Making A HP Frequency Counter More Accurate

[Gerry] built his own high stability timebase add-on for his HP 53131 frequency counter. This project started out after [Gerry] built a rubidium 10 MHz standard for his lab. Upon connecting the standard to the frequency counter for calibration, he found that the HP 53131 had an awful internal oscillator. The official high stability timebase add-on from HP cost about $1000, and he was determined to do better.

Using a second hand OCXO as the oscillator, he designed his own add-on module. OCXO modules pack a crystal oscillator in a thermal chamber. Since temperature fluctuation causes drift in crystal oscillators, an OCXO controls the temperature to keep the frequency constant. They can be bought second hand on eBay for under $30.

The PCB design for the module can accommodate a variety of OCXO modules. It uses a high speed comparator and a high stability 5 volt reference to provide the clock signal to the counter. A DAC is used to calibrate the oscillator. By keeping the same DAC as the original counter, the add-on board can be calibrated using the front panel of the device.

The project is a drop in replacement for HP’s $1000 module for a fraction of the cost. [Gerry]’s write up has all the details you’ll need to build your own.

38 thoughts on “Making A HP Frequency Counter More Accurate

    1. Well oven controlled crystal oscillators are accurate to something like 1 part per billion, so it’s not bad, and it’s a DEFINITE improvement over the internal oscillator in the frequency counter.

        1. @oodain – the calibration only occurs when you run a cal so it does need to be done every now and then, perhaps yearly. However, the swing provided on the cal voltage is 12-bits worth of DAC which in my case at least can adjust the oscillator by +/-0.3ppm which is a very very fine adjustment. I have not done the math but any drift in the DAC and 5V reference will have a minute effect on the frequency drift. The OCXO is really very good – I posted the data sheet in a comment below.

          1. thanks for the great info,
            amazing project, just like your rubidium standard.

            doing the calibration once in a while is a small price to pay for any precision instrument.

          2. You are most welcome. Even without calibration we are still only talking less than 1Hz error, so when you compare that to the timebase shipped with the counter its supremely better, I think thats probably the point some people miss. I am really pleased with the outcome, I powered the meter up from cold this morning and within 15 minutes it was reading 10,000,000.00 which is within 100th of 1 cycle. :)

        1. Tony,

          I guess the question is, do I need it to be certified…which I don’t think I do. Mine is (will be) calibrated against a GPSDO which should get me pretty close (not that its not already very close). How long the RbFS will last is of course another question, they are essentially end of their recommended life – thats why they are $60 on e-bay. Still basing a quality or fit for purpose judgement on price alone is flawed thinking IMHO, this to some degree is like saying the $150 Monster HDMI cable is going to give you better picture quality than the $10 HDMI cable from e-bay, which of course we know is not the case.

          Gerry

          1. I think that you’ve missed the point he was trying to make.

            the reason that they are $30 on ebay is because they are now only really useful as curiosities, they have no place in business or industry, they are old and past it, out of life. can’t be guaranteed as accurate. that’s why they are only $30

            really cool project, but it’s kind of like me setting my watch by asking a mate what the time is.

          2. @Dan, yes I would agree with that basic point but I think you and others may be missing my motivation here. I am not in the calibration business neither am I in industry where I am trying to use this hack as a traceable thing. I am an electronics hacker and the project is on my blog and, for what its worth this site which is called “HACK-a-DAY”, I would find it highly surprising to find anything used in industry that derived from a hack including anything that I do myself.

            For calibrating my own stuff and having a reasonable level of confidence that I am somewhere near the mark is what I was trying to achieve for myself – and I shared what I have done for others who may also find that level of dependability acceptable.

            No one said you should hack a piece of kit you use in industry, and I certainly did not suggest its an alternative to a properly traceable calibration. That seems to be being impled by others which is entirely misplaced – by them!

            The fact though is, by doing this hack I have substantially improved the accuracy and stability of a piece of kit that HP (that world leader in test equipment for industry) had no issue selling to industry something that even as a hacker I find totally unacceptable – which for some purposes industry I am sure was adequate.

            I do find it surprising how aggressive some people are with there often misplaced opinions.

            Gerry

          3. I certainly didn’t mean to come across as aggressive.

            I guess the question is really, how did you decide which device was accurate when hacking the HP generator?

            like I said with setting my watch, if the only information you have is that he says that it’s nine past twelve and I say ten past and the only information that you have is that his watch was previously a very expensive one. which do you trust more?
            clearly either one has gained time or one has lost time, (due to inaccuracies in the clocks) but how do you know which?

          4. @Dan – In my case I assumed the RbFS is my baseline – but I am aware that will almost certainly need calibrating and for that I do have a Trimble Thunderbolt which has slightly better long term stability so a calibration of the RbFS against that should get me to where I want to be. Someone has already lent me another RbFS which has already been calibrated against a Thunderbolt GPSDO so it will be interesting to compare that with my $60 untouched one off e-bay.

            Gerry

          5. While they are old and cannot be guaranteed as being absolutely precise anymore, they are probably still quite accurate. There is a reason we have two words to describe two different characteristics. Although many do overlook the distinction.

          6. The difference is $150 Monster cable will actually work. Of course it’ll be as good as the $10 one but that’s not the point.

            Buying some random garbage off eBay because it used to be highly accurate & reliable is no guarantee that it still is – and the odds are it isn’t – why do you think it’s on eBay in the first place?

            Unless (as you think you do) have a known base you can compare it against, it’s like Dan said; you’re just setting your watch by asking a mate for the time.

            At least we know you’re highly sensitive, that’s a plus in some areas.

          7. @Tony – but there is no guarantee that it is not accurate and reliable either, you are making assumptions saying the “odds are it isn’t”, how would you know that without checking it, you must have some kind of extra sense that the rest of us don’t! Its very clear that these things are on e-bay because they have been recovered from used equipment and may even be end of life, but I did not see anything on the data sheet for the OCXO about expected life, just predictable ageing.

            So put your philosophy aside for a second and be objective. My position is this – the counter is more accurate now with the OCXO mod than it was before the mod. Its both more accurate and more stable than it was before by a large margin. If you cannot refute this then the rest of your argument is void I am afraid. Nowhere have I said that I was making something that could be used as a traceable reference for calibration, but the importance you seem to be placing on this would seem to suggest that I want to re-calibrate the world with it.

            The point is I suppose, I am not sensitive, I am more miffed as to why you would care what I think I have, your somewhat over concern with the potential inaccuracy of the frequency standard and counter I have in my home lab is disturbing, its not that important to me, in fact it would appear to be much more important to you – odd indeed.

            Gerry

    2. As of July 2022 an OCXO can be had new for $40 with an accuracy of 20 ppb, or 2/10^8. Doing a cal with rubidium then gets you to rubidium accuracy for a few months until the ocxo drifts a little. The crystal oscillators have excellent stability over days to weeks and are very good over months to years.

    1. @SYNTRONIKS – for what its worth, I have run the counter for 3 days/nights with a scope on the RFS and the output of the counter and seen no drift at all. The drift/stability of the counter is as good as the OCXO I used which is an Oscilloquarts OCXO 8663-XS, the data sheet for this is here: http://www.sungwhatech.com/product/pdf/aging/8663.pdf, this OCXO has excellent long-term stability specs so for a home lab I think its good enough – don’t you?

      Gerry

  1. Geez, talk about splitting the hairs from some of the people here. Look, I understand all of the philosophical issues that the comments support but the case in point is really about improving the home lab accuracy. I don’t see anyone complaining about dso bamdwidth hacks which are a level above this. Can you guarantee the scope does 200 mhz? With what source and counter?
    Geez, a nice hack and people still bitch and moan. Back to raspi python/.net projects, I guess.

    1. @one well said my friend. Time domain and accuracy stuff seems to really get people fired up for some reason. I am pretty chuffed that I managed to get a significant and practical upgrade to my counter for under $100 :) Oh well, as you say Pi here I come… Gerry

  2. Its interesting, but dude. You have like OCD or something. That HP universial counter is widely used in “the industry” – critical industry – and it’s “not good enough” for your home lab? no… you need a Rubidium standard…. For what?! It’s like having a ferrari – just to to go get groceries, but you know – I just had to have the turbo charger upgraded to get a few more ponys out of it. A lot of effort, a lot of time, zero gain.

    1. Well the original one was out by quite a lot, enough to make the counter undependable. Compare that to a 30 year old Racal counter which was much better. It was not so much out of cal, it was broken. Anyway, very nice balance to the view point here, others have been saying I did not go far enough :)

      1. I would contend that anyone who gets either a Ferrari or a Roller is not getting one on the basis of need (in any context) – “want” is a more apt verb than “need” in these cases, and to be able to afford something that you want is simply rewarding yourself for your efforts/success.

        Man this thread is getting tenuous…

        Gerry

  3. I see a lot of the times that some people here think cal and certification is magic. It isn’t, it just ‘certifies’ that the device has not gone out of specs, though between the checks it might be wildly out.
    It MIGHT be that with this addon the counter has deviated from its derating/aging curve but I personally doubt that.
    As to the overkill point I fully agree. A new trimmer cap and some fiddling around would have brought this unit back in spec, when compared to the rubidium standard. But some things are done for fun and we can all learn from it.
    I would like to see someone flagging an Rb unit from ebay as way out of spec (>10Hz).

    1. @one, indeed, I too would like to see an example of a working Rb that is far out enough to be concerned in the context of a home lab. The theorists and the practitioners very rarely see the world in the same way, thats why some of us do stuff, and some of us don’t! There are the same types of theoretical snobbery in the audio world too, its just certain subjects seem to attract the type.

      I am not sure I agree that a new trimmer would have sorted it, have a look at a video Dave jones done on the same counter, his cap was not broken and his conclusion after trying to trim the thing (with his tongue at exactly the right angle) was just the same – the internal oscillator IS pure rubbish. The counter needs something beyond its stock timebase if you are going to call it a Frequency Counter – its just a shame that with the premium prices HP charge for their kit they did not even consider putting in a TCXO (which having now checked is what the Racal Dana in my video uses, and that is pretty darn good too.), why? because a “Medium Stability Timebase” which is actually a TCXO on a board is actually an *option* for a few hundred dollars….

      Wow i am ranting now – time to put this subject to bed me thinks :)

      1. Are you familiar with this quote? “Artists who seek perfection in everything are those who cannot attain it in anything.” –Gustave Flaubert

        I think it expresses the same idea as part of the first paragraph of your post does. Unless I am misunderstanding you.

  4. I just picked up an hp 5316A Univesal counter. Nowadays you can buy a 10MHz OCXO for $30-40 new off Digi-Key with accuracy of +/- 0.5 ppb. Strikes me as good fun to put such a part into the old machine. What’s nice about these old Universal Counters is that they have a frequency reference output on the back. For me I would get both a better frequency in the HP 5316 and I will have a high quality frequency standard I can use out of the instrument. This is cost effective because the HP 5316 has a power supply already available and the hardware is present for convenient build. As we all know, the housing and power supply is often a costly part of a project. I look forward to leveraging my new device in my developing lab.

    1. Yoy can get an OCXO for $30-40 and you can get a 0.5 ppb OCXO, but you cant get a 0.5 ppb $40 OCXO. Those better units are $500-1000. The $38 special is 20 ppb. Of course, 20 ppb gets you to +/- 0.2 Hz on a 10MHz signal. I’ll be evaluating my HP5316 for accuracy some time this year. The OCXO option for the $5316 was a 10ppb unit, I think. For me the big news is we can get a 20 ppb OCXO for $40 off the shelf quantity 1 from DigiKey delivered in a couple of days. Man, times are good.

Leave a Reply to What's in a name?Cancel reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.