Bob Pease circuit for measuring dielectric absorption

In my current measurements obsession, I've stumbled on this [well-known here] page:
http://www.national.com/rap/Application/0,1570,28,00.html

Wonder if anyone tried the circuit in fig.5.
Was thinking [without the needed skills] about something similar, its so nice to find the job already done 

I haven't done dielectric absorption measurements, although this is on my to-do list too. Here's a paper which might interest you: da.pdf

Also check the Cyrill Bateman articles, IIRC he writes about the topic as well.

Samuel

DA has been well explored by designers of sample and hold circuits. US Audiophiles took their run at reinventing it in the '80s (google Jung, Curl, Marsh, et al). The primary difference between most audio applications and sample and hold applications is that S&H alternately charge from a low impedance, then rest or hold with a high impedance load. This is the worst case for DA expression. Audio circuits typically terminate with a fixed impedance and for DC blocking applications have very little terminal voltage change to interact with the distributed internal RCs.

IMO this is overstated as important in audio DC blocking applications, but can be important for filter components that will experience changing terminal voltage (model looks like several smaller RCs in parallel). DA can also be a source of errors in time constant circuits used inside dynamics processors, but the deviation here from ideal is not audibly problematic unless you are dealing with symmetrical encode/decode circuits like used in companding NR where encode/decode capacitors should be similar to track.

JR

PS: Sam,,, FWIW this is not so much of an issue in the S&H considered for the AGC loop of your oscillator since the voltage being held will not vary significantly from sample to sample. For that application I would be more concerned about leakage, vs time and temp.

DA has been well explored by designers of sample and hold circuits.

Click to expand...

I've wondered about RAP's comment at the end of the article that his writing should be the first article on DA; but indeed I've not found any earlier reference, although I'm sure the effect was well-known before.

IMO this is overstated as important in audio DC blocking applications.

Click to expand...

I've never understud the rant about it--it just has a very slight effect on frequency/phase response.

FWIW this is not so much of an issue in the S&H considered for the AGC loop of your oscillator since the voltage being held will not vary significantly from sample to sample. For that application I would be more concerned about leakage, vs time and temp.

Click to expand...

Indeed, at least for the second stage. There are some nice (and simple) topologies to cancel hold step and leakage.

Samuel

RAP may have been the first to write it up in the popular trade press, but this was already well discussed in the audio hobby press before that ( late 70s), and I even wrote about in my Audio Mythology column around that same time as RAP (early '80s). The Capacitor and instrumentation engineers surely knew about it several decades earlier.

Earlier non-audio references from the Curl/Jung article. 

1-Dow, PC. "An analysis of certain errors in Electronic differential analyzers II, Capacitor Dielectric Absorption" IRE transactions on electronic computers (Vol EC-7) March 1958.
2- Dummer, GWA and HM Nordenburg, "Fixed and Variable capacitors", McGraww-Hill, 1960.
3-Application manual for computing amplifiers, Philbrick/Nexus research, Dedham Ma, 1968. p 86-87.
4- Korn, GA and TM Korm, Electronic Analog and Hybrid Computers, 2nd edition, Mcgraw-Hill, 1972

JR

Thanks a lot, Samuel and John.

Actually, I'm curious now on how [and _if_] the dielectric absorption of various kind of capacitors [the C2s] could influence the behaviour of the VCOs of a modular synth I'm re-building.
[its an Elektor Formant]. VCO

And yes possibly also the filters sections of other audio gear projects I've here, slowly waiting for all the parts needed.

I've a huge amount of recycled/surplus/offers caps, so I'd want to sort them out, depending on use, for value, withstand volts, Q factor, leakage, DA, parasitic inductance, thermal sensitivity, maybe ESR.
So far:
- value/Q factor/parasitic inductance/ESR: an LCR bridge meter
- withstand volts: a cheap "transistor tester" [DY294] feature for measuring V(BR) [200V and 1000V range, current limited to below 1mA, still dangerous btw], and a Sencore LC75
- leakage: Sencore LC75
- DA with variable parameters, if useful: hopefully the circuit in page linked

In general I would try to stick to similar technology capacitors to avoid inadvertently changing some sound characteristic. Cleaner may not be better in a synth.

JR

> I'm curious now on how [and _if_] the dielectric absorption of various kind of capacitors [the C2s] could influence the behaviour of the VCOs of a modular synth

Much less than dozens of other "small" effects. T1 internal resistance and offset voltage, IC1 characteristics, ohmic losses in IC3.....

Reset resistance adds a constant dead-time to every cycle. Exponential ohmic loss bends the top of the exponential curve.

A well-designed VCO will have musically tolerable frequency error over many octaves, always goes flat in the "top octave", with musically annoying pitch error at higher frequency. You design for decent linearity to some high frequency, then use lower frequencies. (At the other end, leakage currents cause error instead.)

That VCO is simple. ARP's were far more complicated, and still had audible error. But mostly the ohmic loss (and maybe DA) is insignificant compared to general drift, unless you specifically test for it.

The tempco of T2 and IC1 seems bound to drift. ARP used a -much- more elaborate comparator. (The early ARPs were perhaps a tutorial in over-design.)

It's a fine VCO as-is. Using fancy caps is not going to make it better. Much of the art of Moog/ARP/eMu work is about managing pitch. Quadrupling complexity (from Elector to ARP) does not eliminate that.