Hello,

Normal design process for this type of LPF is to first calculate the filter and ensure no capacitor is smaller than 0.5pF
Than make a prototype using the calculated capacitors and find the performance is terrible.
Next step is to reduce capacitors and increase inductor to compensate for the PCB parasitic capacitance till you have a acceptable performance.
With the rather low value shunt capacitors at the extremes of the filter one often experimentally determines that no capacitor fitted is the best solution
For a 350MHz low pass filter variable capacitors will be a challenge as anything large size (e.g. above 0603 components) will increase leakage and reduce performance.
Only those that have access to 3D field solvers and all S-models of the used components can calculate the components with some chance to be approximate correct
--
HBTE Files section:?/g/HBTE/files
Erik, PD0EK

Re, some simulation?
It's seem good for a low pass filter 350mhz

Hello,

My guess would be that whoever drew the schematic never entered information for those parts, therefore DNF = Did Not Find, Device Not Found, or Data Not Found.? I have done communications for some 100 mile mountain runs recently, and we use DNF for Did Not Finish.

Yes, inductors can be printed on the circuit board, as can capacitors.? As for form, I assume you mean should the printed inductors be circular, square, hexagonal, etc.?? There are programs you can model that sort of thing with, but my guess would be that circular would be best, as that would eliminate abrupt discontinuities.? I've never done printed inductors, but with microstrip transmission lines, you have to chamfer the corners of bends to avoid impedance continuities, and I assume something similar would apply to printed inductors, at least at higher frequencies.

Hello,

Hello,

Hello Dave and others

I registered a group a few years ago, but never used it. Today I decided to turn it into a group for the design and construction of test equipment.?

/g/Test-Equipment-Design-Construction

The purpose of the group is not amateur radio, but test equipment in general, so something for measuring pH or viscosity would be on topic.

?I announced the group on a few other groups I managed, and made people aware of HBTE too. I think you have picked up about 10 new subscribers in the last couple of hours.

Dave

On Tue, Jul 26, 2022 at 07:56 PM, I_am Wally wrote:

One approach:?
There is a presumption that the effective (loaded by scope and coupled pulse generator) tank capacitor Q is good enough so the inductor Q dominates what you see in the ring-down, something to be tested in your experimental setup.

Another coupling means:? A low capacitance, high speed switching diode can be used to isolate the pulse generator after the pulse.? Generators that allow offsetting the baseline allow the diode then be driven deep into reverse bias if it matters to minimize diode capacitance.?

Wally

Thank you for that. It was not a technique I was aware of.

One approach:?
There is a presumption that the effective (loaded by scope and coupled pulse generator) tank capacitor Q is good enough so the inductor Q dominates what you see in the ring-down, something to be tested in your experimental setup.

Another coupling means:? A low capacitance, high speed switching diode can be used to isolate the pulse generator after the pulse.? Generators that allow offsetting the baseline allow the diode then be driven deep into reverse bias if it matters to minimize diode capacitance.?

Wally

I would like some way of measuring the Q of inductors up to at least 30 MHz - 150 MHz would be nice.?--

My schematic had a print error, but the board was built with J310's .? Sorry for the error.
ANG

Wally, It appears that I used the .01 on the source follower.? That's a long way from 10pf.? Could that stop the oscillation?
see attached files.? Even though I originally put both oscillators on one board, I built them up separately.
Andy KM4 TRT

开云体育

?The 100MHz crystal oscillator? is a very bad idea. A 20MHZ fundamental mode crystal is used in a 100MHz 5th overtone circuit. It is a lottery if the 20MHz crystal shows activity at it's 5th overtone, many fundamental crystals are cut to avoid such activity.

Even if you get it to oscillate it won't be at 100MHz, if you were lucky enough to get some activity at? around 100MHz it would not be that close to 100MHz as the overtone frequency is NEVER an exact multiple of the fundamental frequency, this is a mechanical oscillation inside the quartz crystal slice that has different "end effect" distances for different modes.

A crystal is specifically cut for 3rd, 5th or 7th mode oscillation, normally tuned circuits are needed to suppress unwanted modes. The pulling range of an overtone oscillator is much smaller than a fundamental, an inverse square law is involved, 3rd pulls about 11%, 5th about 4%, 7th about 2% of the fundamental pulling range. Most overtone oscillators have serious high Q resonances HF of the nominal frequency, these need suppression.? Quartz crystal & ceramic? filters can also? have serious resonances HF of the nominal passband too, the effect diminishes with the pole count but is serious in 2 and 4 pole monolithic filters.?

The oscillator is a Butler series mode circuit which is a very predictable and reliable oscillator using bipolar transistors. JFETS are not so suitable as the gm is very variable from device to device. As the crystal is driven in series mode any additional series resistance greatly reduces the resonance width. A series mode overtone crystal should have an internal equivalent resistance of a few ohms, 20 ohms would be very high yet the suggested circuit puts nearly 100 ohms in series with the crystal so loosing a lot of gain ( from the 100/470 potential divider with a drain follower that probably has around 20 ohms of output resistance ( 1/gm).

So it's a bad circuit in that:-

• The designers did not produce a "reproducible with average parts"? circuit
• They assumed that a fundamental mode ( common or garden type) crystal would have overtone activity
• They put excessive resistance in series with the crystal which would seriously degrade phase noise and inhibit oscillation
• They used wide tolerance JFETS where bipolars would work much more reliably
• It was a bodge to get around using an expensive 5th overtone crystal that is not that commonly listed
• They got lucky with the original, others had less luck
  

In the years since the price of? mass market crystals has fallen remarkably low, improvements in mass manufacturing has resulted in generally higher activity, higher Q and sometimes better suppression of unwanted modes. Also packaged oscillators are now very low cost and very reliable. Today it would be very straightforward to use a ready made 100 MHz crystal oscillator package.

Regards, Alan G8LCO.

I'm still reading also, and confused as to which is the real 100 MHz oscillator.? There are two versions of the same topology, one in the large schematic (and in the Google drive clearer image), and another in the article (scroll way down) section 7 description of the 100 MHz oscillator:

10 pf or .01 source follower output loading may or may not matter.? But the reactance difference loads the Q2 source and sure affects the mixer drive level.

But it seems to me that more importantly the two coils should not couple so that the crystal path determines the feedback, not an accidental transformer.? If you enlarge the photo of his construction (the overall photo with shields removed) in the upper left hand section you can see the layout of his 100 MHz oscillator and crystal, with the two coils close but perpendicular forming a T geometry.?

I speculate that minimizing and phasing the coupling of L10 and L11 could be make or break.

Wally