Something to keep in mind in circuits that use multiple power supplies is
power sequencing:? It is important that the 5V comes on before the 3.3V and
turns off after the 3.3V when the 3.3V turns off.? Otherwise you can drive
the 5V unpowered device from the 3.3V driving chip.? If the 3.3V driving
chip can supply a hefty amount of current, you can? possibly damage the
input circuitry of the 5V powered device by forward biasing its protection diode
if it has one.? That protection diode may see more current than it is
designed to handle. Depending on the particular 5V device, it may not
immediately fail but can ultimately fail with repeated power sequencing at turn
on or turn off.? A possible safety measure is to put a small Schottky diode
between the +5V and +3.3V rails in the direction such that the 3.3V supply pulls
up the 5V rail if unpowered, until the 5V supply kicks in and turns off that
diode.? If the 3.3V supply is derived from the 5V supply, that is normally
not a problem. But if the 5V and 3.3V supplies are totally independent, that
problem can easily happen. The input clamp current for the 74ACT family has a
clamp diode which is pretty hefty (current rating of 20mA) so it¡¯s probably not
a problem for the 74ACT chip, but you are momentarily shorting the output of the
SI5351 at power up if there is a power sequence problem.
I haven¡¯t tried
it myself but if you need a larger voltage than that provided by a single +5V
buffer or paralleled buffers, you can wire up the 74ACT04N? to generate a
true and complement output.? A.C. couple them to a 1:1 RF transformer to
get double the output or 10 volts peak to peak.? If you need more than 10
volts and are driving a high impedance load, then increase the turns ratio to 2
to 1 which will provide 20 volts peak to peak.? You can make your own one
to one transformer with a low cost toroid and bifilar winding. Drive one winding
with the true and complement outputs. The other winding is the secondary going
to the load. You can use trifilar winding for two to 1 turns ratio. Drive one
winding with the true and complement outputs.? Put the other two windings
in series aiding for the secondary.?? Winding them bifilar or trifilar
improves the bandwidth.
-Steve K1RF
toggle quoted message
Show quoted text
Sent: Tuesday, December 17, 2019 9:05 PM
Subject: Re: [qrptech] Si5351 output power
?
Maybe
this 74ACT04N, $0.50 plus $10 shipping: ??? Ground
pin 7 with a very short wire (or just bend the pin down to where you can solder
it to a bare copper covered PC board) Also ground the unused inputs at pins
3,5,9,11,13 Put 5.0 volts on pin 14, and add a 0.1uF cap from that pin to
ground, keep the capacitor wires short. Drive the Si5351 output into pin 1
using a twisted wire pair of that signal plus ground. Try driving your
transmitter with the 5.0 volt square wave coming out of pin 2 using another
twisted wire pair. Might work, especially if it almost works with the 3.3
volts from an Si5351. The 74ACT family of parts is unusual in that it is
CMOS but tries to be compatible with old school TTL, so the input threshold
voltage is kept around 1.5 volts regardless of the power supply voltage. That
allows us to drive this part directly from the 3.3 volt Si5351, even when this
part is powered from 5 volts. And at 24ma, the output buffers are relatively
hefty. A bit more output could be had by bumping the supply up as high as 5.5
volts, beyond that is not recommended. If you needed more current available
at the output than 24ma (I doubt it) then multiple inverters could be wired in
parallel. A good idea to ground the inputs as they are high impedance and
might decide to float into the input transition region, at which point the
part can get hot and possibly oscillate. If you need more than a 5v
square wave, an easy pre-packaged solution might be a comparator that can
deal with perhaps a 12v power supply.? But try this first. Jerry,
KE7ER On Tue, Dec 17, 2019 at 01:07 PM, Don Richards wrote:
On a related topic, I built a vfo with a nano and a 5351 to replace the
vfo in a tube rig but I need more voltage to drive it. It will operate the
receiver but is just on the edge for the TX. It will only muster half power at
best. I built the two transistor circuit that is commonly found online but it
doesn't work that well. Does anyone have anything better? Since it is a square
wave, would some flip flop chip work and go rail to rail?
Just thinking out loud
?
73 Don ve3ids
|
Hi all
Loading an Si5351A output at 50-ohms also affects phase noise. I did a study of Si5351A phase noise, particularly in relation to the QCX CW transceiver . One of the motivations for the experiments were suggestions I had received, that more supply line filtering would improve the phase noise of the QCX synthesizer. In the end I found supply cleanliness is certainly important but, in regard to the QCX, the QCX performance is already good and I could not find any way to further improve it.?
Anyway - another conclusion of the experiments was that with a 50-ohm load the phase noise is significantly degraded compared to a higher impedance load; this can be somewhat remediated by using a DC blocking capacitor.?
Application is important: what level of performance is desired depends on the application. If you need highest performance then ideally use a higher impedance load, don't load the Si5351A outputs with 50-ohms. If you must, then be sure to use a capacitor in series with the load.?
Avoiding cross-talk between outputs is another reason to use high impedance loads, in applications where you will be using multiple Si5351A outputs and need low crosstalk.
73 Hans G0UPL? ?
|
If you read the datasheet, the Si5351 claims to have a 50 ohm output.
But yes, it is primarily designed to replace cheap 4 pin crystal oscillators.
Jerry
toggle quoted message
Show quoted text
On Tue, Dec 17, 2019 at 08:24 PM, Bob Clarke wrote:
It¡¯s a clock generator and was designed to drive a logical load, that is, a CMOS or TTL input, and definitely not a 50 ohm resistance. The first rule in a situation like this is to *read* the data sheet and see what the actual spec is, then add an appropriate buffer. You can go the elegant route with an emitter follower or use a bunch of inverters in parallel.?
?
|
Re: What level math is in the RF design books?
Good news:? Algebra for designing and biasing bipolar circuits.
Not so good news:? Partial differential equations for biasing FET circuits, though I¡¯d probably just copy an oscillator circuit rather than design one from scratch.
In W7ZOI¡¯s books just algebra, I think. Generally speaking approximations are used for calculating collector current. It¡¯s all good, don¡¯t worry about the math. I had hard core semiconductor physics in school and never really used it.?
Bob, N1RC
What level of math course would I need to study to learn how to comprehend the equations presented in
design books like Introduction to Radio Frequency Design?? by? W7ZOI.
I have been tinkering with circuits from these books for 50 years but never really learned what is in the math.
There are all sorts of online math course now so.....
What level of math class would teach this gobbledygoop?
and
Do any of you? ACTUALLY use this level of math to build radios or is the math only used by
the guys designing the transistors????
73 Rob KB3BYT
-- Bob Clarke
M: 978.337.2720
|
It¡¯s a clock generator and was designed to drive a logical load, that is, a CMOS or TTL input, and definitely not a 50 ohm resistance. The first rule in a situation like this is to *read* the data sheet and see what the actual spec is, then add an appropriate buffer. You can go the elegant route with an emitter follower or use a bunch of inverters in parallel.?
Bob
On Tue, Dec 17, 2019 at 12:23 PM Michael Maiorana via Groups.Io <zfreak= [email protected]> wrote: I'm just looking for a sanity check (please).
Square wave output from an Si5351 is 3.2V p-p unloaded. Into a 50 ohm resistive load it drops to 1.6V p-p. Average voltage of a square wave is 1/2 p-p value, so Vavg is 0.8V. Power into the resistor (assuming a pure square wave) is (0.8V)^2/50 or 12.8mw. Converting to dBm 10*(Log(12.8/1)) = 11dBm.
11dBm is more than enough to meet the 7dBm drive requirements of a diode ring mixer. However, I'm reading conflicting info on the Si5351's ability to directly drive a diode ring mixer.
Any advice or guidance will be greatly appreciated.
These little Si5351 demo boards from Ebay are amazing for the measly $3.50 that they cost. They have onboard linear LDO regulator and logic level converters so you can drive them direct from an Arduino running at 5V. 4 wires connected to an Uno and some sample code and I was making clocks in less than 10 minutes. What an age we live in!
Mike M. KU4QO
-- Bob Clarke
M: 978.337.2720
|
Re: 7 MHz Oscillator Ideas
I¡¯d grab a 40 meter tank circuit from one of the older handbooks for a cheap and cheerful solution. Or a DDS circuit like this
Good Luck,
Bob
On Sat, Dec 7, 2019 at 10:10 AM JT Croteau via Groups.Io <jt.tobit= [email protected]> wrote: Hi folks, I'd like to build a simple NE612/LM386 receiver to cover
from roughly 7.028 to 7.042.? I'm wondering what I should use for the
oscillator.? I've been poking around on google but it is making my
head spin.? Minimal drift would be ideal as I'd like to park it on
7040 or 7030 for a few hours just to monitor.
Times like this I wish I still had some of my Doug Demaw books.
Thanks
-- Bob Clarke
M: 978.337.2720
|
Re: 7 MHz Oscillator Ideas
Thanks for all the ideas gents, I need to get some parts from Diz and
Rex and start experimenting. Hopefully after Christmas.
toggle quoted message
Show quoted text
On Sun, Dec 15, 2019 at 7:04 AM Chuck Carpenter <w5usj@...> wrote:
OK, last time, I promise.
Jerry and All,
I expect we've been thinking grapefruit and tangerines...related but not the same...8^)
Varactors used to tune an entire AM or FM band with a max C of 350pF or so and a min C of about 30pF at 12 V. I'd expect them to be like you described.
The varactors I'm using, BB910 or MV209, have a max C of about 60pF or so and a min C of about 6pF at 12 V, same percentages but minimal effect in the circuit where used. My choice of these two (or other similar) was related to the 60pF section of a polycon variable often used in VXOs.
Anyway, Fun Stuff...
On Sat, Dec 14, 2019 at 09:14 AM, Jerry Gaffke wrote:
What we are saying is that for most such designs where the varactor might be trying to
give a range of 300khz instead of 3khz, the stability will be 100 times worse than what you experience.
--
Chuck, W5USJ (ex K2OFN)
Point, Rains Co, TX EM22cv
|
That ebay amp can drive 50 ohms with a whopping 10dBm, or 10 milliwatts.
That works out to about 0.707 volts RMS on the output, less than your Si5351.
It's meant for use in front of a receiver, or something like that.
A comparator that would work well at HF with a 10 volt swing might not be cheap.
Do you have QRP transmitter lying about capable of at least 2 Watts?
Drive that QRP transmitter with the Si5351,
put a 50 ohm dummy load on the QRP transmitter so you don't blow it up,
and run coax from that dummy load into your vacuum tube transmitter.
A two watt transmitter into 50 ohms is? squareroot(2*50) = 10 volts RMS?
If it's more than two watts, you could bring it down by a special resistive divider dummy load.
Might work.
Or it might trash the QRP transmitter if you do it wrong.
Wouldn't be a bad idea to seek some help from somebody with experience
in this sort of thing and some test equipment.?
Jerry
toggle quoted message
Show quoted text
On Tue, Dec 17, 2019 at 06:28 PM, Don Richards wrote:
Jerry
Thanks for the info. I need about 10 volts. What are your thoughts on this?
?
?
73 Don ve3ids
|
It has a 14mhz crystal, a transistor, a few passive components, a transformer, and then a two inductor output filter.
So I'm going with Eric that it plugs into a larger assembly, and say that this particular part of the whole is a 14mhz local oscillator.
toggle quoted message
Show quoted text
On Tue, Dec 17, 2019 at 07:16 PM, Eric KE6US wrote:
No idea, but given the stacking header and part designations all in the 20's, it looks to be part of an assembly. A daughterboard of some sort. Maybe that'll jog your memory.
Eric KE6US
?
|
No idea, but given the stacking header and part designations all
in the 20's, it looks to be part of an assembly. A daughterboard
of some sort. Maybe that'll jog your memory.
Eric KE6US
On 12/17/2019 3:40 PM, Kurt Loken
wrote:
toggle quoted message
Show quoted text
Hi all,
I have gotten back into radio after about a decade. ?Finding my
goodies I didn¡¯t complete. ?One kit I found doesn¡¯t have any
documentation or any info on the board. ?The crystal is 14Mhz.
Any idea what kit this is?
Thanks,
Kurt-AE6UJ
|
Jerry Thanks for the info. I need about 10 volts. What are your thoughts on this?
73 Don ve3ids
On Tue., Dec. 17, 2019, 9:05 p.m. Jerry Gaffke via Groups.Io, <jgaffke= [email protected]> wrote: Maybe this 74ACT04N, $0.50 plus $10 shipping:
? ??
Ground pin 7 with a very short wire (or just bend the pin down to where you can solder it to a bare copper covered PC board)
Also ground the unused inputs at pins 3,5,9,11,13
Put 5.0 volts on pin 14, and add a 0.1uF cap from that pin to ground, keep the capacitor wires short.
Drive the Si5351 output into pin 1 using a twisted wire pair of that signal plus ground.
Try driving your transmitter with the 5.0 volt square wave coming out of pin 2 using another twisted wire pair.
Might work, especially if it almost works with the 3.3 volts from an Si5351.
The 74ACT family of parts is unusual in that it is CMOS but tries to be compatible with?
old school TTL, so the input threshold voltage is kept around 1.5 volts regardless of the power supply voltage.
That allows us to drive this part directly from the 3.3 volt Si5351, even when this part is powered from 5 volts.
And at 24ma, the output buffers are relatively hefty.
A bit more output could be had by bumping the supply up as high as 5.5 volts, beyond that is not recommended.
If you needed more current available at the output than 24ma (I doubt it) then multiple inverters could be wired in parallel.
A good idea to ground the inputs as they are high impedance and might decide to float into the input transition region,
at which point the part can get hot and possibly oscillate.
If you need more than a 5v square wave, an easy pre-packaged solution might be a comparator that can deal
with perhaps a 12v power supply.? But try this first.
Jerry, KE7ER
On Tue, Dec 17, 2019 at 01:07 PM, Don Richards wrote:
On a related topic, I built a vfo with a nano and a 5351 to replace the vfo in a tube rig but I need more voltage to drive it. It will operate the receiver but is just on the edge for the TX. It will only muster half power at best. I built the two transistor circuit that is commonly found online but it doesn't work that well. Does anyone have anything better? Since it is a square wave, would some flip flop chip work and go rail to rail?
Just thinking out loud
?
73 Don ve3ids
|
Hi all,
I have gotten back into radio after about a decade. ?Finding my goodies I didn¡¯t complete. ?One kit I found doesn¡¯t have any documentation or any info on the board. ?The crystal is 14Mhz.
Any idea what kit this is?
Thanks,
Kurt-AE6UJ
|
Maybe this 74ACT04N, $0.50 plus $10 shipping:
? ??
Ground pin 7 with a very short wire (or just bend the pin down to where you can solder it to a bare copper covered PC board)
Also ground the unused inputs at pins 3,5,9,11,13
Put 5.0 volts on pin 14, and add a 0.1uF cap from that pin to ground, keep the capacitor wires short.
Drive the Si5351 output into pin 1 using a twisted wire pair of that signal plus ground.
Try driving your transmitter with the 5.0 volt square wave coming out of pin 2 using another twisted wire pair.
Might work, especially if it almost works with the 3.3 volts from an Si5351.
The 74ACT family of parts is unusual in that it is CMOS but tries to be compatible with?
old school TTL, so the input threshold voltage is kept around 1.5 volts regardless of the power supply voltage.
That allows us to drive this part directly from the 3.3 volt Si5351, even when this part is powered from 5 volts.
And at 24ma, the output buffers are relatively hefty.
A bit more output could be had by bumping the supply up as high as 5.5 volts, beyond that is not recommended.
If you needed more current available at the output than 24ma (I doubt it) then multiple inverters could be wired in parallel.
A good idea to ground the inputs as they are high impedance and might decide to float into the input transition region,
at which point the part can get hot and possibly oscillate.
If you need more than a 5v square wave, an easy pre-packaged solution might be a comparator that can deal
with perhaps a 12v power supply.? But try this first.
Jerry, KE7ER
toggle quoted message
Show quoted text
On Tue, Dec 17, 2019 at 01:07 PM, Don Richards wrote:
On a related topic, I built a vfo with a nano and a 5351 to replace the vfo in a tube rig but I need more voltage to drive it. It will operate the receiver but is just on the edge for the TX. It will only muster half power at best. I built the two transistor circuit that is commonly found online but it doesn't work that well. Does anyone have anything better? Since it is a square wave, would some flip flop chip work and go rail to rail?
Just thinking out loud
?
73 Don ve3ids
|
Nick,
Ah, I vaguely remember a discussion on the bitx20 forum about using a logic gate.
Allison kicked it off with a discussion of crosstalk in the Si5351 outputs when they get loaded down:
This thread started around Sept 6 2018, but I didn't find the phrases you quoted.
? ??/g/BITX20/topic/si5351_crosstalk/25213331
That thread is 74 posts long, but worth a read.
The 74ACT family could be powered from 5v and still have an input threshold
around 1.5v.? Most CMOS parts have an input threshold around half of Vdd.
Probably best to power those from 3.3v when driving from a 3.3v Si5351.
Using single gate parts (such as Glen's?74LVC1G04GW)?is preferred, since using
multiple gates on one die will contribute to crosstalk.
Jerry, KE7ER
On Tue, Dec 17, 2019 at 12:20 PM, Nick Kennedy wrote:
I think I would have gone for the power in the sine wave that's under those harmonics and divided the 0.8 Vpk by sqrt(2) to get the power of the fundamental.
?
FWIW and regarding appropriate loading of the Si5351, in this case regarding cross-talk, here's a tidbit I saved from the list a year ago:
?
September 6, 2018
Here¡¯s a bit of info from Allison KB1GMX on cross-talk between the outputs of the Si5351a¡¯s channels which could be an issue if you use one for BFO and one for LO or similar. A possible cure is to drive higher than 50 ? impedances with the outputs:
I also did the measurement with higher than 50 ohm load and got?
remarkably better result.
Hint:? drive the load using a 74LVT04 or one of the friends of that part
or any cmos buffer.??The output is loaded with CMOS gate and the?
CMOS inverter can be loaded heavier.
And AB7VF said: Read somewhere a test that suggested better results with 80 ohms rather than the spec-ed 50. Jim
73, Nick - WA5BDU
|
On a related topic, I built a vfo with a nano and a 5351 to replace the vfo in a tube rig but I need more voltage to drive it. It will operate the receiver but is just on the edge for the TX. It will only muster half power at best. I built the two transistor circuit that is commonly found online but it doesn't work that well. Does anyone have anything better? Since it is a square wave, would some flip flop chip work and go rail to rail? Just thinking out loud
73 Don ve3ids
On Tue., Dec. 17, 2019, 2:49 p.m. Jerry Gaffke via Groups.Io, <jgaffke= [email protected]> wrote: A couple edits are required here:
< With capacitive coupling, the Si5351 is? effectively driving your?50 ohm load into 3.3/2=1.65 volts,
> which reduces the currents by half.
> With capacitive coupling, the Si5351 is driving your?50 ohm load with a square wave centered on ground,
> which reduces the currents by half from what they would be without capacitive couplig.
< When a mixer says it wants 7 dBm, I believe that's defined as a source that can
< drive an 11 dBm sine wave into a 50 ohm resistive load,
> When a mixer says it wants 7 dBm, I believe that's defined as a source that can
> drive an 7 dBm sine wave into a 50 ohm resistive load,
Ryan's 3dB pad is probably a better choice than the 6dB pad typically used
on a diode ring mixer if the 6dB pad is not giving the mixer the required 7dBm of power.
Though the 6dB pad does a better job of terminating that mixer port with 50 ohms,
this prevents unwanted mixer products coming out of the local oscillator port from
reflecting back into the mixer to create additional havoc.
Either one of those pads is terminating the Si5351 with 50 ohms,
I've been assuming that the Si5351 is better off with an easier load.
Jerry
On Tue, Dec 17, 2019 at 10:19 AM, Jerry Gaffke wrote:
Mike,
I haven't made any such measurements, so you're ahead of me on that.
I have heard in other forums that crosstalk and perhaps other forms of distortion
creep in as the Si5351 output buffers get loaded up beyond a few ma.
With capacitive coupling, the Si5351 is? effectively driving your
50 ohm load into 3.3/2=1.65 volts, which reduces the currents by half.
So if you see 1.6v pk-pk, that's 0.8v/50ohms=16ma, which is not just too far
from the programmable 8ma max for the output buffers of AN619.
But those CMOS buffers are working awfully hard.
The datasheet says that the Si5351 has an output impedance of 50 ohms.
So the hypothetical 3.3v square wave voltage source inside the Si5351 is driving
50 ohms internal plus your 50 ohm load.? So a 1.6v square wave makes perfect sense.
I agree, if you are seeing a 1.6v pk-pk square wave into 50 ohms, then that is 11 dBm.
Some of that energy is in the harmonics, but we're in the correct ballpark.
When a mixer says it wants 7 dBm, I believe that's defined as a source that can
drive an 11 dBm sine wave into a 50 ohm resistive load, even though the non-linear
diodes will make that waveform very un-sine-like in both voltage and current.?
Whether a diode ring mixer is better off with?a square wave is perhaps up for debate,
There is only one pin for the 3.3v rail to power the output buffers on the Si5351,
so with two or three outputs loaded up like that you may get worse results.
Your results justify the uBitx approach of driving a resistive pad directly from
an Si5351, and the uBitx does seem to work well enough.? Though it has its birdies.??
Jerry, KE7ER
|
Yet another correction, I should just go back to bed:
Both the 6db and 3db pads are trying to terminate the oscillator and mixer with 50 ohms,
but the 6db pad is doing a better job of it.
This is at the expense providing half the power to the mixer that the 3db pad could.
toggle quoted message
Show quoted text
On Tue, Dec 17, 2019 at 11:49 AM, Jerry Gaffke wrote:
Though the 6dB pad does a better job of terminating that mixer port with 50 ohms,
this prevents unwanted mixer products coming out of the local oscillator port from
reflecting back into the mixer to create additional havoc.
Either one of those pads is terminating the Si5351 with 50 ohms,
|
I think I would have gone for the power in the sine wave that's under those harmonics and divided the 0.8 Vpk by sqrt(2) to get the power of the fundamental.
FWIW and regarding appropriate loading of the Si5351, in this case regarding cross-talk, here's a tidbit I saved from the list a year ago:
September 6, 2018
Here¡¯s a bit of info from
Allison KB1GMX on cross-talk between the outputs of the Si5351a¡¯s channels
which could be an issue if you use one for BFO and one for LO or similar. A
possible cure is to drive higher than 50 ? impedances with the outputs:
I also did the measurement with higher than 50 ohm load and
got?
remarkably better result.
Hint:? drive the load using a 74LVT04 or one of the
friends of that part
or any cmos buffer.??The output is loaded with CMOS
gate and the?
CMOS inverter can be loaded heavier.
And AB7VF said: Read somewhere a test that suggested better
results with 80 ohms rather than the spec-ed 50. Jim 73, Nick - WA5BDU
On Tue, Dec 17, 2019 at 11:23 AM Michael Maiorana < zfreak@...> wrote: I'm just looking for a sanity check (please).
Square wave output from an Si5351 is 3.2V p-p unloaded. Into a 50 ohm resistive load it drops to 1.6V p-p. Average voltage of a square wave is 1/2 p-p value, so Vavg is 0.8V. Power into the resistor (assuming a pure square wave) is (0.8V)^2/50 or 12.8mw. Converting to dBm 10*(Log(12.8/1)) = 11dBm.
11dBm is more than enough to meet the 7dBm drive requirements of a diode ring mixer. However, I'm reading conflicting info on the Si5351's ability to directly drive a diode ring mixer.
Any advice or guidance will be greatly appreciated.
These little Si5351 demo boards from Ebay are amazing for the measly $3.50 that they cost. They have onboard linear LDO regulator and logic level converters so you can drive them direct from an Arduino running at 5V. 4 wires connected to an Uno and some sample code and I was making clocks in less than 10 minutes. What an age we live in!
Mike M. KU4QO
|
A couple edits are required here:
< With capacitive coupling, the Si5351 is? effectively driving your?50 ohm load into 3.3/2=1.65 volts,
> which reduces the currents by half.
> With capacitive coupling, the Si5351 is driving your?50 ohm load with a square wave centered on ground,
> which reduces the currents by half from what they would be without capacitive couplig.
< When a mixer says it wants 7 dBm, I believe that's defined as a source that can
< drive an 11 dBm sine wave into a 50 ohm resistive load,
> When a mixer says it wants 7 dBm, I believe that's defined as a source that can
> drive an 7 dBm sine wave into a 50 ohm resistive load,
Ryan's 3dB pad is probably a better choice than the 6dB pad typically used
on a diode ring mixer if the 6dB pad is not giving the mixer the required 7dBm of power.
Though the 6dB pad does a better job of terminating that mixer port with 50 ohms,
this prevents unwanted mixer products coming out of the local oscillator port from
reflecting back into the mixer to create additional havoc.
Either one of those pads is terminating the Si5351 with 50 ohms,
I've been assuming that the Si5351 is better off with an easier load.
Jerry
toggle quoted message
Show quoted text
On Tue, Dec 17, 2019 at 10:19 AM, Jerry Gaffke wrote:
Mike,
I haven't made any such measurements, so you're ahead of me on that.
I have heard in other forums that crosstalk and perhaps other forms of distortion
creep in as the Si5351 output buffers get loaded up beyond a few ma.
With capacitive coupling, the Si5351 is? effectively driving your
50 ohm load into 3.3/2=1.65 volts, which reduces the currents by half.
So if you see 1.6v pk-pk, that's 0.8v/50ohms=16ma, which is not just too far
from the programmable 8ma max for the output buffers of AN619.
But those CMOS buffers are working awfully hard.
The datasheet says that the Si5351 has an output impedance of 50 ohms.
So the hypothetical 3.3v square wave voltage source inside the Si5351 is driving
50 ohms internal plus your 50 ohm load.? So a 1.6v square wave makes perfect sense.
I agree, if you are seeing a 1.6v pk-pk square wave into 50 ohms, then that is 11 dBm.
Some of that energy is in the harmonics, but we're in the correct ballpark.
When a mixer says it wants 7 dBm, I believe that's defined as a source that can
drive an 11 dBm sine wave into a 50 ohm resistive load, even though the non-linear
diodes will make that waveform very un-sine-like in both voltage and current.?
Whether a diode ring mixer is better off with?a square wave is perhaps up for debate,
There is only one pin for the 3.3v rail to power the output buffers on the Si5351,
so with two or three outputs loaded up like that you may get worse results.
Your results justify the uBitx approach of driving a resistive pad directly from
an Si5351, and the uBitx does seem to work well enough.? Though it has its birdies.??
Jerry, KE7ER
|
Mike,
I haven't made any such measurements, so you're ahead of me on that.
I have heard in other forums that crosstalk and perhaps other forms of distortion
creep in as the Si5351 output buffers get loaded up beyond a few ma.
With capacitive coupling, the Si5351 is? effectively driving your
50 ohm load into 3.3/2=1.65 volts, which reduces the currents by half.
So if you see 1.6v pk-pk, that's 0.8v/50ohms=16ma, which is not just too far
from the programmable 8ma max for the output buffers of AN619.
But those CMOS buffers are working awfully hard.
The datasheet says that the Si5351 has an output impedance of 50 ohms.
So the hypothetical 3.3v square wave voltage source inside the Si5351 is driving
50 ohms internal plus your 50 ohm load.? So a 1.6v square wave makes perfect sense.
I agree, if you are seeing a 1.6v pk-pk square wave into 50 ohms, then that is 11 dBm.
Some of that energy is in the harmonics, but we're in the correct ballpark.
When a mixer says it wants 7 dBm, I believe that's defined as a source that can
drive an 11 dBm sine wave into a 50 ohm resistive load, even though the non-linear
diodes will make that waveform very un-sine-like in both voltage and current.?
Whether a diode ring mixer is better off with?a square wave is perhaps up for debate,
There is only one pin for the 3.3v rail to power the output buffers on the Si5351,
so with two or three outputs loaded up like that you may get worse results.
Your results justify the uBitx approach of driving a resistive pad directly from
an Si5351, and the uBitx does seem to work well enough.? Though it has its birdies.??
Jerry, KE7ER
toggle quoted message
Show quoted text
On Tue, Dec 17, 2019 at 09:23 AM, Michael Maiorana wrote:
I'm just looking for a sanity check (please).
?
Square wave output from an Si5351 is 3.2V p-p unloaded. Into a 50 ohm resistive load it drops to 1.6V p-p. Average voltage of a square wave is 1/2 p-p value, so Vavg is 0.8V. Power into the resistor (assuming a pure square wave) is (0.8V)^2/50 or 12.8mw. Converting to dBm 10*(Log(12.8/1)) = 11dBm.
?
11dBm is more than enough to meet the 7dBm drive requirements of a diode ring mixer. However, I'm reading conflicting info on the Si5351's ability to directly drive a diode ring mixer.
?
Any advice or guidance will be greatly appreciated.
?
These little Si5351 demo boards from Ebay are amazing for the measly $3.50 that they cost. They have onboard linear LDO regulator and logic level converters so you can drive them direct from an Arduino running at 5V. 4 wires connected to an Uno and some sample code and I was making clocks in less than 10 minutes. What an age we live in!
?
Mike M.
KU4QO
|
When building the DC40, Robin AC7LX recommended a 3db pad, and I implemented that. Works great. I wrote up my pad here:
?
There's info for doing your own using parts on hand :)?
On Tue, Dec 17, 2019 at 9:23 AM Michael Maiorana < zfreak@...> wrote: I'm just looking for a sanity check (please).
Square wave output from an Si5351 is 3.2V p-p unloaded. Into a 50 ohm resistive load it drops to 1.6V p-p. Average voltage of a square wave is 1/2 p-p value, so Vavg is 0.8V. Power into the resistor (assuming a pure square wave) is (0.8V)^2/50 or 12.8mw. Converting to dBm 10*(Log(12.8/1)) = 11dBm.
11dBm is more than enough to meet the 7dBm drive requirements of a diode ring mixer. However, I'm reading conflicting info on the Si5351's ability to directly drive a diode ring mixer.
Any advice or guidance will be greatly appreciated.
These little Si5351 demo boards from Ebay are amazing for the measly $3.50 that they cost. They have onboard linear LDO regulator and logic level converters so you can drive them direct from an Arduino running at 5V. 4 wires connected to an Uno and some sample code and I was making clocks in less than 10 minutes. What an age we live in!
Mike M. KU4QO
-- Ryan Flowers W7RLF
https://miscdotgeek.com
|
Bob N1RC
Don VE3IDS
Nick Kennedy