HP 33120A Function generator

[Henry Finley]

This photo is of an HP 33120A 15 mhz function generator. I've boorowed one from a buddy a few times and it is the cats pajamas from what I've been able to tell. The downside is it only goes to 15mhz in a day and time when technology has made such lab quality machines go higher. I don't know when they were made. I'm guessing mid-late 90's or so. Anyway, I have some questions.  My friend's machine has some problems. It boots with an RS-232 error and obviously can't connect with a computer. Also it has a problem of shutting down after it has been on for a while, or not booting at all without repeated attempts. I've done some tests and feel like it is a problem with some leaky electrolytics. But when it works, it works well. Questions: My buddy claims to have paid $300 plus for it. What should i offer him? Secondly, about this 232 error, is that something that can be fixed with reasonably good technical abilities like mine? Third, Since I am mostly really interested as a vintage enthusiest of ham radio gear, wich typicall only goes up to 30 mhz, is it possible to align this gear (like Hammarlunds for instance) on a harmonic of the 15mhz maximum that the generator is capable of? Right now those are the only 3 questions I can thing of at this time. But if this discussion takes off, I may have more. I think it is a way better machine than any of this cheapo Chinese stuff Ive been considering. Thank you.

[nctnico]

These are horrible to use. Been there, done that. Just get the cheap Chinese you have been eyeballing; they will give you a much better user experience. And you don't need to fix it.

[Henry Finley]

These are horrible to use. Been there, done that. Just get the cheap Chinese you have been eyeballing; they will give you a much better user experience. And you don't need to fix it.

Horrible to use? That's an unexpected response. The few times I used it, it was a dream compared to my usual vintage gear (Hickok 288X). I managed to get my Hammarlunds and Hallicrafters right on the button with no problem at all, at least up to 15mhz. How is this HP so bad? I haven't seen it myself.

[nctnico]

These are horrible to use. Been there, done that. Just get the cheap Chinese you have been eyeballing; they will give you a much better user experience. And you don't need to fix it.

Horrible to use? That's an unexpected response. The few times I used it, it was a dream compared to my usual vintage gear (Hickok 288X). I managed to get my Hammarlunds and Hallicrafters right on the button with no problem at all, at lease up to 15mhz. How is this HP so bad? I haven't seen it myself.

Try to setup sweeps and modulation for example. Things get much easier with a clear overview of the settings on the screen.

[Henry Finley]

Thanks. Looks like you may have closed the matter, early on. Although as square wave generators, these Chinese jobs look like a joke. That HP was square as a cube all the way to 15. I never did sweeps. Wasn't necessary for my purpose. But maybe sometime it might become necessary.

[nctnico]

Thanks. Looks like you may have closed the matter, early on. Although as square wave generators, these Chinese jobs look like a joke. That HP was square as a cube all the way to 15. I never did sweeps. Wasn't necessary for my purpose. But maybe sometime it might become necessary.

Square wave depends on how it is generated and the bandwidth of the output amplifier. Remember a square wave has a lot of higher harmonics so it needs a very high bandwidth output amplifier compared to the fundamental tone. For a lot of generators you'll see that the square wave if limited to a much lower frequency so the signal looks like a square wave.

I have a reasonably high-end generator from Tektronix with a 25MHz bandwidth for sine waves. The square waveform only goes to 20MHz.

[J-R]

I paid $130 for an excellent condition, fully working unit 2 years ago.
Granted, the prices are up a bit these days.

Seems like if your buddy is really your buddy he'd just give it to you?  But on the flip side by the time he sells it online he might only make a profit of $100 for it, so maybe that would be a reasonable ballpark offer?  You can see what is available on ebay.  $200 in some cases for parts units with various issues.  But there are fees and shipping costs involved...

For hobbyists, get what you like, since it's just a hobby.  My only function generator for quite some time was the JDR Instruments DFG-600.  Lots of fun knobs to tinker with.

The HP 33120A has been working well for me as a basic function generator.  I'm not needing much more at the moment.  Software and the AWG stuff might be tricky on this unit unless you're motivated?

I also snagged one of the cheap Chinese units nctnico is talking about, the Owon AG1022.  Got it for $125 delivered a couple years back when Arrow was purging their Owon gear.  It has some limitations: for example the square wave function only reaches 5MHz.

These days maybe a scope with a built-in function generator would be a popular choice?

[2N3055]

These days maybe a scope with a built-in function generator would be a popular choice?

It is not.. Those things are good in a pinch when you have nothing else around but are limited.

To OP. Broken things are basically worth as much as a trash. So if you can get it for free or say 50$ then Ok.

300$ buys you a brand new Siglent SDG1032X or Rigol DG812 which are 2 ch, better specs and more functions. And full warranty.

[J-R]

These days maybe a scope with a built-in function generator would be a popular choice?

It is not.. Those things are good in a pinch when you have nothing else around but are limited.

To OP. Broken things are basically worth as much as a trash. So if you can get it for free or say 50$ then Ok.

300$ buys you a brand new Siglent SDG1032X or Rigol DG812 which are 2 ch, better specs and more functions. And full warranty.

The built in function generator can be just fine if it does what you need.  There are advantages to the arrangement.

A function generator costing thousands can do many things your two suggestions can't.  So only stating features is not the only argument.

Dave's recently released video () sports the DSOX1204G, so it seems the setup gets around...

Broken things are worth plenty of money.  You are just plain wrong.  There is a mountain of broken test equipment on ebay ready for repair, especially these days.  Just got a 7.5 digit Keithley 2010 for $500 and it took 2 minutes and just a soldering iron to fix.  Also got a Fluke 8842A with all the options for $100 and again, 2 minutes and just a soldering iron on the transformer and it's perfect.  Win for the sellers, win for me...

[DeepLink]

I have a 33120A (had it for 20+ years) and I like it

Recently I got a Siglent SDG2042x, so now I have 2 Functions generators

Which one do I turn on the most? The HP ~90% of the time

Easier to use for basic stuff. But the HP has it's problems.
 - You can't turn off the output, only turn it down to 50mV
 - Can't see both frequency and amplitude at the same time
 + Display is big and easy to read

If I need any things for modulation, I turn to one of my Signal generators

I see the 33120A as a nice basic Function generator, but if I was to get something today, it would be a new Siglent
If you can have the 33120A for a good price, and you don't mind do the repair, then go for it. You probably also learn something

[2N3055]

These days maybe a scope with a built-in function generator would be a popular choice?

It is not.. Those things are good in a pinch when you have nothing else around but are limited.

To OP. Broken things are basically worth as much as a trash. So if you can get it for free or say 50$ then Ok.

300$ buys you a brand new Siglent SDG1032X or Rigol DG812 which are 2 ch, better specs and more functions. And full warranty.

The built in function generator can be just fine if it does what you need.  There are advantages to the arrangement.

A function generator costing thousands can do many things your two suggestions can't.  So only stating features is not the only argument.

Dave's recently released video sports the DSOX1204G, so it seems the setup gets around...

Broken things are worth plenty of money.  You are just plain wrong.  There is a mountain of broken test equipment on ebay ready for repair, especially these days.  Just got a 7.5 digit Keithley 2010 for $500 and it took 2 minutes and just a soldering iron to fix.  Also got a Fluke 8842A with all the options for $100 and again, 2 minutes and just a soldering iron on the transformer and it's perfect.  Win for the sellers, win for me...

Broken things sold as broken are industrial electronic waste. Fact that you were lucky reclaiming some of it doesn't make any difference  to that fact.
And Keithley 2010 that had some guy poking inside with a soldering iron (and no wrist strap and anti static precautions, maybe?) is still electronic waste, despite it shows some numbers on the screen. Until you send it in for a full adjustment and full factory calibration (or at least full calibration), brand new Fluke 87V is more accurate until proven otherwise...

A function generator costing thousands can do many things that those two costing 300$. Yes that is true. Thank you Captain Obvious.
When Nico says that his Tektronix AFG31000 is better in some respect than a 300$ Rigol or Siglent that is absolutely true.
But we are not talking about those and likes of them.
That 20 years old single channel crap cannot. That particular one doesn't work at all. And even if you resurrect it, it is still very basic single channel signal generator.
And features ARE important. Feature list is a list of things you can and cannot do with a piece of equipment.
There are many cars that cost many times more than a small Cessna. And fancy as they are they cannot fly.
And despite not being fancy F15, Cessna can fly. And quite well.

Op is not buying a new scope. He just need a box with signal generation capabilities. And AWGs built in scopes are wastly limited in capabilities, all to a different extent. But all of them have very limited amplitude for starters. Most of them you cannot terminate into 50 Ohms and use it to drive a gate of a mosfet.

If you collect stuff so you can have cool looking boxes with some numbers glowing sitting on the desk, it's fine to buy old and do some repairs. But if you don't have lots of knowledge and lots of fancy equipment, you cannot repair 7.5 digit meter in a garage and expect that what you have now is actually 7.5 digit meter. 80% percent of the price of the new Keithley is guarantee those numbers on the screen are not coming from random number generator. And even in USA there are maybe dozen of places that can fully adjust and calibrate it to as new condition. And it is not cheap.

As I said many times, on every piece of equipment that was bought from Ebay and repaired, there are dozens that weren't so lucky. Humans are funny creatures, we all like boasting how smart we are, but not many will admit they made a mistake and have bought something by mistake, and that then they weren't "smart enough" to repair it. How many had to buy 2-3 pieces of something until the came up with something made from parts from many "donors"?

So I think it is you that are wrong. You are giving advice to OP but not to the questions he asked but giving him opinion how you do your hobby. OP needs an AWG (or siggen whatever) and he is trying to decide whether to buy 20 year old broken very simple generator from HP or something else.

He also states that he thinks that old thing is better than "some cheapo chinese stuff he is considering".

If he considers to pay 300$ for that old broken crap that he has to repair first (no guarantee he will be able to do so) compared to brand new  superior specifications and features AWG from Rigol or Siglent, that is obvious that he should buy new Siglent or Rigol.

If he considers giving 50$ for old broken crap and chance it to repair it, or one of toy AWG he can get for 50 $, that is a bet that worth considering. Everything more, it's not worth the risk. You just save up a bit and get something that works well.

Those are the logical answers to his questions based on facts.

There are people that need piece of equipment to simply works so they can do what they do. There are people that enjoy buying old broken stuff and repair it.

[nctnico]

As I said many times, on every piece of equipment that was bought from Ebay and repaired, there are dozens that weren't so lucky. Humans are funny creatures, we all like boasting how smart we are, but not many will admit they made a mistake and have bought something by mistake, and that then they weren't "smart enough" to repair it. How many had to buy 2-3 pieces of something until the came up with something made from parts from many "donors"?

You make a good point here. Reminds me of people boasting about how much money they made with buying & selling shares. It is no secret most end up losing their money.

Every now and then I do buy broken gear from Ebay and show how I fixed it on this forum so I guess I'm part of the problem as well  Typically I do a couple of hours of research before buying a piece of equipment to understand what the common problems are on top of having a couple of decades of experience with repairing all kinds of electronic equipment. Fortunately I got paid for gaining that experience.  But in the end it is more for fun than anything else.

He also states that he thinks that old thing is better than "some cheapo chinese stuff he is considering".

If he considers to pay 300$ for that old broken crap that he has to repair first (no guarantee he will be able to do so) compared to brand new  superior specifications and features AWG from Rigol or Siglent, that is obvious that he should buy new Siglent or Rigol.

I checked the specs for the 33120A but it is pretty much on par with the Feeltech / Unit-T units. And spending $300 on Rigol / Siglent doesn't buy you much extra (maybe an external reference input but that is still no guarantee it actually works -as I have found out-). The components to make a digital function generator have become utterly cheap.

Thinking that old gear is better by definition can lead to nasty surprises. Like forgetting how long ago the 80's actually are and that fancy piece of equipment you saw in a magazine is completely outdated and in such a state that it needs a total overhaul in the present day. Recently I let go of a few HP6024A power supplies in favour of a bunch of tiny Gophert PSUs.

[joeqsmith]

Questions: My buddy claims to have paid $300 plus for it. What should i offer him?

Secondly, about this 232 error, is that something that can be fixed with reasonably good technical abilities like mine? Third, Since I am mostly really interested as a vintage enthusiest of ham radio gear, wich typicall only goes up to 30 mhz, is it possible to align this gear (like Hammarlunds for instance) on a harmonic of the 15mhz maximum that the generator is capable of?

They offer a service manual with schematics.   I've used this in order to add my own features.   As for you personally repairing it, not knowing you, I have no way to know if you would do more harm to than good.   Price wise, maybe check used parts units on ebay. 

I saw a video once of an amateur radio hobbyist doing a review of an Arb which consisted of a non-stop rant.  The poor guy was clueless.   It was pretty funny.  Even working it may be a bad choice for your application.  Make sure you understand your requirements and the Arbs capabilities beforehand. 

I bought one of these brand new and still use it a fair amount.  For the comments on the UI and what not, like most of my equipment,  I run it from LabView.

[mawyatt]

A lot of the discussions here are based upon, "Do you make $ with your equipment or just play with it!", and how valuable is your time? A professional makes $ with equipment, a hobbyist plays around. Although a Pro can be a hobbyist at times, when playing around with something just for the fun of it. However, this playing around can lead to some Professional endeavors at times, been there done that!

When I retired some time ago I went to eBay for an analog scope, got a Tektronix 2465 and fixed it, then another. Then got a HP34401A fixed it, then another. As mentioned by 2N3055 what were we reading just a few random numbers that happen to agree in certain digits on these two eBay repaired 34401As??? This forced the decision for a new DMM that had known accuracy, and the decision for the Keysight 34465A. Now the two 34401A readings are known within bounds as compared with the 34465A. Although the scopes were easier to verify since they are more visual instruments that precision instruments, they still had some uncertainty which led to the decision to get the Siglent SDS2000X+ as a "Reference" scope.

Note how the eBay repairs created an opportunity, , excuse, need, requirement to justify the purchase of some new instruments

Best,

[2N3055]

Note how the eBay repairs created an opportunity, , excuse, need, requirement to justify the purchase of some new instruments

Best,

Yeah, sounds familiar.... 

[J-R]

For sure a very short-sighted stance on repair.  A massive part of this EEVblog ecosystem revolves around us fixing our own gear, professional and hobbyist alike. To dump on that is just foolish.  Some repairs could impact the calibration, but many don't.

The OP didn't ask anything at all about buying a different function generator, he only asked about repairing the 33120A.

I actually do have enough calibrated equipment with data to test the repair on the Keithley 2010.  It is within the 2 years DC spec and 1 year AC spec.  If I wanted a 7.5 digit meter that I could keep up with the calibration, I would probably purchase something newer.

[mawyatt]

One of the really nice features of the modern high resolution AWGs is the canned waveforms that are available directly from the generator, some are quite accurate like the SDG2042X we have. We recently used this to verify and compare AC RMS readings of various DMMs when measuring different type waveforms, this was important in the "System" tests that are planned.

Of course the usual sine wave has the peak to RMS value of 1/rt(2), and the square wave has a 1 to 1 ratio. A quality triangle wave (x) has a 1/rt(3) relationship and the x^3 waveform has a 1/rt(7) relationship. If the generator has very accurate and stable peak values then the AC RMS can be inferred from the peak, fortunately the 16 bit SDG2042X and SDG6022X we have are quite good, way better than the +-1% specs. For example: We compared the x^3 waveform with these DMMs and the SDG2042X was set to 2VPP, or 1VP. If the waveform was perfect and the DMM was perfect the reading should be 1/rt(7) or 0.377964VRMS, we got 0.377999, 0.377792, 0.377978, and 0.377754 from KS34465A (2 year old), KS34465A (new), DMM6500 (few months old) and couple decades old AG34401A. Also got good agreement with a 16 bit PicoScope 4262 at 0.3778 VRMS. Similar results with various waveforms, both canned and arbitrary.

These various canned waveforms gave us confidence (as did the arbitrary waveforms we created) that the complex waveforms we encountered were correctly measured and reported.

Just received word an hour ago that the "System" we are working with has just passed first verification with flying colors at another location 1000 miles away (we are not attending the testing, but supplying the system controller), all the beams were spot on!! A tribute to these fine, yet modest cost, instruments that are now available due to the very sophisticated chips being utilized, likely many from Analog Devices and TI, and those folks know what they are doing

BTW the full up "System" is sort of like a Precision 15 bit 180VPP capable 128 Independent Channel Bi-Phase and Offset Modulated AWG, although initial tests are with just 64 channels.

Best,

[2N3055]

One of the really nice features of the modern high resolution AWGs is the canned waveforms that are available directly from the generator, some are quite accurate like the SDG2042X we have. We recently used this to verify and compare AC RMS readings of various DMMs when measuring different type waveforms, this was important in the "System" tests that are planned.

Of course the usual sine wave has the peak to RMS value of 1/rt(2), and the square wave has a 1 to 1 ratio. A quality triangle wave (x) has a 1/rt(3) relationship and the x^3 waveform has a 1/rt(7) relationship. If the generator has very accurate and stable peak values then the AC RMS can be inferred from the peak, fortunately the 16 bit SDG2042X and SDG6022X we have are quite good, way better than the +-1% specs. For example: We compared the x^3 waveform with these DMMs and the SDG2042X was set to 2VPP, or 1VP. If the waveform was perfect and the DMM was perfect the reading should be 1/rt(7) or 0.377964VRMS, we got 0.377999, 0.377792, 0.377978, and 0.377754 from KS34465A (2 year old), KS34465A (new), DMM6500 (few months old) and couple decades old AG34401A. Also got good agreement with a 16 bit PicoScope 4262 at 0.3778 VRMS. Similar results with various waveforms, both canned and arbitrary.

These various canned waveforms gave us confidence (as did the arbitrary waveforms we created) that the complex waveforms we encountered were correctly measured and reported.

Just received word an hour ago that the "System" we are working with has just passed first verification with flying colors at another location 1000 miles away (we are not attending the testing, but supplying the system controller), all the beams were spot on!! A tribute to these fine, yet modest cost, instruments that are now available due to the very sophisticated chips being utilized, likely many from Analog Devices and TI, and those folks know what they are doing

BTW the full up "System" is sort of like a Precision 15 bit 180VPP capable 128 Independent Channel Bi-Phase and Offset Modulated AWG, although initial tests are with just 64 channels.

Best,

Wow, congrats Mike, very good news indeed!!

Best,
Sinisa

[2N3055]

For sure a very short-sighted stance on repair.  A massive part of this EEVblog ecosystem revolves around us fixing our own gear, professional and hobbyist alike. To dump on that is just foolish.  Some repairs could impact the calibration, but many don't.

The OP didn't ask anything at all about buying a different function generator, he only asked about repairing the 33120A.

I actually do have enough calibrated equipment with data to test the repair on the Keithley 2010.  It is within the 2 years DC spec and 1 year AC spec.  If I wanted a 7.5 digit meter that I could keep up with the calibration, I would probably purchase something newer.

I wouldn't call it shortsighted but conservative and not overly enthusiastic. I have my share of Ebay hits and misses. I have Yokogava 6.5 digit DMM that is absolutely perfect and I got it for 150$. I also have an Omega CL8300 V/mA calibrator that was absolute nightmare to repair. It took months... and it was a component level repair on a ppm level calibrator without documentation.

People should repair whatever they like. But not because it is cheaper that way, because it is not always. And sometimes even if you succeed, you get outdated device that is nothing special really..

OP did say: "I think it is a way better machine than any of this cheapo Chinese stuff Ive been considering.."
Hence my answer in that context.

And I respect you and many others like repairing equipment. That is fun and you learn a lot by doing that.
Where we differ is in a concept that I don't see reason to buy a 7.5 digit meter and not care for it's calibration. If I were to get something that is normally worth 3000$  for 300$, then I would give another 500-600$ to have it adjusted and have a 3000$ instrument equivalent. Not free but still good deal. If I had a need for such an instrument. If I didn't have a need for 7.5 digit meter, I wouldn't buy it for repair in a first place. I would buy something else I would have use for.

But that is me. You do your stuff, however you like. That is how it should be. I'm not saying you're wrong with whatever you like doing.

[alm]

Where we differ is in a concept that I don't see reason to buy a 7.5 digit meter and not care for it's calibration.

Maybe they need resolution and linearity, and not absolute accuracy? You can test for linearity without any external reference, for example by matching resistors with each other in a bridge, and then building a divider out of that. Or putting voltage sources like batteries in series. The 7.5 digit meter isn't accurate to 0.1ppm anyhow, so calibration is never going to tell you anything useful about the last two digits or so. They are there for the extra resolution, but are not guaranteed for any absolute accuracy.

I use the 33120A as a function generator, and I quite like it. I find the front panel convenient for simple sine / square / triangle, but I admit I don't often use sweep or modulation. I find it very useful that the output is floating, something that is not available on all competing function gens. Mine had a problem that the RS-232 and GPIB interface didn't work, and neither did the beeper. It was an open diode in the bridge rectifier for the grounded logic board (which contains those circuits). Trivial repair as soon as I started measuring voltages in that circuit.

[bson]

The floating output on the 33120A makes it very easy to bias to just about anything.  Well, modest signal levels at least.

[nfmax]

I have three of these generators at home. One I bought 2nd hand about 12 years ago, working, for about £400. It served me well until I upgraded to a 33522B (also 2nd hand, working) a couple of years back. As a basic generator of sines and squares, and tonebursts, it is still as good as it ever was. Last year, I was given (£0!) a couple of non-working 33120As. One had a failed output amplifier power transistor (easily recognisable symptom: -ve half-cycle output distorts at higher signal levels). The other one had a blown output fuse!

There is a proper service manual, with circuit diagrams and part numbers, readily available. Only a couple of unobtanium parts inside, which don't seem to be particularly prone to failure. Quite easy to troubleshoot and fix.

That said, prices on eBay these days seem completely unrealistic, although asking prices for test gear have gone up massively in the last 18 months. In your case, maybe $100 or so would be worth a punt. If the instrument has the option 001 high-accuracy reference (if it does there will be 4 BNCs on the rear panel, instead of 2) you could probably rip that out and sell it on its own for $100 anyway, even if the rest of the instrument is a write-off.

However, this is a function generator, not a signal generator. For your purposes it may not be the ideal choice of instrument. Phase noise/jitter is on the high side, and the output attenuator is not particularly wonderful. It was introduced in 1994, replacing the 3312A, a traditional analogue circuit based function generator, but adding a basic ARB function. I got my employer at the time to buy one almost as soon as it came out, as we needed a convenient source of pulse trains with programmable delays and repetition rates to simulate a sensor I was working on.

[Henry Finley]

I have three of these generators at home. One I bought 2nd hand about 12 years ago, working, for about £400. It served me well until I upgraded to a 33522B (also 2nd hand, working) a couple of years back. As a basic generator of sines and squares, and tonebursts, it is still as good as it ever was. Last year, I was given (£0!) a couple of non-working 33120As. One had a failed output amplifier power transistor (easily recognisable symptom: -ve half-cycle output distorts at higher signal levels). The other one had a blown output fuse!

There is a proper service manual, with circuit diagrams and part numbers, readily available. Only a couple of unobtanium parts inside, which don't seem to be particularly prone to failure. Quite easy to troubleshoot and fix.

That said, prices on eBay these days seem completely unrealistic, although asking prices for test gear have gone up massively in the last 18 months. In your case, maybe $100 or so would be worth a punt. If the instrument has the option 001 high-accuracy reference (if it does there will be 4 BNCs on the rear panel, instead of 2) you could probably rip that out and sell it on its own for $100 anyway, even if the rest of the instrument is a write-off.

However, this is a function generator, not a signal generator. For your purposes it may not be the ideal choice of instrument. Phase noise/jitter is on the high side, and the output attenuator is not particularly wonderful. It was introduced in 1994, replacing the 3312A, a traditional analogue circuit based function generator, but adding a basic ARB function. I got my employer at the time to buy one almost as soon as it came out, as we needed a convenient source of pulse trains with programmable delays and repetition rates to simulate a sensor I was working on.

Thank you for this post. The 33120A ony goes to 15mhz. That's all I needed to get my Hickok 288X on the button. But the Hickok won't go down in the tape recorder bias oscillator range. But above 100khz or so, the Hickok serves me well. I still think if i could pick up a decent 33120a, I'd have all I need for my purposes. I'm simply not going to pay those crazy ebay prices. and I'm not going to ask my buddy to give me his. That's tacky. Even though he's an 81 year old packrat with 5 of everything he will never use.