Empathy List Archives

EB

ed breya

Tue, Nov 15, 2011 4:21 PM

You can add a crystal filter using a common 10 MHz crystal in series
with a low noise, high impedance amplifier. Terminate the input side
with about 50 ohms, and set the Q of the filter with the load
impedance. Since you're only worried about harmonics, the Q can be
fairly low, making the exact crystal characteristics (frequency
precision especially) less important. Extra components can be added
for more sophistication and compensation for parasitics (especially
parallel Co), if needed. More stages will make it even cleaner, but
the amplifier noises will add up. If there is enough signal power
available in the original output, you can leave out the amplifier and
use an impedance matching transformer instead, avoiding the amplifier noise.

Ed

You can add a crystal filter using a common 10 MHz crystal in series with a low noise, high impedance amplifier. Terminate the input side with about 50 ohms, and set the Q of the filter with the load impedance. Since you're only worried about harmonics, the Q can be fairly low, making the exact crystal characteristics (frequency precision especially) less important. Extra components can be added for more sophistication and compensation for parasitics (especially parallel Co), if needed. More stages will make it even cleaner, but the amplifier noises will add up. If there is enough signal power available in the original output, you can leave out the amplifier and use an impedance matching transformer instead, avoiding the amplifier noise. Ed

JA

John Ackermann N8UR

Tue, Nov 15, 2011 9:18 PM

If the problem is only harmonics (cleaning up a square wave), a simple
LC low pass filter with a cutoff midway between the fundamental and
second harmonic might be a better choice because it minimizes the
tempco-related phase shift that either a high Q filter, or one with a
cutoff near the desired frequency, can exhibit.

John

On 11/15/2011 11:21 AM, ed breya wrote:

You can add a crystal filter using a common 10 MHz crystal in series
with a low noise, high impedance amplifier. Terminate the input side
with about 50 ohms, and set the Q of the filter with the load impedance.
Since you're only worried about harmonics, the Q can be fairly low,
making the exact crystal characteristics (frequency precision
especially) less important. Extra components can be added for more
sophistication and compensation for parasitics (especially parallel Co),
if needed. More stages will make it even cleaner, but the amplifier
noises will add up. If there is enough signal power available in the
original output, you can leave out the amplifier and use an impedance
matching transformer instead, avoiding the amplifier noise.

Ed

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.

If the problem is only harmonics (cleaning up a square wave), a simple LC low pass filter with a cutoff midway between the fundamental and second harmonic might be a better choice because it minimizes the tempco-related phase shift that either a high Q filter, or one with a cutoff near the desired frequency, can exhibit. John ---- On 11/15/2011 11:21 AM, ed breya wrote: > You can add a crystal filter using a common 10 MHz crystal in series > with a low noise, high impedance amplifier. Terminate the input side > with about 50 ohms, and set the Q of the filter with the load impedance. > Since you're only worried about harmonics, the Q can be fairly low, > making the exact crystal characteristics (frequency precision > especially) less important. Extra components can be added for more > sophistication and compensation for parasitics (especially parallel Co), > if needed. More stages will make it even cleaner, but the amplifier > noises will add up. If there is enough signal power available in the > original output, you can leave out the amplifier and use an impedance > matching transformer instead, avoiding the amplifier noise. > > Ed > > > _______________________________________________ > 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.

SM

Sylvain Munaut

Tue, Nov 15, 2011 10:26 PM

Hi,

If the problem is only harmonics (cleaning up a square wave), a simple LC
low pass filter with a cutoff midway between the fundamental and second
harmonic might be a better choice because it minimizes the tempco-related
phase shift that either a high Q filter, or one with a cutoff near the
desired frequency, can exhibit.

Mmm, very interesting.

I was indeed a bit worried about introducing a temperature dependent
element, but didn't think that making a filter with a cutoff far from
the desired frequency might solve that issue.

The unit already has a LC filter but not really good enough I guess.
Has anyone measured it to know what its design parameters are ?

What I still don't get is how bad the signal looks at the RF connector inside.

This is at the DB9 output (measured after a small wire from a DB9
female) : http://i.imgur.com/R33B6.png
There is some distortion near the 0 crossing and the fft shows
harmonics clearly.
This is at the RF connector : http://i.imgur.com/Bg3SK.png
That's just nasty and indescriptible.

The signal before the capacitor leading to that RF connector
doesn't look bad tough ...

(both measurements done with a 500 MHz scope and a 4 GHz FET probe,
with a ground blade to the nearest gnd each time).

Cheers,

Sylvain

Hi, > If the problem is only harmonics (cleaning up a square wave), a simple LC > low pass filter with a cutoff midway between the fundamental and second > harmonic might be a better choice because it minimizes the tempco-related > phase shift that either a high Q filter, or one with a cutoff near the > desired frequency, can exhibit. Mmm, very interesting. I was indeed a bit worried about introducing a temperature dependent element, but didn't think that making a filter with a cutoff far from the desired frequency might solve that issue. The unit already has a LC filter but not really good enough I guess. Has anyone measured it to know what its design parameters are ? What I still don't get is how bad the signal looks at the RF connector inside. * This is at the DB9 output (measured after a small wire from a DB9 female) : http://i.imgur.com/R33B6.png There is some distortion near the 0 crossing and the fft shows harmonics clearly. * This is at the RF connector : http://i.imgur.com/Bg3SK.png That's just nasty and indescriptible. The signal _before_ the capacitor leading to that RF connector doesn't look bad tough ... (both measurements done with a 500 MHz scope and a 4 GHz FET probe, with a ground blade to the nearest gnd each time). Cheers, Sylvain

DV

David VanHorn

Tue, Nov 15, 2011 10:31 PM

This is at the RF connector : http://i.imgur.com/Bg3SK.png
That's just nasty and indescriptible.

Close in impedance discontinuity?

* This is at the RF connector : http://i.imgur.com/Bg3SK.png That's just nasty and indescriptible. Close in impedance discontinuity?

AM

Alan Melia

Tue, Nov 15, 2011 11:06 PM

I may be mistaken but that does not look like harmonic distortion to me.....
the frequency "imposed" on the 10MHz sine-wave is around 140MHz, this sounds
like the VCO frequency in the Rb loop. The corner frequency of the filter
could be a long way away from 10MHz in that case ....with maybe minimal
phase shift tempco at 10MHz ?? Butterworth?

Alan G3NYK

----- Original Message -----
From: "David VanHorn" D.VanHorn@elec-solutions.com
To: "Discussion of precise time and frequency measurement"
time-nuts@febo.com
Sent: Tuesday, November 15, 2011 10:31 PM
Subject: Re: [time-nuts] FE 5680A "new version" - Filtering the 10 MHz

This is at the RF connector : http://i.imgur.com/Bg3SK.png
That's just nasty and indescriptible.

Close in impedance discontinuity?

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.

I may be mistaken but that does not look like harmonic distortion to me..... the frequency "imposed" on the 10MHz sine-wave is around 140MHz, this sounds like the VCO frequency in the Rb loop. The corner frequency of the filter could be a long way away from 10MHz in that case ....with maybe minimal phase shift tempco at 10MHz ?? Butterworth? Alan G3NYK ----- Original Message ----- From: "David VanHorn" <D.VanHorn@elec-solutions.com> To: "Discussion of precise time and frequency measurement" <time-nuts@febo.com> Sent: Tuesday, November 15, 2011 10:31 PM Subject: Re: [time-nuts] FE 5680A "new version" - Filtering the 10 MHz > > > * This is at the RF connector : http://i.imgur.com/Bg3SK.png > That's just nasty and indescriptible. > > > Close in impedance discontinuity? > _______________________________________________ > 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.

JL

Jim Lux

Wed, Nov 16, 2011 1:47 AM

On 11/15/11 3:06 PM, Alan Melia wrote:

Another approach to filtering is a 10.7 MHz IF filter... they're common,
fairly wide band, but not too wide. Can't say much about the tempco.

Minicircuits BLP-10.7 or BLP-15 for instance

or BBP-10.7 (which is about 2 MHz wide)

On 11/15/11 3:06 PM, Alan Melia wrote: > Another approach to filtering is a 10.7 MHz IF filter... they're common, fairly wide band, but not too wide. Can't say much about the tempco. Minicircuits BLP-10.7 or BLP-15 for instance or BBP-10.7 (which is about 2 MHz wide)

time-nuts@lists.febo.com

FE 5680A "new version" - Filtering the 10 MHz

John