Re: [time-nuts] Oleg' s PN test Re: A new member & PN test set
Another potential issue is that the setup should be adjusted for minimum AM sensitivity.Because of diode mismatch and phase mismatch between the internal transformers this doesn't occur when the mixer dc output is zero. To find the correct point a dc offset needs to be introduced at the input of the PLL integrator. The offset is adjusted to minimise AM sensitivity. In order to do this an AM modulator with very little incidental PM is required.This is tricky to implement although it has been done using an unbalanced Mach-Zehnder interferometer plus a pure Phase modulator (easy to do since AM detectors aren't phase sensitive): Microwave Sources of Pure Phase and Amplitude-‐Modulated Signals E. N. Ivanov This adjustment is important as the residual AM of most RF sources isn't negligible.The technique works uses standard components and techniques (3 db hybrids, mixers, voltage controlled phase shifters, lock in amplifiers) and works at far lower frequencies than microwave. At lower frequencies simpler AM modulators may suffice but you would need to show that any incidental PM they produce is insignificant in that it wont have a significant effect on the depth of the AM null at the mixer/phase detector output.Typically a modulation frequency in the audio band would be used. Bruce From: Oleg Skydan <olegsky...@gmail.com> To: Discussion of precise time and frequency measurement <time-nuts@febo.com> Sent: Thursday, 31 March 2016 9:28 PM Subject: Re: [time-nuts] Oleg' s PN test Re: A new member & PN test set -- From: "Bruce Griffiths" <bruce.griffi...@xtra.co.nz> > You actually need to measure the filter > response. OK. It is here (the frequency span is 2..102MHz, the amplitude axis is 10dB/div): http://skydan.in.ua/PNTestSet/PN_LPF1.jpg Sorry, the network analyzer is a bit older than I am :), but it is still in a good condition. At 100MHz we still have more than 40dB attenuation. The inductor I used is low Q axial choke, so I do not expect multiple resonances at higher frequencies - there should be large losses and inductor will look much more like a resistor (at least until we go too high). The capacitors in the pi-LPF are 0805 SMD good quality ones. But, we all like perfect things :), so I tried to make the LPF a bit better adding the BLM31AJ601SN ferrite bead in series with the inductor. Here is the result: http://skydan.in.ua/PNTestSet/PN_LPF2.jpg I like it :) I also did another test checking DC shift at the AD797 output when the mixer was fed with two signal generators (there should be no DC - only different combinations of RF/LO signals). I recalculated all signal levels to LNA input point. Before installing the bead the DC shift was: <150MHz less than 150uV 150..250MHz less than 400uV 250..500MHz less than 1mV After installing the bead: <260MHz less than 80uV There is one big peak near the 300MHz where the DC goes up to 900uV, and several smaller peaks (up to 250uV) higer, up to 500MHz. When the two mixer ports are fed with the same signal (inphase) the DC voltage at LNA input is 130..150mV. With no RF at the mixer ports DC unbalance is 20uV (all voltages recalculated to LNA input). The filter can be made even better by cascading several pi sections using different capacitors and inductors/beads. But as far as I understand, with the current filter even at 500MHz it will not move too far from the quadrature, and at 60MHz everything is definitely OK. All the best! Oleg ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] Oleg' s PN test Re: A new member & PN test set
-- From: "Bruce Griffiths"You actually need to measure the filter response. OK. It is here (the frequency span is 2..102MHz, the amplitude axis is 10dB/div): http://skydan.in.ua/PNTestSet/PN_LPF1.jpg Sorry, the network analyzer is a bit older than I am :), but it is still in a good condition. At 100MHz we still have more than 40dB attenuation. The inductor I used is low Q axial choke, so I do not expect multiple resonances at higher frequencies - there should be large losses and inductor will look much more like a resistor (at least until we go too high). The capacitors in the pi-LPF are 0805 SMD good quality ones. But, we all like perfect things :), so I tried to make the LPF a bit better adding the BLM31AJ601SN ferrite bead in series with the inductor. Here is the result: http://skydan.in.ua/PNTestSet/PN_LPF2.jpg I like it :) I also did another test checking DC shift at the AD797 output when the mixer was fed with two signal generators (there should be no DC - only different combinations of RF/LO signals). I recalculated all signal levels to LNA input point. Before installing the bead the DC shift was: <150MHz less than 150uV 150..250MHz less than 400uV 250..500MHz less than 1mV After installing the bead: <260MHz less than 80uV There is one big peak near the 300MHz where the DC goes up to 900uV, and several smaller peaks (up to 250uV) higer, up to 500MHz. When the two mixer ports are fed with the same signal (inphase) the DC voltage at LNA input is 130..150mV. With no RF at the mixer ports DC unbalance is 20uV (all voltages recalculated to LNA input). The filter can be made even better by cascading several pi sections using different capacitors and inductors/beads. But as far as I understand, with the current filter even at 500MHz it will not move too far from the quadrature, and at 60MHz everything is definitely OK. All the best! Oleg ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] Oleg' s PN test Re: A new member & PN test set
You've missed the point which is that: 1) With 10MHz input frequencies the sum frequency is actually 20MHz which is beyond the first resonance of the inductor used.Something better is required. The sum frequency is the largest unwanted component that exits the mixer IF port. 2) Oleg is restricted in what he has available, rolling ones own conical inductor is an option given the iron powder Piconics claim to use is potentially available to him locally. 3) Oleg indicated he'd used the rig to measure PN with 60MHz inputs resulting in a 120MHz sum frequency output placing even more severe band reject requirements on the filter/diplexer. 4) The specified mixer is usable to 500 MHz with resulting sum frequency of 1GHz. If someone were tempted to use it at those frequencies as is the results would be "interesting" to say the least. Obviously one could select off the shelf components that are satisfactory over a small frequency range or for a particular input frequency of interest.. This means the filter components have to be changed whenever the measurement frequency is changed too much. However its probably more useful to use a filter that works well over the entire range of frequencies for which the mixer is useful.There are many ways to achieve this. Thanks for the confirmation that home made conical inductors can work well. A version that worked from say 1MHz to 1GHz would be very useful. Another potential problem is injection locking:If the VCXO/OCXO under test has inadequate reverse isolation then this can occur leading one to falsely conclude the VCO/OCXO under test has far better performance than it actually has. In effect the injected signal increases the PLL bandwidth far beyond the intended value. Using high reverse isolation amplifiers on the mixer / phase detector input ports is one way of ensuring that injection locking doesn't occur. Bruce From: Gerhard Hoffmann <dk...@arcor.de> To: time-nuts@febo.com Sent: Thursday, 31 March 2016 12:12 PM Subject: Re: [time-nuts] Oleg' s PN test Re: A new member & PN test set Am 30.03.2016 um 21:20 schrieb Bruce Griffiths: > Conical inductors are available that are effectively resonance free to 40GHz > but the largest value is around 10uH. In principle one could wind one's own > conical inductor with a larger value, However an iron powder (carbonyl iron - > available from Ukraine at least via ebay) and epoxy mixture. A cone angle of > about 15 degrees appears to be suitable. > > Failing that, the classical method is to use a series string of inductors of > increasing value. Even then the various resonances need to be damped. > Lossy Ferrites and resistors can be useful, however one has to be careful not > to increase the noise at frequencies of interest. It doesn't take conical inductors to separate the baseband from the carrier at 10 MHz. The world existed before Piconics and their conical L patents. Yes, we used them in our 10 GB/s fiber optic transceivers, just to see what eye diagrams we could achieve. But at €38 a pop they never ever made it into production. It was just too easy to replace them with somewhat more cent stuff. A colleague even rolled his own from wire, epoxy glue and ferrite beads smashed in a mortar. That looked, hmm, ugly, but performed excellently. Now, you get them from MCL and Coilcraft. But for a 10 or 100 MHz lowpass, that's way over the top. Not even if you are nuts. regards, Gerhard ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] Oleg' s PN test Re: A new member & PN test set
Am 30.03.2016 um 21:20 schrieb Bruce Griffiths: Conical inductors are available that are effectively resonance free to 40GHz but the largest value is around 10uH. In principle one could wind one's own conical inductor with a larger value, However an iron powder (carbonyl iron - available from Ukraine at least via ebay) and epoxy mixture. A cone angle of about 15 degrees appears to be suitable. Failing that, the classical method is to use a series string of inductors of increasing value. Even then the various resonances need to be damped. Lossy Ferrites and resistors can be useful, however one has to be careful not to increase the noise at frequencies of interest. It doesn't take conical inductors to separate the baseband from the carrier at 10 MHz. The world existed before Piconics and their conical L patents. Yes, we used them in our 10 GB/s fiber optic transceivers, just to see what eye diagrams we could achieve. But at €38 a pop they never ever made it into production. It was just too easy to replace them with somewhat more cent stuff. A colleague even rolled his own from wire, epoxy glue and ferrite beads smashed in a mortar. That looked, hmm, ugly, but performed excellently. Now, you get them from MCL and Coilcraft. But for a 10 or 100 MHz lowpass, that's way over the top. Not even if you are nuts. regards, Gerhard ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] Oleg' s PN test Re: A new member & PN test set
On Thursday, March 31, 2016 08:20:19 AM Bruce Griffiths wrote: > The first inductor self resonance is much lower than it need be. EPCOS have > a range of inductors which have a much higher first self resonance > frequency. > > The other problem is that inductors have several resonances alternating > between parallel and series resonance. In short, your inductor model is > inadequate and gives false predictions. You actually need to measure the > filter response. > > Conical inductors are available that are effectively resonance free to 40GHz > but the largest value is around 10uH. In principle one could wind one's own > conical inductor with a larger value, However an iron powder (carbonyl iron > - available from Ukraine at least via ebay) and epoxy mixture filling/core is required. > A cone angle of about 15 degrees appears to be suitable. > > Failing that, the classical method is to use a series string of inductors of > increasing value. Even then the various resonances need to be damped. Lossy > Ferrites and resistors can be useful, however one has to be careful not to > increase the noise at frequencies of interest. > > Bruce > Note correction above. Bruce ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] Oleg' s PN test Re: A new member & PN test set
The first inductor self resonance is much lower than it need be. EPCOS have a range of inductors which have a much higher first self resonance frequency. The other problem is that inductors have several resonances alternating between parallel and series resonance. In short, your inductor model is inadequate and gives false predictions. You actually need to measure the filter response. Conical inductors are available that are effectively resonance free to 40GHz but the largest value is around 10uH. In principle one could wind one's own conical inductor with a larger value, However an iron powder (carbonyl iron - available from Ukraine at least via ebay) and epoxy mixture. A cone angle of about 15 degrees appears to be suitable. Failing that, the classical method is to use a series string of inductors of increasing value. Even then the various resonances need to be damped. Lossy Ferrites and resistors can be useful, however one has to be careful not to increase the noise at frequencies of interest. Bruce On Wednesday, March 30, 2016 06:11:51 PM Oleg Skydan wrote: > -- > From: "Bruce Griffiths" <bruce.griffi...@xtra.co.nz> > Sent: Wednesday, March 30, 2016 7:29 AM > To: "Discussion of precise time and frequency measurement" > <time-nuts@febo.com> > Subject: Re: [time-nuts] Oleg' s PN test Re: A new member & PN test set > > > One hidden issue you don't address is that operation of the 40uH inductor > > at > > frequencies above its parallel resonance may allow substantial RF at the > > sum > > of the LO and RF frequencies to appear at the opamp input. > > 120MHz at the 797 input will likely lead to RF rectification effects in > > the > > opamp input stage. The resultant offset will create a number of issues > > including operation away from the quadrature point. > > Unless you use something like a series string of inductors and/or a > > conical > > inductor the first parallel resonance of the 40uH inductor is likely to be > > somewhat below 120MHz. > > Ohhh... I do not like words like "substantial", "much more" and etc. I like > numbers and tests. ;) > > So I looked at the Murrata inductors datasheets, and it appeared 40uH > inductor will have SRF in 10MHz region. But it does not mean that pi-LPF > will not work at the higher frequencies. Actually it mean that our LPF will > have response similar to the elliptic filters. > > So let's draw the model with the inductor with self resonance at 10MHz and > well at 120MHz and 1MHz to see how bad the response is: > 1MHz: http://skydan.in.ua/PNTestSet/1MSRF.png > 10MHz: http://skydan.in.ua/PNTestSet/10MSRF.png > 120MHz: http://skydan.in.ua/PNTestSet/120MSRF.png > > As we can see the RF+LO product will be attenuated more than 60dB in all > cases. So, your comments? Would you like me to measure the RF voltage at the > AD797 input in the real "test set"? > > > However 50 ohms to ground at the LC filter output shouldn't be necessary. > > A somewhat larger value should suffice. > > I made some experiments trying to find the optimal value of the resistor at > the LC filter output. The phase detector gain grew along with the resistor > value, but so did the harmonics level. So I needed to apply more attenuation > to the input signal to stay in the linear region. The resulting "test set" > noise floor was almost identical for 50..300Ohm values (300Ohm was a bit > better at offsets grater then 2kHz and a bit worse closer). Large values > noticeably degraded the performance. > > I suppose the noise floor can be lowered only if better LNA will be used > (currently the LNA noise dominates the PD noise), or if the levels on the > mixer will be increased (this will require higher level mixer and/or new > calibration routine if the mixer will not be in a linear region). > > All the best! > Oleg > > > ___ > time-nuts mailing list -- time-nuts@febo.com > To unsubscribe, go to > https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the > instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] Oleg' s PN test Re: A new member & PN test set
-- From: "Bruce Griffiths" <bruce.griffi...@xtra.co.nz> Sent: Wednesday, March 30, 2016 7:29 AM To: "Discussion of precise time and frequency measurement" <time-nuts@febo.com> Subject: Re: [time-nuts] Oleg' s PN test Re: A new member & PN test set One hidden issue you don't address is that operation of the 40uH inductor at frequencies above its parallel resonance may allow substantial RF at the sum of the LO and RF frequencies to appear at the opamp input. 120MHz at the 797 input will likely lead to RF rectification effects in the opamp input stage. The resultant offset will create a number of issues including operation away from the quadrature point. Unless you use something like a series string of inductors and/or a conical inductor the first parallel resonance of the 40uH inductor is likely to be somewhat below 120MHz. Ohhh... I do not like words like "substantial", "much more" and etc. I like numbers and tests. ;) So I looked at the Murrata inductors datasheets, and it appeared 40uH inductor will have SRF in 10MHz region. But it does not mean that pi-LPF will not work at the higher frequencies. Actually it mean that our LPF will have response similar to the elliptic filters. So let's draw the model with the inductor with self resonance at 10MHz and well at 120MHz and 1MHz to see how bad the response is: 1MHz: http://skydan.in.ua/PNTestSet/1MSRF.png 10MHz: http://skydan.in.ua/PNTestSet/10MSRF.png 120MHz: http://skydan.in.ua/PNTestSet/120MSRF.png As we can see the RF+LO product will be attenuated more than 60dB in all cases. So, your comments? Would you like me to measure the RF voltage at the AD797 input in the real "test set"? However 50 ohms to ground at the LC filter output shouldn't be necessary. A somewhat larger value should suffice. I made some experiments trying to find the optimal value of the resistor at the LC filter output. The phase detector gain grew along with the resistor value, but so did the harmonics level. So I needed to apply more attenuation to the input signal to stay in the linear region. The resulting "test set" noise floor was almost identical for 50..300Ohm values (300Ohm was a bit better at offsets grater then 2kHz and a bit worse closer). Large values noticeably degraded the performance. I suppose the noise floor can be lowered only if better LNA will be used (currently the LNA noise dominates the PD noise), or if the levels on the mixer will be increased (this will require higher level mixer and/or new calibration routine if the mixer will not be in a linear region). All the best! Oleg ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] Oleg' s PN test Re: A new member & PN test set
One hidden issue you don't address is that operation of the 40uH inductor at frequencies above its parallel resonance may allow substantial RF at the sum of the LO and RF frequencies to appear at the opamp input. 120MHz at the 797 input will likely lead to RF rectification effects in the opamp input stage. The resultant offset will create a number of issues including operation away from the quadrature point. Unless you use something like a series string of inductors and/or a conical inductor the first parallel resonance of the 40uH inductor is likely to be somewhat below 120MHz. Bruce On Monday, March 28, 2016 11:04:45 PM Oleg Skydan wrote: > Hi, everybody! > > OK. Let's start. Here is the schematics of the "test set" > http://skydan.in.ua/PNTestSet/PN%20Test%20set.pdf . It consists of three > small > boards: > 1. Mixer board - a simple mixer (500MHz ADE-1+) with 200kHz pi-LPF at the > mixer output. > 2. LNA board - a non-inverting low noise AF amplifier based on AD797 with > switchable 20/40dB gain. > 3. PLL board - contains two TL071 OP amps. One is inverting amplifier, the > other is PLL integrator. The R4,R8,R2,R7,C8 sets the PLL parameters - gain, > passband and damping factor. Loop parameters are also dependent of signal > levels and VCO tuning sensitivity. So you may need to correct them if your > setup differs from mine - VCXO's I use have tuning sensitivity approx > 100Hz/V and I set RF level at mixer near 0dBm with LO level near +7dBm. If > you want to build universal test set you will need to use some switchs to > allow setting different loop parameters (I just use my soldering iron and > change parts if needed :) ). > > The power supply is a simple design based on 7812/7912 regulators. > > I use the E-MU 0202 USB external sound card and laptop PC as the AF spectrum > analyzer. > > You will also need some cables. Different fixed attenuators or switchable > one will be also helpful. > > I also have several homemade low noise VCXOs for some frequencies (7MHz, > 10MHz, 14.318MHz, 60MHz) which I use as the reference signal sources to make > measurements at these frequencies. > > Another option is to test two identical oscillators (or other signal > sources). Assuming that both signals will have identical phase noise > characteristics we can correct the results by 3dB (or just add 3dB > correction during calibration). > > The calibration and use is simple. > 1. Set LNA gain to 20dB. > 2. Set the FFT parameters - flattop window, small (2048..4096) points number > and short averaging in SA software. > 3. Connect reference signal to LO mixer port and signal you are going to > test to RF mixer port through the attenuator. Do not close the PLL yet. > 4. Set the beat level a bit less then the sound card full scale using the > attenuator. Check the beat harmonics levels - they should be at least 30dB > lower then the beat level (add more attenuation if harmonics are higher). > 5. Now set the spectrum analyzer calibration so that beat level is at -27dB > if you measure against low noise reference VXCO, or -30dB if you use two > identical oscillators. > 6. Switch the LNA to 40dB gain. > 7. Set SA software to Blackman window, 131072points/96kHz SR/necessary > averaging, close PLL, wait for the lock, measure the phase noise. > > Why I am calibrating to -27/-30dB: > 20dB because the LNA gain is 20dB less during the calibration (compared to > measurement time) > 1dB because of FFT parameters 96k/131072 = 0.73Hz * 1.73 (Blackman window) = > 1.267Hz, 10 log10(1.267) = 1.03dB > 6dB is the correction inherent to used calibration method > additional 3dB needed in case of testing identical oscillators. > > Now some words about results. The noise floor of this test set depends of > the signals levels, and with the optimal levels it is in -160..-170dBc/Hz > range (depending of the offset from the carrier). It completely satisfies my > needs, better results can be achieved with the higher level mixer and/or > better LNA. I just used parts that I had :). > > Here http://skydan.in.ua/PNTestSet/Screen%20(420)-e.png is an example of the > phase noise measurements results of my homemade low noise 60MHz VCXO (two > identical units were measured). The results at the offsets greater then > 1kHz should be corrected cause the oscillators noise is too close to test > set noise (the real oscillator noise is a bit lower then the displayed > one). The test set noise floor and calibration spectrum is also there. > > The boards also have other use. > > For example I was able to measure my home made 60MHz VCXO harmonic content > http://skydan.in.ua/PNTestSet/Screen%20(414)-e.png using the mixer, LNA > boards and signal generator. I have no spectrum analyzer so it is a big help > to me :). > > Power supply noise can be investigated with the LNA board and sound card. > Look at this screen http://skydan.in.ua/PNTestSet/Screen%20(431)-e.png to > see how bad the LDO regulator noise can be and a great difference in noise >
Re: [time-nuts] Oleg' s PN test Re: A new member & PN test set
Hi Oleg Le 29/03/2016 07:18, Oleg Skydan a écrit : I also tried DL4YHF Spectrum Lab - it works, but lacks of logarithmic frequency scale (or I just did not find how to enable it). Just right-click on the frequency scale, then click on "more ...", and then tick on "logarithmic" in the "Options for the frequency axis" window. Simple, isn't it ? o:) Jean-Louis F6AGR ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] Oleg' s PN test Re: A new member & PN test set
NIST indicate in several early papers that using 50 ohm in series with a capacitor increases the noise over that achieved by capacitively terminating the IF port at the sum frequency (LO + RF) as you have done. NB the RF and LO port match will be improved somewhat with suitable low value series resistors. However 50 ohms to ground at the LC filter output shouldn't be necessary. A somewhat larger value should suffice. Despite assertions by Both minicircuits and Watkins Johnson a mixer connected as a phase detector doesnt act like a current source at the IF output (this is obvious from the plot of the phase detection characteristic as a function of IF port low frequency load resistance on the WJ mixer/PD paper. Also the minicircuits assertion that amplifiers apply a voltage offset to the mixer port that can only be attenuated by a resistor in series with the amplifier input plus a lower value shunt resistance at the output of the IF port LC filter is errant nonsense with an opamp based amplifier. There should be no problems with a parallel connected Buffer amp driving the PLL circuitry at the LC filter output that's the way Wenzel does it for example.Note that the Wenzel low noise amp can be improved significantly using the same components in a slightly different topology. NIST have used an RF noise source for calibration and to measure the frequency response. Bruce On Wednesday, 30 March 2016 12:12 AM, Oleg Skydanwrote: Hi, Bruce, Thank you for the comments and useful link. Probably you did not understand the goal and positioning of this "project" and I did not tell the history of how it was build :) So, the solely goal of making this "test set" was to assist with the design of the synthesizer unit for my HF transceiver. The synthesizer PN goals are to archive PN better then -145..-150dBc/Hz@1kHz offset and better then -150..-155dBc/Hz@5kHz and farther. So I do not need something perfect to measure parts of the synthesizer or the complete unit. Now some words how it was build. Several years ago I experimented with the voltage regulators and needed to measure their noise. So I made an AD797 LNA for my soundcard. Later I added the mixer which I used (along with the signal generator) as a selective meter or primitive spectrum analyzer. Several months ago I started to work at the synthesizer project, so I needed PN "test set". I found a board with two TL071 in suitable configuration in my "junk box" and after several minutes of soldering I had the PLL board :) Usually I am not a fan of such construction methods, but that time it solved problem quickly. > 1) The chosen mixer isnt as low noise as the various Minicircuits phase > detectors. I just used what I have. There are some very bad things here :( (it is way out of the list theme), so buying parts (especially ones not widely used) is not a simple task here. The Minicircuits parts are expensive and exotic here. > 2) The 50 ohm load after the filter merely serves to halve the phase > detector > gain. The IF port is terminated by a 15nF capacitor at RF and LO > frequencies > and their harmonics. This produces a frequency dependent gain, however it > will > likely be relatively flat over the sound card bandwidth. I know it does not terminate mixer correctly, but it is simple and it works.. I tried the termination suggested in the NIST papers (with 50Ohm RF termination and 1kOhm DC/AF one) with no success - the noise floor of the "test set" was higher. As for the gain flatness, I checked it - you can see the results of the quick test here http://skydan.in.ua/PNTestSet/Screen(432)-e.png (it was 60MHz LO + signal generator slowly tuned around 60MHz, the SA was set to peak and hold mode). It completely satisfies my needs. > 3) Saturating both mixer ports increases the phase detector gain > substantially > and has the lowest noise > In this case the simple and reliable calibration method I use will not work, cause the mixer output will not be sinusoidal anymore. Another problem is the signal levels - two good RF LNAs will be needed to amplify signals up to the necessary levels to saturate mixer. > 4) Cascading the PLL circuitry with the preamp causes interaction between > the > Preamp gain settings and the PLL bandwidth. Driving the PLL circuit in > parallel with the preamp input directly from the low pass filtered mixer > output > avoids this issue as well as your 0.1x amplifier in the PLL section. I see no reason to use 20dB preamp gain for measurements (the sound card noise will have too much influence with this setting), so it useful only for calibration or the other LNA use (not in PN test set). On the other hand if the PLL circuit connected to the LNA output we have minimal components/wires/traces/connections in the most sensitive part of the test set, so the chance to pick up some external noise is also minimized. I can add that other good and simple/cheap
Re: [time-nuts] Oleg' s PN test Re: A new member & PN test set
Hi Magnus, Would not GNUradio be a good platform to encode the calibration stuff a little more gift-wrapped? I never used the GNUradio. Basically you can use any SA software/hardware which has the necessary capabilities. What spectrum-analyzer software do you use? (Just curious) It is an old SpectraLAB (which is now as far as I know SpectraPLUS). I also tried DL4YHF Spectrum Lab - it works, but lacks of logarithmic frequency scale (or I just did not find how to enable it). I think, from time to time, about writing specialized software with some special features, but there are always more important things to do :). All the best! Oleg ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] Oleg' s PN test Re: A new member & PN test set
Hi, Bruce, Thank you for the comments and useful link. Probably you did not understand the goal and positioning of this "project" and I did not tell the history of how it was build :) So, the solely goal of making this "test set" was to assist with the design of the synthesizer unit for my HF transceiver. The synthesizer PN goals are to archive PN better then -145..-150dBc/Hz@1kHz offset and better then -150..-155dBc/Hz@5kHz and farther. So I do not need something perfect to measure parts of the synthesizer or the complete unit. Now some words how it was build. Several years ago I experimented with the voltage regulators and needed to measure their noise. So I made an AD797 LNA for my soundcard. Later I added the mixer which I used (along with the signal generator) as a selective meter or primitive spectrum analyzer. Several months ago I started to work at the synthesizer project, so I needed PN "test set". I found a board with two TL071 in suitable configuration in my "junk box" and after several minutes of soldering I had the PLL board :) Usually I am not a fan of such construction methods, but that time it solved problem quickly. 1) The chosen mixer isnt as low noise as the various Minicircuits phase detectors. I just used what I have. There are some very bad things here :( (it is way out of the list theme), so buying parts (especially ones not widely used) is not a simple task here. The Minicircuits parts are expensive and exotic here. 2) The 50 ohm load after the filter merely serves to halve the phase detector gain. The IF port is terminated by a 15nF capacitor at RF and LO frequencies and their harmonics. This produces a frequency dependent gain, however it will likely be relatively flat over the sound card bandwidth. I know it does not terminate mixer correctly, but it is simple and it works. I tried the termination suggested in the NIST papers (with 50Ohm RF termination and 1kOhm DC/AF one) with no success - the noise floor of the "test set" was higher. As for the gain flatness, I checked it - you can see the results of the quick test here http://skydan.in.ua/PNTestSet/Screen(432)-e.png (it was 60MHz LO + signal generator slowly tuned around 60MHz, the SA was set to peak and hold mode). It completely satisfies my needs. 3) Saturating both mixer ports increases the phase detector gain substantially and has the lowest noise In this case the simple and reliable calibration method I use will not work, cause the mixer output will not be sinusoidal anymore. Another problem is the signal levels - two good RF LNAs will be needed to amplify signals up to the necessary levels to saturate mixer. 4) Cascading the PLL circuitry with the preamp causes interaction between the Preamp gain settings and the PLL bandwidth. Driving the PLL circuit in parallel with the preamp input directly from the low pass filtered mixer output avoids this issue as well as your 0.1x amplifier in the PLL section. I see no reason to use 20dB preamp gain for measurements (the sound card noise will have too much influence with this setting), so it useful only for calibration or the other LNA use (not in PN test set). On the other hand if the PLL circuit connected to the LNA output we have minimal components/wires/traces/connections in the most sensitive part of the test set, so the chance to pick up some external noise is also minimized. I can add that other good and simple/cheap additions will be the integrator reset button, two buttons to move integrator in positive or negative direction manually (to speed up the initial lock in some cases, or shift the output voltage into the necessary EFC range), potentiometer for the manual VCO/VCXO frequency control (for the calibration) with the switch to close/open PLL. An OCXO like the 10811A has an EFC gain of around 0.1Hz/volt. The PLL bandwidth should ideally be less than 1/10 of the lowest offset frequency for which the PN is to be measured. If the system frequency response is measured then the PLL bandwidth can be a little higher albeit with a reduction is sensitivity and an increase in system PN at the low offset frequency end of the range. AS is the PN noise of this test set is far too high to measure the PN of state of the art OCXOs or indeed most modern OXCOs. Ohh... I am not a time nut (or maybe not a time nut YET ;). I did not try to make something "state of the art" - my goals were/are different (see earlier). As for the PLL BW of cause one should be aware what the BW is. In my measurements the PLL BW is less then 30Hz. I am not interested in PN closer then 1kHz, so no need for any additional correction of the results. Much more sophisticated system can be made - better ADC, better LNA, RF LNAs to push mixer in saturation, better software, two channels with cross correlation and etc. Or we can even use two high speed ADC and move more things into digital domain. But it can not be done in one evening
Re: [time-nuts] Oleg' s PN test Re: A new member & PN test set
Inversion of gain isn't required, the system will just lock on the opposite slope of the phase detector output. Level shifting to accommodate different EFC ranges nay be useful however. Bruce On Tuesday, 29 March 2016 1:12 PM, Magnus Danielsonwrote: Hi Oleg, I like the simplicity here. Would not GNUradio be a good platform to encode the calibration stuff a little more gift-wrapped? What spectrum-analyzer software do you use? (Just curious) This simple setup would be useful for many purposes. Depending on the oscillator, the EFC can need inversion, so I sketched for a similar design such that I could jumper it for either polarity. In that design I put the inverter after the integrating op-amp. Cheers, Magnus On 03/28/2016 10:04 PM, Oleg Skydan wrote: > Hi, everybody! > > OK. Let's start. Here is the schematics of the "test set" > http://skydan.in.ua/PNTestSet/PN%20Test%20set.pdf . It consists of three > small > boards: > 1. Mixer board - a simple mixer (500MHz ADE-1+) with 200kHz pi-LPF at > the mixer output. > 2. LNA board - a non-inverting low noise AF amplifier based on AD797 > with switchable 20/40dB gain. > 3. PLL board - contains two TL071 OP amps. One is inverting amplifier, > the other is PLL integrator. The R4,R8,R2,R7,C8 sets the PLL parameters > - gain, passband and damping factor. Loop parameters are also dependent > of signal levels and VCO tuning sensitivity. So you may need to correct > them if your setup differs from mine - VCXO's I use have tuning > sensitivity approx 100Hz/V and I set RF level at mixer near 0dBm with LO > level near +7dBm. If you want to build universal test set you will need > to use some switchs to allow setting different loop parameters (I just > use my soldering iron and change parts if needed :) ). > > The power supply is a simple design based on 7812/7912 regulators. > > I use the E-MU 0202 USB external sound card and laptop PC as the AF > spectrum analyzer. > > You will also need some cables. Different fixed attenuators or > switchable one will be also helpful. > > I also have several homemade low noise VCXOs for some frequencies (7MHz, > 10MHz, 14.318MHz, 60MHz) which I use as the reference signal sources to > make measurements at these frequencies. > > Another option is to test two identical oscillators (or other signal > sources). Assuming that both signals will have identical phase noise > characteristics we can correct the results by 3dB (or just add 3dB > correction during calibration). > > The calibration and use is simple. > 1. Set LNA gain to 20dB. > 2. Set the FFT parameters - flattop window, small (2048..4096) points > number and short averaging in SA software. > 3. Connect reference signal to LO mixer port and signal you are going to > test to RF mixer port through the attenuator. Do not close the PLL yet. > 4. Set the beat level a bit less then the sound card full scale using > the attenuator. Check the beat harmonics levels - they should be at > least 30dB lower then the beat level (add more attenuation if harmonics > are higher). > 5. Now set the spectrum analyzer calibration so that beat level is at > -27dB if you measure against low noise reference VXCO, or -30dB if you > use two identical oscillators. > 6. Switch the LNA to 40dB gain. > 7. Set SA software to Blackman window, 131072points/96kHz SR/necessary > averaging, close PLL, wait for the lock, measure the phase noise. > > Why I am calibrating to -27/-30dB: > 20dB because the LNA gain is 20dB less during the calibration (compared > to measurement time) > 1dB because of FFT parameters 96k/131072 = 0.73Hz * 1.73 (Blackman > window) = 1.267Hz, 10 log10(1.267) = 1.03dB > 6dB is the correction inherent to used calibration method > additional 3dB needed in case of testing identical oscillators. > > Now some words about results. The noise floor of this test set depends > of the signals levels, and with the optimal levels it is in > -160..-170dBc/Hz range (depending of the offset from the carrier). It > completely satisfies my needs, better results can be achieved with the > higher level mixer and/or better LNA. I just used parts that I had :). > > Here http://skydan.in.ua/PNTestSet/Screen%20(420)-e.png is an example of > the phase noise measurements results of my homemade low noise 60MHz VCXO > (two identical units were measured). The results at the offsets greater > then 1kHz should be corrected cause the oscillators noise is too close > to test set noise (the real oscillator noise is a bit lower then the > displayed one). The test set noise floor and calibration spectrum is > also there. > > The boards also have other use. > > For example I was able to measure my home made 60MHz VCXO harmonic > content http://skydan.in.ua/PNTestSet/Screen%20(414)-e.png using the > mixer, LNA boards and signal generator. I have no spectrum analyzer so > it is a big help to me :). > > Power supply noise can be investigated with the LNA board and sound > card. Look
Re: [time-nuts] Oleg' s PN test Re: A new member & PN test set
Hi Oleg, I like the simplicity here. Would not GNUradio be a good platform to encode the calibration stuff a little more gift-wrapped? What spectrum-analyzer software do you use? (Just curious) This simple setup would be useful for many purposes. Depending on the oscillator, the EFC can need inversion, so I sketched for a similar design such that I could jumper it for either polarity. In that design I put the inverter after the integrating op-amp. Cheers, Magnus On 03/28/2016 10:04 PM, Oleg Skydan wrote: Hi, everybody! OK. Let's start. Here is the schematics of the "test set" http://skydan.in.ua/PNTestSet/PN%20Test%20set.pdf . It consists of three small boards: 1. Mixer board - a simple mixer (500MHz ADE-1+) with 200kHz pi-LPF at the mixer output. 2. LNA board - a non-inverting low noise AF amplifier based on AD797 with switchable 20/40dB gain. 3. PLL board - contains two TL071 OP amps. One is inverting amplifier, the other is PLL integrator. The R4,R8,R2,R7,C8 sets the PLL parameters - gain, passband and damping factor. Loop parameters are also dependent of signal levels and VCO tuning sensitivity. So you may need to correct them if your setup differs from mine - VCXO's I use have tuning sensitivity approx 100Hz/V and I set RF level at mixer near 0dBm with LO level near +7dBm. If you want to build universal test set you will need to use some switchs to allow setting different loop parameters (I just use my soldering iron and change parts if needed :) ). The power supply is a simple design based on 7812/7912 regulators. I use the E-MU 0202 USB external sound card and laptop PC as the AF spectrum analyzer. You will also need some cables. Different fixed attenuators or switchable one will be also helpful. I also have several homemade low noise VCXOs for some frequencies (7MHz, 10MHz, 14.318MHz, 60MHz) which I use as the reference signal sources to make measurements at these frequencies. Another option is to test two identical oscillators (or other signal sources). Assuming that both signals will have identical phase noise characteristics we can correct the results by 3dB (or just add 3dB correction during calibration). The calibration and use is simple. 1. Set LNA gain to 20dB. 2. Set the FFT parameters - flattop window, small (2048..4096) points number and short averaging in SA software. 3. Connect reference signal to LO mixer port and signal you are going to test to RF mixer port through the attenuator. Do not close the PLL yet. 4. Set the beat level a bit less then the sound card full scale using the attenuator. Check the beat harmonics levels - they should be at least 30dB lower then the beat level (add more attenuation if harmonics are higher). 5. Now set the spectrum analyzer calibration so that beat level is at -27dB if you measure against low noise reference VXCO, or -30dB if you use two identical oscillators. 6. Switch the LNA to 40dB gain. 7. Set SA software to Blackman window, 131072points/96kHz SR/necessary averaging, close PLL, wait for the lock, measure the phase noise. Why I am calibrating to -27/-30dB: 20dB because the LNA gain is 20dB less during the calibration (compared to measurement time) 1dB because of FFT parameters 96k/131072 = 0.73Hz * 1.73 (Blackman window) = 1.267Hz, 10 log10(1.267) = 1.03dB 6dB is the correction inherent to used calibration method additional 3dB needed in case of testing identical oscillators. Now some words about results. The noise floor of this test set depends of the signals levels, and with the optimal levels it is in -160..-170dBc/Hz range (depending of the offset from the carrier). It completely satisfies my needs, better results can be achieved with the higher level mixer and/or better LNA. I just used parts that I had :). Here http://skydan.in.ua/PNTestSet/Screen%20(420)-e.png is an example of the phase noise measurements results of my homemade low noise 60MHz VCXO (two identical units were measured). The results at the offsets greater then 1kHz should be corrected cause the oscillators noise is too close to test set noise (the real oscillator noise is a bit lower then the displayed one). The test set noise floor and calibration spectrum is also there. The boards also have other use. For example I was able to measure my home made 60MHz VCXO harmonic content http://skydan.in.ua/PNTestSet/Screen%20(414)-e.png using the mixer, LNA boards and signal generator. I have no spectrum analyzer so it is a big help to me :). Power supply noise can be investigated with the LNA board and sound card. Look at this screen http://skydan.in.ua/PNTestSet/Screen%20(431)-e.png to see how bad the LDO regulator noise can be and a great difference in noise with the simple transistor filter (sorry there are a lot of power line noise pickup - I needed just to quickly check the power supply noise, so did not pay a lot attention to minimize them). The low noise VCXOs with the combiner and attenuator can be used to measure IMD3 of the receiver. If you add
Re: [time-nuts] Oleg' s PN test Re: A new member & PN test set
That's probably a good example of how not to do it well. 1) The chosen mixer isnt as low noise as the various Minicircuits phase detectors. 2) The 50 ohm load after the filter merely serves to halve the phase detector gain. The IF port is terminated by a 15nF capacitor at RF and LO frequencies and their harmonics. This produces a frequency dependent gain, however it will likely be relatively flat over the sound card bandwidth. 3) Saturating both mixer ports increases the phase detector gain substantially and has the lowest noise 4) Cascading the PLL circuitry with the preamp causes interaction between the Preamp gain settings and the PLL bandwidth. Driving the PLL circuit in parallel with the preamp input directly from the low pass filtered mixer output avoids this issue as well as your 0.1x amplifier in the PLL section. 5) Another method of calibration is desirable in order to account for potentially non flat frequency response. There are a large number of NIST papers on PN measurement including a few on the effect of mixer IF port terminations. A recent one (http://tf.boulder.nist.gov/general/pdf/2556.pdf) compares the PN performance of various mixers used as phase detectors. An OCXO like the 10811A has an EFC gain of around 0.1Hz/volt. The PLL bandwidth should ideally be less than 1/10 of the lowest offset frequency for which the PN is to be measured. If the system frequency response is measured then the PLL bandwidth can be a little higher albeit with a reduction is sensitivity and an increase in system PN at the low offset frequency end of the range. AS is the PN noise of this test set is far too high to measure the PN of state of the art OCXOs or indeed most modern OXCOs. Bruce On Monday, March 28, 2016 11:04:45 PM Oleg Skydan wrote: > Hi, everybody! > > OK. Let's start. Here is the schematics of the "test set" > http://skydan.in.ua/PNTestSet/PN%20Test%20set.pdf . It consists of three > small > boards: > 1. Mixer board - a simple mixer (500MHz ADE-1+) with 200kHz pi-LPF at the > mixer output. > 2. LNA board - a non-inverting low noise AF amplifier based on AD797 with > switchable 20/40dB gain. > 3. PLL board - contains two TL071 OP amps. One is inverting amplifier, the > other is PLL integrator. The R4,R8,R2,R7,C8 sets the PLL parameters - gain, > passband and damping factor. Loop parameters are also dependent of signal > levels and VCO tuning sensitivity. So you may need to correct them if your > setup differs from mine - VCXO's I use have tuning sensitivity approx > 100Hz/V and I set RF level at mixer near 0dBm with LO level near +7dBm. If > you want to build universal test set you will need to use some switchs to > allow setting different loop parameters (I just use my soldering iron and > change parts if needed :) ). > > The power supply is a simple design based on 7812/7912 regulators. > > I use the E-MU 0202 USB external sound card and laptop PC as the AF spectrum > analyzer. > > You will also need some cables. Different fixed attenuators or switchable > one will be also helpful. > > I also have several homemade low noise VCXOs for some frequencies (7MHz, > 10MHz, 14.318MHz, 60MHz) which I use as the reference signal sources to make > measurements at these frequencies. > > Another option is to test two identical oscillators (or other signal > sources). Assuming that both signals will have identical phase noise > characteristics we can correct the results by 3dB (or just add 3dB > correction during calibration). > > The calibration and use is simple. > 1. Set LNA gain to 20dB. > 2. Set the FFT parameters - flattop window, small (2048..4096) points number > and short averaging in SA software. > 3. Connect reference signal to LO mixer port and signal you are going to > test to RF mixer port through the attenuator. Do not close the PLL yet. > 4. Set the beat level a bit less then the sound card full scale using the > attenuator. Check the beat harmonics levels - they should be at least 30dB > lower then the beat level (add more attenuation if harmonics are higher). > 5. Now set the spectrum analyzer calibration so that beat level is at -27dB > if you measure against low noise reference VXCO, or -30dB if you use two > identical oscillators. > 6. Switch the LNA to 40dB gain. > 7. Set SA software to Blackman window, 131072points/96kHz SR/necessary > averaging, close PLL, wait for the lock, measure the phase noise. > > Why I am calibrating to -27/-30dB: > 20dB because the LNA gain is 20dB less during the calibration (compared to > measurement time) > 1dB because of FFT parameters 96k/131072 = 0.73Hz * 1.73 (Blackman window) = > 1.267Hz, 10 log10(1.267) = 1.03dB > 6dB is the correction inherent to used calibration method > additional 3dB needed in case of testing identical oscillators. > > Now some words about results. The noise floor of this test set depends of > the signals levels, and with the optimal levels it is in -160..-170dBc/Hz > range (depending of the
Re: [time-nuts] Oleg' s PN test Re: A new member & PN test set
Hi, everybody! OK. Let's start. Here is the schematics of the "test set" http://skydan.in.ua/PNTestSet/PN%20Test%20set.pdf . It consists of three small boards: 1. Mixer board - a simple mixer (500MHz ADE-1+) with 200kHz pi-LPF at the mixer output. 2. LNA board - a non-inverting low noise AF amplifier based on AD797 with switchable 20/40dB gain. 3. PLL board - contains two TL071 OP amps. One is inverting amplifier, the other is PLL integrator. The R4,R8,R2,R7,C8 sets the PLL parameters - gain, passband and damping factor. Loop parameters are also dependent of signal levels and VCO tuning sensitivity. So you may need to correct them if your setup differs from mine - VCXO's I use have tuning sensitivity approx 100Hz/V and I set RF level at mixer near 0dBm with LO level near +7dBm. If you want to build universal test set you will need to use some switchs to allow setting different loop parameters (I just use my soldering iron and change parts if needed :) ). The power supply is a simple design based on 7812/7912 regulators. I use the E-MU 0202 USB external sound card and laptop PC as the AF spectrum analyzer. You will also need some cables. Different fixed attenuators or switchable one will be also helpful. I also have several homemade low noise VCXOs for some frequencies (7MHz, 10MHz, 14.318MHz, 60MHz) which I use as the reference signal sources to make measurements at these frequencies. Another option is to test two identical oscillators (or other signal sources). Assuming that both signals will have identical phase noise characteristics we can correct the results by 3dB (or just add 3dB correction during calibration). The calibration and use is simple. 1. Set LNA gain to 20dB. 2. Set the FFT parameters - flattop window, small (2048..4096) points number and short averaging in SA software. 3. Connect reference signal to LO mixer port and signal you are going to test to RF mixer port through the attenuator. Do not close the PLL yet. 4. Set the beat level a bit less then the sound card full scale using the attenuator. Check the beat harmonics levels - they should be at least 30dB lower then the beat level (add more attenuation if harmonics are higher). 5. Now set the spectrum analyzer calibration so that beat level is at -27dB if you measure against low noise reference VXCO, or -30dB if you use two identical oscillators. 6. Switch the LNA to 40dB gain. 7. Set SA software to Blackman window, 131072points/96kHz SR/necessary averaging, close PLL, wait for the lock, measure the phase noise. Why I am calibrating to -27/-30dB: 20dB because the LNA gain is 20dB less during the calibration (compared to measurement time) 1dB because of FFT parameters 96k/131072 = 0.73Hz * 1.73 (Blackman window) = 1.267Hz, 10 log10(1.267) = 1.03dB 6dB is the correction inherent to used calibration method additional 3dB needed in case of testing identical oscillators. Now some words about results. The noise floor of this test set depends of the signals levels, and with the optimal levels it is in -160..-170dBc/Hz range (depending of the offset from the carrier). It completely satisfies my needs, better results can be achieved with the higher level mixer and/or better LNA. I just used parts that I had :). Here http://skydan.in.ua/PNTestSet/Screen%20(420)-e.png is an example of the phase noise measurements results of my homemade low noise 60MHz VCXO (two identical units were measured). The results at the offsets greater then 1kHz should be corrected cause the oscillators noise is too close to test set noise (the real oscillator noise is a bit lower then the displayed one). The test set noise floor and calibration spectrum is also there. The boards also have other use. For example I was able to measure my home made 60MHz VCXO harmonic content http://skydan.in.ua/PNTestSet/Screen%20(414)-e.png using the mixer, LNA boards and signal generator. I have no spectrum analyzer so it is a big help to me :). Power supply noise can be investigated with the LNA board and sound card. Look at this screen http://skydan.in.ua/PNTestSet/Screen%20(431)-e.png to see how bad the LDO regulator noise can be and a great difference in noise with the simple transistor filter (sorry there are a lot of power line noise pickup - I needed just to quickly check the power supply noise, so did not pay a lot attention to minimize them). The low noise VCXOs with the combiner and attenuator can be used to measure IMD3 of the receiver. If you add the mixer, LNA and signal generator you can measure the IMD3 of the separate units (mixers, filters, amplifiers and etc.). The low noise VCXO can also be used to test reciprocal mixing DR of the receiver. Other useful combinations are possible. If you like I can post the photos of the boards. They a bit ugly :). Every time I use them I think about mounting them in personal metal boxes, but I always find something more important to do... Best wishes,