# Can I do a quick and dirty calibration on HP 8720D without cal kit?

Status
Not open for further replies.

#### drkirkby

##### Full Member level 6
I am buying a HP 8720D (50 MHz to 20 GHz) VNA. I'll send the money tomorrow and should have it in a week or two, as it has to be shipped from the USA to the UK. I have a week to evaluate this, after which the seller wont take it back. Hence I really want to find out if there are any faults within that time frame.

The problem is, at this point in time, I don't have a cal kit, though I intend buying a 85052B 3.5 mm cal kit as soon as I can get one at a sensible price in good condition.I may or may not get one before the analyzer arrives. The analyzer might be here before I get a cal kit, in which case I'm in trouble.

So my question is, would it be possible to do a quick and dirty cal of the 8720D without have a proper cal kit? On some analyzers, this would be quite easy, as all one needs is an open, short and load. At low frequencies, it should be possible to make a reasonable open and short. But I don't know if an attempt to do that on an 8720D will work, even at 50 MHz, as the cal kit has an offset open, offset short and sliding loads. I can't easily make something which slides if that is really needed.

I have an SMA load that is good to 18 GHz, so that bit is easy, but getting a sliding load is not.

Dave

#### FvM

##### Super Moderator
Staff member
Why do you think that HP8720 can't work with standard calibration elements at low and medium (e.g. up to a few GHz) frequencies? I don't understand the reasoning behind it.

The latest if you do a calibration of your specific measurement setup, you'll rely on standard open/short/load connected somewhere in your circuit.

drkirkby

### drkirkby

points: 2

#### drkirkby

##### Full Member level 6
Why do you think that HP8720 can't work with standard calibration elements at low and medium (e.g. up to a few GHz) frequencies? I don't understand the reasoning behind it.

The latest if you do a calibration of your specific measurement setup, you'll rely on standard open/short/load connected somewhere in your circuit.
Well, if the VNA calibration procedure asks one to connect a sliding load, and one does not do it, I'm not sure if the calibration routine would just not complete. On another VNA I've used, unless one has the open, short and load, one could never complete the calibration. Whilst I can easily cobble an open and short together, if the calibration routine asks for something I don't have, then I might be in a position of not being able to calibrate it at all. That was my concern.

Whilst an open/short/load will be acceptable electrically at 50 MHz (the lower limit of the VNA),it might not be acceptable to the software, which might insist on something else before it will complete.

That was my concern

Dave

Staff member

drkirkby

### drkirkby

points: 2

#### E Kafeman

Without knowing, check if this software allows you to define your own calibration kit: http://www.vnahelp.com/ckman_new2.html
Assumes that the 8720D else have internal functions to define capacitive load and length.

When calculating impedance network do I sometimes do in circuit SOL calibration using a SMD 0402 50 Ohm resistor as I want to include losses related to DUT PCB and a ~50 mm long thin and lossy measurement coaxial cable. Reason for using a less good coaxial cable is that it is very thin and more flexible, which can be needed when working with compact cell phones.
After calibration do I check electric delay by soldering C and L with reactanse around 50 Ohm at intended filter component place to ground.
It is good enough for me up to 2.5 GHz as long as measured result helps me select correct component values from E12 series.
Measured values below 5 Ohm and above 500 Ohm are not very reliable with this method but not including DUT reactive PCB losses will never gives correct readings. At least not in GHz-range.

Last edited:

#### biff44

buy an sma short from pasternack. Just leave the cable end open for the "open" standard. that will allow a response cal for return loss. Get an sma bullet for the thru response cal. that will get you going.

Of course, if you want a really good cal, you will need an official cal kit. If you want to spend 3x as much, you can get those in 3.5 mm too.

#### drkirkby

##### Full Member level 6
buy an sma short from pasternack. Just leave the cable end open for the "open" standard. that will allow a response cal for return loss. Get an sma bullet for the thru response cal. that will get you going.

Of course, if you want a really good cal, you will need an official cal kit. If you want to spend 3x as much, you can get those in 3.5 mm too.
The analyzer, with options 010 and 1D5 (high stability oscillator and TDR function) cost me 16000. I think a used 3.5 mm call kit is going to cost me more than 25% of that. If a calibration can be done with an open/load/short, then I am tempted to buy/make an SMA open and short, and use that for non-critical calibrations at lower frequencies. There is a verification kit, which would be handy, as it verifies the analyzer is performing to its specification, but that that is more than the cal kit. It starts to cost a lot of money, especially as for now at least, I don't have a company, and so are paying this from my personal funds. At a later date, I'll set up a company. I also want an N cal kit, by which time the calibration and versification kits will have probably cost more than the analzyer. Dave V #### volker_muehlhaus ##### Guest If you need a budget cal kit for SMA, have a look here: http://www.hhft.de/index.php?page=vna&subpage=network_analysis_p2b You can see the Rosenberger part numbers of the components here: http://www.hhft.de/index.php?page=competences&subpage=calibration Note that that the required offset length is given for open and short, so that the reference planes of open, short and through are aligned. That detail is important for accurate results. You need to store these values in the VNA cal kit definition. Last edited: #### biff44 ##### Advanced Member level 5 when you do a response cal, it is acceptable for utilitarian type measurements. If you are wondering if your return loss is 10 or 15 dB, a response cal will tell you that. If you want your system calibrated down to a 40 dB return loss, then you need the whole enchilada. Also, if you are going to be working at the high end of the analyzer (15 to 20 GHz) then the full cal will work. You might want to consider an Ecal module. I think some of the firmware versions of the 8720D will work with the Ecal. You can get those on ebay for around1500.

If you are going to buy some fairly expensive but non-HP cal standards, make sure you can load the cal standards information into the network analyzer! Otherwise, they will be relatively useless.

#### drkirkby

##### Full Member level 6
Just for the record, I now have my 8720D it does calibrate fine with just an open, short and load.

Dave

V

#### volker_muehlhaus

##### Guest
Dvae, have you looked at the reference in my post #8? This might help to set the offset length properly, for improved calibration accuracy.

#### drkirkby

##### Full Member level 6
Dvae, have you looked at the reference in my post #8? This might help to set the offset length properly, for improved calibration accuracy.
Hi,
I did see it, but have not acted on it. I only wanted something to give me the ability to check the analyzer was working properly.

I have an HP cal kit on order - just waiting for Fedex to deliver it. It is not the one I wanted (HP/Agilent 85052B, rated to 26.5 GHz)
http://www.home.agilent.com/agilent/product.jspx?cc=US&lc=eng&nid=-536902693.536880729&pageMode=OV

but a lower cost 85033E
http://www.home.agilent.com/agilent/product.jspx?nid=-536902695.536879921.00&lc=eng&cc=US
which is for DC-9 GHz.

But the 85033E is supported by the instrument. (I see this as one of the choices in the firmware when I select a cal kit).

But in the next few days I will have a supported kit, which is OK to 9 GHz, which is all I need just now.

I'll buy the higher frequency kit at some point when I see one at a sensible price. There's a dealer in the UK with the 26.5 GHz cal kit, but it is over half the price of what I paid for the VNA, which was \$16000.

I also got a cheap APC-7 cal kit (HP 11866A) which is only for DC-2 GHz, but I suspect could be pushed quite a bit higher if the HP 909C 2 Ghz load was replaced with a better one such as the HP 909A (18 GHz) load. Unfortunately, 11866A is not supported by the instrument, but it was quite cheap and the 18 GHz load was not that expensive on eBay either. But APC-7 is not something I need much, but the cal kit was quite cheap.

Maybe I am wrong, but in APC-7, I can't see there's going to be much difference between cal kits if they only have the open, load and short.

I've not read much about this, but it would seem the HP VNAs considers the open as having some offset delay in ps, and expresses the fringing capacitance as a 3rd order polynomial. The same is done for the short, but its inductors rather than capacitors in the polynomial. I suspect given one cal kit, it is quite easy to work out the parameters of another cal kit of the same sex, by a simple polymomial curve fit. They don't appear to have done that iwth the economy kit you provided the link to, but I suspect it is possible to improve on that. I would have been more tempted to buy that if it had both sexes, but it did not.

I note some of the coefficients of the HP VNA calibration kits are available in the program "VNA Cal kit manager"

http://na.tm.agilent.com/vnahelp/products.html

but it does not have many of the kits in its database.

Status
Not open for further replies.