Group Actions
Engagement
  • Author Author: michaelkellett
  • Date Created: Date Created
  • Views 1769 views
  • Likes 9 likes
  • Comments 7 comments
Related
Recommended

AD8367 as a wideband true RMS detector.

michaelkellett

In a recent thread   mentioned using the AD8367 as a true RMS detector.

I'd never encountered this chip before and was interested in what it could do. The official eval board from AD costs about £110 which is a little steep for a casual indulgence of curiosity - but fortunately similar eval boards are available form AlieExpress for just over £10 including shipping.

There has been some speculation in another thread as to the provenance of the AD6387 on these boards, from the test data I'm inclined to believe that its a real AD part but it could be second hand (although I can't see the economics of that working for the supplier.)

Whatever - at that price I reckon it was worth a punt.

image

If you want to really understand this blog you need to download the data sheet fo the AD8367 from AD's website. (www.analog.com) or just search "AD8367".

It's described by Farnell as a Linear-in dB variable gain amplifier, AD use more words at https://www.analog.com/en/products/ad8367.html

A concise description might be that its a compressor chip with a true RMS detector and a very wide bandwidth.

It doesn't need any additional active components to use its level detector to close the loop on the variable gain amplifier to give a constant signal output and an analogue control voltage which represents the gain its using and is log scaled.

This means that you can use it to measure the level of an analogue signal over a level range of about 40dB and a very wide frequency range.

I did some initial tests on my board using a TTi TGR2053 signal generator which can reach 3GHz and an R & S FPC1500 as a level detector. The board seems to work up to about 700Mhz.

The TGR2053 won't go below 150kHz so I swapped to using a Keysight 33600A and stuck with max frequency of 120MHz. For my initial tests I tested from 5MHz. The board will need some component changes to work at audio  frequencies.

image

I wrote a programme in VB6 (running on the RF rig PC which uses Windows XP so isn't allowed internet connection.) . The data was processed in MATLAB.

image

I fitted the AD recommended resistors to the input section of the eval board to get 50ohm input impedance.

image

The highest level I used was +5dBmW. The chip worked very well down to -30dBmW but obviously less well with the -35dbmW signal. Didn't have time to investigate further.

The AD8367 looks as if it would work very well as part of a meter circuit, giving very good results over a 20dB range and good results over 35dB.

I've been experimenting with a thermistor based wide-band RMS detector (which will be the subject of a different blog). I don't know how well I shall eventually get that to perform but I can say that the AD8367 is a lot easier to get working.

The next step is to attempt to measure the goodness of the true RMS detector (this is where I expect the thermistor based design to win - if I can ever make it work well enough - it needs a temperature controller that can hold about +/-0.001 K).

MK

Parents
  • The one attribute of AD8367 I don't like is the dB linear response of the XAmp architecture. I would prefer linear V/V, unfortunate that the Analog portfolio does not have a VCA like that.
    Another concern is that 8367 attenuates the signal, then re-amps for the detector. Am afraid it's not helping the noise situation. The HP3400 "architecture" pushes all signals down to the 1mV full scale range, then amplifies for the rms detector. If I insert another attenuation/amplification ramp the noise floor gets even higher. Did not calculate how much, though.
Comment
  • The one attribute of AD8367 I don't like is the dB linear response of the XAmp architecture. I would prefer linear V/V, unfortunate that the Analog portfolio does not have a VCA like that.
    Another concern is that 8367 attenuates the signal, then re-amps for the detector. Am afraid it's not helping the noise situation. The HP3400 "architecture" pushes all signals down to the 1mV full scale range, then amplifies for the rms detector. If I insert another attenuation/amplification ramp the noise floor gets even higher. Did not calculate how much, though.
Children