio oscillator; AWA, Marconi, HP, Tek, others. So look at one of those, or the modern equivalent.

I've never seen function generators as more than a jack of all trades but master of none.  Very convenient though. I'm sure there are megabucks versions which can wrestle an AWA G232  into submission on the noise floor and put trad RF signal generators in their place, but that's NASA stuff, so for the purpose here, irrelevant

If you want an analogue function generator, in my experience, go HP as Oculus suggested.

As to Stanford Research, very impressive, but if it goes wrong you have a hi tech door stop, as I've seen it.

IMHO you wouldn't be disappointed if you bought an FY6800.

My view of buying older test gear is get the service manual first, then consider buying a unit, unless you find one for virtually nothing.

[Calvin]

Hi,

for audio a modern soundcard and some piece of free or lowcost software will do a better job than any of the afore mentioned dedicated generators. The 24Bit resolution allows for higher dynamics and purity of the signals than the best standalone function generators (16Bit max.)
The most prominent limit of soundcards beeing the bandwidth restricted to typically 100kHz with common 24Bit/192kHz cards.
Due to their inputs soundcards offer at the same the advantage of full analyzer functionality.
With a little effort You can achieve a absolutely professional level of measurement quality, very close to a RTX or AudioPrecision or R&S.
If You prefer a dedicated all-in-one solution look for a QuantAsylum QA401.
It is a full fledged Audio Analyzer USB device, featuring 2channels and differential In- and Outputs (BNC connectors) for ~400€. As ADC it utilizes the AK5397, that still is the best ADC for this application. Its software suite is specialised on audio useage and comfortably easy to handle.
It includes of course typical audio measurements like amplitude response, noise, THD, weighing filters, etc.
The support is responsive and most competent and there's a large user community over at the diy-audio forum.

If You need more bandwidth a Digilent AnalogDiscovery2 offers complete Analyzer functionality at 14Bit and up to ~10MHz.

As 'classical' function generator I can recommend the Siglent SDG2042X, a 2-channel 16Bit arbitrary funtion generator. It can provide for Signal levels as low as 2mV for testing Phono input stages. Sine signals are very pure and it can generate low frequency rectangles with rise times of just a few ns.
At ~500€ it offers almost identical functionality to Keysight's generator (~3.000€).
For audio only it is still beaten by the QA401.

regards
Calvin

[Zenith]

Hi,

for audio a modern soundcard and some piece of free or lowcost software will do a better job than any of the afore mentioned dedicated generators. The 24Bit resolution allows for higher dynamics and purity of the signals than the best standalone function generators (16Bit max.)

I don't like this idea. There are all sorts of theoretical and practical objections as well as philosophical and moral ones. The main reason I don't like it is that I like to think in terms of dedicated instruments and preferably a load of old junk. 

Seriously, from frozenfrogz' point of view it has to be an option worth looking into, but there are valid reasons why PC based instruments have never been very popular.

[0culus]

I agree that I prefer dedicated instruments. Not the least of which, if you depend on your lab PC for most of your T&M needs, if something goes wrong with the PC (hopefully you have backups!) it might take some time to get back up and running.

[frozenfrogz]

Thank you all for your suggestions and advice!
I would like to keep the generator a dedicated device for various reasons. One being, that I only have a remote desktop installed via KVM network extender. The SRS looked like a nice deal but I guess I’ll pass on that one then. The FY6800 might be an option or hope for some HP gear to pop up.
As I said my budget is pretty tight around the 200€ mark (student problems), so spending double on a new Siglent does not compute atm. Not saying I don’t want one of those though.

In regards to my actual needs: For most tasks a simple MP3 Player or even my mobile phone might do the job - tracing signal paths without actual quantitative evaluation or testing a circuit against a reference one by feeding it a known waveform and looking at the output with the scope. Other tasks may include PWM signals on an LED driver dimmer input and measuring DC quality at the output. However, all this seems to evolve from a passionate hobby towards something more professional, so I would not mind getting something that enables me to do more intricate work in the future. I hope you can understand what I mean.

[2N3055]

I don't like this idea. There are all sorts of theoretical and practical objections as well as philosophical and moral ones. The main reason I don't like it is that I like to think in terms of dedicated instruments and preferably a load of old junk. 

Well, join the philosophical or moral froum then  :-)  ( No hard feelings, just an attempt of joke.. It should be narrated by  Jeremy Clarkson)

Seriously, from frozenfrogz' point of view it has to be an option worth looking into, but there are valid reasons why PC based instruments have never been very popular.

Not really. They are used plenty, for exact reason stated: tool for the job.
PC-s are cheap as chips (compared with T&M equipment anyways). I have a separate PC that is pretty much dedicated instrument. It boots faster than my Keysight scope. If you keep it in sleep mode it is up in few seconds.

Dedicated instruments are better at many things. PC based ones are better at other things. Tool for the job. I have both and use them both all the time.

[2N3055]

Thank you all for your suggestions and advice!
I would like to keep the generator a dedicated device for various reasons. One being, that I only have a remote desktop installed via KVM network extender. The SRS looked like a nice deal but I guess I’ll pass on that one then. The FY6800 might be an option or hope for some HP gear to pop up.
As I said my budget is pretty tight around the 200€ mark (student problems), so spending double on a new Siglent does not compute atm. Not saying I don’t want one of those though.

In regards to my actual needs: For most tasks a simple MP3 Player or even my mobile phone might do the job - tracing signal paths without actual quantitative evaluation or testing a circuit against a reference one by feeding it a known waveform and looking at the output with the scope. Other tasks may include PWM signals on an LED driver dimmer input and measuring DC quality at the output. However, all this seems to evolve from a passionate hobby towards something more professional, so I would not mind getting something that enables me to do more intricate work in the future. I hope you can understand what I mean.

For all those things you said before and now, I personally would go with FY6800 for your case. Very inexpensive, 2Ch, good enough instrument for most of uses. It is also simple AWG which is also useful sometimes.
If you could splurge 260 €, you could go for new Rigol DG812 that is really a whole different class. If you could afford it, I would suggest that. At lower amplitudes harmonic distortion is very low (much better than specified), you can create dualtones (that is useful for intermodulation distortion test), you have 2 ch etc etc...

But I again think that FY6800 will be more than you need, with 100€ left to buy other much needed things.

[frozenfrogz]

Thanks for suggesting the Rigol. Since I am pretty happy with my scope I might give this a shot. Also, I have some options to get this VAT free (e.g. 219€ via Batronix) what would perfectly fit my budget.

Are these "RigLOLable"?
Not that it matters much in my case but would be nice non the less.

[2N3055]

Thanks for suggesting the Rigol. Since I am pretty happy with my scope I might give this a shot. Also, I have some options to get this VAT free (e.g. 219€ via Batronix) what would perfectly fit my budget.

Are these "RigLOLable"?
Not that it matters much in my case but would be nice non the less.

I don't think anybody unlocked them (so far?). So no free bandwidth.  On the other hand, there is only one option, upgrade of AWG mem from 2M to 8M. Which is not an issue, 2M is plenty. All other features are standard. I also like that you can make signal sequences: 10msec of this and than 200msec of that... Siglents don't seem to have that. They also have PRBS that is nice for testing serial receivers (up to 10Mbps on smallest ones), and quite usable frequency counter to 240 MHz...
If you are a student, you can ask for a quote at Batronix, send them an email, they are quite nice, they might give you few € discount...

Regards,

[radiolistener]

FY6800 and FY6600 unfortunately has too high jitter. That's the real issue for this gen, with no jitter it will be the best gen for the price

For his intended use not an issue.

For arduino stuff it may be not an issue. But he want to use it for audio. This is very important issue for audio applications, at least for good quality audio, because jitter affects THD+N 

[exe]

For arduino stuff it may be not an issue. But he want to use it for audio. This is very important issue for audio applications, at least for good quality audio, because jitter affects THD+N 

So, how much jitter does it have? Is it possible to fix it?

[radiolistener]

So, how much jitter does it have? Is it possible to fix it?

As I know, FY6600/FY6800 has more than 5 nano-seconds jitter. And I think it's impossible to fix.

To understand how much it is for audio, here is a small simulation how ADC sample clock jitter affects audio quality:
1) 10 kHz signal digitized on 16 bit ADC with 44.1 kHz clock with zero jitter:


2) 10 kHz signal digitized on 16 bit ADC with 44.1 kHz clock 2 ns jitter:


3) 10 kHz signal digitized on 16 bit ADC with 44.1 kHz clock 400 ps jitter:


Here is also article about jitter for audio:
https://benchmarkmedia.com/blogs/application_notes/12142221-jitter-and-its-effects

[HalFET]

Thanks for suggesting the Rigol. Since I am pretty happy with my scope I might give this a shot. Also, I have some options to get this VAT free (e.g. 219€ via Batronix) what would perfectly fit my budget.

Are these "RigLOLable"?
Not that it matters much in my case but would be nice non the less.

Don't forget the Rigol clearance sale page, you can get some really nice deals on there: https://www.rigol.eu/clearance/

[2N3055]

So, how much jitter does it have? Is it possible to fix it?

As I know, FY6600/FY6800 has more than 5 nano-seconds jitter. And I think it's impossible to fix.

To understand how much it is for audio, here is a small simulation how ADC sample clock jitter affects audio quality:
1) 10 kHz signal digitized on 16 bit ADC with 44.1 kHz clock with zero jitter:

2) 10 kHz signal digitized on 16 bit ADC with 44.1 kHz clock 2 ns jitter:

3) 10 kHz signal digitized on 16 bit ADC with 44.1 kHz clock 400 ps jitter:

I know you just want to help, but how is that relevant to his use? He won't be clocking A/D or D/A converter with it.  Nor he will us it to measure 24 bit converters parameters...
On the other hand thank you for details for those that might need that information.

Anyways, if OP can afford something better it doesn't matter..

[Calvin]

Hi,

I understand Your reservation against a PC controlled device, but I´d still like to guide Your interest to the Digilent AnalogDiscovery2 again.
As a student You´d be entitled for the reduced academics pricing, which should fall just around 200€ iIrc.
Its a full fledged compact lab well suited for audio applications and up to ~10MHz, including a function generator, a oscilloscope and several analyzers (impedance, spectrum, network, logic) and some more.
With 14 bit of generator as well as oscilloscope resolution its alot better than most cheap benchtop devices and it takes the least lab space.
It´d also be for sure the cheapest, most compact and mobile lab one could think of (together with a laptop).
It´s certainly not a toy or a overprized gimmick and a solid start for a small lab.

regards
Calvin

[radiolistener]

Nor he will us it to measure 24 bit converters parameters...

24 bit audio converters require jitter better than 5 ps.
FY6600/6800 produces 5 ns jitter which spours will be in audible range even for old 16 bit CD quality.
For 8-bit audio FY6800 jitter spours will be near noise floor, but still in audible range. 
For 4-bit audio (such as sound of old cell phones) this jitter can be considered totally inaudible.

Note that topic starter asked generator mainly for audio applications. Not for arduino toys.

I think the good soundcard for PC is the best choice for such purposes, because generator with comparable parameters will cost a lot of money. Also, good soundcard allows to use it as spectrum analyzer with high dynamic range (up to 144 dB).

[Zenith]

Note that topic starter asked generator mainly for audio applications. Not for arduino toys.

The OP gave me the impression that it was for signal tracing and basic electronics and that his mobile phone could almost do the job. He is also interested in applying PWM to LED drivers, and tinkering with Arduino and Raspberry Pi applications, although he could see his interests expanding in the future. He's forced to consider the cost carefully, but he seems tempted to stretch to a Rigol, which is a better instrument. He was happy with two old function generators until they both failed.