EMRFD Message Archive 13516

Message Date From Subject
13516 2017-02-01 23:47:37 Adrian Scripcă RF signal generator
Hi guys,

Part of the test tools that I miss is a RF signal generator able to output
a clean sinewave from HF up to VHF. I tried finding an used one but got
discouraged by the prices so I wondered how hard would it be to diy one. I
started by laying out the desired specs (which of course are subject to
change with respect to how things evolve):

- able to output frequencies between 0.1MHz - 100MHz
- present a stable output level for a load of 50 Ohm which is flat within
1dB or less across all frequency spectrum
- fairly clean sine wave output with harmonics less than -40dBc
- phase noise better than -120 dBc/Hz at 10KHz offset
- ability to enter the desired frequency from a keypad
- ability to tune the output frequency with an encoder
- have digital readout of the generated frequency

I tried to come up with a basic block diagram using EMRF notations -- I
have no design experience whatsoever. In short, an Arduino controls a Si570
osc to output a signal between 100MHz-200MHz which gets downmixed with an
100MHz xtal oscillator. The output of the mixer is filtered with a LPF
having a cutoff of about 150MHz to get rid of harmonics and then gets into
an AGC amplifier whose job is to keep the output flat within the whole
output range. The AGC level is set so that the MMIC amp after it will raise
the output to the desired value. A step attenuator follows, allowing the
output level to be varied.

Parts used: Si570, ADE-1, AD603 for AGC amp, MAR-1 for beefing up the
output.
You can find the block diagram attached to the email.

My questions are as follows:
1) judging by your experience, can the given design meet the specified
requirements or am I walking a dead-end road?
2) are there any obvious pitfalls that you see and I'm missing due to my
lack of experience?
3) does anybody have a simple example of using the AD603 to hold a steady
level output? or is there a simpler way to obtain that?

Thank you,
Adrian, YO6SSW


[Non-text portions of this message have been removed]
13517 2017-02-02 03:04:52 Clint Re: RF signal generator
Hi Adrian,
 
I have built signal generators using the AD9951 DDS chip and the Si570.  Both work very good down to –120 dbm and compare very favorably with my HP 8648D, except the frequency range is not as wide.  I prefer the Si570 – less expensive and easier to build.  It will be a challenge to get the output flat within 1 db for the frequency range that you specified, but the AGC amp can make that happen.  I can send you the construction details (with pictures) on my units.  The Si570 generator uses the kit from SDR-Kits.
 
 
 
73
Clint
W7KEC
 
13518 2017-02-02 03:10:23 Thomas S. Knutsen Re: RF signal generator
Hello Adrian, The block diagram did not come through, can you upload it to the photo section, or to a picture sharing site?

From your text description, I think I can work out what you want to do. It's a approach to a signal generator I tried around 2010 and there are some things you should consider:

1. Building a good quality high preformance signal generator is not easy. The requirements for screening and isolation between the stages is almost as high as when building a spectrum analyzer. 

2. Using a 150MHz LPF will allow your 100MHz product to pass through, the mixer is not ideal, so some of the input products will pass through. For the sketched outline, I would suggest a 70MHz filter to reduce those products.

3. IMHO, the AD603 is overkill in a signal generator, The deviation in the output signal from the generator should not be great, and it should be fairly easy to use a diode detector and a PIN attenuator to control the output level, a op-amp or couple transistor should do the needed regulation. I suspect this will be a point where you need to put in some work if you want the deviation in output power less than 0.1dB.

4. Level 7 mixers may require you to work with levels at -30dBm or so to avoid IMD products by the mixer. This will require a lot of amplification, so you may consider trying higher level mixers. 

Your approach should work well. I would suggest taking a look at the HP 8601A sweep generator manual, they used a approach like this, but by mixing a 200MHz crystal oscillator (made by doubling 100MHz) with a 200MHz-310MHz VCO. 

Building a signal generator like this may not come down cheaper than buying one, but the value in the learning can't be bought.  I'd suggest go for it, and let us know how it works out!


73 de Thomas LA3PNA.


2017-02-02 8:47 GMT+01:00 Adrian Scripcă benishor@gmail.com [emrfd] <emrfd@yahoogroups.com>:
 

Hi guys,

Part of the test tools that I miss is a RF signal generator able to output
a clean sinewave from HF up to VHF. I tried finding an used one but got
discouraged by the prices so I wondered how hard would it be to diy one. I
started by laying out the desired specs (which of course are subject to
change with respect to how things evolve):

- able to output frequencies between 0.1MHz - 100MHz
- present a stable output level for a load of 50 Ohm which is flat within
1dB or less across all frequency spectrum
- fairly clean sine wave output with harmonics less than -40dBc
- phase noise better than -120 dBc/Hz at 10KHz offset
- ability to enter the desired frequency from a keypad
- ability to tune the output frequency with an encoder
- have digital readout of the generated frequency

I tried to come up with a basic block diagram using EMRF notations -- I
have no design experience whatsoever. In short, an Arduino controls a Si570
osc to output a signal between 100MHz-200MHz which gets downmixed with an
100MHz xtal oscillator. The output of the mixer is filtered with a LPF
having a cutoff of about 150MHz to get rid of harmonics and then gets into
an AGC amplifier whose job is to keep the output flat within the whole
output range. The AGC level is set so that the MMIC amp after it will raise
the output to the desired value. A step attenuator follows, allowing the
output level to be varied.

Parts used: Si570, ADE-1, AD603 for AGC amp, MAR-1 for beefing up the
output.
You can find the block diagram attached to the email.

My questions are as follows:
1) judging by your experience, can the given design meet the specified
requirements or am I walking a dead-end road?
2) are there any obvious pitfalls that you see and I'm missing due to my
lack of experience?
3) does anybody have a simple example of using the AD603 to hold a steady
level output? or is there a simpler way to obtain that?

Thank you,
Adrian, YO6SSW

[Non-text portions of this message have been removed]

13520 2017-02-02 03:45:41 Adrian Scripcă Re: RF signal generator
Clint, I acquired a Si570 board from WB6DHW and bought the Si570 chip from sdr-kits.net. I also managed to build the oscillator and write some basic control software. I don't intend to use the Si570 as-is due to the output being a square wave.
I would be grateful if you could share construction details to your units.

Thomas, thanks for the notice. I uploaded the image in the photo section (YO6SSW folder) and also made it avaiable at http://hq.scene.ro/rf-signal-generator.png . Let me know if there are any problems accessing it.

Thanks for your insights, they are very informative! 

1. I understand the screening and isolation requirements are very high and am ok with meeting them, especially since I have also screened my SA modules.

2. I guess I can live with a 70MHz range since most of my hobby work is in HF. I figured I could make use of 100MHz+ for characterizing my W7ZOI SA VHF filter for example, but maybe I can get away with simply low pass filtering the Si570 output there.

3. Using pin diodes in a feedback loop sounds even better. Do you happen to have any maetrial on this? As far as for the 0.1dB deviation, I do not require such a strict flatness, especially since I'm a hobbyist, so simpler is better in this case.

4. Good point. ADE-1 was something that I had already bought but in this case I can try to hunt for a level 13 or 17 mixer.

As for sweeping, I excluded that from the desired feature list since I thought it would complicate the design. I thought about using a POS200 and mixing it with a 100MHz reference (much like in the W7ZOI SA) but that would have made the ditial tuning and frequency stability hard to attain.

Thank you for all the ideas and keep them coming, if you have some more :) Maybe there are others interested in the topic.

73!
Adrian, YO6SSW


13521 2017-02-02 03:59:32 Tony Fishpool Re: RF signal generator

Yes I’m interested Adrian. I’m watching this tread hoping it turns into a project that we can all join in on.

I have a quite old now Altair signal generator. It does most of what I need well beyond 100MHz but the output level is far from constant – even within a range. And of course, the frequency adjusting dial isn’t massively accurate.

 

I also have a spare Si570 left over from the sweeperino project waiting to do some good.

 

Kind regards

Tony G4WIF

>Thank you for all the ideas and keep them coming, if you have some

>
more :) Maybe there are others interested in the topic.

 

73!

Adrian, YO6SSW

13522 2017-02-02 16:50:50 arfghans Re: RF signal generator
Another option to keep in mind as a starting point is an SDR (and you get a receiver as part of the package!). For instance, a cheap and very hack-able SoftRock RXTX might get you most of what you want. The software is free and/or open-source. Some guys are even using their SDRs as vector network analyzers, with excellent results. And of course they make fabulous spectrum analyzers. Also, most VNAs make pretty good signal generators... 

I think the RF instrumentation world is rapidly converging.

Gary NA6O
13523 2017-02-02 22:20:25 Hans Summers Re: RF signal generator
Hi all

I think the experimental suggestions are great and should do fine. For people who don't want to design something themselves but do still want a hand in building it, I want to bring your attention to the QRP Labs Si5351A VFO/SigGen kit for $33 including synth and rotary encoder, see http://qrp-labs.com/vfo ... This has output frequency range 3.5kHz to almost 300MHz, rotary encoder tuning, 16 x 2 LCD, and optional GPS discipline. 

The output is a squarewave. However the next firmware version due in the next week or two (you can have a free firmware upgrade later on request, if you buy it now), supports relay-switched Low Pass Filter selection, using QRP Labs' 6-band relay switch http://qrp-labs.com/ultimatelpf ($16), and six QRP Labs Low Pass Filter kits http://qrp-labs.com/lpfkit ($4.60/ea). LPFs are available for 2200m to 6m bands, and there are designs available for 4m and 2m also. Strategic selection of LPFs and configuration of threshold frequencies for each to be used, should allow a good wide-band sinewave generation. 

QRP Labs also has a GPS Receiver kit $23 http://qrp-labs.com/qlg1 which can optionally be used for GPS discipline (or any other GPS module or other source of with 1 pulse-per-second signal).

There is also an extruded aluminium enclosure available $22 see http://qrp-labs.com/vfobox.html which includes cut/drilled/printed front and rear panels, two toggle switches, two push-buttons, D9 socket, power plug and power socket, BNC socket, 4 self-adhesive feet, and all screws and other mounting hardware. 

So a combination of these kits would also inexpensively get you what you are looking for. 

73 Hans G0UPL


13525 2017-02-03 15:42:31 Dave Nushardt Re: RF signal generator
Adrian just reading your RFGen wishlist, why do you need such a flat output,? Is that a wish or based on a real need ? The reason I ask is what you are asking for may not exist, output levels always varries with frequency, thats just the nature of the animal.

What your asking for is a contradiction in terms.

Only a DDS based device would have the frequency range you desire and none have a flat output, natively, you would have to design a leveling circuit and RF amplifier

Most DDS have low level output.

That said you can make one leveled , as the Tektronix Type 191  constant amplitude RFSignal generator and the more modern replacement ,the Tektronix SG 503,4  the 503 is not phase locked ,the 504 maybe phase locked but doesn't meet your lower frequency specifications.

I have both , they are used to calibrate oscilloscopes to adjust and calibrate the frequency bandwidth networks of the scopes , this is the only application I can think of for such  stringent  spec's  you mentioned.

I know of no hobby type signal generator that even comes close to your wish list , you would have to specifically design such  genderator  and it most likely won't be affordable.

All I can suggest is you study the Type 191 ,and SG 504  manuals, look in the  Theroy of operation and schematics for how the leveling circuits work and RF amplifiers.

Type 191 can be had on ebay but are rare you have to watch for them, thay can be had for under 100 dollars. Prices are going up as people discover their usefullness.

I would have to say that the 191 comes the closest for the money, you will find.

Good luck!

Hope this helps just gogle the Tektronix manuals in pdf they are free.

73

Dave

Sent from TypeApp



13526 2017-02-03 20:29:06 bkuo Re: RF signal generator
Hans,

One thing to mention with the Si5351, which is a great device, is you really DO have to put an LPF on that (don't even think about putting that on an antenna).

I'm building a transceiver right now (CW, QRP 5 watts, for portable use mountain topping -- using the Si5351) and that square wave is definitely a challenge as you switch bands -- you need an LPF per band, for sure. Just in my garage/lab, if I set the signal to 7.100, I can tune a general coverage receiver and hear myself +20 on 7.100, 14.200, 21.300, and 28.400... You could have "all band" signal pollution if you wanted to use the Si5351 for an antenna analyzer, for example, or anything that radiates.

Anyway, the hazards of that square wave, as I'm sure you are aware of!

Ben
KK6FUT
13527 2017-02-03 20:40:00 Ashhar Farhan Re: RF signal generator
adrian,

i guess you need to control the output  to be able to properly plot the tracks. if the tracker is software controlled, you can use do a reference scan, and plot the DUT's output by calculating the relative loss/gain.

the bigger trouble with the hetrodyning approach is that there are a large number of birdies. The clock generators like Si570, on the other hand have a significant harmonic content that will simply not go away for a properly scan. the only way out, really is to use a detector that is tuned to the frequency of interest as well. unfortunately, that means a spectrum analyzer with a tracking generator or a direct conversion VNA.

- f



13528 2017-02-05 01:25:37 Hans Summers Re: RF signal generator

Hi Ben

One thing to mention with the Si5351, which is a great device, is you really DO have to put an LPF on that (don't even think about putting that on an antenna).

Yes agreed! But you almost always need an LPF at the output of a transmitter anyway. But I agree it needs to be a better LPF if squarewaves are involved.

In the Ultimate3S transmitter kit http://qrp-labs.com/ultimate/u3s the Si5351A drives a BS170 MOSFET PA. The plug-in 7-element LPF kit (available for 13 bands 2200m to 6m inclusive) http://qrp-labs.com/lpfkit enables the TX to comfortably meet FCC requirements (all spurious outputs under -43dBc).

73 Hans G0UPL 


13529 2017-02-05 20:11:15 AncelB Re: RF signal generator
While I realize the frugality of the HAM space....
I have always found that having decent gear is worth the cost.
I have 2 of these Ebay # 272523383791

They a pure sine wave units - instrument calibration grade.
13535 2017-02-06 06:30:58 Adrian Scripcă Re: RF signal generator
Hi all,

Thank you all replying and I am for sorry for the late reply; my wife is due to deliver a baby on Wednesday and I am busy with the last minute preparations. I did however find some hours during which I built a rf detector, hooked it up to the Arduino which was running a modified version of the sweeperino software and then wrote some software to command the whole thing. Here's some sweeps with the thing (mind you, the absolute level is not calibrated since I have no means for a proper calibration, but we're somewhere in the ballpark):

1. http://hq.scene.ro/sweeper/sweeper-no-input.png - Here we have a sweep from 10MHz to 200MHz with no input attached to the detector. This proves that the detector works in the expected params 

2. http://hq.scene.ro/sweeper/sweeper-si570-output.png - This is a sweep with the Si570 board output directly connected to the detector input (no pad). It shows the reference level and we can see here that the output level varies with freqency and the total level deviation is somewhere around 16dB

3. http://hq.scene.ro/sweeper/sweeper-70MHz-lpf.png - A sweep with the Si570 ouput coupled to the 70MHz LPF from W7ZOI's SA project. To tell the truth, this left me a bit perplexed; I wasn't expecting that response after 100MHz. Is that normal? Has anyone sweeped their 70MHz filter and can share the sweep?

4. http://hq.scene.ro/sweeper/sweeper-vhf-filter.png - A sweep of the 110MHz VHF filter from the same W7ZOI SA project. While the 3dB bandwidth seems in line with the one presented in the article, the insertion loss is way too high (somewhere around 12dB). I'm still scratching my head over this one as to what I'm doing wrong with the filter.

5. http://hq.scene.ro/sweeper/sweeper-vhf-filter-wide-sweep.png - And finally, a sweep of the same VHF filter but this time on a wider spectrum, from 10MHz up to 200MHz. We can see the false responses from the signals whose harmonics make it in the filter's pass band. This is a good enough reason for me wanting a sine wave output generator and not a square wave one. 

I understand that my requirements are kind of hard to meet but am I too naive to think that enough filtering should fix it? For instance, I would expect that two square wave signals, when mixed, to generate a lot of spurs since their harmonic content is quite rich whereas that should not happen the same if the signals were pure sinusoids. In this case I thought of a compromise, having the square wave switch the DBDM mixer and feed it a clean 100MHz sine for downmix. I could always filter out the harmonics of the Si570 output, given its oscillation range (100MHz-200MHz) before entering the mixer, but will that help? if so, how much? I cannot really say since I do not have the experience.

Another thing I thought of was to simply use Si570 output heavily filtered to get rid of harmonics; the problem with this approach is that Si570 only goes down to 10MHz so in order to get a lower bound I would have to divide its output. The type of Si570 I got goes up to only 280Mhz (they were cheaper) so by division I would get the following ranges:

a) divide by 2: 5MHz-140Mhz
b) divide by 4: 2.5Mhz-70MHz
c) divide by 8: 1.25Mhz-34Mhz

I am not satisfied by any of the options given the initial requirements (frequency range), so that's why I chose the heterodyne approach. Now, some said that since the mixer is not perfect *and* the signals are not pure, I will get a lot of spurs. It would be interesting to make a calculation of how many and how large. Does anybody know a software that can calculate this?

Gary, I bet there are solutions out there but since I started making my own, I would just like to see where this takes me. The goal is to have a solid generator that can output a desired frequency at a specified stable level.

Hans, thanks for the LPF board idea. This could also work, having the software switch in different filters with respect to the generated frequency. 

Dave, I need a flat output out of mere convenience, so that I can perform reliable measurements and calculations without re-measuring things countless times. If I didn't have a flat output, I would constantly need to measure the output level or have a calibration chart before using the output signal as input in the measurement process of a device. Pin diodes could help here but I'm afraid they would distort the signal more than a good agc amp would.

Farhan, thanks for sweeperino and the idea of reference scan  I will implement that in my software. As for the birdies, perhaps they will be reduced if I filter the Si570 output before entering the mixer?

73! de Adrian, YO6SSW


13537 2017-02-06 09:14:08 John Marshall Re: RF signal generator
Adrian,

I offer a couple of comments on your very nice bit of testing and measurement:

Your broadband RF detector responds to the Si570's fundamental AND all its harmonics. This means it will give a result about 4 dB higher than it would for a signal with no harmonics, and this can lead to confusing results. A friend of mine ran into the same thing not long ago < http://zs6rsh.blogspot.com/2016/09/revisit-ad8307-power-meter-calibration.html >.

For example, your 110 MHz bandpass filter shows an output of -12.9 dBm. Compared to your broadband sweep this looks like an insertion loss of 12 dB, but the broadband sweep contains harmonics while the bandpass filter itself removes them. So, the filter's insertion loss is really 8 dB, not 12 dB. The notes I have from Wes say it should be around 5 dB, so it's still not perfect. The -3 dB points look OK but the -6 dB width is much greater. I think the coupling needs work, and perhaps also the shielding between sections.

Your 70 MHz lowpass filter shows a good drop above cutoff but some increase at higher frequencies. I don't know about this design specifically, but inductance in the capacitor leads and stray coupling between filter sections will cause this. There will also be effects from the Si570's harmonics.

Your proposal of a 100 MHz sine wave mixed in a DBM with a Si570 "high-side" LO sounds reasonable and worthy of some experimentation. You will have to keep the sine wave input at a low level, terminate the mixer output properly, and use a good lowpass filter. I think the biggest problem won't be from mixing spurs. The mixer/LPF output will be at a very low level and will need a lot of amplification, and that can generate new harmonics.

An alternative, similar to what Hans suggested, is to use the Si570 directly and filter its output with a bank of lowpass filters, automatically selected by your controller as the Si570 is tuned. You will need to be creative to get down to 100 kHz, perhaps by switching in digital dividers, and you will need a lot of filters! Filters will probably have to span less than an octave, since the Si570 square wave won't meet your -40 dBc requirement at 2F, at least at higher frequencies.

Good luck with this project and with the new baby!

John, KU4AF
Pittsboro, NC

13538 2017-02-06 09:35:37 Ashhar Farhan Re: RF signal generator
th simpler route is to simply use a doubly tuned banpass filter ahead of the detector and manually peak it.
the better alternative is to use an spectrum analyser
- f

13539 2017-02-06 09:43:05 John Marshall Re: RF signal generator
Yes, of course. But a doubly tuned filter that covers 0.1 to 100 MHz presents its own challenges!

John, KU4AF
Pittsboro, NC

13540 2017-02-06 09:57:08 Ashhar Farhan Re: RF signal generator
hehe, true. but you could make two or three to cover the hf spectrum.
- 

13551 2017-02-06 21:19:07 Hasan Murtaza Re: RF signal generator
> One thing to mention with the Si5351, which is a great device, is you
really DO have to put an LPF on that (don't even think about putting that
on an antenna).

Has anyone used a AD9850 DDS (which generates a sine wave output) in this way? Is the harmonic content of the signal low enough to broadcast with, or does it need a LPF to satisfy FCC rules on OOB emissions?

Hasan
13552 2017-02-06 21:30:03 Clint Re: RF signal generator
I have used an AD9951 which has less noise than the 9850.  Low pass filters are needed in the output to get rid of the oscillator harmonics (fundamental is around 100 Mhz).  The output is a very clean sine wave.
 
73
Clint
W7KEC
 
13553 2017-02-07 05:03:16 Clint Re: RF signal generator
I have spectrum analyzer sweeps of the AD9951 showing the fundamental and harmonic content, if anyone is interested.  I used the 9951 in a VFO project.
 
73
Clint
W7KEC
 
13554 2017-02-07 05:39:14 wb8yyy_curt Re: RF signal generator
Hasan

if you transmit with no further amplification, maybe.  if you have an amplifier after the DDS -- it will likely make high level harmonics. 

in addition to the harmonics - a DDS will produce other spurious signals - so it is being a good citizen of the spectrum to add some filtering. 

there are some QRP rigs that transmit from a DDS without frequency conversion -- the PFR3, KX1, DSW-series come to mind.  examining their schematics will provide ideas on good practice. 

also not too long ago, a QST article describes integrating a 9850 or similar DDS with a '49'er' transceiver. 

enjoy your frequency agility responsibly with your home assembled rig. 

73 CUL Curt
13555 2017-02-07 10:27:43 kb1gmx Re: RF signal generator
No.  At best the DDS is clean but maybe 40db range, however that does not include SFDR.

SFDR is spur free dynamic range and the 9850 even with lop pass can can in band or 
just out of band spurs what will move around with tuning.

The 9851 is better (more bits in the D/A) but nor perfect.


Allison
13557 2017-02-07 11:00:24 kb1gmx Re: RF signal generator
Some things:

Very few measurement require gain flatness under 1db, those that do also require measuring 
everything including cable and connector losses.  I use commercial generators a lot and if you
want lab standard you need to at least purchase lab standard measurement (power) capability.
Even a cheap Rigol DS815T is very effective here.

Often frequency generati
13558 2017-02-07 12:11:21 Todd F. Carney / ... Re: RF signal generator
Clint,

Can you reply with a few basic figures, i.e., the relative power of the 2nd and 3rd harmonics in particular, with the fundamental as 0.0dB?

Thanks and
​ ​
73,

Todd
------------------------------------------------------------------------------------
​
K7TFC / Portland, Oregon, USA /
​CN85pl
 / UTC-
​8
 (PST)
------------------------------------------------------------------------------------

13559 2017-02-07 12:26:01 Clint Re: RF signal generator
Sure – let me look through my pictures.  Please keep in mind that the basic AD9951 signal generator is not my design – it comes from:
 
 
I just packaged the basic board in a box, modified the output amp, and added one outboard low pass filter.
 
73
Clint
W7KEC
 
13560 2017-02-07 12:53:22 Tayloe, Dan (Noki... Re: RF signal generator

It seems like such a square wave output generator could use a bank of filters like the HP8640, which had a sharp cut off filter for every “power of 2” frequency band such as 2 to 4 MHz, 4 to 8 MHz, 8 to 16 MHz, 16 to 32 MHz, etc.  For HF, that is not *that* many filters. 

 

The HP8640 used digital logic to divide down starting from something close to 512/1024 MHz, so the output started as square waves for sure. 

 

Wasn’t there plans for a 8640b Jr on this list?  That probably shows what these filters should look like.

 

-          Dan

 

13561 2017-02-07 17:18:19 Clint Re: RF signal generator
I have started adding some pictures under the Photo section.  The first picture is the standard output from the WA1FFL VFO with no low pass filters in the output.  The second picture is the standard VFO with a low pass filter added in the output.
 
73
Clint
W7KEC
 
13564 2017-02-08 03:43:13 Hans Summers Re: RF signal generator
Hi all

I was interested to see how the performance of the QRP Labs VFO/SigGen kit
http://qrp-labs.com/vfo would be, using six relay-switched Low Pass
Filters. This uses the QRP Labs relay kit http://qrp-labs.com/ultimatelpf
and six Low Pass Filter kits http://qrp-labs.com/lpfkit .

Results were quite pleasing. Over the frequency range 2.9 to 34MHz,
sinewave output into 50 ohms is 10.9dBm average, +/- 0.4dBm. Quite flat.
The worst case spurious output is -45dBc.

I can't log in anymore to YahooGroups web interface and I'm not sure if
attachments get through. So I put the measurements here:
http://qrp-labs.com/images/vfo/filtered/Output_2.9MHz_to_34MHz.gif
and a wider measurement from 1MHz to 45MHz:
http://qrp-labs.com/images/vfo/filtered/Output_1MHz_to_45MHz.gif
(below 2.9MHz the harmonic content gets worse than -45dBc because the
lowest filter is the 60m one. Above 34MHz the output amplitude starts to
drop off due to the 10m LPF which is the highest filter used here).

The Low pass filter used were: 60m, 40m, 30m, 20m, 15m and 10m. The latest
beta firmware (available soon) of the VFO kit allows relay selection of an
LPF, based on threshold frequencies. I set the thresholds to 5, 7, 10, 14
and 21MHz. The filters were built WITHOUT measurement (e.g. with LC meter
to check toroid inductance), just counting turns. There is no shielding.
The VFO is open on the bench. RF output to spectrum analyser (50 ohm
input). The results are already good but could surely be further improved
with optimisation and more care.

This system has rotary encoder tuning, digital readout and 16 configurable
preset frequencies. The frequency resolution is 1Hz. Optionally one can add
GPS discipline (http://qrp-labs.com/qlg1) and an aluminium
box/switches/connectors kit (http://qrp-labs.com/vfobox). The basic set-up
described here would cost $33 (VFO kit) + $16 (relay kit) + 6x $4.60 (LPF
kits) = $76.6 total.

73 Hans G0UPL
http://qrp-labs.com


[Non-text portions of this message have been removed]
13565 2017-02-08 14:39:14 billdunsmore Re: RF signal generator

Hans, is that $76.60 price all inclusive or do you need an external microprocessor and software? I am designing a 40/20 cw rig and planned to use an analog colpitts vfo, but your post is interesting!



Sent from my Verizon 4G LTE Smartphone
13568 2017-02-08 21:37:42 Hasan Murtaza Re: RF signal generator
I'd love to see them...I don't have a spectrum analyzer right now and want to get a sense of what the harmonics look like.

Hasan
13569 2017-02-08 22:17:40 Hans Summers Re: RF signal generator

Hi Bill

> Hans, is that $76.60 price all inclusive or do you need 
> an external microprocessor and software? I am designing 
> a 40/20 cw rig and planned to use an analog colpitts vfo, 
> but your post is interesting!

Yes, all inclusive, including the programmed microprocessor. That is the sum of $33 for the VFO kit (includes Si5351A Synth, LCD module, rotary encoder, programmed micro); $16 relay-switched LPF kit, and 6x $4.60 for six plug-in low pass filter kits. Result is the flat, HF signal generator with sinewave output. 

If you are only interested in two bands, then you may not need all that. Depends what you want to design separately and what you want a kit for. You are going to send the output of the VFO to a power amplifier - and the LPFs should come after that, not before it. 

Section 4 of QRP Labs App Note AN005 see http://qrp-labs.com/appnotes details how to construct a 5W CW Transmitter with raised cosine envelope shaping, using the new 5W HF PA kit with RC1 key-shaper chip, http://qrp-labs.com/pa . This is an IRF510-based PA with a discrete component power modulator with foldback current limiting. The modulator is controlled by an 8-bit DAC, which can be controlled by the plug-in RC1 chip to make quite accurate raised cosine key-shaping, for click-free CW. 

The QRP Labs shop items required for this TX would be:

VFO/SigGen kit $33 http://qrp-labs.com/vfo
5W HF PA kit $20 http://qrp-labs.com/pa + $4 RC1 chip option

Total $63. It's a nice set-up, an Si5351A synthesised VFO with 5W PA and click-free raised cosine key-shaping. 

73 Hans G0UPL

13571 2017-02-09 08:16:40 kb1gmx Re: RF signal generator
Dan,

Looked for the 8640jr in the files section and it seems to have disappeared.

A vfo running at say 39-81mhz  and then divided by 2
gets a good start for isolation and stability for the HF range.
it helps if the tuning range is greater than 2:1.
For the 3 to 30 range the number of filters are not too bad.

Working backward from the VFO:

/2 20-40mhz   
/2 10-20mhz
/2 5-10mhz
/2 2.5-5mhz
==== to this point the number of filters are small for the covered range
          and the total of /16 is easily done with one CMOS 74xxx161 or similar.
/2 1.25-2.5Mhz
/2 .625-1.25
/2 .3125-.625mhz

Things that makes this easier is that a symetrical square wave has weaker 2nd 
harmonic than third so a 7 or 9 pole filter will be good.

This can be done with Si570 but them a micro and the NCO are cost items
but the dividers are not needed (they are cheap).  The difference is the divided VFO
will be a very low phase noise  starting source and it improves with every /2 this is 
why the 8640 is notable from the start.

In the end it depends on the intended use.  Many times I've made a signal generator
for a purpose often crude dead bug and it served well.

Allison



13572 2017-02-09 08:51:28 John Marshall Re: RF signal generator
I think it was in the QRP-Tech group. W7ZOI's paper on the 8640-Jr is available at:
< http://cwtd.org/AnalyzeThis/8640jr.pdf >

John, KU4AF
Pittsboro, NC


13577 2017-02-09 16:06:19 Ken Chase Re: RF signal generator
13579 2017-02-10 09:41:22 kb1gmx Re: RF signal generator
HI,

 I knew it was outside yahoo. We used to have it here in the EMRFD files area.

Allison
13581 2017-02-10 23:19:50 Hans Summers Re: RF signal generator


Hi Allison

 Yes 2/3 octave filters would work better but more switching.  Plus I was talking about continuous coverage from say 1.8 to 30 as signal gen rather than ham bands only.

I was able to get worst case harmonics of -45dBc across the wide continous range 2.9MHz to 34MHz. Most frequencies were a lot better than that. Very flat: 10.8 +/- 0.4dBm. That was with Si5351A squarewave VFO/SigGen http://qrp-labs.com/vfo and six filters 60, 40, 30, 20, 15 and 10m. See measurements:
Wide:

73 Hans G0UPL 

13635 2017-02-27 08:48:19 czelusniakd Re: RF signal generator
Hans,

I was interested in the Utimate3S kit.
I am curious however about the Si5351A Synthesizer module.
I prefer not to mess with crystals as I have accurate time bases
to use as REF.
Is there some reason you do not use an Si5351C or have it available?

Thank you for your time,

Dennis A. Czelusniak

I tried your QRP Labs contact but the sales@qrp=labs.com comes up as invalid.
13636 2017-02-27 20:06:08 Hans Summers Re: RF signal generator
 
Hi Dennis
 
I was interested in the Utimate3S kit.
I am curious however about the Si5351A Synthesizer module.
I prefer not to mess with crystals as I have accurate time bases
to use as REF.
Is there some reason you do not use an Si5351C or have it available?

The first reason is just cost. The Si5351C costs more than twice the Si5351A. For reference, the different Si5351 chip versions A, B, C are as follows:

Si5351A: 10-pin chip, 3 outputs, crystal osc 
Si5351A: 20-pin chip, 8-output version, crystal osc
Si5351B: 20-pin chip, 8 outputs, crystal osc with variable voltage crystal oscillator tuning
Si5351C: 20-pin chip, 8 outputs, crystal osc or CLKIN

The Si5351A is the simplest, lowest cost chip of the family. That's why most people are using it. In fact the 8-output variants still only have two internal PLL systems so the large number of outputs is not necessarily massively useful. Most people don't need that many outputs anyway. In a high performance application you maybe wouldn't like the crosstalk, or the jitter (from necessarily abandoning even integer division mode because of so many outputs using two PLLs.

BUT... it sounds as though your only question is about using an external timebase as reference rather than a crystal. The Si5351A supports chip this! The datasheet http://www.silabs.com/documents/public/data-sheets/Si5351-B.pdf shows this in section 5.6 (page 21). The datasheet shows a 1V p-p squarewave being fed into the chip pin 2. Practically speaking I have experimentally confirmed that being a squarewave is not a requirement. And signal amplitudes in the range 0.5V p-p to 3.3V p-p all work fine. 

The difference between the Si5351A and Si5351A is that the Si5351C accepts a clock input in the range 10-100MHz. In the case of the Si5351A it should be in the range 25-27MHz, the same as the crystal. Having said that, people have tried other Si5351A clock inputs with varying degrees of success - it does seem to affect the output signal quality for example if the input reference clock is 30MHz or more. 

There was a batch of QRP Labs Si5351A Synth module kits http://qrp-labs.com/synth where the crystal component supplier made a mistake and supplied a 3rd overtone 27MHz crystal instead of the correct type with 27MHz fundamental mode! So they oscillated in the module at something like 9MHz. The Si5351A still worked, it just gave outputs at 1/3 the desired frequency! If this was taken into account in the calculations for the configuration registers, you could even get the proper output frequencies (though the maximum attainable output frequency I suppose, would not be the datasheet 200MHz value anymore). Around 80 of these kits were shipped out before the problem was realised, and I had to get the new batch of crystals quick, replace the ones in the kits, and ship out 80 replacement crystals to those kits already shipped out! Luckily that was just one batch of 500 kits long ago... now we're into the 9th batch, the crystals are checked *very* carefully every time, and all is well ;-)  Anyway I'm just saying - your mileage may vary - but the 25-27MHz restriction may not be that strict.

The current QRP Labs Si5351A Synth module kit http://qrp-labs.com/synth board revision 5 has provision on the board for:

a) 27MHz crystal (supplied)
b) TCXO PCB pad footprints (TCXO not supplied)
c) Jumper wire configuration to connect chip pin 2 to module pin 2, so that you can apply an external reference clock at module pin 2

73 Hans G0UPL