AD8436 RMS-to-DC

Hi JS,

I've now had time to look at the AD8436 Data Sheet.

In my AD737 post, you wondered why people still have interest in an AVERAGE result. Possibly its because as engineers, we want to be able to flick a switch and actually see how poor the AVERAGE reading is - as in, its for educational, rather than practical reasons.

On the AD737, it is suggested to pre-boost the user supplied input signal by 111%, if an AVERAGE reading is to be done. Is that still the case with the AD8436?

Some data sheet comments.

Page 16 of 20 Figure 41. Slightly below top, left text, bottom line - word RECTIFIED is spelt incorrectly.

Page 11 of 20 FET Input Buffer. Add the word CORE so it reads - the input resistance of the AD8436 core is 8k,

Page 13 of 20 Capacitor Construction - Mention that a Metalized Polyester cap is optimal (it didn't say Metalized Polypropylene MKT - which might be best for timing - yet huge). These types of capacitors in 10u are not so common - and I don't believe any film capacitors come in surface mount. However, in development of the AD8436, the designers probably came across some really "nice" types (smallish, affordable, performed well). Are you able to give information of what those types were?     (Digikey, Mouser...)

Page 15 of 20 Converting to Average Rectified 4th line down, ...appears at the output... I think it could say appears at OUT (I wondered whether it was the core or the output opamp).

Cheers,

Peter

  • Hi Peter,

    Nice input, I saved your text in a Word file for DS update. BTW the AD8436 DS is now rev A, after adding an  option for folks who prefer leaded packages.

    OK on the aveage vs rms and yes, the fundamental difference in value results from the nath expression, not the detector in any way. Your take on the average vs rms preference is interesting, I take it you might be an educator? My recollection from past discussions is that industrial control systems using the large hosepower ac motors need average and rms values, but I don't know why. I believe industrial utilities arer are billed based on average power but I haven't yet connected the dots as to why designers are looking at both modes.

    Thanks for the tips on pg 16,11 and 13. Kemet and AVX have introduced new ceramic styles which overcome piezoelectric effects of prior art ceramic caps. The new styles are called X8L and X8R. They were devloped for automootive applications and so feature higher operating temperpatures as well as higher capacitance values (up to 10uF). I didn't mention polyprop for 2 reasons, first there is no advantage of the polyprop because the averaging cap errors are current induced, not voltage. Polyprop caps are great for low dielrctic absorbtion, which is actually a dc error, and this makes them good for timing, but dominating error source for the averaging capacitor is the is dc (current) leakage and bulk resistance. This takes electrolytic styles out of the competition, as electrolytics are used where lelakage really isn't imporatant (so long as they arene't so leaky as to heat and explode ). None of data sheets for the electrolyte based styles list leakage or dc resistance, although they do specify leakage current. A tantalum cap is a good general purpose stle unless one needs low voltage operpation. I'm writing an applicaiton note for the AD8436 that goes into capacitor styles in more detail. Most people complain that our data sheets are already too long, so we are somewhat constrained to brevity.

    All for now, thanks for reading the DS you've been heard!

    js

  • Hi JS,

    I've now built the AD8436 RMS-TO-DC converter module for evaluation. I've been "parallel" comparing it against the AD737 module.

    The AD737 appears to work correctly. It does require a trimpot adjustment on the Cc pin (pin-1), to trim the output to get the appropriate DC level.

    The AD8436 doesn't have any adjustment. I am finding that for 200mV RMS 1kHz at the module input, it gives 190mV DC at the module output. That is within 5% and probably quite reasonable.

    1) However, where would you add fine level adjustment to the AD8436 circuit - to get it more precise?

    Should I assume that there is no difference between adding it BEFORE or AFTER the RMS core?

    You might even suggest that the AD8436 needs no such adjustment.

    That question needs to be taken in regards to this next RMS versus AVERAGE issue.

    On both modules, I apply 200mV RMS 1kHz and monitor the DC output on an oscilloscope (see below for why a DMM is not used).

    For RMS mode, I should get 200mV DC (for 200mV RMS input). The AD737 is fine - but I'm already reading slightly low (190mV) for the AD8436 module (described above).

    Yet, when I select AVERAGE mode on both modules, I actually seem to get the correct output voltages on both modules.

    On the AD737 module, if I select AVERAGE (pre-Amplifies input signal by 1.11), then remove any capacitors from the Cav pin (pin-5), I get 200mV - which seems to be what I'd expect.

    On the AD8436 module, if I select AVERAGE (pre-Amplifies input signal by 1.11), then remove any capacitors from the Cav pin (pin-19), I get 200mV - which also seems to be what I'd expect.

    Both front-end amplifiers show identical levels (Av=+1.00 and AV=+1.11).

    2) Should the AVERAGE mode on the AD8436 work as I expect (similar to the AD737)?

    ------

    As an aside, I generally connected the AD8436 output to an Oscilloscope and the trace is reasonably quiet and smooth. But when I also hook up a bench or handheld true DMM (no-name brand), I get a lot of noise on the oscilloscope trace due to the DMM input stage switching. Therefore the AD8436's output must be fairly high impedance?

    Regards,

    Peter

  •   

    2. Re: AD8436 RMS-to-DC

     

    Currently Being Moderated

     

    Hi JS,

    I've now built the AD8436 RMS-TO-DC converter module for evaluation. I've been "parallel" comparing it against the AD737 module.

    The AD737 appears to work correctly. It does require a trimpot adjustment on the Cc pin (pin-1), to trim the output to get the appropriate DC level.

    You're seeing an error because you've adding quite a lot of impedance in series with the averaging capacitors when you use bipolar transistors to switch the caps. In the AD737 data sheet, you'll find a circuit in fig 26 I actually built and tested. It uses a discrete MOS device (very low ON-resistance which basically adds no error. What's worse, the error will be non-libnear so you can't calibrate it out. You need to use a relay of some sort, or a solid state approach - a single channel MUX sould be OK for you mult-averaging cap circuit. You should see no more than 0.5% error, typically the AD737 delivers << 0-2%.

    The AD8436 doesn't have any adjustment. I am finding that for 200mV RMS 1kHz at the module input, it gives 190mV DC at the module output. That is within 5% and probably quite reasonable.

    190mV is pretty awful, we'd stop shipment if the parts were really that bad. It seems you made the same switch, so try a different switch type as recommended above.

    1) However, where would you add fine level adjustment to the AD8436 circuit - to get it more precise?

    Should I assume that there is no difference between adding it BEFORE or AFTER the RMS core?

    You might even suggest that the AD8436 needs no such adjustment.

    Neither part should require an adjustment, especially if you think 5% is good .

    That question needs to be taken in regards to this next RMS versus AVERAGE issue.

     

    On both modules, I apply 200mV RMS 1kHz and monitor the DC output on an oscilloscope (see below for why a DMM is not used).

    For RMS mode, I should get 200mV DC (for 200mV RMS input). The AD737 is fine - but I'm already reading slightly low (190mV) for the AD8436 module (described above).

    Yet, when I select AVERAGE mode on both modules, I actually seem to get the correct output voltages on both modules.

    On the AD737 module, if I select AVERAGE (pre-Amplifies input signal by 1.11), then remove any capacitors from the Cav pin (pin-5), I get 200mV - which seems to be what I'd expect.

    On the AD8436 module, if I select AVERAGE (pre-Amplifies input signal by 1.11), then remove any capacitors from the Cav pin (pin-19), I get 200mV - which also seems to be what I'd expect.

    Both front-end amplifiers show identical levels (Av=+1.00 and AV=+1.11).

     

    2) Should the AVERAGE mode on the AD8436 work as I expect (similar to the AD737)?

     

    As an aside, I generally connected the AD8436 output to an Oscilloscope and the trace is reasonably quiet and smooth. But when I also hook up a bench or handheld true DMM (no-name brand), I get a lot of noise on the oscilloscope trace due to the DMM input stage switching. Therefore the AD8436's output must be fairly high impedance?

    Peter the averaging cap connections create such an error that I wouldn't care to comment further until these are corrected. I can see you've put a lot of work into this project so you deserve to have it come out with shining success. Please check out the options you have for changing to some sort of 0 (or near 0) Z connections for your switching and let's see what happens. I believe you'll be delighted with the results.

    BTW I take it you're a student, maybe I can commend you r efforts to you professor? You're work is quite impressive so don't discouraged with a few errors.

    Best,

    JS

  • This question has been assumed as answered either offline via email or with a multi-part answer. This question has now been closed out. If you have an inquiry related to this topic please post a new question in the applicable product forum.

    Thank you,
    EZ Admin