Problem with HP 3312A Function generator

[jgalak]

Complete newbie here, so quite possibly this is a PEBCAK error.  Just got an HP 3312A function generator off ebay.  Tried hooking it up to my scope.  Scope doesn't have 50 termination, so I used a BNC Tee with one leg going to the 3312A, one leg to scope, and one leg to terminator.  I think that should be ok?

Settings of the 3312A are visible in the photo.  1kHz range, freq dial midway, 10 Vpp, sine output, no modulation.  The resulting waveform is weird.  See attached.

Playing with the nobs, offset and amplitude controls seem to do the expected.  Sym. makes tiny changes in the "width" of the waveform, not much else.  Trigger Phase seems to do nothing.  Most damning, the Freq dial seems to only have a small effect in the range of about 1 to 3, and then no effect above that.

So.  Bad unit?  Incompetent user? It's within the return period, I can send it back if needed.  Or try to fix it?  It came with service manual, which is nice, but I'm not sure I'm really qualified to try to troubleshoot this....

EDIT: Oops, failed to attach photos.

[Paul Moir]

Even without pictures, it sounds broken.  For this termination is not going to matter.

"Sym" is "Symmetry".  It only works when the blue button in the centre of the knob is popped out (push it).  Frequency should adjust .1-13 so 130x change across the span.

Trigger phase - leave that off for now.  It depends on how you have the switches set on the back and how the 2nd channel (the modulator) is configured.

EDIT:  shut the unit off and exercise all the switches and knobs to see if it makes any difference.  Ensure the line voltage is set correctly.

[jgalak]

Line voltage set correctly for 110V.  The other switch in back, trigger, tried on both single and multiple, no effect.

Will try to exercise the buttons and knobs in the morning.

[jgalak]

Also, note frequency of 54 Hz when generator is in the 1 KHz range....

[VK5RC]

My 3312 is a bit twitchy, if I jump frequency range by more than one decade at a time I occasionally get no (or very low) output - a dodgy gang switch. If I double push the button if usually works

[bd139]

Push the blue button in and out on sym on the primary generator, left side. Do this quite hard a few times. Then set the frequency control to 10 and then exercise all the range switches a few times. Sounds like it's a bit claggy after storage. If this fixes it, buy some proper switch cleaner, open the top half of the enclosure and give the range switches a squirt then exercise them and it'll be as good as new.

[jgalak]

Push the blue button in and out on sym on the primary generator, left side. Do this quite hard a few times. Then set the frequency control to 10 and then exercise all the range switches a few times. Sounds like it's a bit claggy after storage. If this fixes it, buy some proper switch cleaner, open the top half of the enclosure and give the range switches a squirt then exercise them and it'll be as good as new.

That (mostly) did it.  Now I understand what the Sym control does.  Analog freq dial obviously isn't calibrated, but it's reasonably close.

It's now working over half of the ranges.  Unfortunately, the lowest 4 frequency ranges (.1 - 100 Hz) still don't work.  I just get a flatline when choosing those.  If exercising them some more doesn't fix it, I'll probably grab some contact cleaner and open it up just to try that.

[bd139]

Dial is only specified within 5% which is normal for function generators. You can calibrate the dial by setting it mid-range, measuring the output frequency (do this from the sync output!) and undoing the grub screw and then setting it to the right point. Should be spot on after that. Mine was definitely better than 5%.

Keep at the low range switches for 30 seconds each. That sometimes works too. If not, contact cleaner. Definitely should work though if it has got that far.

[jgalak]

One other thing.  At the highest frequencies, I'm getting some roll off in voltage.  I'm assuming this is normal?  With aptitude set to 1, I get 1.6-1.8 Vpp from, about 40 kHz to about 1MHz.  At 5 MHz it's 1.13.  At 8 it's 1.1.  At 10, it's 1.04.  At 14.6 MHz, the highest the dial goes, the Vpp is .792 V.

Is this just hitting the 3dB bandwidth limit of the generator?

[bd139]

How are you connecting to and measuring it?

The generator is flat to 12MHz.

[jgalak]

How are you connecting to and measuring it?

The generator is flat to 12MHz.

It's a little cludgy.  BNC cable from generator into a BNC tee.  One leg of tee has a 50 terminator.  The final leg goes to scope, via a bnc-to-sma and then sma-to-bnc adapter.  Didn't have correct gender Tee or a BNC-BNC adapter.

I have a 50 through terminator on order.

[bd139]

Ok should be fine still if properly terminated. Try ramming a 10x probe in the output as well and see if that has a difference. That should isolate it to the measurement or the instrument.

[jgalak]

Good news/bad news.   Good news is that contact cleaner took care of the frequency problem entirely.  Every range now works.  And wow, .01Hz takes a long time to trace out on the scope....

However, the high freq rolloff is still there.  I tried sticking a 10x probe directly into the output jack, and I tried using a pass-through terminator (no more cludge).  Same result.  Fine up to about 6MHz, and starts to roll off around 6-7MHz, ends up about 3dB down by 14MHz.

[bd139]

Can you try measuring the rise time of the square wave output at full amplitude and frequency. That should give some clues.

[jgalak]

At 1 MHz (before rolloff) and max amplitude (11.4Vpp), the rise time, as measured by the scope, is about 15.9ns (bouncing between about 15.6 and 16.2).

At max freq, 14.28 MHz, the square wave hardly looks square (see image), and the rise time is reported by the scope as about 17ns (bouncing between about 16.6 and 17.4).

The deformation of the wave from square to this lumpy thing starts at the same time as the rolloff in amplitude starts.

[bd139]

Ok that scope has enough bandwidth to show a better representation of the square than that (rise time is about 3.5ns on the 1054Z). It won't be that good but rise time should be around 10ns. Worth running through the calibration steps in the manual and seeing if there is a problem with frequency compensation somewhere next.

[jgalak]

Ok that scope has enough bandwidth to show a better representation of the square than that (rise time is about 3.5ns on the 1054Z). It won't be that good but rise time should be around 10ns. Worth running through the calibration steps in the manual and seeing if there is a problem with frequency compensation somewhere next.

Do you mean the scope's calibration steps or the generator's?  I ran the scopes auto-calibration a few weeks ago, when I first got it, so I assume it's good.

I'll take a look at the generator's calibration instructions, but frankly, a) they are kind of intimidating and b) I don't know that I have all the equipment they call for.  But hey, let's go learn something

[jgalak]

Here's what the square wave looks like in several steps from 1MHz to 14 MHz.  In all cases the vertical scale was kept the same, but the horizontal scale was adjusted to keep several waveforms on screen.

Vpp and Rise time can be seen on bottom, freq can be seen in top right.

Due to forum limitations, posting this as 2 separate messages.

[jgalak]

And the rest of the series:

[bd139]

Ok that's not looking too bad there. You can see the RC response at the higher frequencies.

You're not going to get a perfect square at maximum frequency on one of these. Even my 6 month old DG1022Z for example will only kick out a good square up to 5MHz as, due to Fourier's theorem (all squares are composed of sines at different frequencies), the thing would have to have 7-11x the bandwidth of the maximum frequency output to reach it (275MHz).

Here's a 25MHz square wave out on the DG1022Z connected to my DS1054Z as an example. This has twice the bandwidth of the HP3312A so I've shown it running at twice the speed:



I just referred back to the service manual instead of my brain as that's more reliable. The frequency response check (sine flatness check) is defined as follows so this is worth checking first. I've simplified this a bit:

1. Set output to sine, max amplitude, zero offset. Connect to scope via 50 ohm termination (your tee jobby should be ok).
2. Set frequency to 1KHz and measure p-p amplitude.
3. Set frequency to 100kHz and measure p-p amplitude. This should be within 3% of the reference amplitude from (2)
4. Set frequency to 10MHz and measure p-p amplitude. This should be within 10% of the reference amplitude from (2)

If this fails, something is up with either the measurement set up or the output amplifier. I'd suggest changing the measurement set up first. If it passes, you're good.

Also worth checking the output impedance:

1. Set output to sine, max amplitude, zero offset, 1KHz. Connect scope via straight cable.
2. Measure p-p amplitude on scope.
3. Attach 50 ohm terminator T/inline.
4. Mesaure p-p amplitude on scope. This should be approximately half of (2).

If that passes, it's probably ok. If it doesn't someone buggered up the output attenuator resistors. This is a common problem on these when previously owned by idiots. Look for burned ones on the board - they're pretty obvious and they cause most of the reduced output problems themselves:



Got to shoot out now for a few hours but will reply when I get a few minutes.

[jgalak]

You're not going to get a perfect square at maximum frequency on one of these. Even my 6 month old DG1022Z for example will only kick out a good square up to 5MHz as, due to Fourier's theorem (all squares are composed of sines at different frequencies), the thing would have to have 7-11x the bandwidth of the maximum frequency output to reach it (275MHz).

Yes, I'd forgotten that.  Certainly makes sense.  So shape of square not an issue.  Amplitude roll off still is, maybe.

1. Set output to sine, max amplitude, zero offset. Connect to scope via 50 ohm termination (your tee jobby should be ok).
2. Set frequency to 1KHz and measure p-p amplitude.
3. Set frequency to 100kHz and measure p-p amplitude. This should be within 3% of the reference amplitude from (2)
4. Set frequency to 10MHz and measure p-p amplitude. This should be within 10% of the reference amplitude from (2)

Ok, trying that (got an actual pass through terminator now).

@1kHz, 11.5 Vpp
@100kHz, 11.6Vpp
@ 10MHz, 10.3Vpp.

So no drop at 100kHz at all (I'm assuming the increase is measurement error), and 10.4% drop at 10MHz.  Given that I'm only measuring 3 significant digits on the scope, I'm going to call that within tolerance.

Also worth checking the output impedance:

1. Set output to sine, max amplitude, zero offset, 1KHz. Connect scope via straight cable.
2. Measure p-p amplitude on scope.
3. Attach 50 ohm terminator T/inline.
4. Mesaure p-p amplitude on scope. This should be approximately half of (2).

@1KHz, no terminator, Vpp is 23.6V.
With terminator, Vpp is 11.8V. 

So exactly half, at this measurement resolution.

[bd139]

Back again. That's pretty close so looks fine to me then. Think you got a winner 

[Paul Moir]

That looks pretty consistent with my 3312A.  I'll try to get some pictures up in the next couple days.  If you need a cleaner square wave, the one coming from the SYNC output is quite a lot faster.

[Paul Moir]

1MHz, 10MHz and 14MHz.

The Bell National Research tag  on the side says it's not calibrated and requires verification though.