@ 1audio
It is correct that the design is basically a Wien bridge and is not unique. But it seems to work well. I will try to check the amplitude stability though.
Some of the reasons for the good performance could be:
1. The choice of op-amp.
2. The use of an LDR.
3. The very small regulation area, hopefully keeping the LDR linear.
4. The slow amplitude regulation.
5. The impedance level chosen. As Vojtěch Janásek writes, it is possible to use lower resistor values to reduce the noise, but at the expense of increasing the distortion.
Of course polystyrene capacitors were used.
There is of course the (remote) possibility of a measurement error.
The injection-lock option is basically just a resistor connected to the negative input of the first op-amp. Hopefully by adding a high value resistor here I can inject enough signal to lock it without disturbing the basic performance. And then there is a (low value) resistor on the connector to ground to keep it quiet it when nothing is connected.
It is correct that the design is basically a Wien bridge and is not unique. But it seems to work well. I will try to check the amplitude stability though.
Some of the reasons for the good performance could be:
1. The choice of op-amp.
2. The use of an LDR.
3. The very small regulation area, hopefully keeping the LDR linear.
4. The slow amplitude regulation.
5. The impedance level chosen. As Vojtěch Janásek writes, it is possible to use lower resistor values to reduce the noise, but at the expense of increasing the distortion.
Of course polystyrene capacitors were used.
There is of course the (remote) possibility of a measurement error.
The injection-lock option is basically just a resistor connected to the negative input of the first op-amp. Hopefully by adding a high value resistor here I can inject enough signal to lock it without disturbing the basic performance. And then there is a (low value) resistor on the connector to ground to keep it quiet it when nothing is connected.
Hello Jens,
woult it be possible to get a balanced (180 degrees out of phase as well) output from your excellent measureing oscillator with the same low THD for both phases?
/S
woult it be possible to get a balanced (180 degrees out of phase as well) output from your excellent measureing oscillator with the same low THD for both phases?
/S
@ 1audio
It is correct that the design is basically a Wien bridge and is not unique. But it seems to work well. I will try to check the amplitude stability though.
Some of the reasons for the good performance could be:
1. The choice of op-amp.
2. The use of an LDR.
3. The very small regulation area, hopefully keeping the LDR linear.
4. The slow amplitude regulation.
5. The impedance level chosen. As Vojtěch Janásek writes, it is possible to use lower resistor values to reduce the noise, but at the expense of increasing the distortion.
Of course polystyrene capacitors were used.
There is of course the (remote) possibility of a measurement error.
The injection-lock option is basically just a resistor connected to the negative input of the first op-amp. Hopefully by adding a high value resistor here I can inject enough signal to lock it without disturbing the basic performance. And then there is a (low value) resistor on the connector to ground to keep it quiet it when nothing is connected.
Actually you don't need another inverter - did you measure the THD at the output of the unity gain inverting stage? Levels at the opamp outputs should be reasonably well matched, too.
L.
That one could of course also be used. I have not measured the THD at the inverter output. I assume that the noise level will be higher at this point than on the output, but that will also be the case with an additional op-amp inverter, at least if we use an LME49710 here as well. Perhaps a different op-amp, driven with a low input resistance could be an option? But then we also have to keep an eye on the THD from the normal oscillator output. Whether a load on the existing inverter output will influence the performance of the whole design would also need to be checked.
A buffered inverted output would probably be the best, but what OPamp has such low THD values as your sine wave generator?
If one instead could take the inverted output from within the sine wave generator, then we are all settled Then, if it could drive 600 Ohm loads, we can be satisfied.
Could you please also measure THD at the inverting output from within the sine wave generator?
Br,
S.
If one instead could take the inverted output from within the sine wave generator, then we are all settled
Then, if it could drive 600 Ohm loads, we can be satisfied.Could you please also measure THD at the inverting output from within the sine wave generator?
Br,
S.
If this is the LME49710 used, it is only specified for -130dB.
So how is it possible to do better than this in a Wien Bridge.
It seems to me that this is a key ingredient... when changing freqs - keep the R's in the optimum range - so distortion (and noise) stays lowest possible.
Thx-RNMarsh
Looks like good work!
The open question is of course how the notch filter influences the measurement. As far as I can tell from your notes you're probably running the passives in it at the same AC level as in the oscillator. This means that their distortion contribution will be similar. With a 50:50 chance the notch filter distortion contribution either adds or subtracts from the actual oscillator performance, and there's no way to tell which one's going on.
As noted earlier, one IMHO good way to get around this is to build a carefully designed attenuator in front of the notch filter. This reduces the AC level in the filter, and thus pushes its distortion contribution way down. The attenuator could be built e.g. out of 10 1k resistors (1/2 or 1/4 W metal film), with the last tap forming the -20 dB output into the notch filter.
Samuel
You are right about the levels. I did connect the oscillator output directly to the notch filter input.
Perhaps I should try the method you describe. One potential problem could be that the distortion may disappear into the noise floor. In my measurement the distortion was only around 5 dB above the noise floor. But perhaps it can be improved by using a longer FFT and average over a larger number of measurements.
In any case, it would be good to verify the result with different test set up's.
I measured the distortion with the notch filter of the R&S UPL and it was not nearly as good. I do suspect though, that it was a limitation of the UPL.
With the UPL I got 2nd and 3rd harmonics of -136.5 dB and -132 dB relative to the fundamental. That was with 30 dB gain in the notch filter.
Perhaps I should try the method you describe. One potential problem could be that the distortion may disappear into the noise floor. In my measurement the distortion was only around 5 dB above the noise floor. But perhaps it can be improved by using a longer FFT and average over a larger number of measurements.
In any case, it would be good to verify the result with different test set up's.
I measured the distortion with the notch filter of the R&S UPL and it was not nearly as good. I do suspect though, that it was a limitation of the UPL.
With the UPL I got 2nd and 3rd harmonics of -136.5 dB and -132 dB relative to the fundamental. That was with 30 dB gain in the notch filter.
Just as a DUT (amplifier) can increase its distortion with high Z input sources, so might other circuits -- osc and notch filters. And the Zo of the notch filter (unbuffered) going into the DUT can be giving 'false' numbers which wouldnt exist with a very low Z source. Like-wise the osc and notch can interact to produce poorer results than anticipated.
Here is something also interesting.... I have compared the THD of two analyzers using the same source generator. They gave similar THD figures but as you can see in the photos.... one is showing a 2H dominant and the other its a 3H !!! Now how can i tell there isnt some sort of timing error in setup display (didnt change any settings)...
Both can indicate the level of the individual harmonic I choose. And as the residual display showed, the analyzer indicated the Harmonic which was highest was the one shown via the residual monitor output port.
Impedance interaction with generator z on the analyzer inputs?? I tried adding 1K in series with each analyzer input but the Harmonic level and freq disnt change.
Got any ideas... cockpit error or ??
Thx-RNMarsh
Here is something also interesting.... I have compared the THD of two analyzers using the same source generator. They gave similar THD figures but as you can see in the photos.... one is showing a 2H dominant and the other its a 3H !!! Now how can i tell there isnt some sort of timing error in setup display (didnt change any settings)...
Both can indicate the level of the individual harmonic I choose. And as the residual display showed, the analyzer indicated the Harmonic which was highest was the one shown via the residual monitor output port.
Impedance interaction with generator z on the analyzer inputs?? I tried adding 1K in series with each analyzer input but the Harmonic level and freq disnt change.
Got any ideas... cockpit error or ??
Thx-RNMarsh
I get the same thing with my measurements. All it takes is making a small insignificant change in the measurement path and the second and third H can change places in terms of there level. Sometimes the second melts away proportionally with the rise in 3rd. Explain this. At these infinitesimal levels I'm never sure of what I'm measuring.
There is some phase shifting apparently happening somewhere.... between the notch phasiness and harmonics a confusion in the dsp analyzer portion at certain phase angles. ?
But note - mine is from 2 different analyzers..... they are each fixed on their harmonic and it didnt change... each analyzer sees something the other doesnt. Might be related to your phenomenon, though. Or is the 2H beating with the residual 1H to produce a 3H?
-RNM
But note - mine is from 2 different analyzers..... they are each fixed on their harmonic and it didnt change... each analyzer sees something the other doesnt. Might be related to your phenomenon, though. Or is the 2H beating with the residual 1H to produce a 3H?
-RNM
Last edited:
Hi Rick,
I re calibrated the EMU0204 left input for 0.1Vrms FS.
The displayed spectrum spectrum is a lot cleaner. The garbage I had with 0dB FS is gone and all I'm seeing is the generators disto.
My take on this is it's not accurate when trying to measure near the bottom of the ADC's DR. Shifting the scale up to where the measurement is closer to the ADC's mid DR is better.
I re calibrated the EMU0204 left input for 0.1Vrms FS.
The displayed spectrum spectrum is a lot cleaner. The garbage I had with 0dB FS is gone and all I'm seeing is the generators disto.
My take on this is it's not accurate when trying to measure near the bottom of the ADC's DR. Shifting the scale up to where the measurement is closer to the ADC's mid DR is better.
This is why the 10 to 1 principle in measurement is so important. You really have no idea what is dominant and if there are phase interactions in the harmonics. I would suggest the three corner hat approach which you have enough hardware to explore- measure and note/plot the harmonics of each generator on each analyzer. Since all the distortions are so low making a matrix would help identify which oscillators/analyzers show the same distortions consistently.
You can also take the passive B&K filter, pass the signal through it, feed the output into the 30 mV input on the 725D after attenuating the fundamental a bunch and see if you have the same harmonic mix (remember to use the harmonic corrected mode). However I suspect you will still have a -120 dB confidence level in measured harmonics at best. Below that +/- 5 dB could be attributable to a number of things and hard to be sure of.
Toroid core material
Here is a clip I found in the interwebs:
Material: Unisil M089-27N (27M4)
Maximum loss 0.89w/kg @ 1.5T.50Hz, 1.7T.60Hz.
Widths 10 - 80mm
All material supplied to BS EN 10107:2005
Toroidal cores available from stock
Maximum OD 200mm Maximum ID 100mm
G.O.S.S. has been manufacturing toroidal cores since 1996 on high speed automatic core winding machines (shown above).
transformer core manufacturers,toroidal cores,manufacturer of strip wound cores,manufacturer of magnetic cores,magnetic cores
I don't know if it gives a clue into the core or its ability to retain a field. It seems they use M4 and it won't have a gap. It mentions 1.7T flux so thats a lot (and way over JN's limit). Hopefully JN can derive some meaning from this.
Here is a clip I found in the interwebs:
Material: Unisil M089-27N (27M4)
Maximum loss 0.89w/kg @ 1.5T.50Hz, 1.7T.60Hz.
Widths 10 - 80mm
All material supplied to BS EN 10107:2005
Toroidal cores available from stock
Maximum OD 200mm Maximum ID 100mm
G.O.S.S. has been manufacturing toroidal cores since 1996 on high speed automatic core winding machines (shown above).
transformer core manufacturers,toroidal cores,manufacturer of strip wound cores,manufacturer of magnetic cores,magnetic cores
I don't know if it gives a clue into the core or its ability to retain a field. It seems they use M4 and it won't have a gap. It mentions 1.7T flux so thats a lot (and way over JN's limit). Hopefully JN can derive some meaning from this.