Re: Specifications
27 views
Skip to first unread message
GLIC
Dec 21, 2010, 9:22:53 AM12/21/10
to Open counter
Some comments interspersed.
>Initial Spec:
>Powered from external 12-24Vdc supply.
I understand that each board will have its own power supply, derived
from this one.
>Runs stand alone or via character based command line interface
>available via serial (RS 232) or ethernet (telnet
or web
>) (Choice TBD by
>builder). Internal interface is TTL serial. Interface slot can hold
>one card providing one of:
> RS232 level shifter,
> USB<-> Serial interface (FTDI or equiv)
> Serial <-> ethernet card (telnet interfce)
> Serial <-> ethernet card (http interface).
> GPIB controller
I built a web interface for a PTS synthesizer. I regret the decision.
It is not practical. However, the decision is up to the builder.
>Uses internal timebase. Low cost option is xtal bases osc module. TCXO
>option or external 10mhz reference standard, sine or ttl as
>alternative to internal timebase.
I favor a cheap xtal - most time-nuts will use a house reference.
>Four inputs: External timebase (above), Arm, A & B.
> Arm input may be used to disable either/both inputs.
>Inputs are ttl compatible,
at 20 GHz??? AFAIK, the limit is around 1 GHz (http://
www.potatosemi.com/). I don’t have any experience with their products,
nor with prices.
Instead, some flavor of PECL?
>rising edge triggered. Later additions may
>include input filtering such as: AC/DC coupling, 50ohm/1Mohm
>impedance, variable attenuation, prescalers, trigger on edge or level,
>over voltage protection, <add list here>
>Provides the following functions:
> Time interval A->B
> Frequency A
> Width A
> Events A during B
> Period A
> Count A
> Count A +/- Count B
Perhaps a bit too much for one module. I suggest a basic time interval
module, with additional functions as options.
>Possible additional functions:
> Rise/fall time A
Use a scope…
> Average of any of above
> Ratio A/B
Statistical functions should also be considered as options
>Goals:
> Frequency: 20Ghz w/o external prescaler.
It sounds a bit too high. There are cheap components capable of
handling 10-12 GHz, and that cover the 10GHz ham band. Higher
frequencies may also have problems with FR4.
> Time Interval resolution: 25ps
> Build cost < $200 not including power supply, case, or interface
>other than RS-232 for initial run of 25 units.
Quite optimistic, but I hope you achieve it.
> Use only SMT, SMD packaging. Avoid through hole when possible. No
>BGA packaging. Must be assemble-able by skilled amateur with soldering
>iron, 20x magnification, solder/solder wick, other common tools.
OK for leaded components, but QFN is really hard so solder in a
repeatable way. Please restrict QFNs to one or two per board. Size
0603 is doable, but seems to be the limit.
> All components must be available internationally. Vendors who will
>sample parts are preferred.
> Built with conventional, 2 layer pcb wherever possible.
Four layers are not that expensive. BatchPCB do prototypes at $8.00
per square inch for 1 unit, but usually you receive more - in my last
order of one board I received five. For 10 GHz, it seems unavoidable.
>Design guidelines:
> Modular design. Separate control and front panel from core
>counter, counter from input. Modules interconnect with commonly
>available connectors such as 0.1" space inline.
Don´t forget space for coaxial connectors for high frequency and noise
sensitive signals. Boxed modules are a good idea. Hammond prices are
too high, but there are cheap Chinese alternatives: http://www.rfsupplier.com
has boxes in the range of $6.00, with reasonable sizes. I have some,
and quality is good.
> Code for internal processors to be written in C or VHDL
> Counting should be direct where possible, via hardware division
>where possible, interpolation elsewhere.
> Schematic capture and PCB design with EAGLE.
I use tinyCAD and FreePcb, with good results. They are open source, no
restrictions on size or number of layers.
> PCB's to be made available through group buy, parts list to be
>available on group, build instructions on group.
And yes, I am/was thinking of undertaking a project like this myself.
>Initial Spec:
>Powered from external 12-24Vdc supply.
I understand that each board will have its own power supply, derived
from this one.
>Runs stand alone or via character based command line interface
>available via serial (RS 232) or ethernet (telnet
or web
>) (Choice TBD by
>builder). Internal interface is TTL serial. Interface slot can hold
>one card providing one of:
> RS232 level shifter,
> USB<-> Serial interface (FTDI or equiv)
> Serial <-> ethernet card (telnet interfce)
> Serial <-> ethernet card (http interface).
> GPIB controller
I built a web interface for a PTS synthesizer. I regret the decision.
It is not practical. However, the decision is up to the builder.
>Uses internal timebase. Low cost option is xtal bases osc module. TCXO
>option or external 10mhz reference standard, sine or ttl as
>alternative to internal timebase.
I favor a cheap xtal - most time-nuts will use a house reference.
>Four inputs: External timebase (above), Arm, A & B.
> Arm input may be used to disable either/both inputs.
>Inputs are ttl compatible,
at 20 GHz??? AFAIK, the limit is around 1 GHz (http://
www.potatosemi.com/). I don’t have any experience with their products,
nor with prices.
Instead, some flavor of PECL?
>rising edge triggered. Later additions may
>include input filtering such as: AC/DC coupling, 50ohm/1Mohm
>impedance, variable attenuation, prescalers, trigger on edge or level,
>over voltage protection, <add list here>
>Provides the following functions:
> Time interval A->B
> Frequency A
> Width A
> Events A during B
> Period A
> Count A
> Count A +/- Count B
Perhaps a bit too much for one module. I suggest a basic time interval
module, with additional functions as options.
>Possible additional functions:
> Rise/fall time A
Use a scope…
> Average of any of above
> Ratio A/B
Statistical functions should also be considered as options
>Goals:
> Frequency: 20Ghz w/o external prescaler.
It sounds a bit too high. There are cheap components capable of
handling 10-12 GHz, and that cover the 10GHz ham band. Higher
frequencies may also have problems with FR4.
> Time Interval resolution: 25ps
> Build cost < $200 not including power supply, case, or interface
>other than RS-232 for initial run of 25 units.
Quite optimistic, but I hope you achieve it.
> Use only SMT, SMD packaging. Avoid through hole when possible. No
>BGA packaging. Must be assemble-able by skilled amateur with soldering
>iron, 20x magnification, solder/solder wick, other common tools.
OK for leaded components, but QFN is really hard so solder in a
repeatable way. Please restrict QFNs to one or two per board. Size
0603 is doable, but seems to be the limit.
> All components must be available internationally. Vendors who will
>sample parts are preferred.
> Built with conventional, 2 layer pcb wherever possible.
Four layers are not that expensive. BatchPCB do prototypes at $8.00
per square inch for 1 unit, but usually you receive more - in my last
order of one board I received five. For 10 GHz, it seems unavoidable.
>Design guidelines:
> Modular design. Separate control and front panel from core
>counter, counter from input. Modules interconnect with commonly
>available connectors such as 0.1" space inline.
Don´t forget space for coaxial connectors for high frequency and noise
sensitive signals. Boxed modules are a good idea. Hammond prices are
too high, but there are cheap Chinese alternatives: http://www.rfsupplier.com
has boxes in the range of $6.00, with reasonable sizes. I have some,
and quality is good.
> Code for internal processors to be written in C or VHDL
> Counting should be direct where possible, via hardware division
>where possible, interpolation elsewhere.
> Schematic capture and PCB design with EAGLE.
I use tinyCAD and FreePcb, with good results. They are open source, no
restrictions on size or number of layers.
> PCB's to be made available through group buy, parts list to be
>available on group, build instructions on group.
And yes, I am/was thinking of undertaking a project like this myself.
Reply all
Reply to author
Forward
0 new messages