Circuit to Convert Pulse Width of 15-30kHz Pulsed Sine Wave to Voltage Output

Can anyone help me with a circuit that will provide a voltage equivalent to the pulse width of a pulsed low voltage sine wave? The typical pulse width is in the range of 15-30 Khz and the Pulse Frequency can vary between 2 kHz and 20 Khz. I've looked at several circuits that will measure the pulse width of a continuous sine wave, but have not been able to find one that will work with a pulsed sine wave. 
Thanks,Brian  

Hi Brian- I think I understand what you want to do, but let's talk about the terminology first. The term "pulse width" is generally expressed as units of time (millisecond, microseconds, etc.) rather than frequency.The term "pulse" is often used for square (on-off) waves.  For sine waves the term "cycle" is more frequently used.In the drawing you provided, each "on" period consists of one cycle.  Is this actually the case, or can there be more than one cycle in each "on" period?If there is more than one cycle in each "on" period, it is frequently referred to as a "burst".The time between bursts is usually not called out as a frequency, but as units of time. If you look at it in terms of "how many of these busts happen in a second" then it is referred to as "repetition rate" which can be specified as a frequency. Now- after having totally confused you- back to the original question- One approach would be to use an oscilloscope to measure the pulse by moving cursors on the screen to highlight the area of interest and then you can read the amount of time between the cursors.  -Rick   

BrianI have several questions including some asked by Rick:1. Is there only ever one cycle in each "on" period?2. What response time do you need- from measured input to voltage output? 3. If there is more than one cycle in the "on" period, is the "on" period a fixed width?4. Is there a fixed repetition rate?If you have multiple cycles you may be able to measure the frequency (the more the better)- and with a fixed on period and/or repetition rate you can sample and hold each reading. Once that is done, converting to a voltage is a secondary matter.If the answer to 1 is yes (or if there are only a few cycles), then you are probably going to have to measure the cycle time and calculate the frequency from that (the inverse, as I am sure you know).In either event you are going to have to used a zero-crossing detector to digitise the waveform, but if you only have one cycle you are going to have some inaccuracies- you will only be able to detect a certain amount above the 0V line and probably have to work with 1/2 the period to measure the first crossing. You could probably compensate for this in software (I am just about to get to that). Just a reminder +/-10V will probably kill any micro or logic you connect it to.IMHO the easiest way to implement this, for either case, is to use a micro with suitable internal hardware. I can testify that the PSoC can do this and has the versatility to modify the circuit as if you were working with hardware alone. But many other processors will be able to realise the function as well. If you are using the cycle time approach, then you may want to average several (hence question 2) to generate the voltage out. Of course voltage out may be generated from a PWM or D/A (internal or external). There will be a direct relationship from cycle time (again if you take that approach), so you may not have to invert to get the frequency.

Thanks for your response.  I appreciate it.  
Each “ON” period does consist of just one cycle, not a burst.  The sine waveshown is generated by a thyristor controlled series resonant circuit.  The thyristor is self commutating so each time it is gated it gives one cycle and then turnsoff, waiting for the next gating signal.
The repetition rate of the gating signal can vary at any time between about 50-500us.
Cycle width can vary between 20-50us depending on the inductance of the resonantcircuit.  This inductance is dynamic and can change from cycle to cycle.  I can watch this variation on the scope with no problem.
I’m trying to generate a voltage that is proportional to the cycle width so that I can usethis voltage to control a related circuit.  I’ve been able to develop a circuit that  works fine for a fixed repetition rate but it can’t handle the variable repetition rate.  (I rectifythe sine wave, clip it to a fixed amplitude then read the area of the wave using anRMS to DC converter.)  If I know the repetition rate I can generate a voltage representativeof the cycle width.
Any thoughts that you have would be appreciated. I’ve attached the basic circuit that I’veused successfully with a fixed repetition rate.

Thanks for your response.  I appreciate it.  
Each “ON” period does consist of just one cycle, not a burst.  The sine waveshown is generated by a thyristor controlled series resonant circuit.  The thyristor is self commutating so each time it is gated it gives one cycle and then turnsoff, waiting for the next gating signal.
The repetition rate of the gating signal can vary at any time between about 50-500us.
Cycle width can vary between 20-50us depending on the inductance of the resonantcircuit.  This inductance is dynamic and can change from cycle to cycle.  I can watch this variation on the scope with no problem.
I’m trying to generate a voltage that is proportional to the cycle width so that I can usethis voltage to control a related circuit.  I’ve been able to develop a circuit that  works fine for a fixed repetition rate but it can’t handle the variable repetition rate.  (I rectifythe sine wave, clip it to a fixed amplitude then read the area of the wave using anRMS to DC converter.)  If I know the repetition rate I can generate a voltage representativeof the cycle width.
Any thoughts that you have would be appreciated. I’ve attached the basic circuit that I’veused successfully with a fixed repetition rate.

BrianLet's see if I can explain this more clearly than I did before. I assume you cannot get at the thyristor trigger.Take a comparator and set the trip point for just above ground. Let's assume the reference voltage is connected to the negative and the input signal to the positive. When your signal starts and goes positive the comparator will produce a positive signal at the output until the input signal drops below the reference voltage. If you measure the pulse width Tpw of this signal it is almost 1/2 the period of the full signal. It is possible to use hardware to then take this number and covert it to your output signal, but it is much easier to use a micro. You haven't precluded this option so I am going with that. Many micros have the ability to measure elapsed time. A known (and normally convenient) frequency F is fed to a counter and the incoming pulse (from the comparator) is used to gate the counter. Obviously the period Tpw is the count N multiplied by the period of the incoming frequency. So Tpw=N*(1/F). Because of the non-zero comparator trip voltage, this is approximate and you can use the micro to adjust the signal, and it is even possible to calibrate for more accurate readings by working with known values.You can then use the transition to low level of the incoming comparator signal to initiate the calculation of the output signal. There must be some relationship between the pulse width of the single cycle and the output voltage. You have not provided any details of that relationship but in any event you can calculate what the output voltage will be.  Some micros have an internal D/A or you could use an external D/A. You could also use a poor man's D/A- the filtered PWM output.You will also need to monitor for the next input pulse (whilst holding the current output) to restart the process. There surely must be some timeout on the repetition of the pulses beyond which the output is zeroed. Let me return to the comparator input. If you are using a single supply comparator, the comparator may have a problem with input voltages around zero, and so the input reference would have to be, say 0.6V and you would be out by about 12% on your measurement. If you use a dual supply on the comparator (perhaps using a switched capacitor DC/DC converter like the ICL7660) you can drop the input reference to something closer to zero.Brian, I  am not sure of your experience, so please forgive me if I told you stuff you already know.