Keyboard Shortcuts
Likes
Search
346B verification?
开云体育Would anybody have a recommendation on where to send a 346B noise source for calibration, or verification? Generate a new table if necessary. ? It appears Keysight does still support the 346B, just wondering if there are other options that may be more hobbyist friendly. Gary WA2OMY |
开云体育Gary,I strongly recommend Keysight, but you might give Noisecom a shot. They did some noise source calibrations for me at ham frequencies and reasonable prices a number of years ago.? Steve WB0DBS EM17iq09ao On Oct 22, 2024, at 10:03?AM, Gary Hitchner via groups.io <talgarth@...> wrote:
|
|
Hi Gary Noisecom. Inc actually designed and makes the HP346B, HP just re badges it. The Noisecom model number is NC364B. George G6HIG On Tuesday 22 October 2024 at 16:03:29 BST, Gary Hitchner via groups.io <talgarth@...> wrote: Would anybody have a recommendation on where to send a 346B noise source for calibration, or verification? Generate a new table if necessary. ? It appears Keysight does still support the 346B, just wondering if there are other options that may be more hobbyist friendly. Gary WA2OMY |
开云体育Yes, I remember when I was working, the Noisecom sources cost was about 2/3 compared to the HP.? Good quality also. ? Thanks, Gary ? ? From: [email protected] <[email protected]> On Behalf Of george edmonds via groups.io
Sent: Tuesday, October 22, 2024 11:31 AM To: [email protected] Subject: Re: [HP-Agilent-Keysight-equipment] 346B verification? ? Hi Gary On Tuesday 22 October 2024 at 16:03:29 BST, Gary Hitchner via groups.io <talgarth@...> wrote: ? ? Would anybody have a recommendation on where to send a 346B noise source for calibration, or verification? Generate a new table if necessary. ? It appears Keysight does still support the 346B, just wondering if there are other options that may be more hobbyist friendly. Gary WA2OMY |
|
I do have some background information on this topic. George, you are incorrect. The 346 noise sources were designed at HP in Stanford Park (Palo Alto). As far as I know, they have never been made by an outside vendor. The only thing that changed since the original design is the fab that makes the noise diodes. HP obsoleted the Si process in Santa Rosa and started buying diodes from an outside fab. At least some of the founders of that outside fab were former HP employees.
?
The N4000 noise sources were designed at Agilent U.K. in South Queensferry, as were NFA noise figure meters which are based on the E440X ESA economy spectrum analyzers.
?
Noisecom made look-alike noise sources and even called them 346 to leave no doubt about what market they were after. There was another company, now called Noisewave, which made a similar product. I can't recall for sure, but Noisecom may have been started by a Noisewave employee after some disagreements. I also can't remember what Noisewave was called back then.
?
Another manufacturer of noise sources was Ailtech, later part of Eaton, and after that Maury Microwave. Ailtech did not use the 346 designation. The Ailtech sources I opened didn't look anything like HP's. Some of them struck me as hobby projects. I think Maury gave up on noise sources. They now make calibration systems for noise sources.
?
However, I just went to the Maury Microwave site and it appears that Noisecom is now part of them!
?
Vladan
?
|
|
Gary, the most hobbyist-friendly approach would be to borrow a noise source with a trusted ENR table and use a noise figure meter to transfer the calibration to your noise source. That's in principle what Keysight does when you send your source in for calibration. There is a little bit of math involved which you can take care of in Excel.
?
Keysight uses a noise figure meter, a noise source they vouch for, and a N2002A test set. They also check return loss before they calibrate the source. The N2002A is not essential, it helps automate the process. It contains some relays, attenuators and isolators.
?
Vladan
?
On Tue, Oct 22, 2024 at 08:03 AM, Gary Hitchner wrote:
|
|
toggle quoted message
Show quoted text
On Oct 22, 2024, at 9:40?PM, pianovt via groups.io <pianovt@...> wrote:
|
|
At Keysight, the noise source measurement bench uses two noise sources to calibrate the customer's noise source.
And of course the measuring bench in question includes the isolators as on the N2002A in order to guarantee the best possible accuracy.
For me and for many of the experts in this field, as VSWR is the biggest contributor to uncertainties in noise measurements, it is obvious that calibration without isolators is absolutely impossible. This can only be a check of proper operation, but in no way can this be called a calibration.
Even if it means paying, you might as well pay to get the most accurate values as possible.
Obviously Noisecom is certainly as precise as Keysight but I strongly advise against any other provider that is not seriously accredited. To determine which is the best, it is sufficient to request the technical annex of the accreditation in the field of noise measurements from the testing laboratory and to check whether the calibration uncertainties are satisfactory or not.
If the lab isn't accredited, then drop it, you'll pay for nothing.
|
|
Hi Vladan, Thank you very much, That looks like it will answer every question one could ever dream up. I naively thought it would be much easier?to do. Dave VE7HR On Tue, Oct 22, 2024 at 11:35?PM pianovt via <pianovt=[email protected]> wrote:
--
72 de Dave VE7HR |
|
I think a lot depends on what frequency range you want to use the noise source over. If you plan to use it above a few GHz then sending it to Keysight (or Noisecom) is the only realistic option if you want reliable results. By reliable, I mean within reliably quoted uncertainty limits.
?
If the upper frequency limit is less than 3GHz (especially less than 1GHz) then I think it can be done acceptably at home if you have access to the suitable test gear and a known good noise source to check against. This assumes that you aren't planning to use the noise source in a 'noise figure shootout' where you will be comparing your VHF or UHF LNA measurements in a competition for lowest noise figure etc.
?
At work today I cherry picked two 346A noise sources that have current cal certs and I also took copies of the ENR tables from the latest cal certificates in each case. I picked the two with the lowest VSWR when measured on an E5080A VNA up to 9GHz. I'll try and cross check them against each other and against my old HP 346A up to about 3GHz. Obviously, I can't compete with Keysight here, but even a Keysight 'calibration' will have a degree of uncertainty and so will the noise source itself.
?
If I was really fussed about this stuff I could have done this cross check numerous times since I bought the 346A many years ago. But I have never bothered because I think it is already close enough and stable enough to meet my needs up to about 3GHz. But I suppose it might be fun to try it anyway...
?
? |
开云体育How does one guarantee isolators are as good as good attenuators, other than to select with a VNA?
|
|
Hi Ed!
Easy one! IL low (<1 dB) S12 hi (> 20 dB). A pad might introduce too much loss, reducing the noise level too much, impacting measurement accuracy.
Jeff Kruth
In a message dated 10/23/2024 12:45:31 PM Eastern Standard Time, ed@... writes: ?
How does one guarantee isolators are as good as good attenuators, other than to select with a VNA?
?
|
|
Look at the reverse isolation spec. Isolators or circulators offer low insertion loss and should offer > 20 dB of reverse isolation. This means that any reflected power ends up in the isolator and not in the circuit.? Mike On Wed, Oct 23, 2024 at 9:50?AM Jeff Kruth via <kmec=[email protected]> wrote:
|
开云体育I think my focus was on the noise source itself, and assuming a non-defective attenuator would give better return loss. I wasn't thinking of the NFA itself and it's presumably non-reflectionless input with mixers and all that.
Thank you for encouraging me to dig deeper.
|
|
Dave, it's actually not very complicated. Much of the document dwells on which buttons to push on the NFA. At its core, it's just a Y-method measurement. The formula also simplifies if you assume T=290K for the reference source and DUT. The OP didn't say much about his requirements, he may not even need ENR numbers above a GHz.
?
People can get obsessed with these numbers, similar to reference standards for voltage, resistance, etc. However, there are other sources of error that can enter into NF measurements and overwhelm ENR uncertainty. Occasionally, I see people claiming NF measurements below 0.1dB with two digits precision. In truth, at best they can make relative measurements to compare competing designs, and only when measured with the same meter and same source.
?
Vladan
?
?
On Wed, Oct 23, 2024 at 08:03 AM, Dave Miller wrote:
|
|
My first attempt at comparing the two (calibrated by Keysight) 346A noise sources gave excellent results at 10MHz and 100MHz.
?
I used an old Agilent E4440A PSA analyser with the built in noise figure personality to measure the Y factor but I used my own external preamp instead of the internal preamp in the E4440A. Note that my external preamp has a noise figure of about 2.8dB and a VSWR of <1.02:1 across 10-100MHz.
?
Using the first 346A as ENR1 and Y1 I was able to predict the ENR of the second 346A to be within 0.01dB of what was stamped on it. This was a bit of a surprise, I didn't expect it to agree that well!
However, when tested at 1GHz, the prediction was off by 0.04dB but the VSWR of my preamp degrades to 1.15:1 at 1GHz. I could try again with a 2dB pad in front of the preamp but this will have to wait until later.
?
When I tested my own 346A against one of the calibrated 346A noise sources I saw a worst case ENR error of 0.1dB. This isn't bad considering it hasn't been formally calibrated in about 20 years.
?
All three 346A noise sources show excellent VSWR at these three test frequencies. They are all about the same. The plot below is the VSWR of my 346A up to 3GHz in both hot and cold states.
?
I used the equation given in the N2002A manual to work out ENR2 from ENR1 and Y1 and Y2. It's probably best I repeat all this a few times to make sure I get consistent results. I'll use more averaging the next time although 4 averages seems to work quite well.
?
?
?
|
开云体育Viadan, ok to e-mail privately and take this off-line? ?I would like to know more about what SP 9.1 -9.4 is measuring, and what it means. ? Good information on this entire thread. Gary WA2OMY ? From: [email protected] <[email protected]> On Behalf Of pianovt via groups.io
Sent: Wednesday, October 23, 2024 1:38 PM To: [email protected] Subject: Re: [HP-Agilent-Keysight-equipment] 346B verification? ? Dave, it's actually not very complicated. Much of the document dwells on which buttons to push on the NFA. At its core, it's just a Y-method measurement. The formula also simplifies if you assume T=290K for the reference source and DUT. The OP didn't say much about his requirements, he may not even need ENR numbers above a GHz. ? People can get obsessed with these numbers, similar to reference standards for voltage, resistance, etc. However, there are other sources of error that can enter into NF measurements and overwhelm ENR uncertainty. Occasionally, I see people claiming NF measurements below 0.1dB with two digits precision. In truth, at best they can make relative measurements to compare competing designs, and only when measured with the same meter and same source. ? Vladan ? ? On Wed, Oct 23, 2024 at 08:03 AM, Dave Miller wrote:
|
|
I tried again using 16 averages and got similar results. The manual suggests using 128 averages so maybe I'll try this over the weekend. The manual also suggests that the noise source should not be used for an hour before these tests such that it can thermally stabilise. I guess this means I don't handle them either as my hands will warm them up. So I'll have a fresh go over the weekend and try and follow the procedure in more detail.
?
When I currently compare against either of the calibrated noise sources, I'm seeing a worst case ENR error of <0.03dB for my old noise source at 10MHz and 100MHz and 0.1dB at 1GHz. I've not tried the 2dB pad yet. I'm not expecting it to make much difference.
?
? |