Please, advice cheap DDS up to 200MHz

[radiolistener]

Maybe You're right... I'm looking for a cheap solution for receiving on a fixed frequency (or tuning within very small range) on VHF
Can really use two convertions by one si5351 chip?
For example, IF1 80 MHz - filtering - and already low frequency

yes, you can do 3 conversions by using single si5351 chip. It has 3 channels. But it has only two PLL, so if you want to use all 3 channels it requires to share one PLL for 2 channels, as result third channel frequency setting is a little bit limited, because you can configure only divider for it since PLL multiplier needs to be set the same for two channels.

you can buy si5351 module on aliexpress (magenta color) they are very cheap now, but these cheap Chinese modules uses cheap oscillator with high temperature dependency and high initial frequency error. I have also Adafruit version of si5351 module (blue color), it has better TCXO which has less temperature dependency and less initial frequency error.

Also notice, that there is some parasitic crosstalk between channels, so if you want to get the best spurious performance it's better to use separate si5351 for each carrier.

I'm using shortwave transceiver, it uses two channels of single si5351for superheterodyne and it works good enough.

I tested the same transceiver with frequency synthesizer on AD9952 (400 MSPS 14-Bit DDS) and can approve that si5351 version has much more clean output with much less spurs.

[HF8V]

My friend, first you should learn how to properly choose IFs to your radiosystem. 1IF = 80 MHz is nonsense. It is too high and second harmonics lies within maritime band. There are simple rules for that, your maritime band is 156-162 MHz, so is 6 MHz wide, 1IF should be approximately more than twice higher. Typical set is 1IF=10.7 MHz / 2IF=465 kHz, because there are available filters for such freqs.
 

[radiolistener]

My short wave transceiver on si5351 uses IF=45 MHz, works ok

[mag_therm]

Today I ran a calibration check on a DDS.
It is the AD9850 with 125MHz crystal clock, from Nooelec.
I am using this to injection lock the local oscillator of an old sw receiver.
The receiver has a demodulator with an Epson SG8018 oscillator programmed at 1820.000 kHz for a 455kHz Gilbert demodulator

First I ran a calibration on WWV Colorado to get an overall correction of -5.6Hz for the combo of the DDS and the Gilbert.
This cal was entered as fixed frequency offset in the DDS control software for all tests.
The DDS calibration was by the latest release of FLDIGI which now has a Frequency Measurement Test Mode.
A measurement of various other time stations  was run using FLDIGI.

A check frequency measurement was then done in WSJT_X  using the Hermes 14 TXCVR with master clock, GPSDTCXO (Leo Bodnar Mini) .
WSJT_X has an automated test mode which scans a list of frequency standards entered by user.
Result:
Time Signal   Freq MHz   DDS_Error   WSJTX_/GPSDO_Error   (Errors are in Hz, DDS errors were visually from the time trace on FLDIGI )
CHU          3.330           +1.0                  -0.02
CHU          7.850            +0.1                 -0.05
WWV          10.000        -0.05(cal)            -0.04
CHU          14.670          -0.8                 +0.02
WWV          15.000          (-)                -0.04

Edit: Through evening the house heating has started.Radio room ambient increased by ~ 3 C.
 The DDS_Error on CHE 7.850 MHz has increased from  the earlier +0.1 Hz @ ambient, to  +1.8 Hz @ ambient +3 C
Next I will try to get a thermocouple in close to the DDS XO and repeat the test.

[mag_therm]

Addition to Post#53: Next morning, the frequency of CHU 7.850 MHz is drifting between +1.1 and + 1.3 Hz above DDS setting.
So the DDS without temperature control  seems more accurate than most of the ordinary ham radio  transceivers
I see many of them  up to +/- 30Hz on the digi modes.

For the upcoming eclipse in USA I am setting up a second transceiver, it is a HackRF One Software Defined Radio
software is Gqrx + FLDIGI, + other stuff. It is running on an old i686 (32 bit computer) . Using the last Fedora 32 bit version Fc25 of about 2018.
Fedora stopped releasing 32 bit after that.

The transmission during eclipse are basically using ham radio data to infer transient changes in refractive index of ionosphere.
That is why GPSDTCXO master clocks are needed.

The new transceiver will take 10MHz master clock, same as the existing  (Hermes14) transceiver,
The GPSDTCXO is bolted to the basement wall in another unheated basement room.
I have made a constant K 50 Ohm filter/splitter to give two SMA outputs from GPSDTCXO.
Cables 10MHz to transceivers are 8 metre long . Cable to the GPS antenna is about 15 metre.

There is another Leo Bodnar GPSDTCXO with two programmable outputs.(eg 10MHz + 125MHz)
I thought of getting one for the old general coverage receiver's DDS, but after these tests showing the good accuracy
of the DDS,  I don't think it necessary.

[radiolistener]

The DDS calibration was by the latest release of FLDIGI which now has a Frequency Measurement Test Mode.
A measurement of various other time stations  was run using FLDIGI.

FT8 and other digi modes don't require very clean carrier, it just want to get some average frequency over long time. DDS phase errors don't affect it much.