Testing the TK Tar Baby

[TinselKoala]

Making progress there are we, .99? I see someone dodged your question about the resistor that was sent, while confirming that they did send one. Of course someone can't be arsed to get up, go find it, and LOOK AT IT and tell you what it is.

[mrsean2k]

@poynt


I think there's room for confusion with your answer to...


"IS it the rule that the measured rate of current flow through a circuit must be consistent.  In other words it cannot be greater at one element than at another?"


I realise you're answering with a specific context in mind there (AC at a particular frequency range, in a particular circuit), but is it safe to assume you're sharing enough context with RA that obvious exceptions like DC parallel resistance won't be thrown back into the mix when it suits?

[picowatt]

@poynt


I think there's room for confusion with your answer to...


"IS it the rule that the measured rate of current flow through a circuit must be consistent.  In other words it cannot be greater at one element than at another?"


I realise you're answering with a specific context in mind there (AC at a particular frequency range, in a particular circuit), but is it safe to assume you're sharing enough context with RA that obvious exceptions like DC parallel resistance won't be thrown back into the mix when it suits?

I think .99 has been about as clear as possible without overcomplication that would surely only confuse.  He started out the discussion with DC conditions before moving on to AC.  Whether or not she can understand and visualize the difference between the two is another question.

The point is that even if Q2 is turned off with regard to DC, the AC current current path is ALWAYS there via the MOSFET capacitances (i.e., Ciss and Coss).  Stated otherwise, AC current can flow thru the MOSFET capacitances irregardless of the on or off state of the MOSFET.

In fact, when the voltage at the drains of the MOSFET's is at their minimum (less than 25 volts), the input capacitance (Ciss) for a single IRFPG50 MOSFET increases to over 7,000pF and the output capacitance (Coss) increases to over 5,000pF.  For the four Q2's in parallel, those numbers are 28nF and 20nF.  In other words, the MOSFET capacitances are not a singular fixed value, but are dynamic, and the capacitance values typically increase as the drain to source voltage decreases.  This is why the waveform in FIG4 slows down when Vds is below 25 volts (i.e., the longer charge/ramp time at the negative trough of the waveform in FIG 4).

Regarding inductance, she needs to understand that the amount of AC current flow is determined by, and limited by, the impedances at the observed AC frequencies, which requires adding or accounting for the inductive reactance of the components and wiring.

A 2" length of 12 gauge wire has about 40nHy of inductance, which has a reactance of about 1 ohm at 2MHz.  So, even if a .25 ohm non-inductive resistor is used for the CSR, if the 'scope probes are connected to a point 1" on either side of the CSR, it will cause the CSR to actually be a 1.25 ohm resistor at 2MHz.

A wire is not always a wire, and in the instance above, measurements made using just a 1" length of wire connected to each end of a perfectly non-inductive CSR will cause a very large error, indicating 4 amps when in reality only .8 amps are flowing (and that assumes using 12 gauge wire!).

Therefore, very short lead lengths are required.  In fact, it would be advisable to increase the CSR value to between .5 and 1 ohm to decrease the percentage error caused by lead inductance.  The best non-inductive resistor I have found has 10nHy when measured .2" from the package, which would add .112 ohms of reactance to a .25 ohm resistor at 2MHz, which represents a very significant percentage error.  Increasing the CSR value to .5 or 1 ohm would reduce the percentage of error caused by the .112 ohms of lead reactance (and again, that's .2" from the resistor package).

Also, as the NERD waveform contains a multitude of frequencies and harmoonics, power calculations must be made individually for all of the observed frequencies (at least the five most dominant frequencies, as would be observed in an FFT spectral plot).  One cannot simply use the indicated trigger frequency as the singular frequency contained in the waveform.  This would only be the case if the waveform were a perfect sine wave with 0% THD.  At the higher frequencies and harmonics contained in the waveform, inductive reactance becomes a more significant source of measurement error.     

As an aside, someone should let her know that if she is going to discuss resistor values using European notation, the "R" replaces the decimal point.  A 50 ohm resistor is 50R, not R50.  R50 would be a .5 ohm resistor.  More typically, a .5 ohm resistor would be written as a 0R5.  A 50 ohm resistor would be 50R, or 50R0, depending on the resistor"s specified precision.     

PW

ADDED:

An alternate method of reducing wire related inductance is to use larger diameter wires or use parallel runs of wire.

In the 2" of wire equals 40nHy used above, paralleling two 2" lengths of wire will cut the inductance in half, or to 20nHY.  Paralleling two more wires, that is, a total of four 2" lengths of wire in parallel, will again halve the inductance, now totaling only 10nHY.

In fact, an infinite number of parallel wires would represent a geometric plane with a given thickness, so, in order to reduce inductance in the ground and supply circuits of high speed printed circuit boards (PC motherboard, etc) entire circuit board layers are dedicated as the ground and power planes with as much solid copper area as possible on these layers (six to eight layer circuit boards are very common because of this).  For the most part, these solid ground and power plane layers act as infinite parallel runs to reduce inductance to a minimum.


 

   

[mrsean2k]

He started out the discussion with DC conditions before moving on to AC.  Whether or not she can understand and visualize the difference between the two is another question

That's what prompted me to comment really. That and the fact that .99 restricted the response to just answering yes to a summary in RA's own words, rather than stating the position in less ambiguous terms (but I suppose even Sisyphus fancies a rest now and again).

[TinselKoala]

And of course she has seen and commented upon my short video demonstration of a"measurement pitfall" using the simple Joule Thief circuit.

I show a dramatic difference in the scoped waveform, simply by moving the scope probe's reference lead to the other end of a 3" bit of wire in the circuit.

Her considered analysis of this video is... that it is faked in some way.

http://www.youtube.com/watch?v=KWDfrzBIxoQ