ADF5355 - Good Enough for 24 GHz!


Dave G8GKQ
 

Today Noel G8GTZ and I completed a 2-way digital ATV contact on 24 GHz over 29 km.  What was notable was that I was using a free-running ADF5355 as my local oscillator.


Since an earlier failed contact, I had been working on reducing the phase noise and I would like to thank Brian GM8BJF for his ScatterPoint article and Andy G4JNT for his programming suggestions (increasing the charge pump current and using the double ref freq option).  These modifications reduced the phase noise down to a usable level.  You can see how the phase noise affects the constellation in this image on the BATC Forum: http://batc.org.uk/forum/viewtopic.php?f=15&p=14395#p14395 


What surprised me was that Bell Hill had a reasonable tone even when the phase noise prevented DATV decoding; it did however increase the noise level significantly either side of the frequency.


My (green) ADF5355 board has also surprised me with its frequency accuracy and stability.  Bell Hill always seems to be just 3 KHz high on the dial (both last week at +2 degrees and today at +12 degrees).  So the bottom line is that an ADF5355, once modified, is good enough for 24 GHz.


Dave, G8GKQ


PS Thanks also to Paul G8KFW for his help with equipment and to Brian G4NNS for loaning Noel a 24 GHz setup.


 


Dave G8GKQ
 

Apologies - here's the correct link:
(unless Yahoo mangles it again...)
Dave


Brian Flynn GM8BJF
 

I echo the sentiment that the ADF5355 (or the ADF4351) should be run at high charge pump current levels to keep the close to carrier PN down. I have also found that the highest possible PD frequency helps. Certainly going from 25 to 50 MHz helps noticeably. The improvement going to 100 MHz was not so obvious. The effect of varying other parameters was not discernible.

I have expanded on the Scatterpoint article and and added some more plots which may be of interest.

https://gm8bjf.joomla.com/past-talks/15-reducing-phase-noise-pn-on-chinese-adf5355-boards

I have also put links to the schematics of the "black" PCBs. (Found on various Chinese websites)

73s
Brian Flynn
GM8BJF


Andy G4JNT
 

Was just about to post a query,   then this message appeared.

Ref [3] in the Scatterpoint article , the circuit of the black board, points to a URL that requires a login or sign up.  WHich I'm not prepared to do.

In the article you just refer to bypassing the regulator with 3300uF low ESR.   By that, I assume the 5V regulator, it doesn't specifically say so.

Andy  G4JNT



Virus-free. www.avg.com

On 22 December 2017 at 11:26, Brian Flynn brian.flynn@... [ukmicrowaves] <ukmicrowaves@...> wrote:
 

I echo the sentiment that the ADF5355 (or the ADF4351) should be run at
high charge pump current levels to keep the close to carrier PN down. I
have also found that the highest possible PD frequency helps. Certainly
going from 25 to 50 MHz helps noticeably. The improvement going to 100
MHz was not so obvious. The effect of varying other parameters was not
discernible.

I have expanded on the Scatterpoint article and and added some more
plots which may be of interest.

https://gm8bjf.joomla.com/past-talks/15-reducing-phase-noise-pn-on-chinese-adf5355-boards

I have also put links to the schematics of the "black" PCBs. (Found on
various Chinese websites)

73s
Brian Flynn
GM8BJF



Brian Flynn GM8BJF
 

Hi Andy,

If you look at my website I have updated the URL and put the schematic onto my site. Saves going to dodgey foreign sites! ;-). I have added a couple of addenda to the article and they are on my site.
https://gm8bjf.joomla.com/past-talks/15-reducing-phase-noise-pn-on-chinese-adf5355-boards
They should answer your queris.

I have ened up decoupling all three rails on the board ie the 3,3 analogue, 3.3 digital and the 5V with 1000uF 6.3 V low ESR types.  (old PC MoBos are good sources of suitable parts).

73s
Brian.


Paul G8KFW
 

Hi Brian

 

Thank you for your posting so can I assume we need to use a 50 mhz  reference for lowest phase noise the problem is most of user have 10 mhz off air reference

I assume the  reference must be very clean for the best / lowest phase noise

 

You say “I have also put links to the schematics of the "black" PCBs. (Found on
various Chinese websites)”
is it worth putting the actual document in the file section of the forum

 Regards Paul B


From: ukmicrowaves@... [mailto:ukmicrowaves@...]
Sent: 22 December 2017 11:27
To: ukmicrowaves@...
Subject: [ukmicrowaves] Re: ADF5355 - Good Enough for 24 GHz!

 

I echo the sentiment that the ADF5355 (or the ADF4351) should be run at
high charge pump current levels to keep the close to carrier PN down. I
have also found that the highest possible PD frequency helps. Certainly
going from 25 to 50 MHz helps noticeably. The improvement going to 100
MHz was not so obvious. The effect of varying other parameters was not
discernible.

I have expanded on the Scatterpoint article and and added some more
plots which may be of interest.

https://gm8bjf.joomla.com/past-talks/15-reducing-phase-noise-pn-on-chinese-adf5355-boards

I have also put links to the schematics of the "black" PCBs. (Found on
various Chinese websites)

73s
Brian Flynn
GM8BJF

No virus found in this message.
Checked by AVG - www.avg.com
Version: 2016.0.8013 / Virus Database: 4791/15241 - Release Date: 12/21/17


Neil Smith G4DBN
 

I use a decent 100MHz OCXO in a thermally isolated enclosure as the clock for my 5355s. I phase lock the OCXO to my 10MHz Rb reference using a very very low bandwidth PLL with an HMC1031. That keeps the OCXO just about as clean as when it is free running, although it takes 10-15 minutes to settle completely. I use 47 microfarad ceramic SMD caps in the loop filter.

For /P use, I have a U-Blox LEA-M8F GPS chip and use that to discipline a 10MHz Morion OCXO directly, then use that to lock the 100MHz OCXO as at home. I am trying some tests to see if I can get the LEA-M8F to lock a 100MHz OCXO directly, by dividing the output by five to give a 20MHz comparator frequency.

I was rather alarmed at the noise performance of my ADF5355 boards, so Brian's work on this is very timely and welcome. I was wondering about doing some micrsurgery and installing some 0402 caps near some of the supply pins to see if that helped, but I'll try the low-esr supply bypass caps first. I note that some of the other manufacturers recommend using SMD ferrite beads to bring the supply to the bypass caps as well as ultra-low noise regulators and use of multilayer boards to keep noise off the supply lines.

Neil G4DBN



On 22/12/2017 14:36, 'Paul Bicknell' paul@... [ukmicrowaves] wrote:

Hi Brian

Thank you for your posting so can I assume we need to use a 50 mhz reference for lowest phase noise the problem is most of user have 10 mhz off air reference

I assume the reference must be very clean for the best / lowest phase noise




Christopher Bartram <cbartram@...>
 

Hello Dave,

That's very impressive, particularly for PSK. I looked at the use of QPSK for a data link in the 24GHz area which I was being asked to design some years ago, and came to the conclusion that it wasn't a particularly good system choice. I recommended the use of OFDM on the basis of the inherent instabilities in the transmission path. The client didn't go ahead with the project as he felt that the costs of writing or licencing the software needed to generate and decode ODFM made the project uneconomic.

At 24GHz after a few km under normal conditions, the signal has spreads of arrival times of the order of microseconds, or greater, which effectively phase modulate even a very stable signal. These could easily be degrading the received QPSK, while they would be inaudible as degradation of 'tone' (close-in PN). The constellation shown in your post looks very much like that I'd expect to see.

Generally the single chip synths can have pretty reasonable close-in phase noise, as the loop tends to cancel oscillator noise within its bandwidth. For QPSK you will be looking for good PN performance at offset frequencies comparable to the bit rate of the modulation. I agree with Brian and Andy, using the highest phase comparison frequency, and opening-up the loop bandwidth of the synth. will help, albeit at the risk of increased discrete signal spurs., but the loop gain is finite!

73

Chris GW4DGU


Andy G4JNT
 

Neil -
Be sure your 47uF ceramic caps are not microphonic or exhibit non-linear V/C characteristic

This text was sent to me, recently, and will form part of Feb's RC column

[on a  group posting]  relating to microphonic effects of high value ceramic capacitors ... quotes from various posts:

One weird characteristic of X5R/X7R MLCC capacitors is how capacitance changes with voltage.  When using a modern capacitor that packs the maximum C into the smallest physical size, the capacitance as the DC voltage approaches the rated maximum voltage for the capacitor might be half the rated capacitance.  The rated capacitance is only seen near zero volts DC.

 “… They are piezoelectric and therefore microphonic.   They can sing when used in power supply circuits.   Be careful using them in low level audio or microphone input circuits as when mechanically stressed they can add voltages to your input.   Tap on the board, and hear the tapping in your output.”


Andy  G4JNT




On 22 December 2017 at 16:47, Neil neil@... [ukmicrowaves] <ukmicrowaves@...> wrote:
 

I use a decent 100MHz OCXO in a thermally isolated enclosure as the clock for my 5355s.  I phase lock the OCXO to my 10MHz Rb reference using a very very low bandwidth PLL with an HMC1031. That keeps the OCXO just about as clean as when it is free running, although it takes 10-15 minutes to settle completely. I use 47 microfarad ceramic SMD caps in the loop filter.

For /P use, I have a U-Blox LEA-M8F GPS chip and use that to discipline a 10MHz Morion OCXO directly, then use that to lock the 100MHz OCXO as at home.  I am trying some tests to see if I can get the LEA-M8F to lock a 100MHz OCXO directly, by dividing the output by five to give a 20MHz comparator frequency. 

I was rather alarmed at the noise performance of my ADF5355 boards, so Brian's work on this is very timely and welcome.  I was wondering about doing some micrsurgery and installing some 0402 caps near some of the supply pins to see if that helped, but I'll try the low-esr supply bypass caps first. I note that some of the other manufacturers recommend using SMD ferrite beads to bring the supply to the bypass caps as well as ultra-low noise regulators and use of multilayer boards to keep noise off the supply lines.

Neil G4DBN



On 22/12/2017 14:36, 'Paul Bicknell' paul@... [ukmicrowaves] wrote:
 

Hi Brian

 

Thank you for your posting so can I assume we need to use a 50 mhz  reference for lowest phase noise the problem is most of user have 10 mhz off air reference

I assume the  reference must be very clean for the best / lowest phase noise

 




Virus-free. www.avg.com


Christopher Bartram <cbartram@...>
 

Andy, Neil,

< Be sure your 47uF ceramic caps are not microphonic or exhibit non-linear V/C characteristics

I'd be reasonably relaxed about the C - V characteristics of ceramic capacitors used for decoupling, and have used them extensively, choosing the suitable material mixes for different applications. All ceramic capacitors are not created equal! BUT, I'd be very cautious about using many ceramic caps in PLL filters. It really can be a great way of making a radio microphone!

There is one exception: NPO/COG materials are pretty immune to microphony. All others are are, to some extent, not! Surface mount NPO capacitors of >10nF and reasonable size start to become unobtanium. Good practice is to use film dielectric parts, such as the Panasonic ECHU series, but even they are not entirely trouble-free. Many plastic materials - pfte is an exception - suffer from an effect called dielectric absorption, which is a a memory effect where charge is stored in the dielectric material. That can wreak havoc with the performance of loop filters.

73

Chris

GW4DGU


Neil Smith G4DBN
 

I tried bashing the boards, and sat them on top of a speaker playing loud music and pure tones.  There was definitely a bit of microphony, I wasn't sure what the cause was, but I slathered the board with Araldite.  That seemed to help a lot.  The electronics are inside a foam block which is compressed inside a solid milled temperature-controlled outer box which itself is enclosed in foam, and then glued into a sturdy rack case.  I haven't gone all time-nutty and encased it in a buried concrete block (yet).

I used a tantalum to start with, but it seemed a bit unstable, hence the swapout for a ceramic.  The ones I am using were TDK, nearly four quid each when I bought them, and they are in 1812 packages.  Their V/C slope performance is reasonable, but the main reason I chose them was that the dielectric absorption effects are reasonably low at the 2-3V level, despite all that Barium titanate.  I am guessing that the V/C slope and dielectric absorption would only affect the speed and stability of the loop as it locked-in, and the OCXO needs 15 minutes to stabilise anyway.  Once it is locked and everything has settled down, the V/C and polarisation effects shouldn't really matter, should they?  I should probably have stuck with the tantalums and let it settle down for a week, but I was impatient and (as usual) had no idea what I was doing.

Anyway, it seems to behave impeccably in this application, given the mechanical and thermal isolation.  When I am /P, I just take the rack case out of the radio room rack and put in into the motorhome rack.  Everyone *does* have a 19 inch rack in their motorhome for radio kit, right?

Back on topic, I'll report on what improvement I see in the 5355s with Brian's mods (and if I can fit some 0402 bypass caps, whether that helps at all)

Neil G4DBN


On 22/12/2017 17:55, Christopher Bartram cbartram@... [ukmicrowaves] wrote:
 

Andy, Neil,

< Be sure your 47uF ceramic caps are not microphonic or exhibit
non-linear V/C characteristics

I'd be reasonably relaxed about the C - V characteristics of ceramic
capacitors used for decoupling, and have used them extensively, choosing
the suitable material mixes for different applications. All ceramic
capacitors are not created equal! BUT, I'd be very cautious about using
many ceramic caps in PLL filters. It really can be a great way of making
a radio microphone!

There is one exception: NPO/COG materials are pretty immune to
microphony. All others are are, to some extent, not! Surface mount NPO
capacitors of >10nF and reasonable size start to become unobtanium. Good
practice is to use film dielectric parts, such as the Panasonic ECHU
series, but even they are not entirely trouble-free. Many plastic
materials - pfte is an exception - suffer from an effect called
dielectric absorption, which is a a memory effect where charge is stored
in the dielectric material. That can wreak havoc with the performance of
loop filters.

73

Chris

GW4DGU



Paul G8KFW
 

Hi Neil

 

Thank you for your reply as this is a hot topic and most do not use a 50 or 100 Mhz reference I was wondering if you could Write up the cct and components used so all could benefit from your recherché

As I personally use 198 Khz  for my 10 mhz reference

 

Regards Paul


From: ukmicrowaves@... [mailto:ukmicrowaves@...]
Sent: 22 December 2017 16:47
To: ukmicrowaves@...
Subject: Re: [ukmicrowaves] Re: ADF5355 - Good Enough for 24 GHz!

I use a decent 100MHz OCXO in a thermally isolated enclosure as the clock for my 5355s.  I phase lock the OCXO to my 10MHz Rb reference using a very very low bandwidth PLL with an HMC1031. That keeps the OCXO just about as clean as when it is free running, although it takes 10-15 minutes to settle completely. I use 47 microfarad ceramic SMD caps in the loop filter.

For /P use, I have a U-Blox LEA-M8F GPS chip and use that to discipline a 10MHz Morion OCXO directly, then use that to lock the 100MHz OCXO as at home.  I am trying some tests to see if I can get the LEA-M8F to lock a 100MHz OCXO directly, by dividing the output by five to give a 20MHz comparator frequency. 

I was rather alarmed at the noise performance of my ADF5355 boards, so Brian's work on this is very timely and welcome.  I was wondering about doing some micrsurgery and installing some 0402 caps near some of the supply pins to see if that helped, but I'll try the low-esr supply bypass caps first. I note that some of the other manufacturers recommend using SMD ferrite beads to bring the supply to the bypass caps as well as ultra-low noise regulators and use of multilayer boards to keep noise off the supply lines.

Neil G4DBN

On 22/12/2017 14:36, 'Paul Bicknell' paul@... [ukmicrowaves] wrote:

 

Hi Brian

Thank you for your posting so can I assume we need to use a 50 mhz  reference for lowest phase noise the problem is most of user have 10 mhz off air reference

I assume the  reference must be very clean for the best / lowest phase noise  _._,_.___


Posted by: Neil


Reply via web post

Reply to sender

Reply to group

Start a New Topic

Messages in this topic (7)


Have you tried the highest rated email app?

With 4.5 stars in iTunes, the Yahoo Mail app is the highest rated email app on the market. What are you waiting for? Now you can access all your inboxes (Gmail, Outlook, AOL and more) in one place. Never delete an email again with 1000GB of free cloud storage.


.


No virus found in this message.
Checked by AVG - www.avg.com
Version: 2016.0.8013 / Virus Database: 4791/15244 - Release Date: 12/22/17


Neil Smith G4DBN
 

Hi Paul, happy to share the circuit for the 100MHz disciplined OCXO, I am doing a new version of the PCB (hopefully without the layout errors this time!) but it is simple enough to make using an SMD breakout board for the HMC1031 and use dead-bug construction for the rest. I made a version using discrete high speed SMD logic, but the HMC1031 works just as well, with faster lock-up, and has a lower component count.


X1 in OCXO in, X6 is OCXO out, X2 is 10MHz in, X3 goes to the OCXO. X5 is a lock indication LED, X4 is +12V in .

The 8613 op-amp is only required if your OCXO needs more than 2.4V at nominal frequency (my Valpey-Fisher locks at 1.74V, so I just jumper out the op amp). NM=not mounted.

The Ublox thing hardly needs a circuit. I just bought the LEA-M8F chip on a tiny PCB and wired it up to a PIC and OCXO. If I can get the thing to work OK with the 100MHz OCXO, I'll write that up as well.

The 100MHz OCXOs I use were cheapish from ebay, and they are remarkably good standalone, although not in the same class as a Morion MV89. My best Morion has been running for 11 months standalone, and has only drifted 3ppb relative to my GPSDO since then (currently 2.5Hz LF at 23cm) and most of that was in the first month.

For comparison, the discrete logic version looks like this:

The MC12080D is set to divide by 20. Half of 7474 is used as a /2 and the 7486 is the usual XOR. X1 feeds the OCXO. The LTC1719s are there to square up the 10MHz ref and the ECL output from the 12080.

Board for this version looks like this:

Neil G4DBN


On 23/12/2017 00:15, 'Paul Bicknell' paul@... [ukmicrowaves] wrote:

Hi Neil

Thank you for your reply as this is a hot topic and most do not use a 50 or 100 Mhz reference I was wondering if you could Write up the cct and components used so all could benefit from your recherch

As I personally use 198 Khz for my 10 mhz reference

Regards Paul


From: ukmicrowaves@... [mailto:ukmicrowaves@...]
Sent: 22 December 2017 16:47
To: ukmicrowaves@...
Subject: Re: [ukmicrowaves] Re: ADF5355 - Good Enough for 24 GHz!

I use a decent 100MHz OCXO in a thermally isolated enclosure as the clock for my 5355s. I phase lock the OCXO to my 10MHz Rb reference using a very very low bandwidth PLL with an HMC1031. That keeps the OCXO just about as clean as when it is free running, although it takes 10-15 minutes to settle completely. I use 47 microfarad ceramic SMD caps in the loop filter.

For /P use, I have a U-Blox LEA-M8F GPS chip and use that to discipline a 10MHz Morion OCXO directly, then use that to lock the 100MHz OCXO as at home. I am trying some tests to see if I can get the LEA-M8F to lock a 100MHz OCXO directly, by dividing the output by five to give a 20MHz comparator frequency.

I was rather alarmed at the noise performance of my ADF5355 boards, so Brian's work on this is very timely and welcome. I was wondering about doing some micrsurgery and installing some 0402 caps near some of the supply pins to see if that helped, but I'll try the low-esr supply bypass caps first. I note that some of the other manufacturers recommend using SMD ferrite beads to bring the supply to the bypass caps as well as ultra-low noise regulators and use of multilayer boards to keep noise off the supply lines.

Neil G4DBN

On 22/12/2017 14:36, 'Paul Bicknell' paul@... [ukmicrowaves] wrote:

Hi Brian

Thank you for your posting so can I assume we need to use a 50 mhz reference for lowest phase noise the problem is most of user have 10 mhz off air reference

I assume the reference must be very clean for the best / lowest phase noise _._,_.___


Posted by: Neil


Reply via web post

Reply to sender

Reply to group

Start a New Topic

Messages in this topic (7)


Have you tried the highest rated email app?

With 4.5 stars in iTunes, the Yahoo Mail app is the highest rated email app on the market. What are you waiting for? Now you can access all your inboxes (Gmail, Outlook, AOL and more) in one place. Never delete an email again with 1000GB of free cloud storage.


.


No virus found in this message.
Checked by AVG - www.avg.com
Version: 2016.0.8013 / Virus Database: 4791/15244 - Release Date: 12/22/17



Andy G4JNT
 

At rallies and roundtables I've often seen good quality TCXOs and OCXOs for sale at frequencies other than 10MHz that don't sell - simply because they are not 10MHz!  They are usually significantly cheaper and open to a haggle on price.    Some 13MHz OCXOs I bought were really very good.    Also seen large high quality TCXOs in the several tens of MHz region.    If such units have a voltage control input then they can be locked to a master reference one way or another.

OCXOs at VHF are around too and for most practical purposes, any OCXO is going to be stable and accurate enough for our purposes - there really is no need agt all to go to GPS stability for microwave operation.  the few parts in 10^-8 is more than adequate up to 24GHz even for the frequency accuracy needed for WSJT.

Using Fract-N synthesizers it doesn't matter what the actual value of reference frequency input is.

Andy  G4JNT


Virus-free. www.avg.com

On 23 December 2017 at 00:15, 'Paul Bicknell' paul@... [ukmicrowaves] <ukmicrowaves@...> wrote:
 

Hi Neil

 

Thank you for your reply as this is a hot topic and most do not use a 50 or 100 Mhz reference I was wondering if you could Write up the cct and components used so all could benefit from your recherché

As I personally use 198 Khz  for my 10 mhz reference

 

Regards Paul


From: ukmicrowaves@... [mailto:ukmicrowaves@yahoogroups.com]
Sent: 22 December 2017 16:47
To: ukmicrowaves@...
Subject: Re: [ukmicrowaves] Re: ADF5355 - Good Enough for 24 GHz!

I use a decent 100MHz OCXO in a thermally isolated enclosure as the clock for my 5355s.  I phase lock the OCXO to my 10MHz Rb reference using a very very low bandwidth PLL with an HMC1031. That keeps the OCXO just about as clean as when it is free running, although it takes 10-15 minutes to settle completely. I use 47 microfarad ceramic SMD caps in the loop filter.

For /P use, I have a U-Blox LEA-M8F GPS chip and use that to discipline a 10MHz Morion OCXO directly, then use that to lock the 100MHz OCXO as at home.  I am trying some tests to see if I can get the LEA-M8F to lock a 100MHz OCXO directly, by dividing the output by five to give a 20MHz comparator frequency. 

I was rather alarmed at the noise performance of my ADF5355 boards, so Brian's work on this is very timely and welcome.  I was wondering about doing some micrsurgery and installing some 0402 caps near some of the supply pins to see if that helped, but I'll try the low-esr supply bypass caps first. I note that some of the other manufacturers recommend using SMD ferrite beads to bring the supply to the bypass caps as well as ultra-low noise regulators and use of multilayer boards to keep noise off the supply lines.

Neil G4DBN

On 22/12/2017 14:36, 'Paul Bicknell' paul@... [ukmicrowaves] wrote:

 

Hi Brian

Thank you for your posting so can I assume we need to use a 50 mhz  reference for lowest phase noise the problem is most of user have 10 mhz off air reference

I assume the  reference must be very clean for the best / lowest phase noise  _._,_.___


Posted by: Neil <neil@...>


Reply via web post

Reply to sender

Reply to group

Start a New Topic

Messages in this topic (7)


Have you tried the highest rated email app?

With 4.5 stars in iTunes, the Yahoo Mail app is the highest rated email app on the market. What are you waiting for? Now you can access all your inboxes (Gmail, Outlook, AOL and more) in one place. Never delete an email again with 1000GB of free cloud storage.


.


No virus found in this message.
Checked by AVG - www.avg.com
Version: 2016.0.8013 / Virus Database: 4791/15244 - Release Date: 12/22/17



Andy G4JNT
 


The LMX2541  synth chip has six separate DC supply feeds into the chip, powering different parts of the Fract-N and PLL circuitry.  The data sheet specifies each one being fed via a separate choke with its own bypass capacitor.
A circuit of how that chip is used can be seen at http://www.g4jnt.com/LMX2541_Synth_Module.pdf.

My experience of that chip is that it generally has lower sidebands and spurious (shan't call it phase noise) than any of the AD equivalent products supplied on Chinese PCBs.     Quite how good the AD ones could be if decoupled to the same extent as the LMX device: who knows?

As an aside, my very first breadboard of the LMX2541, on a homebrew PCB, missed out one of those inductors, one feeding part of the logic circuitry.  But instead of the device not working at all, it generated wrong frequencies, and misbehaved in odd ways.  SO much so I thought I'd damaged it having reworked the mounting of just about every other component on the PCB.    It was only after playing with subsequent devices that I went back and revisited that board, found the missing inductor and all worked properly.


On 22 December 2017 at 16:47, Neil neil@... [ukmicrowaves] <ukmicrowaves@...> wrote:
 

 ..., but I'll try the low-esr supply bypass caps first. I note that some of the other manufacturers recommend using SMD ferrite beads to bring the supply to the bypass caps as well as ultra-low noise regulators and use of multilayer boards to keep noise off the supply lines.

Neil G4DBN



On 22/12/2017 14:36, 'Paul Bicknell' paul@... [ukmicrowaves] wrote:
 

Hi Brian

 

Thank you for your posting so can I assume we need to use a 50 mhz  reference for lowest phase noise the problem is most of user have 10 mhz off air reference

I assume the  reference must be very clean for the best / lowest phase noise

 




Virus-free. www.avg.com


Christopher Bartram <cbartram@...>
 

And again its important to appreciate that any old ferrite bead won't do! There are a large number of different internal geometries and materials in use, and reference to data sheets is essential.

Many of the synthesisers have internal LDO regulators, and HF decoupling, but others require multi-octave bypassing and you won't know until you read the literature ... Too much decoupling is (usually!) better than too little!

73

Chris

GW4DGU


--
Christopher Bartram MIET
Christopher Bartram RF Design
Llygad yr Haul
Carno
CAERSWS
Powys SY17 5LE

phone: 01686237732
mobile: 07545094490
email: cbartram@theiet.org
URL: www.christopherbartramrfdesign.com

Note: This message is for the addressee of this email only. It may contain confidential, proprietary,or legally priviledged information. Noconfidentiallity or privilege is waived or lost by any mistransmission. If you receive this message in error, please immediately delete it and all copies of it from your system, destroy any hard copies of it, and notify the sender. You must not, directly, or indirectly, use, disclose, distribute, print or copy anpart of this message if you are not the intended recipient Christopher Bartram RF Design reserves the right to monitor all email communications through its network. Any views expressed in this message are those of the individual sender, except where the message states otherwise and the sender is authorised to state them to be the views of any such entity.


Brian Flynn GM8BJF
 

Hi Paul,
Thanks for the suggestion of putting the files up on this reflector. For some reason that had not occured to me.
They are there now. Not sure how exactly the component values on the board follow the schematic though. the R values I can read in the loop filter on my one black specimen are different. I have also uploaded a HiRes photo of the PCB pinched from one of the Ebay adverts. I have found it useful (printed out on A4 ! ) when working on the PCB. I have not investigated the RF de-coupling of the board. The designer seems to have followed AD's recomendations there. The major departure was the the use of two LT "low noise" regulators as opposed to three of the AD "ultra low noise" parts recomended, one on each rail, AVDD, DVDD and VP5V. Incidentally the recommended part. the ADM7150 is 10 times lower noise than the LT part if the data sheet is to be beleived. I have now got some the will need a proper PCB to use them as the have the dreaded Pin 0 on the base of the package.

73s
Brian GM8BJF


Andy G4JNT
 

The greenPCB at https://groups.yahoo.com/neo/groups/ukmicrowaves/files/ADF5355_Green_PCB.jpg  looks to be remarkably similar, although there are minor differences in screen print positions and some track widths.  At first sight, component references look to be the same.

'jnt


Virus-free. www.avg.com

On 23 December 2017 at 16:20, brian.flynn@... [ukmicrowaves] <ukmicrowaves@...> wrote:
 

Hi Paul,
Thanks for the suggestion of putting the files up on this reflector. For some reason that had not occured to me.
They are there now. Not sure how exactly the component values on the board follow the schematic though. the R values I can read in the loop filter on my one black specimen are different. I have also uploaded a HiRes photo of the PCB pinched from one of the Ebay adverts. I have found it useful (printed out on A4 ! ) when working on the PCB. I have not investigated the RF de-coupling of the board. The designer seems to have followed AD's recomendations there. The major departure was the the use of two LT "low noise" regulators as opposed to three of the AD "ultra low noise" parts recomended, one on each rail, AVDD, DVDD and VP5V. Incidentally the recommended part. the ADM7150 is 10 times lower noise than the LT part if the data sheet is to be beleived. I have now got some the will need a proper PCB to use them as the have the dreaded Pin 0 on the base of the package.

73s
Brian GM8BJF



Paul G8KFW
 

Hi Brian

 

Thank you for putting the files on the reflector   we might have to start a file structure soon on the reflector

Please note I only received my first ADF 5355  4 days ago and haven’t fired it up yet but

Just one point the ADF 5355 I have has 2 inputs on the left hand side and 3 on the right hand side all in line 

so making at least 3 variants  available

regards Paul B


From: ukmicrowaves@... [mailto:ukmicrowaves@...]
Sent: 23 December 2017 17:18
To: ukmicrowaves@...
Subject: Re: [ukmicrowaves] Re: ADF5355 - Good Enough for 24 GHz!

 

 

The greenPCB at https://groups..yahoo.com/neo/groups/ukmicrowaves/files/ADF5355_Green_PCB.jpg  looks to be remarkably similar, although there are minor differences in screen print positions and some track widths.  At first sight, component references look to be the same.

 

'jnt

 

 

Virus-free. www.avg.com

 

On 23 December 2017 at 16:20, brian.flynn@... [ukmicrowaves] <ukmicrowaves@...> wrote:

 

Hi Paul,
Thanks for the suggestion of putting the files up on this reflector. For some reason that had not occured to me.
They are there now. Not sure how exactly the component values on the board follow the schematic though. the R values I can read in the loop filter on my one black specimen are different. I have also uploaded a HiRes photo of the PCB pinched from one of the Ebay adverts. I have found it useful (printed out on A4 ! ) when working on the PCB. I have not investigated the RF de-coupling of the board. The designer seems to have followed AD's recomendations there. The major departure was the the use of two LT "low noise" regulators as opposed to three of the AD "ultra low noise" parts recomended, one on each rail, AVDD, DVDD and VP5V. Incidentally the recommended part. the ADM7150 is 10 times lower noise than the LT part if the data sheet is to be beleived. I have now got some the will need a proper PCB to use them as the have the dreaded Pin 0 on the base of the package.

73s
Brian GM8BJF

 

No virus found in this message.
Checked by AVG - www.avg.com
Version: 2016.0.8013 / Virus Database: 4791/15246 - Release Date: 12/23/17


Brian Flynn GM8BJF
 

Hi Paul,
Regarding the inputs and outputs. You can have a complimenmtary drive for the referemce hence the two SMA sockets . The onboard reference is done that way, . I have used just one to feed a 100 MHz reference to the chip. The data sheet claims there are advatages in PN performance to using a complimentary drive for the reference but I cannot say I noticed any. I used the REF- connector, added a bridge to R8 and disconnected  L1. The fundamental VCO outputs are also complimentary and I  have only ever used one.I mainly used the "OB" output as that is the frequency doubled output.

As the boards arrive you will find the PN performance disapointing.  The main source of the noise is "baseband" noise from the LT voltage regulators. It shows on a spectral plot as large noise shoulders around the carrier. I attach a plot of what mine looked like before adding bypass caps. The noise is within the bandwidth of the PLL and modulates the VCO and so appears on the output. Until you bypass that and get rid of it you cannot see the influence of more subtle noise sources on the PLL In my experience  a "green" Board and a "black" board .both showed similar behaviour.
Hope that helps.

73s, Brian GM8BJF.