Topics

si5351a CLK drive levels in uBITX #ubitx

 

I noticed that in both Ashar and Ian's uBITX code that the CLK outputs of the si5351a are being set to 4ma.

uint8_t  si5351bx_drive[3] = {1, 1, 1};  // 0=2ma 1=4ma 2=6ma 3=8ma for CLK 0,1,2

My understanding is that this translates to an output signal level of about 6dBm. 

What I have previously read is that most passive Double Balanced Diode ring mixers need a LO signal of 
7dBm or greater for optimal performance. 

Has anyone tried increasing one or more CLK output levels to 6mA to see what effect this has on receiver performance ?
A 6mA setting would translate to about 8dBm.

This would be an interesting experiment for someone with the necessary test equipment to verify whether this would be a positive
or a negative change. 

Cheers

Michael VE3WMB 

Jim Sheldon
 

Michael and all,
I set my signal generator to .5 microvolt @ 7.000 MHz and listened to the sound of it (no accurate S meter here yet).  Compiled my own group's software, Version 6.00R (for the BITeensio card we sell) at 6 and 8 ma.  6 ma gave an increase in sensitivity by a noticeable amount (not a lot but noticeable) and 8 ma brought it up again by a quite small amount but I noticed a marked increase in extra signal artifacts in the RX Audio.  There were a few added artifiacts @ 6 ma also, but not nearly as pronounced as with the drive set to 8 ma.  

This may very well be different for each different uBITX due to the variable nature of the transistors and mixer diode arrays used in the rig.  Not a scientific measurement but I did see some RX gain from setting it to 8 ma.  I left mine at that setting for now and will run it for a while to see what happens.  Might have to get the adapter out and see what it did on TX to the spurious and harmonic output.  I don't TX with this particular uBITX except into a dummy load so may let someone with better test equipment check the TX spurs & harmonics @ 8 ma clock drive.

Jim Sheldon, W0EB

------ Original Message ------
From: "Michael Babineau" <mbabineau.ve3wmb@...>
Sent: 7/24/2018 10:27:01 AM
Subject: [BITX20] si5351a CLK drive levels in uBITX #ubitx

I noticed that in both Ashar and Ian's uBITX code that the CLK outputs of the si5351a are being set to 4ma.

uint8_t  si5351bx_drive[3] = {1, 1, 1};  // 0=2ma 1=4ma 2=6ma 3=8ma for CLK 0,1,2

My understanding is that this translates to an output signal level of about 6dBm. 

What I have previously read is that most passive Double Balanced Diode ring mixers need a LO signal of 
7dBm or greater for optimal performance. 

Has anyone tried increasing one or more CLK output levels to 6mA to see what effect this has on receiver performance ?
A 6mA setting would translate to about 8dBm.

This would be an interesting experiment for someone with the necessary test equipment to verify whether this would be a positive
or a negative change. 

Cheers

Michael VE3WMB 

ajparent1/KB1GMX
 

Actually 5 to 10dbm the performance is about the same...

However!  There are attenuators before each of the mixers so there is
what the 5351 puts out and what actually gets delivered which is 3 to 6 db less.

Myself, 4ma into a 50 ohm load does not translate to 6dbm.  Math says maybe -1dbm.
My scope and Spec analyzer both agree within about 1db.

To hit +7Dbm in a 50 ohm system you need 1.41V PP or about 28ma PP.


Allison

ajparent1/KB1GMX
 

I haven't looked at spurs or harmonics related to this save for it makes sense.
The drive is way too low and the sensitivity increase is consistent with drive increase.

If the drive is increased the spurs due to mixer overload should improve, harmonics
may stay the same.

The problem is even at 8ma drive your barely at 5dbm out out and most of the
mixers have 6db attenuators ahead of them.... So drive is -1dbm.  The mixers
are running starved.

Allison

Jim Sheldon
 

Sort of what I thought Allison I'm going to try it at max for a while and see just what transpires.  I have the SA on and will run rig into a dummy load with the SA pickoff in the line - we'll see what I find as far as harmonics.  Unfortunately my SA is an old GW Instek and I can't get enough span at the normal HF frequencies involved to see all the harmonics at once.  I have to carefully increase the center frequency to see 'em all.  Makes it sort of unscientifice and I'm not sure how well calibrated this thing is. 

Wish I could have afforded a RIGOL or Sigilent but I got this one off ITT Tech when they close a NJ branch for just under $400.  Better than no SA at all - LOL - but sometimes the readings (even though it had a recent calibration sticker on it when I got it) can be suspect.

OK, before sending this, I hooked up the SA and ran a harmonic / garbage scan on it.  Set the TX freq first at 3.5 MHz, output 5 watts (where I normally run it).  Set the Spectrum Analyzer input sensitivity to put the signal from the pick-off so that the fundamental was at 0 dBm on the screen and started looking.  2nd Harmonic 55 dB down, 3rd harmonic 62 dB down and all the others were each slightly better than that so I think everything is still OK.  

With the transmitter keyed at 5 watts out, I ran the analyzer through the freq range up into the VHF range and I didn't see any garbage that wasn't at least 60 + dB down from the fundamental.  Remember though this was with the fundamental at 3.5 MHz.  I'm out of time for right now, but will try and run scans with the TX on the higher bands and look for the "spurs" as well.  

I am pleased though that with the clock drive set to the max available in the software, the sensitivity of the RX has increased enough to make it worthwhile and the Harmonics/Spurs are still within FCC boundaries.  (At least on this particular V4 uBITX).

Jim Sheldon, W0EB



------ Original Message ------
From: "ajparent1/KB1GMX" <kb1gmx@...>
Sent: 7/24/2018 12:30:38 PM
Subject: Re: [BITX20] si5351a CLK drive levels in uBITX #ubitx

I haven't looked at spurs or harmonics related to this save for it makes sense.
The drive is way too low and the sensitivity increase is consistent with drive increase.

If the drive is increased the spurs due to mixer overload should improve, harmonics
may stay the same.

The problem is even at 8ma drive your barely at 5dbm out out and most of the
mixers have 6db attenuators ahead of them.... So drive is -1dbm.  The mixers
are running starved.

Allison

Don, ND6T
 

I'd never really considered running it other than full output. Since I run my own firmware all of the time, I never bothered to try it at lower levels and have had great results. Measuring with my old HP power meter I get between +3.2 dBm to +3.8 dBm from 3 MHz to 150 MHz. The slight dip near 52 MHz I attribute to the blocking capacitor that I use. The spectrum analyzer shows it as +3.17 dBm on 10 MHz where my power meter shows +3.8 dBm. That pretty much coincides, since the SA is looking at narrow-band energy and the old bolometer power meter is looking at overall energy (including all the harmonics).
This is using just the Si5351 in a signal generator project.
The mixer depends upon the peak voltage to switch those diodes so I wouldn't think that there would be much of a performance change unless the output was dropped appreciably. I see no reason to run the synthesizer at any level other than full throttle. Did anyone find a measurable increase in either receive or transmit birdies?
-Don

 

Allison et al :

I am not sure where I got the dBm numbers corresponding to the si5351a level settings, but I found them written in my notebook when I was looking 
back for something else I noted. Yes I certainly didn't take into consideration the attenuators before the mixers, that makes it into a whole different
"ball-game". 

I am currently running the KD8CEC V1.095 beta with a Nextion display tacked on so until Ian releases this code and puts it in GITHUB
I don't have any access to source code to play with the settings myself. I am anxious to see how much this might  improve the sensitivity on the 
higher bands as my uBITX seems very "quiet" on 20m and up. 

From what I remember seeing in the code, it appears that the output level parameter for the clocks gets sent to the si5351a every time that there is a change in frequency.
If this is the case then there is even the option of varying the level according to dial frequency, if that is desirable (i.e. cranking it up so that there is more gain
on the higher frequencies).

Cheers

Michael VE3WMB 

ajparent1/KB1GMX
 

I ran the Raduino bare and 3.1 dbm is what I'd seen however put a 6DB pad after than and your at -3dbm.
Mixers will work at lower than 1mW but the overload performance goes down greatly and you really want
about 3mW (+5dbm) into the mixer to make the diode conduct hard or you get non linear results and 
increased spurs.  

Based on that I pumped it up to 8ma.... it helps but after the attenuator your still way low.
With a 6db pad I got +0.1 dbm on the analyzer or about +6.1 dbm into the attenuator. 

Keep in mind 8ma into 50 ohms is only .25V p-p! which for a square wave is the same as RMS.
The square wave makes the diodes switch more abruptly but the current still has to be there.

Allison

ajparent1/KB1GMX
 

Don,

I agree full throttle and cut the attenuators out or down.  If anything I'd expect the birdies 
to go down with the mixers run at a reasonable level rather than underdriven.

I have to assemble everything to measure that but I'd expect better results.

Allison

Don, ND6T
 

By "full throttle" I meant the programmed drive levels. I've never tried running the Si5351 directly into a diode mixer. Instead, I have always used an amplifier to increase the level and match impedance. I must say that the uBITX sounds surprisingly good with just the pads. But yes, I am really curious to see what results if just using the synthesizer output. Documents list the output impedance at 85 ohms at default levels and that is a reasonable termination, I believe.
Taking the resources and time to re-program the synthesizer for a lower level just never seemed worthwhile. Not including output levels results in running at the default full power. So programming time is reduced, too. -Don

Jerry Gaffke
 

Yup.
In addition to under-driving the mixer, we are over-loading the si5351 outputs.
Some previous speculation regarding the si5351 attenuators can be found here:
    https://groups.io/g/BITX20/message/46122



On Wed, Jul 25, 2018 at 01:05 PM, ajparent1/KB1GMX wrote:
I agree full throttle and cut the attenuators out or down.  If anything I'd expect the birdies 
to go down with the mixers run at a reasonable level rather than underdriven.

I have to assemble everything to measure that but I'd expect better results.

VK3HN
 

I have several homebrew projects with Minicircuits diode DBMs and si5351 VFO/BFO/carrier oscillators.

I first tried si5351 clocks at all output levels via a 3dB 50 ohm pad into an SBL1 with unconvincing results. I seem to recall a scope on the clock pin showed significant loading when attached to the mixer LO port. 

I then inserted a 74HC04 squarer and a bigger pad with much better results, but it stops squaring at the higher end of HF. 

In my last build, a G6LBQ Bitx, I settled on a single stage 2n2222 common emitter buffer/amp (with broadband bifilar tapped (10T FT37-43) inductor) between the si5351 clock (running at 2mA) and an ADE-1, which gave good results. 
 
Just sharin', YMMV.  

 

At 26/07/2018, you wrote:

I then inserted a 74HC04 squarer and a bigger pad with much better results, but it stops squaring at the higher end of HF.
Maybe the frequency is going beyond the limit of the device ?

Raj