[sdiy] Those AS33xx chips again...

[Nils Pipenbrinck]

The discussion of temperature coefficient of capacitors just reminded me
of something.

Last year I was doing some RF development and in a high resonance LC
resonator even the temperature coefficient of C0G caps showed up as
detuning of the resonance frequency. I'm only speaking of a drift of 10°
to 15° degrees here. That was quite a surprise for everyone because the
textbook sais C0G don't do this. Well - they did nonetheless.


To solve the problem I did some research on temperature compensating
capacitors. There are a lot of choice with different coefficients from
single digit picofarads up to half a microfarad.

If your design has a systematic but small temperature drift you can
often just put one of them in parallel to your integration cap.

(we ended up with a solution of just directing the heat away from the
cap by doing a ground-plane split in the PCB. That was good enough for us).

/Nils



On 11/29/2017 02:49 PM, Roman Sowa wrote:
> thank you for your comments, here are mine:
> 
> 1. I know it's not the only source, that's why I put a little note at
> the bottom I plan to build another board witch few changes, in specific:
> get rid of trimmers, and use lower voltage reference for current
> reference in expo pair as there's not great variety of 25ppm resistors
> in megaohm range. Also will try different capacitors. Currently there is
> Panasonic ECH-U which is PPS film.
> According to their poorly scaled graph in datasheet I can assume tempco
> is around 250 ppm. For some reason I thought they were far better. Maybe
> I'll just replace with COG, or build another one with 20 times more
> expensive mica.
> Funny thing, I was already holding the COG 1n cap in order to solder it,
> then I thought "naaah, let's make it better and use super fancy ECHU cap!".
> 
> 2. I don't think I'll ever test more than a handful of chips, so we'll
> leave that uncertainity open for future generations if the ringbump is
> trademark of AS33 or simply manufacturing variation. Anyway, the bump is
> small, merely 120mV for half a microsecond. Easily flatten by following
> opamp. What is more disturbing is the step at lower frequencies. Here
> triangle corner is missed by 40mV, while CEM - 20mV. That also could be
> chip-to-chip variation and may turnout the other way.
> 
> 3. couldn't agree with you more
> 
> Roman
> 
> W dniu 2017-11-29 o 10:45, Steve Lenham pisze:
>> Very interesting and useful - thanks Roman! My immediate thoughts:
>>
>> 1. Re the tempco, the 3340 is not the only potential source of
>> variation. What dielectric are you using for the timing cap? C0G/NPO
>> is fairly blameless, but others have significant tempco (I remember
>> the tempco of a polystyrene timing cap being used to cancel an
>> opposite tempco in a classic National Semiconductor appnote). Even
>> trimpots have thermal drift.
>>
>> 2. Re the different voltages at the bottom of the triangle, this will
>> be set by the offset voltage of the internal comparator. One would
>> need to test multiple examples of each type to determine whether there
>> really is a difference between the CEM and AS parts or whether it is
>> simply chip-to-chip variation. The original CEM3340 datasheet
>> specifies that the triangle bottom point can vary between +15mV and
>> -15mV.
>>
>> 3. I don't know how much they matter in real applications, but the
>> larger glitches on the triangle midpoint and bottom look like the
>> biggest differences between the two devices. They look like the sort
>> of thing you get when the opamp/comparator has slightly too much
>> bandwdth - a case of modern IC technology being a bit too good? For
>> example, there are certain vintage sawtooth VCO designs where you have
>> to use a 741 for the main opamp, otherwise you get a colossal spike at
>> the sawtooth reset; the sluggish 741 simply isn't capable of doing it.
>>
>> Cheers,
>>
>> Steve L.
>> Benden Sound Technology
>>
>>
>> On 29/11/2017 02:04, David G Dixon wrote:
>>> Interesting!  I would have expected better performance.  The
>>> waveforms look a little dicey and the tempco is definitely not too
>>> great.
>>>
>>>    
>>> ------------------------------------------------------------------------
>>>     *From:* Synth-diy [mailto:synth-diy-bounces at synth-diy.org] *On
>>>     Behalf Of *Roman
>>>     *Sent:* Tuesday, November 28, 2017 2:34 PM
>>>     *To:* synth-diy at synth-diy.org
>>>     *Subject:* [sdiy] Those AS33xx chips again...
>>>
>>>     During few last days I've spent wonderful time with AS3340 VCO,
>>>     giving it a closer look, and also a few measurements. Nothing
>>>     spectacular, just something that any geeky nerd would find
>>> interesting.
>>>
>>>     It's about vintage chips, so it seems appropriate to have it written
>>>     in ancient vintage static html, basicaly the 90's style.
>>>     http://www.sowa.synth.net/synthchip
>>>
>>>     Roman
>>>
>>>
>>>
>>> _______________________________________________
>>> Synth-diy mailing list
>>> Synth-diy at synth-diy.org
>>> http://synth-diy.org/mailman/listinfo/synth-diy
>>>
>>
>> _______________________________________________
>> Synth-diy mailing list
>> Synth-diy at synth-diy.org
>> http://synth-diy.org/mailman/listinfo/synth-diy
>>
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at synth-diy.org
> http://synth-diy.org/mailman/listinfo/synth-diy
>