I've seen the warnings on the DigiLite website about excessive oscillator injection causing various problems and suggestions that the Ultram oscillator's MMIC should be removed so I was curious to see if my set up was OK or if it need optimising.
I ran a test on my DigiLite 23cm version that is fitted with the onboard 12dB attenuator (R28 etc) to find out whether there is an optimum oscillator drive level. I've fitted SMA edge launch PCB connectors to the input and output of my v5.6 board and used them to connect to a HP sig gen and spec analyser.
With the onboard 12dB attenuator I can drive the input between -6dBm to +17dBm (the top limit of the sig gen) at 1249MHz with no difference to the output level at all. It stays put at -6dBm.
Below -6dBm input the output starts to drop, but with decreasing drive the noise floor starts to rise and by the time the input is down to -15dBm the noise floor has risen by over 15dB. That behaviour is mentioned in the datasheet for the AD8346 modulator.
I also noticed that there are some sprogs about 100 and 200 MHz out that are minimised when the drive level is 0dB or above. With weak drive they can get to -50dBc.
The bottom line is that if you use the Ultram oscillator with its MMIC left in circuit to give around +10dBm output (mine is +9.2dBm) and fit the onboard input attenuation you will be spot on. Obviously, SMA connectors will do a better job than coax pigtails so far as screening is concerned but if you are careful you should be able to get the drive levels right.
Clive
DigiLite oscillator power
Re: DigiLite oscillator power
Interesting. I must try increasing my drive level from -10dBm to -6dBm next time I have access to an analyser. How was the noise floor at -10dBm?
Brian
Brian
Re: DigiLite oscillator power
Brian
Here are some screen shots of the SA looking at the output of my v5.6 board with the onboard 12dB attenuator with reducing drive levels. Not pictured is the 10dBm as there appears to be a limit on the attachments; it's much the same as the OdBm but with further reduced noise 'bumps' at +/= 200MHz. Note how the output level drops when drive falls below 0dBm and the noise floor rises, particularly at +/- 200MHz where it only -35dBc, with -15dBm drive.
0dBm drive
-10dBm drive
-15dBm drive
There's clearly a need for a filter on the DigiLite board output. I've got a 3 pole 1/4 wave filter from an old RSGB handbook design that cleans it up but it's not that good despite being made from chunks of copper.
Clive G3GJA
Here are some screen shots of the SA looking at the output of my v5.6 board with the onboard 12dB attenuator with reducing drive levels. Not pictured is the 10dBm as there appears to be a limit on the attachments; it's much the same as the OdBm but with further reduced noise 'bumps' at +/= 200MHz. Note how the output level drops when drive falls below 0dBm and the noise floor rises, particularly at +/- 200MHz where it only -35dBc, with -15dBm drive.
0dBm drive
- 0dBm.jpg (72.07 KiB) Viewed 7996 times
- -10dBm.jpg (67.1 KiB) Viewed 7996 times
- -15dBm.jpg (66.78 KiB) Viewed 7996 times
Clive G3GJA
Re: DigiLite oscillator power
I don't use the Ultram vco myself, so I haven't seen the spike problem. Have you done the mod below?
http://www.batc.org.uk/forum/viewtopic. ... t=10#p2521
Could you do a plot at -6dBm injection please. That's probably what I'll aim for.
Just to clarify, the input levels you quote are at the modulator LO pins, taking into account the onboard attenuator?
Brian
http://www.batc.org.uk/forum/viewtopic. ... t=10#p2521
Could you do a plot at -6dBm injection please. That's probably what I'll aim for.
Just to clarify, the input levels you quote are at the modulator LO pins, taking into account the onboard attenuator?
Brian
Re: DigiLite oscillator power
I think he's using a signal generator other than the Ultram board to do these tests. But it would be helpful for the rest of us to learn where the "optimal" situation with the Ultram VCO is. Taking the stock VCO, adding the 68pf shunt cap to ground and then feeding it through there -12dB attenuator would be a good test. If going to -10 or the other way -15dB attenuation is better with that osc, that's good to know too.
I have to apologize I've been very busy the last couple of weeks and have not worked on the project. But I'm about ready to package it up (between Christmas and New Year's) and I'd like to mess with the Ultram board in particular as little as possible since it's so small and everything on it is so small! It's bad enough that I'll have to remove the PIC on it but I want to eliminate mods that are not necessary.. if it turns out removing the MMIC isn't necessary I'd rather not do it because then it makes the board more useful for other things as well.
Fred W0FMS
I have to apologize I've been very busy the last couple of weeks and have not worked on the project. But I'm about ready to package it up (between Christmas and New Year's) and I'd like to mess with the Ultram board in particular as little as possible since it's so small and everything on it is so small! It's bad enough that I'll have to remove the PIC on it but I want to eliminate mods that are not necessary.. if it turns out removing the MMIC isn't necessary I'd rather not do it because then it makes the board more useful for other things as well.
Fred W0FMS
Re: DigiLite oscillator power
That's right Fred, I am using the HP 8657B signal genrator to run the tests so the mod for the Ultram board doesn't apply. Please note that I changed the values of the input attenuator to give -12dB and a better match to 50R. (I used 1 x series 93R5 and 2 x parallel 82R5.) The original attenuator gives around -10dB.
With my Ultram producing +9.2dBm the level applied to the modulator chip should be around -6dBm allowing for the quoted 3dB insertion loss of the balun. The recommended level is from -6 to -12dBm.
However, looking at the design of the DigiLite oscillator input, I have to question the inclusion of the 50R load in the balun circuit; The datasheet says that the differential input of the chip is a good match to 50R (typically 1.9:1) and the test circuits shown all have the balun 'secondary' connected directly to ground and not via 50R. Guanella baluns don't need that resistor according to all references I've found on the net so that may skew the drive level to the AD8346 if you're trying to work from datasheet values.
With my 12dB attenuator in circuit the noise floor starts to rise noticeably when the input drops below 0dBm. That equates to -12dBm drive level without the attenuator so Brian's -6dBm should be good allowing for loss in the balun. The rate at which the noise floor rises is 2-3 times the decrease in local oscillator drive.
With regard to Alain's observed effect, I'm wondering whether his mod is increasing the drive level and that's fixing the noise bumps I've seen rather than being a defect in the Ultram oscillator. I've run 5 of the Ultram oscillators into my SA (3 @ 437MHz and 1 each @ 1249 and 1288MHz) and they are all within the claimed spec. for spurious outputs although some of the 70cm ones seem a lot better than others for close in sprogs. The second harmonic is quite high though, being only about -15dbC.
All the above does make me question whether it's wise to remove the MMIC on the Ultram board as has been suggested in this forum. It's supposed to have 22dB gain so without it I assume you will get -13dBm; apply that to the DigiLite board without the attenuator and include the balun loss and you end up with -16dBm which will definitely cause the noise floor to be higher than it should be.
Clive G3GJA / G8EQZ
With my Ultram producing +9.2dBm the level applied to the modulator chip should be around -6dBm allowing for the quoted 3dB insertion loss of the balun. The recommended level is from -6 to -12dBm.
However, looking at the design of the DigiLite oscillator input, I have to question the inclusion of the 50R load in the balun circuit; The datasheet says that the differential input of the chip is a good match to 50R (typically 1.9:1) and the test circuits shown all have the balun 'secondary' connected directly to ground and not via 50R. Guanella baluns don't need that resistor according to all references I've found on the net so that may skew the drive level to the AD8346 if you're trying to work from datasheet values.
With my 12dB attenuator in circuit the noise floor starts to rise noticeably when the input drops below 0dBm. That equates to -12dBm drive level without the attenuator so Brian's -6dBm should be good allowing for loss in the balun. The rate at which the noise floor rises is 2-3 times the decrease in local oscillator drive.
With regard to Alain's observed effect, I'm wondering whether his mod is increasing the drive level and that's fixing the noise bumps I've seen rather than being a defect in the Ultram oscillator. I've run 5 of the Ultram oscillators into my SA (3 @ 437MHz and 1 each @ 1249 and 1288MHz) and they are all within the claimed spec. for spurious outputs although some of the 70cm ones seem a lot better than others for close in sprogs. The second harmonic is quite high though, being only about -15dbC.
All the above does make me question whether it's wise to remove the MMIC on the Ultram board as has been suggested in this forum. It's supposed to have 22dB gain so without it I assume you will get -13dBm; apply that to the DigiLite board without the attenuator and include the balun loss and you end up with -16dBm which will definitely cause the noise floor to be higher than it should be.
Clive G3GJA / G8EQZ