Budget?  Freq range?

HP 8657As are good general-purpose SGs.  Reliable, serviceable, and
reasonably clean.  Avoid 8656s and 8660s IMHO unless you can get one for
next to nothing.

If you don't mind getting your hands dirty there have been some inexpensive
8662As on the surplus market lately.  They are cleaner than most, have an
actual tuning knob, and include sweep capability.  They are complex,
sometimes maintenance-intensive, and very hard to ship safely, and many of
the ones you find on eBay have lived hard lives.

-- john, KE5FX

Thanks John

With regard to frequency and price, low frequency is fine for me and I
am hoping to buy something for < $250-Patrick

John Miles wrote:

I just bought an HP 3325A synthesizer/function generator that I really like
(for some things anyway) in about that price range.  What you get is
probably dependent on what YOU need though.  This thing is pretty limited
but this particular one has the high voltage option so the output goes up to
40 volts up to 1MHz.  It only goes to 20.99999999 Mhz (at up to 10V I
think) but for 99% of what I do that's just fine.  It has a 10MHz external
reference which I hook to either an Rb osc or a Thunderbolt depending on
what I'm doing.  It's pretty neat to see all the digits match between the
3325A and the 5335A counter.  Of course, that's being clocked by the same
oscillator too.

Also, I noticed on some signal generators, dropping the output down to a few
mV distorts the signal.  What I learned to do was send my signal out at
about 1 volt ((RMS or p2p, doesn't matter) and go through a step attenuator
to drop it back down so it comes out clean.

-Bob, N3XKB

On Fri, Jun 19, 2009 at 6:04 AM, Patrick optomatic@rogers.com wrote:

Hey Robert

Great tip about the attenuator.

I looked up some models on the internet and some look fairly expensive.
I know that I will always be injecting low voltage signals, do you think
it would be wise to buy a cheaper fixed attenuator, let's say 20dB?, and
then just depend on the variable rate that the signal generator?

Thanks-Patrick

Robert Darlington wrote:

No distortion at low levels will occur with any signal generator worth
owning.  Nothing from HP will do that.

-- john, KE5FX

We use a lot of 3325As in the lab at JPL (they used to have dozens of them at the deep space network, so there are lots of them around).  Some have the rear panel option that puts out a sine wave up to 60MHz, which is fairly convenient.
The settability with lots o' digits is nice (that's why they were used in DSN..).. Really a function generator with sine square triangle, and some modulation capability. You can also lock multiple 3325s together which is nice for generating things like I/Q signals with known errors in phase/amplitude, or with frequencies offset by 0.1 Hz (so it sweeps through all relative phases in 10 seconds), or clock signals with known skew (although something like an 8110 or it's newer brethren is nicer).
They're not super quiet

Depending on your needs, there might be better solutions.  For a bit more money($600), you can get the TAPR Vector Network Analyzer from TenTec, which has a calibrated signal generator mode.  In the few hundred dollar range, I've used a lot of eval boards from one source or another.  Analog Devices has a whole bunch of DDS eval boards that take an external reference (always important for timenuts use) and with a small amount of work, you can calibrate them. National Semi has a whole bunch of PLL eval boards, if you need GHz kinds of frequencies.

On 6/19/09 6:13 AM, "Robert Darlington" rdarlington@gmail.com wrote:

I just bought an HP 3325A synthesizer/function generator that I really like
(for some things anyway) in about that price range.  What you get is
probably dependent on what YOU need though.  This thing is pretty limited
but this particular one has the high voltage option so the output goes up to
40 volts up to 1MHz.  It only goes to 20.99999999 Mhz (at up to 10V I
think) but for 99% of what I do that's just fine.  It has a 10MHz external
reference which I hook to either an Rb osc or a Thunderbolt depending on
what I'm doing.  It's pretty neat to see all the digits match between the
3325A and the 5335A counter.  Of course, that's being clocked by the same
oscillator too.

Also, I noticed on some signal generators, dropping the output down to a few
mV distorts the signal.  What I learned to do was send my signal out at
about 1 volt ((RMS or p2p, doesn't matter) and go through a step attenuator
to drop it back down so it comes out clean.

-Bob, N3XKB

On Fri, Jun 19, 2009 at 6:04 AM, Patrick optomatic@rogers.com wrote:

time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

On 6/19/09 6:38 AM, "optomatic@rogers.com" optomatic@rogers.com wrote:

How precise does your level have to be? How stable?
There are surplus attenuators available all over the place, some variable
ones too.
MiniCircuits has VAT-nn attenuators which are relatively inexpensive (not by
"found it at a ham-fest 30 years ago" standards, though)

Building your own attenuator using chip resistors is another possibility.

Hi James

Thanks for your feedback on this.

My needs are probably very privative compared to those of the people on
this list. I service a lot of the time without a schematic so I spend a
lot of time figuring out how the circuit works. I was thinking that if I
injected a signal with a known waveform I could follow it around the
amplification circuits and such. My only real need is to create
something that does not appear to already be there. For instance I don't
see many triangle waves, if I produced one I could following it around
with my oscilloscope. So in terms of frequency just a few hertz would
do, heck even 1 would probably be fine. My only concern is that some of
the circuits are high impedance and have low voltages. It might be a
good idea if I could get down somewhere into the uV range.

Thanks again!-Patrick

Lux, James P wrote:

I took a look at the phase noise of a 3325B and as Jim said, the phase
noise is not great -- about -60 dBc at 1 Hz offset.  An external
reference cleans up the close in (<10 Hz) signal, but beyond about 10 Hz
the synthesizer noice drowns out the reference; from the noise from 10
Hz to about 3 kHz is about -115 dBc, then it drops off to a floor of
around -135 dBc at 100 kHz.  Some plots are at
http://www.febo.com/pages/hp3325b/

That said, they are awfully nice boxes for audio and low HF use.

John

Lux, James P wrote:

Patrick,
Using a large fixed attenuator should work fine.  I learned the attenuator
trick when dealing with some variable gain amp chips from Analog.  Instead
of varying the gain, they run at some fixed level that produces a clean
output, and vary the attenuation level to prevent non-linear effects and
distortion.  HP might be doing this, then again, maybe not.  Some of my
instruments are very sensitive to input overloading and I just got in the
habit of using a step attenuator between the signal source and the VNA or
spectrum analyzer, or what have you.  It helps to prevent stupid mistakes
that seem to happen at all the wrong times.

As John Miles said, no HP signal generator should produce a distorted output
at low level.  My Tektronix does!

-Bob

On Fri, Jun 19, 2009 at 7:38 AM, optomatic@rogers.com wrote:

Robert Darlington wrote:

It depends on how they lower the signal level.  If the signal level is
lowered using a traditional attenuator, the distortion will be unchanged,
but if the signal level is lowered by biasing a pin diode attenuator, or
by changing the bias on a gain stage, it will change the distortion all
over the place.

Basically, if there is a pot to adjust the signal level, you can
expect the distortion to change too.  If there is a selector switch
like mechanism, the distortion will probably not be affected.

-Chuck Harris

Hi Patrick:

There are a bunch of signal sources that could be used.

The rack sized HP signal generators that weigh 50+ pounds were designed to have
very low phase noise and although now many years old are still very good for that.

The HP 3325() is the only signal generator that has adjustable amplitude that's
calibrated to a small fraction of a dB.  They still are the specified
instrument for amplitude calibration of many other pieces of lab equipment and
there is no newer instrument to replace them.  They don't have good specs in
the frequency domain.

The HP 8648() series are reasonably small and light synthesized signal
generators.  I got the 8648A Option 1EP which is specific for pager testing and
includes many modulation enhancements. The prior versions of the 8648 did not
have good enough specs to test pagers.
http://www.prc68.com/I/HP8648.shtml

A problem with injecting a signal into operating equipment is you may burn out
the signal generator.  For example just touching a DC point feeds a step change
back into the sig gen.  So it's good to have a series resistor and blocking cap
to protect the sig gen.

You might be able to use a probe that generates something like narrow pulses at
an audio rate as a universal signal source.

Have Fun,

Brooke Clarke
http://www.prc68.com

Patrick wrote:

Patrick wrote:

The HP 3336 (A,B,C) is similar to the 3325 but somewhat lesser
capabilities and can be a bit cheaper.  There is one on eBay "in working
condition" for $150.

Frequency range is 10 Hz to 21 MHz.
Frequency is settable to uHz or mHz steps and it has frequency sweep
capability.
The output attenuator has 0.01dB steps and the output goes down to about
-71 dBm.
It is sine wave output only.  If you want square or triangle, this won't
do.  If you need to go below about 10 Hz you will need a function
generator.

My impression from your description is that you do not need the
capabilities of a synthesized signal generator.  Perhaps an old analog
test oscillator such as the HP 654A would be adequate.  This is just an
analog oscillator combined with a step attenuator that goes down to
about -90dBm in 1 dB steps.  There is one on eBay for $85 OBO, "tested".
Or, for function generator capabilities (square, triangle, lower
frequencies) an HP 3310A with an external attenuator.  RF attenuators
can be somewhat expensive, but audio ones are usually much less.
I'll let others comment on whether anything that old might have too many
problems.

One advantage of the old analog stuff is that they are acoustically
quiet (no fan).

73,
Gary