Forums

Improving my best diode detector

Started by amdx April 19, 2017
   Can't get/see that schematic; in any case, for low(er) level 
sensitivity and "linearity", try adding a small forward DC bias to move 
the average OP up the curve.

On 4/21/2017 3:33 AM, Robert Baer wrote:
> Can't get/see that schematic; in any case, for low(er) level > sensitivity and "linearity", try adding a small forward DC bias to move > the average OP up the curve. >
Still works for me. https://www.dropbox.com/s/3oyjn2teglg47b7
> https://www.dropbox.com/s/3oyjn2teglg47b7/Diode%20detector%201N34A%20500ohm%20adjusted%20R1%20and%20R2.jpg?dl=0
--- This email has been checked for viruses by Avast antivirus software. https://www.avast.com/antivirus
On 4/21/2017 4:33 AM, Robert Baer wrote:
> Can't get/see that schematic; in any case, for low(er) level > sensitivity and "linearity", try adding a small forward DC bias to move > the average OP up the curve.
The bias would then need to be subtracted out again after rectification. -- Rick C
On 4/21/2017 3:26 AM, Martin Brown wrote:
> On 20/04/2017 15:26, amdx wrote: >> On 4/20/2017 7:37 AM, Martin Brown wrote: >>> On 19/04/2017 16:13, amdx wrote: >>>> Hi Guys, >>>> I have three Diode Detectors built as the Figure 4 schematic in this >>>> article, using different number diodes. >>>> >>>>> https://www.rf-microwave.com/uploads/diodes/a_temperature_compensated_linear_diode_detector.pdf >>>>> >>>> >>>> I used Bat62, 1n5711 and some unknown germanium diodes I pulled of of >>>> some old pcbs. >>>> The best is the unknown germainums, which are probably 1N34A diodes. I >>>> have some on order to test. >>>> >>>> Here is a graph here showing the input at 1MHz and the DC output >>>> curves. >>>>> https://www.dropbox.com/s/63hsukb7jyryeb1/Diode%20Deetector%201N34a%2C%20Bat62%2C1N5711.jpg?dl=0 >>>>> >>>> >>>> It is actually better than I expected, but, >>>> The question; Is there any change I could make to improve it? >>> >>> Possibly. The thing to do to make life easier is subtract off the >>> least squares best straight line and show the residuals x10 on the >>> same scale. >>> >>> You perhaps want to concentrate on the fit to 0.7v-3v being most linear. >>> >>> Bat62 looks pretty good to me if you allow it to cross the y axis with >>> a DC offset and ignore the region <0.5v entirely. >>> >> >> The Range to optimize is 3Vrms to 5Vrms. I couldn't run my graph out >> to 5Vrms, I didn't have the capability, but I'm getting there. >> My RF millivolt meter only goes to 3volts, I have 100 to 1 adapter, but >> it seem to cause a 10% error. >> When using the Q meter the drive can be adjusted to keep the meter in >> that 3V to 5V range. I also think that will be the most linear range. > > Why are you so concerned about linearity of your diode detector if you > are merely using the detector to look for a resonance peak?
That's not the purpose of the circuit.
> All you require of its amplitude detection is that it is monotonically > increasing with input voltage and not too frequency dependent.
Amplitude *measurement*.
> I think you really should be checking how it behaves at constant input > amplitude over the range of frequencies that you expect to use...
You are not up to date. -- Rick C
On 4/21/2017 8:44 AM, rickman wrote:
> On 4/21/2017 3:26 AM, Martin Brown wrote: >> On 20/04/2017 15:26, amdx wrote: >>> On 4/20/2017 7:37 AM, Martin Brown wrote: >>>> On 19/04/2017 16:13, amdx wrote: >>>>> Hi Guys, >>>>> I have three Diode Detectors built as the Figure 4 schematic in this >>>>> article, using different number diodes. >>>>> >>>>>> https://www.rf-microwave.com/uploads/diodes/a_temperature_compensated_linear_diode_detector.pdf >>>>>> >>>>>> >>>>> >>>>> I used Bat62, 1n5711 and some unknown germanium diodes I pulled of of >>>>> some old pcbs. >>>>> The best is the unknown germainums, which are probably 1N34A >>>>> diodes. I >>>>> have some on order to test. >>>>> >>>>> Here is a graph here showing the input at 1MHz and the DC output >>>>> curves. >>>>>> https://www.dropbox.com/s/63hsukb7jyryeb1/Diode%20Deetector%201N34a%2C%20Bat62%2C1N5711.jpg?dl=0 >>>>>> >>>>>> >>>>> >>>>> It is actually better than I expected, but, >>>>> The question; Is there any change I could make to improve it? >>>> >>>> Possibly. The thing to do to make life easier is subtract off the >>>> least squares best straight line and show the residuals x10 on the >>>> same scale. >>>> >>>> You perhaps want to concentrate on the fit to 0.7v-3v being most >>>> linear. >>>> >>>> Bat62 looks pretty good to me if you allow it to cross the y axis with >>>> a DC offset and ignore the region <0.5v entirely. >>>> >>> >>> The Range to optimize is 3Vrms to 5Vrms. I couldn't run my graph out >>> to 5Vrms, I didn't have the capability, but I'm getting there. >>> My RF millivolt meter only goes to 3volts, I have 100 to 1 adapter, but >>> it seem to cause a 10% error. >>> When using the Q meter the drive can be adjusted to keep the meter in >>> that 3V to 5V range. I also think that will be the most linear range. >> >> Why are you so concerned about linearity of your diode detector if you >> are merely using the detector to look for a resonance peak? > > That's not the purpose of the circuit. > > >> All you require of its amplitude detection is that it is monotonically >> increasing with input voltage and not too frequency dependent. > > Amplitude *measurement*. > > >> I think you really should be checking how it behaves at constant input >> amplitude over the range of frequencies that you expect to use... > > You are not up to date. >
He's has a point there, I already know T1 and Q2 start to roll off at 12 or 13 MHz. That is well above where my interest lies, but, I'll try to improve that later, I need to see something work before I lose interest. I have the first section which is mostly an impedance changer (from very high to low) with a gain of 0.96. I have a working detector I'm happy with, it has a better frequency response than my impedance hanger. Now all I need is a amp with a gain of 4.5 and a 500 ohm collector resistor. I have the circuit just need to get the gain and bias right and it should be ready for testing. On a different note, I have had the AD8307 log amplifier recommended. It would save several parts, but I don't understand it's output in relation to it's input. It says 25mV/db. I want to drive an analog meter, will that work? Or is is scaled in db, am I confused? Mikek --- This email has been checked for viruses by Avast antivirus software. https://www.avast.com/antivirus
On 4/21/2017 12:14 PM, amdx wrote:
> On 4/21/2017 8:44 AM, rickman wrote: >> On 4/21/2017 3:26 AM, Martin Brown wrote: >>> On 20/04/2017 15:26, amdx wrote: >>>> On 4/20/2017 7:37 AM, Martin Brown wrote: >>>>> On 19/04/2017 16:13, amdx wrote: >>>>>> Hi Guys, >>>>>> I have three Diode Detectors built as the Figure 4 schematic in this >>>>>> article, using different number diodes. >>>>>> >>>>>>> https://www.rf-microwave.com/uploads/diodes/a_temperature_compensated_linear_diode_detector.pdf >>>>>>> >>>>>>> >>>>>> >>>>>> I used Bat62, 1n5711 and some unknown germanium diodes I pulled >>>>>> of of >>>>>> some old pcbs. >>>>>> The best is the unknown germainums, which are probably 1N34A >>>>>> diodes. I >>>>>> have some on order to test. >>>>>> >>>>>> Here is a graph here showing the input at 1MHz and the DC output >>>>>> curves. >>>>>>> https://www.dropbox.com/s/63hsukb7jyryeb1/Diode%20Deetector%201N34a%2C%20Bat62%2C1N5711.jpg?dl=0 >>>>>>> >>>>>>> >>>>>> >>>>>> It is actually better than I expected, but, >>>>>> The question; Is there any change I could make to improve it? >>>>> >>>>> Possibly. The thing to do to make life easier is subtract off the >>>>> least squares best straight line and show the residuals x10 on the >>>>> same scale. >>>>> >>>>> You perhaps want to concentrate on the fit to 0.7v-3v being most >>>>> linear. >>>>> >>>>> Bat62 looks pretty good to me if you allow it to cross the y axis with >>>>> a DC offset and ignore the region <0.5v entirely. >>>>> >>>> >>>> The Range to optimize is 3Vrms to 5Vrms. I couldn't run my graph out >>>> to 5Vrms, I didn't have the capability, but I'm getting there. >>>> My RF millivolt meter only goes to 3volts, I have 100 to 1 adapter, >>>> but >>>> it seem to cause a 10% error. >>>> When using the Q meter the drive can be adjusted to keep the meter in >>>> that 3V to 5V range. I also think that will be the most linear range. >>> >>> Why are you so concerned about linearity of your diode detector if you >>> are merely using the detector to look for a resonance peak? >> >> That's not the purpose of the circuit. >> >> >>> All you require of its amplitude detection is that it is monotonically >>> increasing with input voltage and not too frequency dependent. >> >> Amplitude *measurement*. >> >> >>> I think you really should be checking how it behaves at constant input >>> amplitude over the range of frequencies that you expect to use... >> >> You are not up to date. >> > > He's has a point there, I already know T1 and Q2 start to roll off > at 12 or 13 MHz. That is well above where my interest lies, but, I'll > try to improve that later, I need to see something work before I lose > interest. > I have the first section which is mostly an impedance changer (from > very high to low) with a gain of 0.96. I have a working detector I'm > happy with, it has a better frequency response than my impedance hanger. > Now all I need is a amp with a gain of 4.5 and a 500 ohm collector > resistor. I have the circuit just need to get the gain and bias right > and it should be ready for testing.
The gain of the circuit can be modified by adding an emitter resistor where G = Rc/Re. The gain to Re is always about 1 and the current in the two resistors is nearly equal, so the gain at the collector will be proportional by Rc/Re.
> On a different note, I have had the AD8307 log amplifier recommended. > It would save several parts, but I don't understand it's output in > relation to it's input. It says 25mV/db. I want to drive an analog > meter, will that work? Or is is scaled in db, am I confused?
The voltage on the output goes up by 25 mV with each dB of increase on the input. dB is a scale based on logarithms : dB = 20 log(V/Vref) where V is the input voltage and Vref is the input voltage for a zero dB output. That means the Q will be on a logarithmic scale which you would have to mark on the meter face. You may need to add an op amp circuit to adjust the zero point of the meter or you might be able to use a voltage source and a pot on the ground leg of the meter like the original Boonton circuit, but with a much lower voltage. This would crowd the numbers on the high end of the meter and space them further apart on the low end. Maybe this is good, maybe not. BTW, if you are only interested in a given voltage range at the output of the diode circuit, you can connect the meter ground leg to the adjustable voltage source (a pot) and make the zero current point on the meter correspond to a 3 Vrms input and the max reading on the meter face correspond to a 5 Vrms input. Then you have the entire range of the meter to measure the Q more accurately. There is no reason why the low end has to correspond to a Q of 1. -- Rick C
On 4/21/2017 12:25 PM, rickman wrote:
> On 4/21/2017 12:14 PM, amdx wrote: >> On 4/21/2017 8:44 AM, rickman wrote: >>> On 4/21/2017 3:26 AM, Martin Brown wrote: >>>> On 20/04/2017 15:26, amdx wrote: >>>>> On 4/20/2017 7:37 AM, Martin Brown wrote: >>>>>> On 19/04/2017 16:13, amdx wrote: >>>>>>> Hi Guys, >>>>>>> I have three Diode Detectors built as the Figure 4 schematic in >>>>>>> this >>>>>>> article, using different number diodes. >>>>>>> >>>>>>>> https://www.rf-microwave.com/uploads/diodes/a_temperature_compensated_linear_diode_detector.pdf >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>> >>>>>>> I used Bat62, 1n5711 and some unknown germanium diodes I pulled >>>>>>> of of >>>>>>> some old pcbs. >>>>>>> The best is the unknown germainums, which are probably 1N34A >>>>>>> diodes. I >>>>>>> have some on order to test. >>>>>>> >>>>>>> Here is a graph here showing the input at 1MHz and the DC output >>>>>>> curves. >>>>>>>> https://www.dropbox.com/s/63hsukb7jyryeb1/Diode%20Deetector%201N34a%2C%20Bat62%2C1N5711.jpg?dl=0 >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>> >>>>>>> It is actually better than I expected, but, >>>>>>> The question; Is there any change I could make to improve it? >>>>>> >>>>>> Possibly. The thing to do to make life easier is subtract off the >>>>>> least squares best straight line and show the residuals x10 on the >>>>>> same scale. >>>>>> >>>>>> You perhaps want to concentrate on the fit to 0.7v-3v being most >>>>>> linear. >>>>>> >>>>>> Bat62 looks pretty good to me if you allow it to cross the y axis >>>>>> with >>>>>> a DC offset and ignore the region <0.5v entirely. >>>>>> >>>>> >>>>> The Range to optimize is 3Vrms to 5Vrms. I couldn't run my graph out >>>>> to 5Vrms, I didn't have the capability, but I'm getting there. >>>>> My RF millivolt meter only goes to 3volts, I have 100 to 1 adapter, >>>>> but >>>>> it seem to cause a 10% error. >>>>> When using the Q meter the drive can be adjusted to keep the meter in >>>>> that 3V to 5V range. I also think that will be the most linear range. >>>> >>>> Why are you so concerned about linearity of your diode detector if you >>>> are merely using the detector to look for a resonance peak? >>> >>> That's not the purpose of the circuit. >>> >>> >>>> All you require of its amplitude detection is that it is monotonically >>>> increasing with input voltage and not too frequency dependent. >>> >>> Amplitude *measurement*. >>> >>> >>>> I think you really should be checking how it behaves at constant input >>>> amplitude over the range of frequencies that you expect to use... >>> >>> You are not up to date. >>> >> >> He's has a point there, I already know T1 and Q2 start to roll off >> at 12 or 13 MHz. That is well above where my interest lies, but, I'll >> try to improve that later, I need to see something work before I lose >> interest. >> I have the first section which is mostly an impedance changer (from >> very high to low) with a gain of 0.96. I have a working detector I'm >> happy with, it has a better frequency response than my impedance hanger. >> Now all I need is a amp with a gain of 4.5 and a 500 ohm collector >> resistor. I have the circuit just need to get the gain and bias right >> and it should be ready for testing. > > The gain of the circuit can be modified by adding an emitter resistor > where G = Rc/Re. The gain to Re is always about 1 and the current in > the two resistors is nearly equal, so the gain at the collector will be > proportional by Rc/Re. > > >> On a different note, I have had the AD8307 log amplifier recommended. >> It would save several parts, but I don't understand it's output in >> relation to it's input. It says 25mV/db. I want to drive an analog >> meter, will that work? Or is is scaled in db, am I confused? > > The voltage on the output goes up by 25 mV with each dB of increase on > the input. dB is a scale based on logarithms : dB = 20 log(V/Vref) > where V is the input voltage and Vref is the input voltage for a zero dB > output. > > That means the Q will be on a logarithmic scale which you would have to > mark on the meter face. You may need to add an op amp circuit to adjust > the zero point of the meter or you might be able to use a voltage source > and a pot on the ground leg of the meter like the original Boonton > circuit, but with a much lower voltage. > > This would crowd the numbers on the high end of the meter and space them > further apart on the low end. Maybe this is good, maybe not.
Then it doesn't work with my plans for an analog meter.
> > BTW, if you are only interested in a given voltage range at the output > of the diode circuit, you can connect the meter ground leg to the > adjustable voltage source (a pot) and make the zero current point on the > meter correspond to a 3 Vrms input and the max reading on the meter face > correspond to a 5 Vrms input. Then you have the entire range of the > meter to measure the Q more accurately. There is no reason why the low > end has to correspond to a Q of 1. >
If biasing the diode detector up by 3 volts will make it more linear on the bottom end, That's good. I can bias the other end of my meter up by three volts to remove the 3 volts. The Q meter already is setup with an adjustable bias on the negative side of the meter. I'll keep this in mind. I would just as soon keep the zero to 250 scale. Thanks, Mikek --- This email has been checked for viruses by Avast antivirus software. https://www.avast.com/antivirus
On 4/21/2017 1:53 PM, amdx wrote:
> On 4/21/2017 12:25 PM, rickman wrote: >> >> BTW, if you are only interested in a given voltage range at the output >> of the diode circuit, you can connect the meter ground leg to the >> adjustable voltage source (a pot) and make the zero current point on >> the meter correspond to a 3 Vrms input and the max reading on the >> meter face correspond to a 5 Vrms input. Then you have the entire >> range of the meter to measure the Q more accurately. There is no >> reason why the low end has to correspond to a Q of 1. >> > If biasing the diode detector up by 3 volts will make it more linear on > the bottom end, That's good. I can bias the other end of my meter up by > three volts to remove the 3 volts. The Q meter already is setup with an > adjustable bias on the negative side of the meter. > I'll keep this in mind. > I would just as soon keep the zero to 250 scale.
What I am talking about doesn't affect the diode circuit, it changes the range of your meter range from 0-5 to 3-5 or whatever numbers you want. Because you can change your power input level to bring the output of the circuit to whatever level you want there is no reason to work with the low end of the output range, so no need to linearize the entire output range. Besides, your circuit is pretty durn linear as it is. If you want to keep the 0 to 250 scale, consider using an adder (+250 say) before you use the multiplier. That would make the true Q scale 250 to 500. -- Rick C
amdx wrote:
> On 4/21/2017 3:33 AM, Robert Baer wrote: >> Can't get/see that schematic; in any case, for low(er) level >> sensitivity and "linearity", try adding a small forward DC bias to >> move the average OP up the curve. >> > > Still works for me. > https://www.dropbox.com/s/3oyjn2teglg47b7
* Nope; get: "Download and save to your Dropbox by creating a free account, or sign in" I do not have a dropbox...
> > >> https://www.dropbox.com/s/3oyjn2teglg47b7/Diode%20detector%201N34A%20500ohm%20adjusted%20R1%20and%20R2.jpg?dl=0
* No schematic..
>> > > > --- > This email has been checked for viruses by Avast antivirus software. > https://www.avast.com/antivirus >
rickman wrote:
> On 4/21/2017 4:33 AM, Robert Baer wrote: >> Can't get/see that schematic; in any case, for low(er) level >> sensitivity and "linearity", try adding a small forward DC bias to move >> the average OP up the curve. > > The bias would then need to be subtracted out again after rectification. >
Too complicated??