Here are some thoughts on frequency calibration solutions.
A GPS Disciplined Oscillator (GPSDO) is probably the easiest way to have a high precision frequency reference at home that does not need calibration.
The GPS system uses precise clocks aboard satellites sending signals towards the earth. Receivers on earth can use the signals from several satellites to determine their position with respect to the satellites themselves using a mathematical tool called "triangulation". If you know your position with respect to the satellites, and know the position of the satellites themselves (the satellites broadcast their position in the signal they transmit), you can find your position on earth. This involves a lot of complicated math, but fortunately the user of GPS receivers don't have to be math wiz to use them (good thing!!!)
Now, if you know your position on earth, you can use the satellite signal simply as a precise time reference. That's what a GPSDO (GPS Disciplined Oscillator) does. Most GPSDO starts by performing a self-survey, a period of time during which they determine their exact position. Once self-survey is done (it can take from 20 minutes to several hours), the GPSDO switches to timing mode and from that point will provide timing pulses (and a reference frequency) precisely aligned to the frequency references in the satellites themselves.
I was lucky to be able to purchase a used Trimble Thunderbolt GPSDO (Ref 1).
This is a good product, but not the latest generation. The receiver has only 8 channels (Note 1) and is a little less sensitive than current products. It is however an excellent time and frequency standard for most applications. For comparison, I also have a Jupiter GPS receiver (Ref 6) which I intend to use for my own upcoming GPSDO project. The Jupiter is a 12 channel receiver and is more sensitive. I don't want to sound negative about the Thunderbolt, there is nothing wrong with the Thunderbolt receiver, it's just not the current state of the art.
If you want to put $300 or $400 in a GPSDO, you can probably get a complete GPSDO from eBay. Alternately, the Brooks Shera GPSDO (Ref 4) is a solid design for a hobby frequency standard and will cost much less, even if you have to buy all the parts.
The next step in professional GPSDO would be the Fury by Jackson-Labs (Ref 8). A high performance, reasonably priced GPSDO.
Please note that while GPSDOs do not require adjustment (when they are working normally), they typically cannot be called primary frequency standards because the signal reference generated by the GPS satellites is not controlled by NIST (Ref 7). Most of us can ignore this subtle difference and for all practical hobby applications, a GPSDO is a fine frequency standard.
Rubidium oscillators are very stable and precise frequency references. Unlike the GPSDO, they have to be calibrated once in a while, but they are significantly more stable than crystal oscillators, and they achieve their stable frequency output relatively rapidly after power up (typically around 10 minutes).
A number of Rb oscillators are available on the surplus market (eBay for instance) at prices between $100 and several hundred dollars depending on model, condition and the time of day. I bought an LPRO-101 on eBay for $90, but this was a while ago. I was lucky that it was in excellent condition with plenty of life left (according to the lamp voltage). As of 2015, it seems like you could get a working unit for about $200 on eBay. Rb oscillators are complicated beasts with a number of failure modes. They main aging characteristic is the Rb lamp itself becoming darker due to Rb deposit on the glass envelop. After a while, not enough signal passes through to achieve lock.
Be careful that many of the surplus Rb oscillators have actually bad phase noise and spurs. That means that while their long term stability may be excellent, the short term stability is not. If you intend to use an Rb oscillator as the reference for a multiplier chain used in a microwave transverter for instance, you may be disapointed unless you use a clean-up loop with a clean crystal oscillator.
If you do not need the kind of precision and accuracy provided by a GPSDO, the next best thing I can recommend is to find an HP 3586 Selective Voltmeter (Ref 3).
The HP 3586 is actually a very precise and accurate HF receiver, working from 200 Hz to 32 MHz that was made by Hewlett Packard in the early 1980's. Working units often show up on eBay. Most have the optional high stability timebase (an HP 10811 or an equivalent from Ovenair) and all have a built-in counter with 0.1 Hz resolution.
If you set the HP 3586 to 5, 10 or 15 MHz (the WWV frequencies), hook it up to an antenna and press the Counter button, it will display the WWV frequency with 0.1 Hz resolution. Please note you get this resolution with several readings per second, not like some counters that have to count for 10 seconds before giving you 0.1 Hz resolution. It is easy then to adjust the time base until you read exactly 5.0000000, 10.0000000 or 15.0000000 MHz and you will know the time base is adjusted to within 0.1 Hz of WWV. You can then use the HP 3586 time base as an external reference for your other gear. I have been doing that for a while before I got the Thunderbolt and it works great.
Please note that variations in propagation can affect the WWV signal by several 1/10th of Hz (if you are located within the continental US), so even though you can adjust the time base to within 0.1 Hz of the received signal, the accuracy will be a little less, but still much better than 1 Hz. I live in Florida and I have observed as much as +/- 0.3 Hz variations over a few days.
Obviously, the HP 3586 is more than a 50 lbs time base. You can have a lot of fun with it. As a very accurate receiver, it will display the level of a received signal in its passband with 0.01dB resolution and about 1 dB accuracy down to about -120 dBm, it puts the S meter on my HF receivers (FT-1000 MP Mk 5 and TS-440S/AT) to shame.
The HP 3586 comes in 3 models, A, B and C, and several options. All will work for this purpose but only the C model comes with a BNC antenna connector from the factory. Mine is an A model, but it was easy to replace the input connector with a BNC. The input module comes out easily and I just had to file the hole in the module and the front panel to let the BNC go through. If you looked at it, you would not know it did not come from the factory like that. If you decide you want to buy an HP 3586, try to get one with the high stability time base option (option 004).
Check the resolution bandwidth options too. All models have 20 Hz and 400 Hz, but the A and B models' third bandwidth is 1.8 and 2 kHz respectively instead of the more common 3.1 kHz, unless they have option 003 Impairement Functions, which includes the 3.1 kHz bandwidth filter. The 3.1 kHz filter is great for SSB, 2 kHz is not very useful.
Bottom line: more useful models will be 3586A or B with options 003 (3.1 kHz) and 004 (High Stability Timebase) and 3586C option 004 (High Stability Timebase).
HP 3586s can be bought on eBay. They usually sell for around $400 (2014). However, it is a 19" wide, 50 lbs sample of Hewlett Packard technology, so it is not as easily carried as the Thunderbolt :-) I have the full maintenance manual for the HP 3586 (and others) on my Manuals page (Ref 5).
The least expensive way to calibrate a frequency counter's timebase to WWV is by using an HF receiver and a PC's sound card.
Download and install Spectrum Lab (Ref 2) on your PC. Connect the audio output from the receiver to the line input of your PC's sound card. Tune the receiver to 10 MHz and watch the signal on the screen. The frequency where the receiver is tuned is unimportant, as long as the signal is in the passband of the receiver and the signal level is good.
Then put a short wire in the counter's 10 MHz time base output (the idea is to radiate a weak signal at 10 MHz which will be captured by the receiver) and watch the signal on the PC's screen. If the wire is the right length, you should be able to see the two signals at about the same strength and close in frequency. Alternately, you can build a small coupler with a T connector placed in series with the antenna connection and a high value resistor to attenuate the signal from the generator.
You can set the Spectrum Lab software to have great resolution on screen and then you should be able to adjust the counter's time base on top of the WWV signal and you are done. This is a manual process, but it will cost you nothing and will get you to within a fraction of 1 Hz from WWV.
There are numbers of other ways to have a precise and accurate frequency reference for the home lab. These include Hydrogen masers (well, maybe not everyone's home lab :-), and Cesium frequency standards.
Hydrogen masers and Cesium frequency standards are considered "primary" frequency standard. A primary frequency standard, when operating properly, does not need to be calibrated.
Rubidium oscillators are considered secondary frequency standards because their initial precision is not as good and they are susceptible to external factors such as temperature and magnetic field. However, once calibrated, they keep accurate timing for a long time as their drift is hundreds of times smaller than typical crystal oscillators.
Rubidium oscillators can be used in a GPSDO and they offer superior performance to Quartz oscillators during those times when the GPS signal may be unavailable or impaired (holdover).
Keep in mind that Hydrogen masers, Rubidium and Cesium oscillators use significantly more power to operate than even an Oven Controlled Crystal Oscillator (OCXO) and that they use a special type of vacuum tube that has limited life. You may find used units on eBay, and if they do not lock or lock slowly, the tube may have exceeded its useful life and need replacement.