Anyone got any mileage, manuals, rumors, etc. of the Wavetek 273 signal generator?
A customer is testing a PLL that I designed. The PLL is designed to go into the far sub-Hz bandwidths, and they're having trouble getting it to lock onto the 15kHz signal from a Wavetek 273. I've got it running off of a crystal oscillator divided down and made into a sine wave, and it's perking along quite happily at a bandwidth of 1/50th Hz.
They can't run it at a bandwidth below about 5Hz.
All I can find for that signal generator are places selling manuals, and a picture of the front panel. It has a digital readout and appears to be synthesized -- yet it's actual performance doesn't seem to jibe with the performance my customer is seeing.
So I'm wondering if its broken, or if it's just a multivibrator oscillator behind a dolled-up digital panel, or what.
Maybe it has so much jitter that your loop can't lock to it? Sometimes you have to have a wider acquisition BW, then ease into narrow. ...Jim Thompson
--
| James E.Thompson, CTO | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |
I love to cook with wine. Sometimes I even put it in the food.
It has so much low-frequency jitter (wander? -- call it what you will) that the loop locks, then after a while unlocks. The signal that they'll be locking onto is going to be very stable, so one would like to test the loop with a test source whose stability matches the ultimate one.
They're waiting to cut me a PO for more work until they've verified that it works, which is reasonable. I know darn well it works just fine, so I want to help them achieve their test goals!
What is the point of an extremely narrow bandwidth? What you need is a very high-Q VCO but with a bandwidth adequate to not lose the source. ...Jim Thompson
--
| James E.Thompson, CTO | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |
I love to cook with wine. Sometimes I even put it in the food.
Well, depending on how much things are corrupted by noise I suppose you're right. I may, indeed, be gilding the lily on this particular part of the system. But their test source is still miles less stable than their ultimate source, and it's causing me grief!
What are you using for a phase detector? ...Jim Thompson
--
| James E.Thompson, CTO | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |
I love to cook with wine. Sometimes I even put it in the food.
If the jitter/FM is on the order of 5 Hz out of 15 KHz, you should be able to see that on an oscilloscope, analog or digital. Decent analog scopes had timebase jitter of a part in 30,000 or so.
You could also zoom up the zero crossings and look for uglies there.
An old Wavetek could easily have that sort of jitter. Even a modern DDS generator can have period jitter in the ballpark of 1 part in
10,000 at low frequencies, where the output lowpass filter doesn't affect the DAC quantization much. Worse if the DAC doesn't have a lot of bits. We do interpolation in our DDS signal generators, which helps some, but not everybody does that.
if you are doing the phase comparison ON the output frequency, i.e. there is no division, then the "Dynamic range" of phase jitter that the loop can tolerate without cycle slip is the same as the linear range of the phase detector itself which is typicaly 180 to 360 deg. If the jitter execeeds this value, the loop will cycle slip and no longer be "phase locked"
The way to improve this is to divide the variable and ref frequency before they are phase compared. For example if you divide both by 10, then you still have a loop operating on the same frequency but the loop can tolerate
10 times the amount of phase jitter without cycle slipping. (Of course you have to modify the loop filter accordingly as well)
I came across this problem first a long time ago when designing a PLL FM transmitter. We wanted the VCO to be FM'd with the audio but want the PLL to keep the average frequency correct. In order to get this to work correctly, we had to divide the comparison frequency down to so that at the max freq deviation at the lowest audio freq, the phase change at the phase detector was within it's dymaic range.
$3 Digikey packaged crystal oscillators have typical jitters of a few ns per second... parts in 1e9, or 1e8 for a really bad one. Tempcos are usually a fraction of a PPM per degree C. They are usually within
10 PPM of advertised frequency, at least initially.
A decent SC-cut OCXO has one-second jitter of a few parts in 1e12.
I'm not talking about acquisition, that works just fine (it starts with a discriminator, then switches to a wide-open PI filter once the phase stops hopping around, then when the phase settles in it stops down the gain to whatever the preset value is).
The problem is holding lock -- they can't get it to stay locked unless the bandwidth is above one or two Hz, while mine stays locked just fine on my 12.5kHz crystal derived signal. Since this is very symptomatic of the signal generator not being in good shape, I'm thinking in that direction.
I suppose they could have issues with their processor crystal -- but it'd have to be exceedingly sick.
--
Tim Wescott
Control system and signal processing consulting
www.wescottdesign.com
I've been searching on eBay for a function generator and I will soon = receive=20 an HP3312A which I bought for $80. In the process I located and = downloaded=20 quite a few manuals, but I did not come across the 273. I found manuals = on=20 the 275 and the 278 =
formatting link
and perhaps the 273 is similar.
Another resource that has quite a few manuals, including the 275, is:=20
formatting link
But not the 273.
As you probably found, there are manuals for sale, such as this one for = $75:
formatting link
or here for $55 (or $25/week rental):
formatting link
5
I found three of these for sale on eBay, as low as $399:
formatting link
I don't have much experience with synthesized waveform generators, = except=20 long ago the company I worked for (EIL) made a frequency test set for = line=20 voltage frequency relays (50-60 Hz and 400 Hz), where the waveform was=20 synthesized using a D/A converter and a binary counter. But IIRC it used =
16=20 pots which were set to the sine function and CMOS multiplexers were used = to=20 produce the output. This was around 1980, and perhaps EPROMs were = relatively=20 new. Now I am building a waveform synthesizer with three-phase output, = using=20 a PIC and a USB connection for selecting the waveform shape, frequency,=20 amplitude, phase, and duration.
Using such a method it is theoretically possible to produce a waveform = of=20 arbitrarily low frequency, by using step-wise adjustment of output to=20 whatever precision is desired, although most DACs are limited to 14 or =
16=20 bits. For your purpose of phase locking, I wonder if it is a problem of=20 sensing the zero crossing or a particular signal level. My experience = with=20 phase angle meters showed that waveform distortion of sine waves made it =
difficult to get accurate readings, and symmetry and DC offset were=20 particularly critical.
I have had some experience with the 74HC4046 PLL, but mostly at power = line=20 frequencies trying to read 60.000 Hz with a 1000:1 multiplier. I do = recall=20 some problems with stability and noise, but this was also almost 30 = years=20 ago.
No wavtek 273 knowledge, Sorry. Does your customer have another signal generator? I was comparing different digital signal generators, triggering the 'scope off one DSG and watching the square wave crossing of the other. This cheap Protek DSG would have a hic- up, every second or two. Maybe they could check out the 273 with a similar method.
The Wavetek 273 appears to date to about 1980. So, it is about 30 years old, maybe more. It would be no surprise that the power supply filter caps are shot, and the whole instrument is running with lots of ripple. It may also be that some tantalum caps have croaked somewhere in it. That is about the vintage where the sick tantalum syndrome shows up.
That's about the vintage where tantalum caps were widely regarded as the Coolest Caps in the World. Given that the 1980s was the period when I was coming of age as an engineer, the whole "tantalum cap == good thing" mantra is stuck in my brain; I have to regularly dislodge it.
After seeing what a failed tantalum cap did to a board in a graphics display system, and hearing a few other horror stories, I avoid them like the plague. I have had to replace many of them in various systems, especially test gear that is used for a few months and then put in storage until the next time it is needed. That is a disaster for the old Tantalums, that intermittent usage.
A friend overhauled a Schlumberger communications analyzer, and it had dozens of failed tantalums in it. He just replaced them all.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.