Function generator with sweep function

[WaveyDipole]

I recently aligned a Roberts R300 radio using the procedure described in the service information a multimeter and signal generator but having completed it, I get all the stations in the correct place but it just doesn’t seem to quite ‘feel right’. I’m not sure quite how to describe the problem except that stations do not seem to snap into tune as cleanly as they are supposed to and the volume and clarity of stations is a little below par. They sound a bit harsh. Anyway I thought one way forward might be to sweep the IF cans in turn and view the output on an oscilloscope so as to get a better understanding of what is happening. However this requires a sig gen with a sweep function, which my Thandar TG2001 does not have.

The Thandar does have an input for a VCO and I have experimented with this and determined the ramp voltage range required. I found some basic ramp generator circuits online to use as a basis for experimentation and tried a basic three transistor one in a simulator. I also added a way to generate a sync pulse and vary the sweep speed. However I still need a method set the upper and lower sweep limit as well as generate markers so this is still a work in progress. I also have a couple of op amp based circuits to try. However I am also considering the possibility of buying a function generator with a sweep function as I think this will be a useful addition to my current equipment.

I have seen a number of these mentioned on this forum such Adavnce 63 and Taylor 62 as well as various HP and Philips models. In fact, I recently saw a HP3325A on eBay for £150. This does not power up so may be a simple fix, but my gut felling is that this is too much for a dead unit. I don’t wand to spend more that £200 on a good working one, but ideally much less than that.

I have also considered a more modern Rigol DG1022U that I have seen for just over £200, however there are caveats. For example, my TG2001 can deliver a 20v p2p over the entire frequency range of the generator, whereas this Rigol will deliver 20v p2p only up to 5MHz, then it goes down to 10v and finally to 5v p2p by the time it reaches 20MHz, not to mention that the shape of the waveform begins to suffer considerably after a point. The TG2001 gets up to at least 10MHz before that starts to happen. The second channel has nowhere near the same spec as the first although the unit is perhaps more configurable than the TG2001. It also has FM modulation although I’m not sure how much use that is going to be below 25MHz. The more I think about it, the less I am inclined to buy this in favour of an older piece of equipment as mentioned above. The Sigilent apparently comes out much better but costs more brand new and nothing available used at present. Compared to those I don’t suppose those £50 ones you see on eBay are particularly good? For radio work I don’t believe I need an arbitary signal generator anyway.

So I would like to ask for advice. Modern entry level device or older lab grade device? Or make my own VCO? What is a sensible and realistic budget for a sig gen with sweep for casual repairs and hobby work?

Regarding making my own VCO, does anyone have anything on making a marker generator? Alternatively, if an oscilloscope has cursors, could a cursor be used in some way instead?

[MrBungle]

The HP3325 is notoriously unreliable. I wouldn’t poke it with a 6 foot long stick. DG1022U is also naff. If they break down then they send you a DG1022Z as a replacement as they aren’t repairable and have some nasty design flaws.

The HP3312A is pretty good but suffers from the usual analogue complaints of dicky switches galore, inability to set anything accurately and drift. It does have a sweep generator built in and will hit 12Mhz.

I have been here with a dead HP3312A on my hands though and just foot the bill for a Rigol DG1022Z instead. Money well spent. It’s not perfect but it does the job and has a sync output on the back you can use as a frequency marker. Does AM, FM, ASK, PSK, all wave shapes and has 25MHz of bandwidth on both outputs at the same time. Is accurate to 1Hz always. You can set the output level in dBm into 50 ohms as well. It replaces a pulse generator, two function generators with sweep, RF signal generator, all sorts. At around £350 it was a lot more expensive than some options but the Keysight equivalent is about £2000. By the time I’d found it I had probably spent as much as it on cruddy generators and hours trying to cheap out. Annoying however that it isn’t a bit cheaper.

As for output amplitude, this is normal for modern generators. The amplifiers after the DDS generator can’t slew quickly so they reduce amplitude to keep the quotable bandwidth up.

[IanG3XYV]

I think you are being rather unfair on the venerable HP3325 Mr Bungle. I have two 3325A and a 3325B (I know I have a problem regarding test gear) and don't recall any issues although they don't get a lot of use.

Ian

[Radio Wrangler]

For my sins, I had the job of redesigning the synthesiser in the 3325A. It is big, complex and the sheer number of parts makes it unreliable. I wouldn't advise anyone to get one. And they rely on several unobtainable custom ICs

I would question why they wanted a function generator in the first place.

For aligning radios, go for an RF signal generator. They're a LOT cleaner and have more useful modulation facilities, they are a lot more stable.

For audio work, function generators aren't good for low distortion. Use a wien bridge oscillator (HP 339A low distortion osc/ distortion meter)... or go the full hog for an HP8903A/B.

I'm not really sure what a function generator is FOR.

Jack of several trades, master of none.

David

[MrBungle]

I have a similar problem with test gear

I worked at a company which had a pile of 12x 3325A units that were dead or dying in various ways. Apparently they were repaired at least 5-6 times each before they were abandoned, usually first repaired within a month of being powered up. They got replaced with the much more reliable 33120A in the end.

That’s a good option as well. If you can get a 33120A. Very nice. Also floating outputs.

Function generator is great for low level low speed analogue design. Control loops, opamp wrangling and the likes. Also audio, if you ignore the awful THD and just use it as a signal source for testing filters for example. I find them rather useful. No good for RF. too drifty even if it’ll go high enough up the frequency scale.

[WaveyDipole]

Originally Posted by MrBungle

DG1022U is also naff. If they break down then they send you a DG1022Z as a replacement as they aren’t repairable and have some nasty design flaws. ...

... just foot the bill for a Rigol DG1022Z instead. Money well spent. It’s not perfect but it does the job and has a sync output on the back you can use as a frequency marker. Does AM, FM, ASK, PSK, all wave shapes and has 25MHz of bandwidth on both outputs at the same time. Is accurate to 1Hz always. You can set the output level in dBm into 50 ohms as well. It replaces a pulse generator, two function generators with sweep, RF signal generator, all sorts. At around £350 it was a lot more expensive than some options but the Keysight equivalent is about £2000. By the time I’d found it I had probably spent as much as it on cruddy generators and hours trying to cheap out. Annoying however that it isn’t a bit cheaper.

That's an interesting observation. I had a very quick look at the specs of the DG1022Z this morning and indeed they don't make any distinction between channel 1 & 2 like on the DG1022A (I couldn't find specs for the 'U' version). The harmonic distortion figures on the Z version are much better and its 200MSa/s vs 125MSa/s, however amplitude is 10vpp only to 10MHz as opposed to 10vpp to 20MHz (on paper at least) on the U. On the other hand the U version is only 3Vpp on CH2.

I also had a look at the equivalent Sigilent - SDG1025? and it seems that although this was better specified than the DG1022U, Rigol seem to have upped their game with the DG1022Z. On the other hand the SDG1032X seems to at least match the Rigol DG1022Z for THD and improve upon amplitude which is 20vpp to 10MHz and then 10vpp above and they have done this with 150MSa/s and at a similar, if slightly lower price point.

I need to have a ponder over this. Anyone have experience with the Sigilent gear?

Originally Posted by Radio Wrangler

I would question why they wanted a function generator in the first place.

For aligning radios, go for an RF signal generator. They're a LOT cleaner and have more useful modulation facilities, they are a lot more stable.

For audio work, function generators aren't good for low distortion. Use a wien bridge oscillator (HP 339A low distortion osc/ distortion meter)... or go the full hog for an HP8903A/B.

I'm not really sure what a function generator is FOR.

Jack of several trades, master of none.

David

I am curious about this comment. The models I have mentioned all have AM and FM modulation. The Rigols and Siglents have other modulation types as well (e.g. PM, ASK, FSK, PSK, PWM, noise). PWM is perhaps of interest for reasons other than radio, however the reason I looked at these is because they seem to be frequently mentioned. However, I am simply looking for something suitable primarily for radio work that is within my budget so if there are any others I should consider then if you can suggest what they might be I will do my research. I have noticed that when the description 'RF' is included, this seems to put the equipment into a higher price category, but I am by all means interested to hear more about what I should look for.

[MrBungle]

I played with a Siglent before I got the Rigol and the interface on it was awful. Can’t remember which model it was.

The moment someone adds RF it goes into 4 digits. The new DDS generators are perfectly fine for that to be honest. You can always add external attenuators to get low dBm powers.

Also worth looking at is the TTi generators. They’re pretty nice. Full 20v at full frequency. TGP3121 is bottom end at €1300 or so. However I have seen some of these go for virtually nothing on eBay as well. In fact I got myself a TTi TF930 counter a few weeks back for £50 and it certainly didn’t look like it had even been used. Same with their power supplies.

[WaveyDipole]

The Sigilent interface was one of the negatives mentioned by the reviewer in the article/video I saw. Apparently you had to bring up the mnu for everything. On the positive side it displayed all the key parameters at once, although that is something that has now been addressed on the DG1022Z.

https://core-electronics.com.au/vide...glent-sdg1032x

The TGP3121 is way out of my budget unfortunately, but I will keep an eye on eBay.

[MrBungle]

Yes it is very menu driven. The main things on the rigol are direct entry or one button press then entry or use the encoder.

For example to set the frequency you turn it on then 7, 0, 3, 0, KHz, output. And it’ll kick that out. Set amplitude press ampl, 2, vp-p. Once you’re in the mode you can use the encoder to move it around. Only the complex modulation functions and source impedance controls are in menus. Sweep controls you just hit sweep and then low, high frequencies and period using similar methods.

Also to note the Rigol has a half decent 200MHz frequency counter built in that works entirely independently to the generator.

It was a much better purchase for me than their scope was. It’s also very accurate so it serves as a good time mark and constant amplitude generator for calibrating scopes.

Only thing to note is that on very low pulse speed mode there are quite horrible abberations on the leading and trailing edges. Siglent does this too.

Crazy thought, non sweep generator and a varicap* coupled to the radios local oscillator via a small (10pF) cap driven from the 'scopes X output. No X output? Use a 50Hz sinewave from a small transformer with the 'scope in XY mode.

OK this will detune the front end (relatively) but that wont matter.

*almost any small diode will work

[MrBungle]

Not as crazy as it sounds. From Sprat 121:



I suppose it's a spectrum wave meter really. Tuned circuit which is tuned by the scope's X output and a diode detector and low pass filter.

I'm continually impressed with what people manage to do when necessity is high and funds are low.

[dseymo1]

Could you take a different approach, and use your scope's X output, if it's available externally, to sweep your sig gen?
Edit: Sorry, just noticed Merlinmaxwell's similar suggestion.

[mhennessy]

I must admit that when it comes to function generators, I like knobs

My "daily driver" is a Wavetek 162. It has a certain amount of sentimental value, being as I've owned it since I was 16 or 17, and it's had one or two minor repairs in that time, but it's been with me for so long that it feels like it's part of me. I'd say the same about my Telequipment DM63 'scope, which came from the same source. I've used countless others in that time, and all have had their pros and cons, but the Wavetek remains central on my bench, and it does get a lot of use.

I have a Philips PM5134, and that is a worthy competitor. It's smaller, but runs warm and has a (very quiet) fan - this means you have to leave space above it, so that negates its compact size. A lot of BNCs are located on the back, so I've had to extend them to the front (I have a 1U panel with a load of BNC barrels, and this gives me access to the back panels of several bits of gear rack-mounted on the bench). The Philips has the advantage of a second internal oscillator so AM and FM can be done with no other gear. Personally, I prefer to use the Wavetek to generate the sweeping waveform, which I feed to both the 'scope and whatever generator I'm using. That's because the Wavetek can be "stopped" at any arbitrary point in the waveform, and that does away with the need for markers. I set the Wavetek output to be 0V, put the spot in the centre of the 'scope screen and set the oscillator that's being modulated to 10.7MHz or 470kHz as needed. Then, I move the Wavetek until the output frequency has gone to 10.8MHz or whatever, then set the X-gain of the 'scope to move a whole number of divisions - hence that calibrates the display. Then I set the Wavetek sweeping, and away I go! It takes more time to describe than actually do...

Arguably, the "distorted triangle wave" that an analogue function generator produces isn't terribly clean, but it doesn't seem to make any odds for domestic radio alignment, where sharp IF filters remove the harmonics. Some function generators are cleaner than others, but don't expect less than around 1% THD. But that's fine for most applications. To generate cleaner sine waves in the analogue domain, you need a Wien bridge or state-variable oscillator, but these usually have amplitude bounce, so quickly sweeping an amplifier or filter to get a quick feel for what it's doing before detailed measurements is so much slower.

What is a potentially a problem for radio work is the range of the output attenuator. That's certainly a problem with the Philips. I need to buy or build some external attenuators.

In the past, I've used a Marconi TF2008 for this sort of work, but this is massive and very heavy. I still had to sweep it with the Wavetek, as the built in modulating oscillator wasn't flexible enough. On the "to-do pile" I have a Marconi TF2015 which is much smaller and should be able to be swept by the Wavetek, but it doesn't go below 10MHz, so no good for AM sets.

It's worth saying that all function generators do FM because the central oscillator applies equal and opposite currents to the timing capacitor, and these currents are derived from a voltage. But not all function generators do AM (my Wavetek doesn't) as AM involves a VCA, which is a relatively tricky thing to design, especially given the wide bandwidth required.

I have a couple of other generators on the bench - a TTi TG1010 and a HP 3314A. Both were bargains (bought as faulty, but easy enough to get going), and both are pleasantly competent, but I hardly use them. And that's purely because of the user interface. In both cases, too many buttons and only one rotary encoder. And the HP has an extremely noisy fan. A modern Rigol or equivalent will have the same problem. Slow, slow, slow. Function generators are about "quick and dirty" work, and for that, you really can't beat knobs!

Back to the radio, for an AM set, you can usually "peak" the cores (not so with FM). And you can manually sweep your existing generator - that's worth doing for fun. The main learning point will be the effect of the AGC circuit. For that, you might find you need an attenuator. And of course, the audio stage might not be blameless. Feeding in your own audio helps there...

Sorry to ramble - hope it helps

Mark

[Radio Wrangler]

Originally Posted by WaveyDipole

I am curious about this comment. The models I have mentioned all have AM and FM modulation.

I do a lot of receiver performance measurement as part of the day-job. Yes there are func generators offering fairly comprehensive types of modulation, but you need to be wary of their cleanliness and modulation distortion.

A plain analogue function generator is usually too imprecise and too drifty to make simple alignment of broadcast radios easy. You can do it at a pinch but you wind up fighting the test gear to get the job done.

There are DDS based function generators which are really arbitrary waveform generators. With appropriate waveform files (sometimes built in) they can do all sorts of tricks, but their bit depth and sampling rate and file length give them some significant limitations in the ability to make signals as clean as an old school cheapo analogue RF generator.

For radio work, a straightforward RF sig gen is a hard act to follow. You can pick them up cheaply and they're eminently fixable. Big, though, not very modern looking, but they have their strong points.

If I want ramps and triangles, I reach for a function generator, if I want square waves, a pulse generator unless the risetimes needed are very easy.

David

[Argus25]

A very good option is the PM5326. It goes to 100MHz too, has a great VFO with a very good digital counter in it based on an Intersil IC, and it sweeps as required and it is usually reasonably priced and they have a very good linear attenuator assembly which can reduce the signal below -80dB. The only faults common in these are the power supply electros, otherwise they are pretty good and very well made and more reliable than the Philips test gear of the previous generation, which I also have for comparison. It is more of an RF generator suitable for TV & radio work & IF sweeping, than it is a "function generator" which usually only go to a couple of MHz and have much higher output levels.

[WaveyDipole]

Unfortunately there is no X output on any of my oscilloscopes. However I will look at the transformer option when I have a bit of time. Thanks for the ideas and the circuit. I don't think I have a varicap but I do have various other diodes I can experiment with.

I am fairly confident in the audio part of the radio as feeding a signal into the volume control gives a loud clear sound on the speaker. I am curious about that comment regarding the Rigol being slow. Could you give me a little more detail on that please?

I have added the PM5326 to my lookout list. Stewarts of Reading have a PM5134 for sale. I have inquired about the price.
Incidentally, how much of an output level does one actually need for RF sweeping?

[MrBungle]

Rigol is not particularly slow. No slower than the TG210 it replaced

Also worth looking out for a Philips PM5324. That is an all band RF signal generator with a wobbulator built in and a sweep output. It can wobble over IF bands

Not sure I'd buy anything from Stewarts of Reading these days. Prices are quite frankly ridiculous when you add packaging and VAT.

[Radio Wrangler]

The Marconi TF2008 is a lab grade sig gen which does AM, FM and can go down to 0.2uV for sensitive stuff. BUT you can set it to sweep over narrow or wide ranges. It has a horizontal output for scopes along with crystal markers. It's a good sig gen, but also a sweeper and a wobbulator. 10kHz to 510MHz.

No triangles or square waves, though.

David

[WaveyDipole]

Originally Posted by MrBungle

Not sure I'd buy anything from Stewarts of Reading these days. Prices are quite frankly ridiculous when you add packaging and VAT.

From my own earlier inquiries I also got the impression their prices were a bit on the high side, but I do not have a basis for comparison for everything I asked about.

The TF2008 and PM5324 look like quite big beasties, but I will add them to my list for now.

[mhennessy]

The PM5324 is tiny compared to the TF2008. And less than a quarter of its weight

Oh, in case the comments about speed came about as a result of my earlier post (and sorry if not), I'll just clarify that I've not used a Rigol or similar, and I wasn't referring to how responsive or otherwise the UI might have been. What I meant is that it's just so much slower to look at the display to see what mode it's in, change that if required, and then twizzle the rotary encoder, before then having to look back to put it in another mode so that the same encoder then does something else (repeat). The HP I mentioned is better in that regard than the TTi TG1010 I also mentioned, but both are much slower than a box with a brace of potentiometers and switches. I drive my Wavetek without actually looking at it most of the time! Both eyes on the 'scope screen

But yes, I wouldn't be at all surprised if a modern Rigol or similar had a "laggier" UI than either my HP or TTi, so that would only add insult to injury.

Oh, and I hate typing in numbers. Yes, it's nice to know that what you get is (hopefully!) what you just typed, and occasionally that is valuable, but most of the time it's not - you adjust until you observe some effect. For example, the -3dB point on the output of a DUT - having found that, you'd read off the frequency on a counter or the 'scope.

But in the same way that people like to have very precise voltage and time measurements, some people are uncomfortable with the analogue nature of traditional signal generators. It's not something that bothers me most of the time, but it's very much "horses for courses". The fact that I have a modern DDS generator in the TG1010 that I never use despite it being right in front of me speaks volumes. Recently I rearranged my bench, and after a bit of nice woodwork, I now have both the HP and TTI literally "in my face" in an attempt to make me use them a bit more - so no-one can accuse me of not trying

Until you have a few generators and actually use them, you won't be able to work out what to choose - and nor can we: so much of what's been written here is based on our personal experiences, but we're all different, and all doing different things, or similar things in different ways. You'll never find a single one that does everything - a function generator is a great general-purpose workhorse, but it needs to be supplemented by other, more specialist gear.

Sorry to waffle