Re: [time-nuts] Dead 5065A - Some comments - Some questions... (III)
Ulf I suspect we are the same "Get it done". Not a science project. Though science comes into it. I just find the various new devices may be more easily obtained. L are getting hard to come by. So far I don't have failed FETs or anything else. But that day will come. Regards Paul On Wed, Aug 26, 2020 at 3:14 AM Ulf Kylenfall via time-nuts < time-nuts@lists.febo.com> wrote: > Paul, > > A 60 MHz xtal oscillator divided by 12 and comparedwith the incoming 5 > MHz. Then there is this thing > how to acheive a good symmetrical FM injecting > the 137 Hz signal without affecting the loop filterparameters. > > Keeping the original design with modern typesof semiconductors is less > work/effort, provided that the > unit keeps working... > The A7 original design used a non selective2nd harmonic detector. And uA > 709's.That one was easy to motivate an upgrade. > > I am a bit lazy otherwise. > > As mentioned, the semiconductors that have failed > in A3 were all dated 1968. > At the time without knowing the exact types of the JFET'sI replaced them > with 2N4416 only to discoverthat that was the type HP specified in a > laterrevision of the A3. > Best Regards > Ulf Kylenfall > ___ > time-nuts mailing list -- time-nuts@lists.febo.com > To unsubscribe, go to > http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com > and follow the instructions there. > ___ time-nuts mailing list -- time-nuts@lists.febo.com To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com and follow the instructions there.
Re: [time-nuts] Dead 5065A - Some comments - Some questions... (III)
Hi > On Aug 27, 2020, at 10:35 AM, Poul-Henning Kamp wrote: > > > Bob kb8tq writes: > >> Common wisdom is that there is a filter in the cavity match that >> is specific to 60 MHz. Putting in 120 MHz apparently does not work >> well. I haven’t tried it so I’m just passing on what I’ve heard. > > Driving step-diodes is a bit of an art, and the relationship between > driving frequency, AC power and DC bias is very far from trivial, > as the "real" parameter, if I have understood it right, is the > slew-rate through the "magic" voltage. > >> The “other” signal (normally 5.3x MHz) is mixed with the multiplied 60 MHz. > > As far as I have been able to make out, both the 137Hz and 5.3x MHz > are "mixed" through their amplitude modulation of the DC-bias. At least in the 5065, the modulation signal feeds into the (A3) multiplier board and hits a phase modulator running at 5 MHz. It’s that phase modulator that generates the PM on the output. In my manual the stage is labeled “10 MHz Multiplier and Phase Mod”. The dirty work is done by CR3,4,5, and 6. They are driven by the “mod level” pot. Since the modulator is early in the string of stages. I doubt that a lot of AM is *supposed* to get through to the output. Bob > > My Personal DDS plan is to simply drive the step diode at 59.953... MHz > +/- the modulation pattern, and forget entirely about the 5.3x MHz and > 137 Hz. > > -- > Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 > p...@freebsd.org | TCP/IP since RFC 956 > FreeBSD committer | BSD since 4.3-tahoe > Never attribute to malice what can adequately be explained by incompetence. ___ time-nuts mailing list -- time-nuts@lists.febo.com To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com and follow the instructions there.
Re: [time-nuts] Dead 5065A - Some comments - Some questions... (III)
Magnus Danielson writes: > No, the modulation frequency would still be the same, but the modulation > width scale up as the carrier scales up. Thus, the frequency deviation > scales up. If we are going to (as I hope) DDS synthesis, then there are better modulation *patterns* than the current sine-wave. (I have no info/opinion on the choice of modulation *frequency*.) -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 p...@freebsd.org | TCP/IP since RFC 956 FreeBSD committer | BSD since 4.3-tahoe Never attribute to malice what can adequately be explained by incompetence. ___ time-nuts mailing list -- time-nuts@lists.febo.com To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com and follow the instructions there.
Re: [time-nuts] Dead 5065A - Some comments - Some questions... (III)
Bob kb8tq writes: > Common wisdom is that there is a filter in the cavity match that > is specific to 60 MHz. Putting in 120 MHz apparently does not work > well. I haven’t tried it so I’m just passing on what I’ve heard. Driving step-diodes is a bit of an art, and the relationship between driving frequency, AC power and DC bias is very far from trivial, as the "real" parameter, if I have understood it right, is the slew-rate through the "magic" voltage. > The “other” signal (normally 5.3x MHz) is mixed with the multiplied 60 MHz. As far as I have been able to make out, both the 137Hz and 5.3x MHz are "mixed" through their amplitude modulation of the DC-bias. My Personal DDS plan is to simply drive the step diode at 59.953... MHz +/- the modulation pattern, and forget entirely about the 5.3x MHz and 137 Hz. -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 p...@freebsd.org | TCP/IP since RFC 956 FreeBSD committer | BSD since 4.3-tahoe Never attribute to malice what can adequately be explained by incompetence. ___ time-nuts mailing list -- time-nuts@lists.febo.com To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com and follow the instructions there.
Re: [time-nuts] Dead 5065A - Some comments - Some questions... (III)
Magnus agree on the bandwidth tripling. I thought the A3 generated 90 MHz. I was thinking the PLL at 90 MHz with a small power amplifier to drive the multiplier. I did see the pictures of the modernized A3 so thats a great answer. Ultimately my concern has been the SRD multiplier. There is some real magic going on in that unit as has been mentioned on time-nuts. If the diode goes there are no replacements accept for another unit. Regards Paul. On Thu, Aug 27, 2020 at 8:19 AM Magnus Danielson wrote: > Hi Paul, > > On 2020-08-26 17:11, paul swed wrote: > > Ulf I was thinking the 137 Hz was injected differently. But yes its a > > varicap on the 5 Mhz signal. Still that same effect can be accomplished > in > > the PLL feedback loop. I would simply go try these things but can not > hack > > my one and only 5065. Its working just fine. > Arn't you curious about making it work finer? :) > > A question > > The 137 is injected at 5Mhz isn't it multiplied to a FM swing of 2466 Hz > at > > 90 MHz? Granted close in sidebands would still contain 137 Hz. > > No, the modulation frequency would still be the same, but the modulation > width scale up as the carrier scales up. Thus, the frequency deviation > scales up. > > As you are a ham, I will make you recall the old days. In the old days > as people aimed to reach the 70 cm band, they used their 2 m equipment > and trippled the frequency. This makes the 144 MHz go to 432 MHz, the 25 > kHz deviation go to 75 kHz deviation and the modulated voice stays the > same pitch as it had on 144 MHz and does not pitch up about 1.5 octaves. > > Cheers, > Magnus > > > > ___ > time-nuts mailing list -- time-nuts@lists.febo.com > To unsubscribe, go to > http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com > and follow the instructions there. > ___ time-nuts mailing list -- time-nuts@lists.febo.com To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com and follow the instructions there.
Re: [time-nuts] Dead 5065A - Some comments - Some questions... (III)
Hi Unless you are going to dig in and rebuild the microwave part of the physics package, you are limited in terms of what you can or can’t do. The SRD setup seems to be a bit unique in the 5065 so messing with it probably isn’t a good idea. Common wisdom is that there is a filter in the cavity match that is specific to 60 MHz. Putting in 120 MHz apparently does not work well. I haven’t tried it so I’m just passing on what I’ve heard. The 60 MHz is multiplied to the 6 GHz region. That increases it’s phase noise by the well known 20 log N. With N > 100, noise on that side goes up by > 40 db. The “other” signal (normally 5.3x MHz) is mixed with the multiplied 60 MHz. The net phase noise is the sum of the noise on the two carriers. You get the noise contribution on both carriers at 137 Hz offset *and* the far remove noise on the 60 MHz at a 5.3 MHz offset. All of that is just basic mix and multiply stuff. If a mixer didn’t transfer modulation that way, a super heterodyne radio would have a really hard time doing it’s thing … ( = think of the 5.3x as the signal from the antenna and the 6.x GHz as the VFO input to the mixer). Fun !! Bob > On Aug 27, 2020, at 4:36 AM, Magnus Danielson wrote: > > Hej Ulf, > > On 2020-08-26 09:09, Ulf Kylenfall via time-nuts wrote: >> Paul, >> >> A 60 MHz xtal oscillator divided by 12 and comparedwith the incoming 5 MHz. >> Then there is this thing >> how to acheive a good symmetrical FM injecting >> the 137 Hz signal without affecting the loop filterparameters. >> >> Keeping the original design with modern typesof semiconductors is less >> work/effort, provided that the >> unit keeps working... >> The A7 original design used a non selective2nd harmonic detector. And uA >> 709's.That one was easy to motivate an upgrade. >> >> I am a bit lazy otherwise. > > OK, If you do a PLL loop of 60 MHz oscillator, well then you can just > inject the 137 Hz into the loop just at the phase detector, assuming > it's above the PLL bandwidth. If you the PLL bandwidth is higher, then > you are on a 6 dB/Oct slope, but sure, for smaller range you can just > compensate with the drive level. This is a known technology to do > phase-modulation. You can be pretty lazy doing it after you gone through > the trouble of doing a PLL-loop. > > Using a higher frequency oscillator with low noise can reduce noise > compared to the doubler/trippler chain. One could consider if not an > even higher frequency should be considered. > > If one plays that game, one should also consider the strategy of > synthesizing two frequencies that then is combined at 6.8 GHz, as the > noise of the two signals adds. Synthesizing a single frequency would be > preferred, but then again, working within the limits of the current > system might just work well enough. We can improve however by leaving 60 > MHz and see if we can find a more suitable higher frequency that would > work equally well, as the need for low noise at higher frequencies has > increased for similar enough purposes. > >> >> As mentioned, the semiconductors that have failed >> in A3 were all dated 1968. >> At the time without knowing the exact types of the JFET'sI replaced them >> with 2N4416 only to discoverthat that was the type HP specified in a >> laterrevision of the A3. > > Good to know. Thanks. > > Cheers, > Magnus > > > > ___ > time-nuts mailing list -- time-nuts@lists.febo.com > To unsubscribe, go to > http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com > and follow the instructions there. ___ time-nuts mailing list -- time-nuts@lists.febo.com To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com and follow the instructions there.
Re: [time-nuts] Dead 5065A - Some comments - Some questions... (III)
Hej Ulf, On 2020-08-26 09:09, Ulf Kylenfall via time-nuts wrote: > Paul, > > A 60 MHz xtal oscillator divided by 12 and comparedwith the incoming 5 MHz. > Then there is this thing > how to acheive a good symmetrical FM injecting > the 137 Hz signal without affecting the loop filterparameters. > > Keeping the original design with modern typesof semiconductors is less > work/effort, provided that the > unit keeps working... > The A7 original design used a non selective2nd harmonic detector. And uA > 709's.That one was easy to motivate an upgrade. > > I am a bit lazy otherwise. OK, If you do a PLL loop of 60 MHz oscillator, well then you can just inject the 137 Hz into the loop just at the phase detector, assuming it's above the PLL bandwidth. If you the PLL bandwidth is higher, then you are on a 6 dB/Oct slope, but sure, for smaller range you can just compensate with the drive level. This is a known technology to do phase-modulation. You can be pretty lazy doing it after you gone through the trouble of doing a PLL-loop. Using a higher frequency oscillator with low noise can reduce noise compared to the doubler/trippler chain. One could consider if not an even higher frequency should be considered. If one plays that game, one should also consider the strategy of synthesizing two frequencies that then is combined at 6.8 GHz, as the noise of the two signals adds. Synthesizing a single frequency would be preferred, but then again, working within the limits of the current system might just work well enough. We can improve however by leaving 60 MHz and see if we can find a more suitable higher frequency that would work equally well, as the need for low noise at higher frequencies has increased for similar enough purposes. > > As mentioned, the semiconductors that have failed > in A3 were all dated 1968. > At the time without knowing the exact types of the JFET'sI replaced them with > 2N4416 only to discoverthat that was the type HP specified in a laterrevision > of the A3. Good to know. Thanks. Cheers, Magnus ___ time-nuts mailing list -- time-nuts@lists.febo.com To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com and follow the instructions there.
Re: [time-nuts] Dead 5065A - Some comments - Some questions... (III)
Hi Paul, On 2020-08-26 17:11, paul swed wrote: > Ulf I was thinking the 137 Hz was injected differently. But yes its a > varicap on the 5 Mhz signal. Still that same effect can be accomplished in > the PLL feedback loop. I would simply go try these things but can not hack > my one and only 5065. Its working just fine. Arn't you curious about making it work finer? :) > A question > The 137 is injected at 5Mhz isn't it multiplied to a FM swing of 2466 Hz at > 90 MHz? Granted close in sidebands would still contain 137 Hz. No, the modulation frequency would still be the same, but the modulation width scale up as the carrier scales up. Thus, the frequency deviation scales up. As you are a ham, I will make you recall the old days. In the old days as people aimed to reach the 70 cm band, they used their 2 m equipment and trippled the frequency. This makes the 144 MHz go to 432 MHz, the 25 kHz deviation go to 75 kHz deviation and the modulated voice stays the same pitch as it had on 144 MHz and does not pitch up about 1.5 octaves. Cheers, Magnus ___ time-nuts mailing list -- time-nuts@lists.febo.com To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com and follow the instructions there.
Re: [time-nuts] Dead 5065A - Some comments - Some questions... (III)
Ulf I was thinking the 137 Hz was injected differently. But yes its a varicap on the 5 Mhz signal. Still that same effect can be accomplished in the PLL feedback loop. I would simply go try these things but can not hack my one and only 5065. Its working just fine. A question The 137 is injected at 5Mhz isn't it multiplied to a FM swing of 2466 Hz at 90 MHz? Granted close in sidebands would still contain 137 Hz. Regards Paul On Wed, Aug 26, 2020 at 3:14 AM Ulf Kylenfall via time-nuts < time-nuts@lists.febo.com> wrote: > Paul, > > A 60 MHz xtal oscillator divided by 12 and comparedwith the incoming 5 > MHz. Then there is this thing > how to acheive a good symmetrical FM injecting > the 137 Hz signal without affecting the loop filterparameters. > > Keeping the original design with modern typesof semiconductors is less > work/effort, provided that the > unit keeps working... > The A7 original design used a non selective2nd harmonic detector. And uA > 709's.That one was easy to motivate an upgrade. > > I am a bit lazy otherwise. > > As mentioned, the semiconductors that have failed > in A3 were all dated 1968. > At the time without knowing the exact types of the JFET'sI replaced them > with 2N4416 only to discoverthat that was the type HP specified in a > laterrevision of the A3. > Best Regards > Ulf Kylenfall > ___ > time-nuts mailing list -- time-nuts@lists.febo.com > To unsubscribe, go to > http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com > and follow the instructions there. > ___ time-nuts mailing list -- time-nuts@lists.febo.com To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com and follow the instructions there.