Testing the TK Tar Baby

[Rosemary Ainslie]

Indeed Rosemary,

I've read parts 1 & 2 as well.

Presently (and formerly as well), my concern is regarding the use of two different load types, and the direct temperature measurement on the load elements.

Ideally, both loads should be the same part number, and both loads should be immersed in some fluid that won't evaporate at moderately high temperatures. Then the temperature probe is placed in the fluid rather than in close contact with the load element.

Since identical loads can not be accommodated, for a true comparison it is equally, if not more important that the loads be immersed in fluid for the temperature measurements.

Fair comment Poynty Point

I think we'll run the batteries down as detailed - and then conclude with a final test showing the resistors in some kind of material to show that the temperature rise is equivalent over both tests.  But that will be the concluding test.  If there's no gain in the first test then there's probably no point in that final test.

Regards again
Rosie

added
I take it that you see the relevance in those subtle temperature changes to reflect the energy actually being dissipated?  It's a good parameter.

As ever,
R

[picowatt]

Before I comment, I'd like to ask for some input from the folks here. Thanks.

Poynt,

I am considering using an aluminum cased wirewound power resistor mounted to an aluminum thermal mass/heat sink and mounted in a vertical chimney with a CPU fan and controller maintaining constant air flow.  Thermister would be buried in the thermal mass and its immediate surroundings/wires insulated to ensure the temp measured represents the bulk of the thermal mass.  Inlet air temp and humidity would need to be maintained fairly constant throughout all tests.

As to the equivalent load applied to the battery, I do not like this fixed load and separate resistor idea.  Will a flooded lead acid battery (or any chemistry) deliver more amp hours if the applied load is a 10ohm load applied at a 10% duty cycle or if the applied load is a constant 100 ohm?  If this is the planned test, than the assumption that the amp hour rating is similar for both loads must be confirmed via rigorous testing.

I would use the same load throughout all tests.  For a comparison battery run down test, I would pulse a very well controlled, low RDSon MOSFET at a duty cycle that provides an equivalent dissipation in the load as was indicated by the NERD circuit tests.  This way the battery sees the same pulsed load but without the oscillations and MOSFET losses inherent to the NERD circuit. 

As for measuring the assumed 10.6V per battery cutoff voltage, the circuit used to measure/monitor this must be filtered fairly heavily.  First, all oscillations must be removed from this measurement to ensure they do not affect the cut-off point.  Second, as the batteries approach 10.6V, their internal resistance will increase and cause voltage drops during pulsed loads which must also be filtered from the measurement.  The use of similarly pulsed loads as described above would reduce this issue a bit.

I agree with TK in that the proposed test does not lean in the NERD's favor, as MOSFET dissipation is not accounted for.

What are your thoughts?

PW

   

[picowatt]

Indeed Rosemary,

I've read parts 1 & 2 as well.

Presently (and formerly as well), my concern is regarding the use of two different load types, and the direct temperature measurement on the load elements.

Ideally, both loads should be the same part number, and both loads should be immersed in some fluid that won't evaporate at moderately high temperatures. Then the temperature probe is placed in the fluid rather than in close contact with the load element.

Since identical loads can not be accommodated, for a true comparison it is equally, if not more important that the loads be immersed in fluid for the temperature measurements.

Poynt,

If the load resistor must be some "special" type, I have considered using fluid immersion in a cylindrical container and then mounting that in the chimney discussed in my last post.  The use of a fluid filled container could be made to accomodate all manner of odd shaped resistors, such as the March "demo" resistor.

PW

[Rosemary Ainslie]

I am considering using an aluminum cased wirewound power resistor mounted to an aluminum thermal mass/heat sink and mounted in a vertical chimney with a CPU fan and controller maintaining constant air flow.  Thermister would be buried in the thermal mass and its immediate surroundings/wires insulated to ensure the temp measured represents the bulk of the thermal mass.  Inlet air temp and humidity would need to be maintained fairly constant throughout all tests.

It would be more interesting if you actually ever performed any tests at all - with respect.  You've been promising this from your early adoption of that 'picowatt' name and have, thus far, given us no confirmation of your receipt of those multiple FET's that you apparently ordered - let alone any indication that you are in a position to move yourself off your armchair to build anything at all.  But if you prefer to consider things - then I see that you also need much time to complete all that deliberation.

As to the equivalent load applied to the battery, I do not like this fixed load and separate resistor idea.  Will a flooded lead acid battery (or any chemistry) deliver more amp hours if the applied load is a 10ohm load applied at a 10% duty cycle or if the applied load is a constant 100 ohm?  If this is the planned test, than the assumption that the amp hour rating is similar for both loads must be confirmed via rigorous testing.

Where gains are marginal there is no question the concern will be whether the battery is advantaged by a switched cycle.  But we're anticipating unequivocal gains. 

I would use the same load throughout all tests.  For a comparison battery run down test, I would pulse a very well controlled, low RDSon MOSFET at a duty cycle that provides an equivalent dissipation in the load as was indicated by the NERD circuit tests.  This way the battery sees the same pulsed load but without the oscillations and MOSFET losses inherent to the NERD circuit.

We are NOT interested in the heat dissipated at the MOSFETs.  At best it will only ADD to the proof of gains.  We're satisfied that the heat dissipation at the resistors will be equivalent - ELSE there's NOTHING dependable in standard measurements.  And we'll be able to prove this is an final test - if such a final test is ever required.  Then the resistors can be immersed in an equivalent quantity of water or oil to prove an equivalent temperature rise.

As for measuring the assumed 10.6V per battery cutoff voltage, the circuit used to measure/monitor this must be filtered fairly heavily.  First, all oscillations must be removed from this measurement to ensure they do not affect the cut-off point.  Second, as the batteries approach 10.6V, their internal resistance will increase and cause voltage drops during pulsed loads which must also be filtered from the measurement.  The use of similarly pulsed loads as described above would reduce this issue a bit.

This is ridiculous.  How does one test a switched circuit if there's an imposed requirement is to prevent that switch?    Please! or 'Dear God help us all'  or LOL.  (MileHigh - take your pick)

Rosie Pose

[picowatt]

It would be more interesting if you actually ever performed any tests at all - with respect.  You've been promising this from your early adoption of that 'picowatt' name and have, thus far, given us no confirmation of your receipt of those multiple FET's that you apparently ordered - let alone any indication that you are in a position to move yourself off your armchair to build anything at all.  But if you prefer to consider things - then I see that you also need much time to complete all that deliberation.
Where gains are marginal there is no question the concern will be whether the battery is advantaged by a switched cycle.  But we're anticipating unequivocal gains. 
We are NOT interested in the heat dissipated at the MOSFETs.  At best it will only ADD to the proof of gains.  We're satisfied that the heat dissipation at the resistors will be equivalent - ELSE there's NOTHING dependable in standard measurements.  And we'll be able to prove this is an final test - if such a final test is ever required.  Then the resistors can be immersed in an equivalent quantity of water or oil to prove an equivalent temperature rise.
This is ridiculous.  How does one test a switched circuit if there's an imposed requirement is to prevent that switch?    Please! or 'Dear God help us all'  or LOL.  (MileHigh - take your pick)

Rosie Pose

Maybe you should try to understand what people are saying before dragging out words like "ridiculous".

Why is Q1 not turning on in FIG 3?