Calibrating an RF detector


terry.bailey@...
 

Hi all, I have one of these

http://www.ebay.co.uk/itm/AD8317-1MHz-to-10GHz-RF-Power-Meter-Logarithmic-Detector-Controller-f-Amplifier-/172248001491?hash=item281ac7ebd3:g:SSoAAOSwM4xXZ9NN

I it works fine but I don't have any idea how to calibrate the readings I am getting from it. I don't have a reference source. I do have a spectrum analyser which is good to 10GHz, but it is totally uncalibrated and somewhat temperamental.

Can anyone point me in the right direction?

Cheers

Terry



Andy G4JNT
 

First of all, download the datasheet for the AD8317 and study that to see what you've got

As the evaluation board you have "probably" has no additional DC conditioning on the output or RF gain or attenuation,  until you can prove otherwise, it'll probably do what it says in the datasheet in terms of Vout vs Pin

'jnt


On 26 August 2017 at 15:50, terry.bailey@... [ukmicrowaves] <ukmicrowaves@...> wrote:
 

Hi all, I have one of these

http://www.ebay.co.uk/itm/AD8317-1MHz-to-10GHz-RF-Power-Meter-Logarithmic-Detector-Controller-f-Amplifier-/172248001491?hash=item281ac7ebd3:g:SSoAAOSwM4xXZ9NN

I it works fine but I don't have any idea how to calibrate the readings I am getting from it. I don't have a reference source. I do have a spectrum analyser which is good to 10GHz, but it is totally uncalibrated and somewhat temperamental.

Can anyone point me in the right direction?

Cheers

Terry




GORDONJ REASON <gordonj.reason@...>
 

to give me a 0dBm , (1 mW ) source at 50 Mhz ..............

I use a Xtal oscillator ........ and buffer .......... followed by a single , +V ,  pin diode attenuatror .

the rf output is rectified with a shottky diode .

That DC is fed to one input of an opamp .

The other input is a DC reference from a TL431 .

The output + feeds back to the PIN Diode attenuator , via a small resistor , and the opamp has a multiturn pot in the feedback path ..... 

Works like a charm ...... 

It's not inmportant to use a Crystal osc .......... the frequency is your choice .....

On 26 August 2017 at 15:54 "Andy Talbot andy.g4jnt@... [ukmicrowaves]" <ukmicrowaves@...> wrote:

 

 

First of all, download the datasheet for the AD8317 and study that to see what you've got

As the evaluation board you have "probably" has no additional DC conditioning on the output or RF gain or attenuation,  until you can prove otherwise, it'll probably do what it says in the datasheet in terms of Vout vs Pin

'jnt


On 26 August 2017 at 15:50, terry.bailey@... [ukmicrowaves] <ukmicrowaves@...> wrote:
 

 

Hi all, I have one of these

http://www.ebay.co.uk/itm/AD8317-1MHz-to-10GHz-RF-Power-Meter-Logarithmic-Detector-Controller-f-Amplifier-/172248001491?hash=item281ac7ebd3:g:SSoAAOSwM4xXZ9NN

I it works fine but I don't have any idea how to calibrate the readings I am getting from it. I don't have a reference source. I do have a spectrum analyser which is good to 10GHz, but it is totally uncalibrated and somewhat temperamental.

Can anyone point me in the right direction?

Cheers

Terry


 

 

 


 


 


Andy G4JNT
 

Feedback and PIN attenuators are fine, but how do you calibrate your 1mW in the first place?

You need an accurate power level or meter to compare with.

CMAS gates can work OK ish.   Use AC series gate which at 50Mhz will be giving a reasonable squarewave its ouptut resistance is about 10 ohms so use 39R reistor and connect to 50R load.  You now have a 2.5pk-pk square across your 50R load, that is delivering 

At the fundamental that is 2.5 * 4/pi = 3.2V pk-pk or 25mW = 14dBm
Use resistors / attenuators appropriately.

But remember that level is at the fundamental only.  If unfiltered, harmonic components will add.

'jnt


On 26 August 2017 at 17:58, GORDONJ REASON gordonj.reason@... [ukmicrowaves] <ukmicrowaves@...> wrote:
 

to give me a 0dBm , (1 mW ) source at 50 Mhz ..............

I use a Xtal oscillator ........ and buffer .......... followed by a single , +V ,  pin diode attenuatror .

the rf output is rectified with a shottky diode .

That DC is fed to one input of an opamp .

The other input is a DC reference from a TL431 .

The output + feeds back to the PIN Diode attenuator , via a small resistor , and the opamp has a multiturn pot in the feedback path ..... 

Works like a charm ...... 

It's not inmportant to use a Crystal osc .......... the frequency is your choice .....

On 26 August 2017 at 15:54 "Andy Talbot andy.g4jnt@... [ukmicrowaves]" <ukmicrowaves@...> wrote:

 

 

First of all, download the datasheet for the AD8317 and study that to see what you've got

As the evaluation board you have "probably" has no additional DC conditioning on the output or RF gain or attenuation,  until you can prove otherwise, it'll probably do what it says in the datasheet in terms of Vout vs Pin

'jnt


On 26 August 2017 at 15:50, terry.bailey@... [ukmicrowaves] <ukmicrowaves@...> wrote:
 

 

Hi all, I have one of these

http://www.ebay.co.uk/itm/AD8317-1MHz-to-10GHz-RF-Power-Meter-Logarithmic-Detector-Controller-f-Amplifier-/172248001491?hash=item281ac7ebd3:g:SSoAAOSwM4xXZ9NN

I it works fine but I don't have any idea how to calibrate the readings I am getting from it. I don't have a reference source. I do have a spectrum analyser which is good to 10GHz, but it is totally uncalibrated and somewhat temperamental.

Can anyone point me in the right direction?

Cheers

Terry


 

 

 


 


 



geoff Blake
 

First, you need to decide exactly how accurate you need to calibrate the sensor. RF power is notably fickle in this aspect. All but the best, most well equipped calibration labs are hard put to measure much better than +/- 5% over the range you speak of. Before other folk raise their pens to criticise me, at certain spot frequencies better figures are attainable.  In amateur circles I cannot see that any better than +/- 10% is necessary. Of course your amateur SWR/power meter will be much better than this :-)

OK, to sensibly and reasonably accurately calibrate a RF power sensor, you need to:

1) check the RF sensitivity of the device, i.e. RF in to outputted reading. This is the primary step and is best done at 0dB or 1mW. This is probably best done with the 1mW 50MHz reference fitted on many HP and other power meters. This power meter should have been recently calibrated - the specified power drift of the reference source is 0.7%/yr IIRC.

2) check the return loss of the sensor. Obviously, any reflected power is going to reduce the readout and make the sensor read low. This needs to be done at several frequencies over the range.

3) look at the linearity of the sensor, i.e. if you reduce the power by 10dB, the output reading should reduce by 10dB.

4) Check the noise floor of the sensor, i would expect this to be about -60dBm and would increase the error at -50dB by about 10% or so.

In practice, and in an amateur world, I would choose Francois's option first, although I would go easy on the good beer, at least until the measurements are all finished :-)

Andy's proposal, just to use the spec. sheet figures, would probably produce significantly better results that those of a typical SWR/power meter or dare I say it, a Bird model 43 system.

Good luck

Geoff G8GNZ

#################################################
Geoff Blake,   G8GNZ    JO01fq:   Chelmsford,  Essex,  UK
<geoff@...>    or   <melecerties@...>
Using Linux: Ubuntu 14.04 on Intel or Debian on UltraSparc
    and  Apple  OS  X  El Capitan  on  my  Macbook  Pro.    
                Avoiding Micro$oft like the plague.
#################################################

On 26 August 2017 at 15:54, Andy Talbot andy.g4jnt@... [ukmicrowaves] <ukmicrowaves@...> wrote:


First of all, download the datasheet for the AD8317 and study that to see what you've got

As the evaluation board you have "probably" has no additional DC conditioning on the output or RF gain or attenuation,  until you can prove otherwise, it'll probably do what it says in the datasheet in terms of Vout vs Pin

'jnt


On 26 August 2017 at 15:50, terry.bailey@... [ukmicrowaves] <ukmicrowaves@...> wrote:
 

Hi all, I have one of these

http://www.ebay.co.uk/itm/AD8317-1MHz-to-10GHz-RF-Power-Meter-Logarithmic-Detector-Controller-f-Amplifier-/172248001491?hash=item281ac7ebd3:g:SSoAAOSwM4xXZ9NN

I it works fine but I don't have any idea how to calibrate the readings I am getting from it. I don't have a reference source. I do have a spectrum analyser which is good to 10GHz, but it is totally uncalibrated and somewhat temperamental.

Can anyone point me in the right direction?

Cheers

Terry







Mike Willis
 

Hi Terry,

I have of of those and also one of the OLED versions. I calibrated it against a signal generator which I believe is fairly accurate. It was within 2dB and much better in relative accuracy rather than absolute accuracy, i.e. measuring changes in power, so for example very useful to determine an amplifiers 1 dB compression point. You said you don't have one of those but they are often made available for testing at microwave round tables, as are good quality power meters that you can compare against. The next RT near you can be determined from the UKUG website. There may also be a local amateur with the right gear that can help.

Bear in mind that these chips have some thermal drift so you also need to calibrate against temperature. In practice, they are not really for accurate metrology but as very useful when used with a directional coupler to verify the TX is more or less working.

Mike 


Paul G8AQA
 

Just a few experiences with the AD8310 which is the low frequency (440MHz) version of this range of devices.

Linearity was good but there was quire a large variation of output voltage with frequency so we havr ro calibrate at each frequency of interest.  The 50MHz calibrators are usually used for thermal power meters. 

The filter capacitor CLPF should not be omitted as rectified RF can otherwise appear on the output.

We encountered problems with the regulation of the supple and ended up using one of the A/D inputs of the Arduino processor to compensate for variations.

Examination of Fig 15 on the AD8317 data sheet will show the variation of input impedance with frequency.

It will pay to read the data sheet carefully to see how well it will suit your application.

Paul G8AQA


On 27/08/2017 08:17, willis.mj@... [ukmicrowaves] wrote:
 

Hi Terry,


I have of of those and also one of the OLED versions. I calibrated it against a signal generator which I believe is fairly accurate. It was within 2dB and much better in relative accuracy rather than absolute accuracy, i.e. measuring changes in power, so for example very useful to determine an amplifiers 1 dB compression point. You said you don't have one of those but they are often made available for testing at microwave round tables, as are good quality power meters that you can compare against. The next RT near you can be determined from the UKUG website. There may also be a local amateur with the right gear that can help.

Bear in mind that these chips have some thermal drift so you also need to calibrate against temperature. In practice, they are not really for accurate metrology but as very useful when used with a direct ional coupler to verify the TX is more or less working.

Mike 


g4bao
 

If anyone wants to get a reasonable calibration of their power detectors next week I'll bring my 435  and 436 power meters to the Crawley round table. They have a 50MHz 0dBm reference built in and they are within a decimal of a dB of each other. I also have a reference 30dB 50MHz attenuator. They have been measured by my 432 meter which is a DC calibrated unit so should be pretty good. 
Just say the word and I'll bring them.
If you are local to me and want to pop in and use the kit, just give me a bit of notice by email.
I'm of the view that we get a bit hung up on accuracy of measurement for power and noise figure. As long as you can measure improvement in your tweakings of kit that's all you need. 
I for one measure noise figure  in "dBddk" in that I calibrated my meter against Sam's system just so that we can compare measurements against each other. Power and NF are only about bragging rights after all.

73 John 


On 29 Aug 2017 00:25, "Paul Nickalls paulnick@... [ukmicrowaves]" <ukmicrowaves@...> wrote:
 

Just a few experiences with the AD8310 which is the low frequency (440MHz) version of this range of devices.

Linearity was good but there was quire a large variation of output voltage with frequency so we havr ro calibrate at each frequency of interest.  The 50MHz calibrators are usually used for thermal power meters. 

The filter capacitor CLPF should not be omitted as rectified RF can otherwise appear on the output.

We encountered problems with the regulation of the supple and ended up using one of the A/D inputs of the Arduino processor to compensate for variations.

Examination of Fig 15 on the AD8317 data sheet will show the variation of input impedance with frequency.

It will pay to read the data sheet carefully to see how well it will suit your application.

Paul G8AQA




On 27/08/2017 08:17, willis.mj@... [ukmicrowaves] wrote:
 

Hi Terry,


I have of of those and also one of the OLED versions. I calibrated it against a signal generator which I believe is fairly accurate. It was within 2dB and much better in relative accuracy rather than absolute accuracy, i.e. measuring changes in power, so for example very useful to determine an amplifiers 1 dB compression point. You said you don't have one of those but they are often made available for testing at microwave round tables, as are good quality power meters that you can compare against. The next RT near you can be determined from the UKUG website. There may also be a local amateur with the right gear that can help.

Bear in mind that these chips have some thermal drift so you also need to calibrate against temperature. In practice, they are not really for accurate metrology but as very useful when used with a direct ional coupler to verify the TX is more or less working.

Mike 



Andy G4JNT
 

I found that AD8310 device to be more than flat enough in its frequency response at least up to a couple of  hundred MHz - I suspect if you found sensitivity with frequency variation at frequencies significantly below its specified maximum, you have a layout issue.   Any variation in the 50R input impedance over frequency will give apparent reading changes, as the chip measures RF voltage across a load, not actual power.

I used 0805 components and the SOIC packaged version of the chip witj teh input R and C immediately adjacent to the pads.   Wire ended resistors and a DIL chip really can't be asked to cope with more than up to, say, 144MHz

The data sheet calibration is good enough as a first approximation but unfortunately the volts per dB has some latitude in its setting.  And that is the thing that can leads to high initial errors.   You can take that out if you have a couple of accurately calibrated attenuators.  You know the response is a straight line (Volts / dB) so only two, or three points of attenuated Rf versus measured volts  are needed.   Absolute power calibration, (equivalent to a constant voltage offset rather than the slope of the line) can be left 'til later as it can't be too far out once you've got the slope spot-on.

See RadCom  May 2015 for more on these family of Log Power Detector chips

Andy  G4JNT

Virus-free. www.avg.com

On 29 August 2017 at 00:25, Paul Nickalls paulnick@... [ukmicrowaves] <ukmicrowaves@...> wrote:
 

Just a few experiences with the AD8310 which is the low frequency (440MHz) version of this range of devices.

Linearity was good but there was quire a large variation of output voltage with frequency so we havr ro calibrate at each frequency of interest.  The 50MHz calibrators are usually used for thermal power meters. 

The filter capacitor CLPF should not be omitted as rectified RF can otherwise appear on the output.

We encountered problems with the regulation of the supple and ended up using one of the A/D inputs of the Arduino processor to compensate for variations.

Examination of Fig 15 on the AD8317 data sheet will show the variation of input impedance with frequency.

It will pay to read the data sheet carefully to see how well it will suit your application.

Paul G8AQA


On 27/08/2017 08:17, willis.mj@... [ukmicrowaves] wrote:
 

Hi Terry,


I have of of those and also one of the OLED versions. I calibrated it against a signal generator which I believe is fairly accurate. It was within 2dB and much better in relative accuracy rather than absolute accuracy, i.e. measuring changes in power, so for example very useful to determine an amplifiers 1 dB compression point. You said you don't have one of those but they are often made available for testing at microwave round tables, as are good quality power meters that you can compare against. The next RT near you can be determined from the UKUG website. There may also be a local amateur with the right gear that can help.

Bear in mind that these chips have some thermal drift so you also need to calibrate against temperature. In practice, they are not really for accurate metrology but as very useful when used with a direct ional coupler to verify the TX is more or less working.

Mike