Precision Chopper Circuit

  • The attached file describes a chopper circuit that has the provision of precisely measuring power. It is based on the fact that a low resistance (~2 Ohm) heating resistor has a <1 uH inductance and therefore appears resistive at a~20 Khz switching frequency. Since the chopper is powered from a DC supply, it is straightforward to measure both voltage and current (and therefore power) to a high degree of accuracy. I have built this circuit, and it functions as designed.


    It has advantages over triac or phase control circuits in that both the supply voltage and duty cycle can be precisely controlled. Power output can be controlled either by varying the duty cycle or by adjusting the power supply voltage. I have configured it to do the latter because supply voltage is more easily measured and controlled than duty cycle. As designed, the circuit generates 75% duty cycle, 21 KHz pulses at any voltage up to the limit of the supply which is 40V.

  • Jeff,


    This circuit is fine and a good way to go but you have to be very careful in calculating power.


    In fact Rossi did this wrong and got a 2.5X uplift in apparent COP...


    You are measuring (certainly) average power and (I guess) average voltage. If you use a true RMS voltmeter you will get inconsistent results based on the sample rate and the 20kHz pulse frequency. But even if consistent you will get different results using an average voltmeter and a true RMS voltmeter. The best bet is voltage measurement with a scope in which case you just need to be carefuly about scope and (if X10) probe calibration.


    Once you have got average voltage (Vav), and average current (Iav) the true average power delivered is:
    (Vav * Iav) / delta


    where delta is the duty cycle between 0 and 1.


    This is a bit counterintuitive, and can easily lead to large underestimate of input power if you don't take it into account.


    To check this works take 50% duty cycle and 1V / 1A peak voltage / current - the averages are 0.5V, 0.5A => 0.25W. But the actual power is 1W for 50% of time => average power is 0.5W.

  • See the attached file. It demonstrates that the voltage and current sense points shown in the previous post yield identical averaged power values when compared to taking instantaneous I and V values in series with and across the load plus the switch MOSFET.

  • See the attached file. It demonstrates that the voltage and current sense points shown in the previous post yield identical averaged power values when compared to taking instantaneous I and V values in series with and across the load plus the switch…


    I have no doubt of that. But I hope you agree with me that without the 1/delta adjustment the calculated average power will be totally wrong?


    I'm getting a bit worried now that there may be a few people misled by this into thinking they have COP of 2 or so... But maybe no-one is using this circuit yet.

  • I made a decision some time ago to avoid the whole potential mares nest of problems associated with AC measurement as LFH is lacking a good Fluke power meter (or similar) bu using switched-mode DC power supplies readily available and inexpensive. Fan-cooled and capable of giving out 400W at up to 48V continuously. This way (since V is very very stable) all I have to measure is current. What simp;licity, what joy!

  • From the attached circuit diagram it seems you are measuring the power supply voltage behind the switch so it should (theoretically) be steady. The supply current would be a chopped signal. If D is the duty cycle (between 0 and 1), assuming Iavg = Ipk*D, Pinstant = Ipk*Vdc, Pavg = Pinstant * D then Pavg = Ipk*Vdc*D which is Pavg = (Iavg/D) * Vdc * D which is Pavg = Iavg * Vdc. Please check the calculation but it should be possible to record the actual power going into the resistive load assuming that you have an accurate average current. Therein lies the issue. Can you trust a meter to give you an accurate average current? I assume the lowpass filter on the current measurement is intended to aid with that? As Alan Smith pointed out, steady DC voltages and currents are far easier to measure. Since you already have a variable DC power supply, why the extra complexity of a chopper?

  • Yes, I agree that unless you know what you are doing it is easy to obtain incorrect power readings for phase controlled AC, or for that matter, for chopped DC. I would advise anyone who wants to use such techniques to get a copy of SPICE (Linear Technology has a free version called LTSPICE) and simulate the circuit first.

  • The SPICE model I developed only simulates the power MOSFET, the heater resistance, and external components required to monitor power via I/V measurements. In other words, the ICs necessary to generate the pulses and drive the MOSFET are not modeled but are instead abstracted into an ideal pulse generator. What I can do, however, is post both the complete schematic and the SPICE model for the power MOSFET circuit.


    Jeff

  • Quote

    As Alan Smith pointed out, steady DC voltages and currents are far easier to measure. Since you already have a variable DC power supply, why the extra complexity of a chopper?


    As glowfish points out above, I use steady DC inputs. And the output terminals of the DC PSU is the point where I measure the power input to the reactor. Any special magic frequencies, PWM, and my current project, a pseudo 3-phase square wave unit come after this point. Losses here are relatively small, but as the reactor system LFH designed and uses only ever compares 'control' with 'active' hot zones in the reactor these losses are not important.

  • I often wonder why folks don't just measure the power where it goes into the apparatus from the wall outlet or whatever. Should be far less noisy there.
    The extra power for the control should not be so much, and really is part of the required effect if it is needed.

  • @Paradigmnoia


    The issue with power measurement is the accuracy, resolution and capability of whatever it is that you are using for measurements. Accurate measurements of AC power require true RMS meters which can be pricey. Added to the complexities is that depending on what you have connected (and the quality thereof), even switchmode supplies, triac choppers etc draw "spikey" currents from your wall socket which are very tricky to measure. Normal run-of-the-mill multimeters which are commonly available might not give you good readings under these conditions. The best is if both the current and the voltage that are being measured are steady DC. If you add choppers etc to the system then you need to add filters to the supply. Just measuring at the wall plug with no filtering could give you variable results. It is also easier to measure DC with a (slow) computer connected ADC which is what you would probably need if you want to keep a streaming record of experiment results.

  • Quote

    Just measuring at the wall plug with no filtering could give you variable results. It is also easier to measure DC with a (slow) computer connected ADC which is what you would probably need if you want to keep a streaming record of experiment results.


    And for Arduino-based data-logging it is important to have a 'quiet' voltage - AC SCR's etc drive them crazy. As for measuring at the wall socket- Killawatt type digital power meters are cheap enough but I'm not sure how reliable they are. And another thing - my lab space is at the other end of a 50M length of 60A armoured cable coming from very close to the grid connection. Random switching of the lab electrical space heating system causes a mains voltage variation at the lab end of the cable between 232V and 224V. While this 8V shift in supply level leaves a switched mode DC PSU entirely unmoved, when driving the reactor heater coils directly from the main supply via an SCR you can actually see the glow of the red-hot coils varying.

  • As long as your equipment is decent, it should catch wandering V, ect. My coils see external loads like refrigerators, especially at high powers, where they are rather sensitive to extra current.
    Big Momma coils like Lugano won't care too much.
    Specifications: http://www.steam-engine.org/coil.asp?a=true&p=roundmulti&tp=2.2&r=0.4&hfnw=310&str=2&awg=14&wl=748.8186500359949&id=10&ll=200&ws=20
    (I couldn't find 15 Ga wire, but still looking).

  • The chopper circuit was designed to operate either in a chopped or a DC mode. The purpose of using a chopper is to be able to drive harmonic rich power to the heater element. There is some evidence that doing so may enhance LENR activity.

  • @Alan Smith
    I have some 16 Ga already, thanks.
    15 Ga was mentioned in one of the IH-Rossi patent applications, so I was looking for some of that size for a while.
    Seems to be special order stuff. And a pain to use as heater in short lengths.
    I was just attempting to see if a Lugano Coil as described in the application was feasible. I managed with 14 Ga, so it is.
    Except for the 2.650 Ω/ft specifcation, which does not seem possible in wire of anything like 15 Ga.

  • New to this thread, but I saw several inaccuracies from the previous posts.
    Power measurements cannot use the average (mean) for current or voltage unless it is DC or a perfect square wave. Only RMS values can be used because power is I-squared * R or Vsquared/R.
    For instance, the average of a triangle wave is .5, but the RMS value is 1/sqrt3 = .577. Using .5 instead of .577 to compute power input would give you 15% apparent excess power. The apparent excess depends on the actual waveform.
    The best way to measure is to simultaneously measure V and I, and then average the V*I samples. That is what the power meters do. But the sample rate must be much higher than the highest frequency you are measuring.


    It is also not correct to multiply RMS voltage times RMS current unless the load is purely resistive. If the load is inductive, the current and voltage are out of phase and Vrms*Irms does not equal the actual power delivered.


    The Spice model shows some series inductance, but is not accurate because it assumes that resistance remains constant. The resistance of a heater element changes by a large amount as it heats or cools and you cannot use the cold resistance alone to determine power.


    Also, for a heater, the resistance of the leads to the heater element should not be neglected. The voltage should be measured with a separate set of test leads connect at the ends of the heater element. This effect also makes measurement of wall outlet power inaccurate, although at least this error is in the right (conservative) direction - it will overestimate input power by the amount of power delivered to the heater leads and the electronics.

  • @Robert Horst


    While I basically agree with all you say, the error from ignoring the coil inductance is very likely OK. That is because:


    (1) It is conservative: P = Vrms*Irms*cos(phi)
    (2) For a small reactive phase shift the correction is second order (cos(phi))


    You do have to be careful because the power waveform has significant components well above 100Hz due to the Triac drive. And I agree that for the same reason any digital sampling meter (which includes all true RMS meters) needs to have a high enough sampling rate to capture the spikes.

  • Using a programmable power supply and measuring the current and voltage (at the load) would give the most accurate measure of applied power but would not achieve your requirement for high harmonic content. Measuring the (IV) power coming out of the power supply would give a reasonable measure of the power supplied to the pulse circuit. However, if you are pulsing the load current you will probably have some unexpected loss of power due to the ripple current flowing in the output capacitors on the power supply due to their ESR.

  • I use a home brew 4 channel power supply- up to 2.5kW at 50V max. 3 channels have variable voltage current-limiting PWMs, which read-outs that give output (conveniently) in watts as well as VA. The 4th channel is a fixed 9V output dedicated to power my data-logging stuff.


    The pictures show 2 channels feeding into H-Bridges which create square wave AC over a huge range of frequencies. The square waves go in turn to supply the heater coils in two closely matched reactors.


    All good fun, even better than watching football.