Crystal frequency and broadcast signal quality
Phil Pilgrim (PhiPi)
Posts: 23,514
I've been vacillating these past few weeks about which crystal to use with a new Propeller board I'm working on. 10MHz crystals are nice, because they can be obtained in small SMD packages; but this frequency is outside the published specs for the Propeller when using the PLL (notwithstanding that it's used with the Spin Stamp). 5MHz crystals are within spec, but small SMD packages for this frequency are well nigh impossible to source. The board is big enough (barely) to accommodate a 5MHz crystal in a through-hole HC49US package, so I've had an opportunity to try both frequencies.
My experiments with the 5MHz crystal (ECS-50-20-4X 5.0) used the following settings:
The 10MHz crystal (FOXSLF/100-20) settings were:
So, in both cases, the system clock frequency was 80MHz.
Digital performance-wise, at room temperature, they're equal. This also extends to baseband video, where stability and color fidelity are indistinguisable between the two. Where I saw profound differences was in broadband video. With the 5MHz crystal, reception on channel 3 was clear, well-synced, and with good color fidelity. With the 10MHz crystal, reception was, in a word, crappy. Color hues wandered through each horizontal line, sync was hit-or-miss, and there were a couple ghostly vertical lines along the left border.
To investigate this further, I looked at the frequency spectra of the two signals, while displaying a blank screen and modulated with a mute 4.5MHz audio subcarrier. Images of these are attached below. As the spectra show, the 5MHz crystal produces the cleanest 60MHz carrier frequency. So I guess I'm stuck with it.
-Phil
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
'Still some PropSTICK Kit bare PCBs left!
Post Edited (Phil Pilgrim (PhiPi)) : 7/17/2008 4:23:32 AM GMT
My experiments with the 5MHz crystal (ECS-50-20-4X 5.0) used the following settings:
_clkmode = xtal1 + pll16x _xinfreq = 5_000_000
The 10MHz crystal (FOXSLF/100-20) settings were:
_clkmode = xtal1 + pll8x _xinfreq = 10_000_000
So, in both cases, the system clock frequency was 80MHz.
Digital performance-wise, at room temperature, they're equal. This also extends to baseband video, where stability and color fidelity are indistinguisable between the two. Where I saw profound differences was in broadband video. With the 5MHz crystal, reception on channel 3 was clear, well-synced, and with good color fidelity. With the 10MHz crystal, reception was, in a word, crappy. Color hues wandered through each horizontal line, sync was hit-or-miss, and there were a couple ghostly vertical lines along the left border.
To investigate this further, I looked at the frequency spectra of the two signals, while displaying a blank screen and modulated with a mute 4.5MHz audio subcarrier. Images of these are attached below. As the spectra show, the 5MHz crystal produces the cleanest 60MHz carrier frequency. So I guess I'm stuck with it.
-Phil
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
'Still some PropSTICK Kit bare PCBs left!
Post Edited (Phil Pilgrim (PhiPi)) : 7/17/2008 4:23:32 AM GMT
800 x 300 - 33K
Comments
·
I'm curious what a 6 MHz crystal would do for you?·· ... 5 MHz and 10 MHz are going to have similar characteristics because of their divisibilities of one another, and the relationship to your desired frequency.· 6 MHz might be a better fit for broadcast TV.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe
IC Layout Engineer
Parallax, Inc.
You're right about 6 MHz being a better fit for broadcast TV, since the channels are at 6 MHz intervals. Because this is meant to be a commercial product, though, I'm reluctant to push the system clock to 96 MHz. The results I'm getting at 60 MHz (channel 3) with the 5 MHz crystal are adequate, and I don't care so much about channels 2 and 4. It's just the fact that I have to use a thru-hole crystal to fit the PCB form factor that's a bit nettlesome. I suspect that the Prop's PLL is less stable with a 10 MHz input, and that that's what's causing the decline in broadcast carrier quality with the higher frequency crystal.
BTW, in an earlier post, you published an LC circuit for boosting broadcast range. Do you have any recommendations for how to implement it? I'd like to be able to build such a circuit within the confines of a 1/8" monaural phone plug, which also serves as the base for a whip antenna. But actual dimensions and number of coil turns are a mystery. It would be nice to be able to base load the antenna, too, so it doesn't have to be a full 1/4 wave in length.
Thanks,
-Phil
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
'Still some PropSTICK Kit bare PCBs left!
I just thought I might share our experiences with 6MHz crystals...
We have seen absolutely no issues with our Hybrid and PropGFX Lite which both run at 96MHz with a 6MHz crystal.
Even on a long term soak test at higher temperature ranges and with all cogs running at full pelt we haven't seen any issues at all.
The same cannot be said however using higher speeds where we have had mixed results.
I know Hippy had had some success with 7.3728 MHz but we have found this unreliable when pushing the prop to extremes.
Regards,
Coley
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
PropGFX Forums - The home of the Hybrid Development System and PropGFX Lite
Thanks for sharing your experience. The high temp tests are particularly reassuring.
But, checking with DigiKey, it doesn't look like a 6 MHz crystal will buy me anything over the 5 MHz device. The only 6 Mhz SMD units available have bigger footprints than that of HC49US thru-hole. (It kinda makes me wonder how they're able to make 32.768 KHz crystals so dang small, when anything in the MHz range under 10 MHz or so has to be so large.)
-Phil
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
'Still some PropSTICK Kit bare PCBs left!
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
The circuit is about as simple as it gets... a series inductor and capacitor.... the I/O pin goes to the CAP, while one side of the inductor goes to ground.· The point where the Cap and Inductor are shared go to your antenna.
1/32th wavelength = approx. 6 inch antenna
A 1uH inductor with a variable inductor ranging from 4pf to 10pf cap will provide a tunable range of 80MHz to 50MHz.
A 7pF cap with a variable inductor ranging from 0.5uH to 1.5uH will provide a tunable range from 85MHz to 49MHz.
If using a fixed inductor, I would try to find an air-core design as it will tend to have a higher "Q"
Also, take into consideration how much capacitance the circuit·board contributes.
If you wanted to try rolling your own inductor.... (These are air-core calculations)
8 turns of #26 wire on a 1/4 inch form will give you about 1uH· (1.08uH)
7 turns of #28 wire on a 1/4 inch form will give you about 1uH· (.93uH)
7 turns of #30 wire on a 1/4 inch form will give you about 1uH· (.98uH)
7 turns of #32 wire on a 1/4 inch form will give you about 1uH· (1.02uH)
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe
IC Layout Engineer
Parallax, Inc.
-Phil
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
'Still some PropSTICK Kit bare PCBs left!
http://www.pulsedpower.net/Applets/Electromagnetics/RfCoil/RfCoil.html
That's·the same formula that I used... In my calculations however,·H and P are determined·based on the diameter of the wire.
Attached is a coil calculator Excel spread-sheet that I·use... it's work in progress, but accuracies are better than 5%
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe
IC Layout Engineer
Parallax, Inc.
Post Edited (Beau Schwabe (Parallax)) : 7/17/2008 8:17:47 PM GMT