Agilent 33120A / Option 001 Phase-Lock + external 10mhz standard

[linux-works]

does anyone have any info on whether this option is worth anything?

what its supposed to do is to lock onto your reference 10mhz (in my case, a rubidium freq std) and use that when synthesizing user-set frequencies.

what I'm wondering is: for $250 or so (used), is this a worthwhile upgrade to the 33120 that does not have opt 001?  reading the specs, it seems that the PLL will keep the freq to 1ppm, but the Rb standard is far better than that.  maybe having a 10mhz standard that is better than 1ppm is wasted on this config?

there is a local XO can at 40mhz.  its not ovenized, etc; and so will this end up being my weakest link in the stability/accuracy test?

[Bored@Work]

Please, it is MHz, not mhz.

[linux-works]

in case you didn't notice, I type most of my words in lowercase (the carpal tunnel makes hitting shift harder than it should be and I default to all lowercase except for 'I', which is one of the only exceptions when I type).

that said, your post sure did contribute a lot to my question (sigh).

[ve7xen]

I could be wrong, but reading the specsheet it looks like Option 001 includes a high stability timebase as well, which is what's specified at +/- 1ppm. The phase lock doesn't have a specification, but I believe a parts-per-x value for a PLL doesn't make a lot of sense, since the frequency error will be effectively zero, pretty much by the definition of phase locking. You'll always have some phase noise, which will contribute to the output, but it doesn't appear to be specified for this device.

If you're only concerned about frequency accuracy, the phase locking will ensure the output is locked to your standard. There should be no additional error contribution.

[Rigby]

in case you didn't notice, I type most of my words in lowercase (the carpal tunnel makes hitting shift harder than it should be and I default to all lowercase except for 'I', which is one of the only exceptions when I type).

that said, your post sure did contribute a lot to my question (sigh).

you must understand that everyone is allowed their idiosyncrasies.  just as you have carpal tunnel and find capitalization painful, others have whatever they have that makes them do whatever they do.   don't bother making a big deal of it, none of it matters.

[linux-works]

I could be wrong, but reading the specsheet it looks like Option 001 includes a high stability timebase as well, which is what's specified at +/- 1ppm.

ah, that makes sense, thanks.  I didn't realize this was a fallback for when the external 10mhz ceases.

[jpb]

Some time ago someone reported measurements on a Rigol DG4162 (I think) where they had used an external reference and got much better phase noise than with the internal reference. I've spent a little time trying to find the thread but can't, it was at least a year ago.

So having the ability to use an external reference is probably a plus from the phase noise point of view rather than absolute frequency accuracy. Given that, it might be better to use an OCXO rather than a Rubidium source.

[HighVoltage]

. Given that, it might be better to use an OCXO rather than a Rubidium source.

Why is that OCXO better than a Rubidium source?

[VK5RC]

The accuracy will probably mostly depend on the time base the generator is connected to, for short term "accuracy" which is often termed "phase noise" a good  ovenised  crystal will beat many other time base sources, these can be purchased for ~AUD $150. The longer term accuracy (over 100s of seconds or more) GPS disciplined oscillators will do better. Whether you need to have that accuracy in frequency output is the real question, if you do need that sort of accuracy is it worth considering another sig gen that has an external ref input?
Recently I installed in one of my transmitters Icom 706MK2G (for amateur radio) a small circuit that allowed you to lock the radio (which normally has a temperature compensated crystal) to any good 10MHz reference. Interestingly the results showed much greater frequency accuracy but not as good as the external time base. Some long term (over minutes) errors in frequency generated were noted and were consistent in the error, probably due to 'errors' in the maths of the phase locked loop, these were 50 or so mHz (small m Hz) when generating a 144MHz signal.
The point is if you need real accuracy you will need to look at all elements from the time base to the synthesiser/amplifier.
If you are not aware the time-nuts is a great resource for time frequency information (pretty serious though)
Dr Frank and G0HZU are pretty hot on this topic.
Re original question, an additional $250 seems a lot as an add on, I think it may be better to source something that already has ext ref in as standard.
Re OCXO vs Rubidium the mechanism of generating the 10MHz frequency in the rubidium produces quite a bit more phase noise (can be +10-20dB in the 10-100Hz away from carrier region) than a OCXO. Long term (seconds) the Rubidium will win out, very long term GPSDO.
Also see .https://www.eevblog.com/forum/testgear/my-venture-into-the-rubidium-domain/msg412067/#msg412067

[linux-works]

the 33120 is the only FG I have that -can- take an external 10mhz input.  all the rest that I have cant' do this.  the 33120 was not expensive as a base model (used) and so $200 more is not going to break the bank.  looking at the circuit, its certainly not $200 worth but the functionality it gives is nice to have.  I think I paid $300 or in that range, for the 33120.  hard to find a digital dds FG of that (reasonable) quality for that price.

(the opt 001 can be bought for about $200 if you find the right deal)

its possible I could go with an external crystal oven osc later on, but the first thing is to modify the FG so that it -can- sync to external clocks.  I have a freq counter (a few, actually) that can do this and I'd like my FG to be able to, as well.

[VK5RC]

Sorry I misunderstood your OP re the 33120, I thought you were considering buying it as well.
The product used to replace a crystal in some amateur radio transmitters can be found on this website
http://www.vk3hz.net/XRef/XRef_Home.html
The phase noise is not fantastic but OK.
I don't know whether it can be converted to your needs, and the price is getting close to the 001 option, ( I recall it being cheaper!).
In this circumstance it may be nicer to stay with all HP gear (and go with the 001 option), rather than modify too much for only a little saving and potential to degrade the 33120's performance.

[linux-works]

yes, I already have the 33120 and while its not the ultimate FG (I hate that triangles top out at 100khz while sine/sq can go to a full 15mhz) its clean enough and remotely controllable (one of the big reasons I went with this one) that I might as well just upgrade it if its 'only' a $200 cost option.

and while that's not a small amount of money, options for DMMs and such often are outrageous.  gpib interfaces can sometimes be very expensive as add-ons, etc.  finally, when its time to sell this unit, having the opt001 installed sure will add extra worth.

anyway, my main concern is resolved; I was not clear on what the XO module was doing and if the external 10mhz signal was being degraded by the local XO or not.  it seems that the XO is not used if the 10mhz signal is present and so that relieves most of my concern.

[Bored@Work]

in case you didn't notice, I type most of my words in lowercase (the carpal tunnel makes hitting shift harder than it should be and I default to all lowercase except for 'I', which is one of the only exceptions when I type).

that said, your post sure did contribute a lot to my question (sigh).

In other words, you choose to behave like an idiot. Hint: Almost the only thing a full carpal tunnel syndrome doesn't hinder is typing. So next time, come up with a better excuse for your sloppiness.

[Rigby]

In other words, you choose to behave like an idiot. Hint: Almost the only thing a full carpal tunnel syndrome doesn't hinder is typing. So next time, come up with a better excuse for your sloppiness.

Let it go, dude.  You're the one nagging him, not the other way around.  Don't act so indignant when you're the one stirring shit up.

[electronic_eel]

I'm sure there are schematics for the pll-option of the 33120A available in the service manual. Take a look at them. Maybe it is just simple stuff on there so you could make a board yourself for far less than the $200.

[linux-works]

its possible I could make it.  I do have the schematic since the service manual is available online.  but if I plan to sell this later on (maybe, who knows) then it adds value to have proper boards that are known to work and be reliable vs a home made hack.

it could make a few days to build this board, start to finish.  for that effort, its probably just more expedient to buy the damned thing

if it was $1k, I would not consider it.  but its in the 'expensive but not impossible to justify' price range.  I don't know what they go for new, though.

[jpb]

  I don't know what they go for new, though.

The official Agilent web store have been trying to sell one (with the option) for $1,299 :

http://www.ebay.co.uk/itm/Agilent-33210A-Function-Arbitrary-Waveform-Generator-10-MHz-incl-Opt-001-/190885490556?pt=LH_DefaultDomain_0&hash=item2c71a96f7c

But they have had it up for months (I'm a regular visitor to the web-store) and I doubt they'll sell it at anywhere near that price. The reason the price is so high is that when Agilent brings out new series (in this case the 33500 series and then the 33600) they seem to have a policy of making the new, better specced model cheaper than the old worse specced model to encourage customers to move to the new tech and to make a good profit from those customers who have to have the old model to fit in with their existing setups.

The problem though is that with second hand models they are not bought by big companies who buy the new (old) models at inflated prices and then the prices are too high to be bought by anyone else so they just sit on the Agilent store.

So I think $1299 is grossly overpriced but I think you could sell yours with the option for probably around half that say $600 or $700 but that is just a guess on my part.

[linux-works]

that's the 33210.  I have the 120   at least 1 gen apart (maybe more).

[echen1024]

that's the 33210.  I have the 120   at least 1 gen apart (maybe more).

210 is one gen apart I think. I'm not sure if OPT 001 is 1 ppm stability, as the manual tells me 2 ppm. The ability to sync up to an external 10MHz is helpful though. My 33120a doesn't have this option installed.

[jpb]

that's the 33210.  I have the 120   at least 1 gen apart (maybe more).

Sorry - I do too many crosswords and think in anagrams, or perhaps am a bit dyslexic.

Remembering Agilent product numbers can be difficult - they would make life a lot easier if they used a combination of letters and numbers, e.g. Rigol uses DG for digital generator and DS for digital 'scope though the latter should be DO for digital oscilloscope.

I suppose there is logic to some of the digits, the third being the generation number and the fourth being the number of ports.

[HighVoltage]

I have one Agilent 33210A and one 33220A

The 33210A start the serial number with "43"
This one had lots of problems and I would never buy it again.
(I still have a broken one laying around, awaiting repair)

The 33220A that I have, starts the serial number with "44" and is a completely new generation and should have been called 33220B
But for some weird reasons, Agilent kept the"A"
This 33220A has the option 001 installed and I highly recommend it.

[linux-works]

I read (on this forum, too, I think) that the 33120 has a problem when taking in the external 10mhz signal.  someone was saying something about it having to be 18hz (?) or some other number off from the 10mhz just to be able to create a 10mhz output on the front panel?  anything like that ring a bell with anyone?

[jpb]

I read (on this forum, too, I think) that the 33120 has a problem when taking in the external 10mhz signal.  someone was saying something about it having to be 18hz (?) or some other number off from the 10mhz just to be able to create a 10mhz output on the front panel?  anything like that ring a bell with anyone?

I've not heard that, but looking at the data sheet the lock range is +/-50Hz while the accuracy of the internal crystal is 20ppm in one year which is 200Hz so if the internal crystal is out by more than 50Hz does it still achieve lock?

EDIT : I realise that this is why the external reference option comes with a TCXO as well, just to give it a chance of being within lock range. The spec. for the TCXO is 2ppm for the first 30days but then 0.1ppm/month ageing so it could be another 1ppm/year so if it wasn't calibrated for 3 years it might be out by 50Hz and not lock.

[linux-works]

81hz, not 18hz

here's the link:

https://www.eevblog.com/forum/testgear/testing-the-hp-5315b-with-a-rubidium-frequency-standard/msg141328/#msg141328

does that make sense that you'd have to set the 10mhz standard HIGHER just to get 10mhz out of it?

or maybe the poster was just not very clear.  maybe there was amplitude or shaping problems using a weak 10mhz signal.  I don't want to buy opt001 and then find that it can't do what I'm hoping it will do.

[jpb]

81hz, not 18hz

here's the link:

https://www.eevblog.com/forum/testgear/testing-the-hp-5315b-with-a-rubidium-frequency-standard/msg141328/#msg141328

does that make sense that you'd have to set the 10mhz standard HIGHER just to get 10mhz out of it?

or maybe the poster was just not very clear.  maybe there was amplitude or shaping problems using a weak 10mhz signal.  I don't want to buy opt001 and then find that it can't do what I'm hoping it will do.

Looking at that thread and looking at the service manual I think that there was nothing wrong with their 33120A s and they were confused about the frequency calibration.

The 81Hz out on 10MHz is much less than the spec of 200Hz for the standard instrument (without the 2ppm option with the frequency locking). The person who did have frequency locking mentioned that their instrument was also out but didn't say by how much.

They then tried to calibrate it by changing values by 1Hz BUT the calibration frequency (from the manual) that is generated by the 33120A is 1 kHz NOT 10MHz so a change of 1Hz is 1 part in a thousand which is very large so I'm not surprised that they got an out of range error message.

The person with the frequency locking option stated that they could lock easily.

So I'd say the option works and given that it also includes a TCXO is a good upgrade and worth getting for $200.