Hi Best guess:
Your 10 MHz (for whatever reason) is being turned into a fast edge square wave. That’s dumping current spikes into the supply and ground on your board. You have a really wide spectrum as a result ( usually well up into the GHz region). That “stuff” is going here / going there. Some of it is getting into your amplifier / filter chain by one route or the other. It looks like modulation because of the filtering in your multiplier chain. Since the PLL chip does the same thing ( when it generates 10 MHz into the phase detector, there are two possible sources of the fast edges and current spikes. Yes this makes tracking things down a bit more fun. Either way, the answer is board layout / bypassing / decoupling related. If you want to go with a different “low frequency” phase detector chip, the ADF4002 has been a go to part for quite a while. Bob > On Feb 21, 2022, at 9:27 PM, g...@hoffmann-hochfrequenz.de wrote: > > Hi! > > I'm building a clock source for my LMX2594 15 GHz PLL. > > I want to max out the LMX2594 phase comparator frequency; > that means 300 MHz for fractional operation. The VCXO is a > 100 MHz ECOC-2250 crystal oven by ECS because I have it, > and it's tripled. > > Dumping its output into a pair of 1G125 line drivers gets me > this spectrum: (100 MHz....png) The 300 MHz is quite prominent > with not much of a loss. I dumped it into a 300 MHz filter, > 3 poles, C-coupled, then a sot-89 MMIC and another 3 poles. > That cleaned up the harmonics quite good. Sorry, there were > no 300 MHz SAW-filters available. All obsoleted. > The filter is 6 Murata 0603 SMD inductors and fixed Cs. Still a > bit to the low side; that can be fixed. > > BUT - there is a problem. The 100 MHz can be locked to an external > 10 MHz reference, and the 10 MHz is modulated onto the 300 MHz. > I do not want this to be multiplied to 15 GHz. > > There is a 74lvc163 counter that is visible as well. (100/10 = 10 MHz) > The PLL chip is a 74lv4046, the phase comparator part. When the > reference is != 10 MHz and the prescaler is running, OMG, grass like > on an African steppe. Only the zebras are missing. > Without ext ref: sx3fQ8.png > > I don't think that the VCXO has thus a large modulation bandwidth > and I'd like to replace the 4046 with a 9046, but I've found no > way to switch off its internal oscillator. Having not-vanishing loop > gain at lock is fine, but getting the 8046's undefined on-chip VCO > on the output spectrum would be no improvement. > > Any ideas? > > Cheers, Gerhard > > p.s. > How can the spectrum analyzer (89411A) encode such sharp screen dumps in just > 3 KB? > > <100MHz_from 2 * > lvc125.png><300mhzmal2-sky17.png><sx3fQ8.png>_______________________________________________ > time-nuts mailing list -- time-nuts@lists.febo.com -- To unsubscribe send an > email to time-nuts-le...@lists.febo.com > To unsubscribe, go to and follow the instructions there. _______________________________________________ time-nuts mailing list -- time-nuts@lists.febo.com -- To unsubscribe send an email to time-nuts-le...@lists.febo.com To unsubscribe, go to and follow the instructions there.