I programmed Spartan chip to produce serial data ( main_Clock, Data
Clock, Data and Tag ) stream for a DAC. I need to make this system
work with out wires. The main clock is 3MHz, Data clock is 1.5MHz. Can
anybody advice me how to make it digital wireless ( Transmitter plus
receiver). I have not done anything like this before. The DAC will be
at 6 feet away from the Spartan chip.
The best way is to modulate your signal at 5.8GHz and embed the phase
correction data into the data stream. I assume the data flow is one-
Yes, The data is flowing in one way. Would you please suggest some
hardware. I am also thinking about bluetooth. Please advice!
Bluetooth will not give you 1.5MHz data rate. In fact, 5.8GHz might
not be enough, but the next ISM band is 25GHz.
The Atmel ATR2820 RF Transceiver would be a good start, but I can't
design your apps without knowing the full spec.
One-off project or a few dozen or a few hundred thousand? Per week?
A hobby/learning/grow-your-brain project or for sale (and use) by a
Will UL/CSA/EC/FCC/etc. certification be required?
Are the consequences of a communications failure: none, benign, pretty
bad, or someone dies?
Is this for a medical application (noting that your IP block is from
Wayne State University Medical Center)?
With that out of the way, if you can grind down the data rate a bit
(no pun intended), you're close to being able to use an off-the-shelf
solution like an XBee module, advertised to top out at 1 Mbps.
Rich Webb Norfolk, VA
Take a look at the RF data rate of 250Kbps. Does it mean the same for
me as for you?
What do you think about Texas Instrument's IEEE 802.16 (WiMax) chip
set ( TRF1X/2X RF Chipset ).
> set ( TRF1X/2X RF Chipset ).
> Please advice!
Forget about Bluetooth, WiFi, XYZBee, WiAnything. If you want
slicings or channelings, you need 25GHz. To get 1.5MHz data rate, you
need the full bandwidth for 5.8GHz.
Why do I need the bandwidth of 5.8GHz. Please advice!
Radio Freguency Modulations and Demodulations.
How many bits at the 1.5MHz rate? In other words, how many Mbps?
Regular old 802.11g wireless routers will get you >50Mbps; channel bonding
(40MHz occupied bandwidth) gets you up to >80Mbps in the best cases... which
is just as good as anything you get on 5.8GHz WiFi. (These are all *average*
throughputs, taken from the "wireless charts" at www.smallnetbuilder.com...)
48 bits at 1.5Mhz rate?
Never mind, Bluetooth 2.0 would be fine. I thought you mean 1.5Mbps
You don't need 5.8 GHz of bandwidth, and no one has said that you do. But
because you asked to be able to transmit continuously at 1.5 Mbps, you are
going to need a high carrier frequency. Your responses to various replies
indicate you really don't understand what you are asking for. So if this is
anything but a hobby project, I'd "advice" you to hand this task to someone
with RF experience. What you are trying to do is non-trivial because you
want the equivalent of an unswitched "coax cable in the sky". Most chipsets
that are available assume a much different environment, where data is
packetized, pipeline delays are acceptable, retransmissions on errors or
lost packets are allowed, channel equalization is dynamic, data travels in
two directions (at least for acknowledgements) etc. Accounting for the
overhead of these other things so you can use some vendor's OTS chip set
will only increase your total bandwidth requirements. I think you'll need a
custom solution or else you'll have to scale back your requirements.
That's 72Mbps, so you can just manage it with the best WiFi routers out there.
Go visit smallnetbuilder.com for the particular models...
Not sure if he meant 48 bits per second, per millisecond or per
> That's 72Mbps, so you can just manage it with the best WiFi routers out there.
Not so, Wifi are rated at burst data rate, not sustained rate.
Sure, the advertised data rate is something absurd like "300Mbps!" But go
read the link I provided -- the best routers absolutely do better than 70Mbps
*on average*. Obviously the speeds gets reduced as you move further from the
router or have other interferers that reduce SNR, but close-in something like
a Belkin N1 will do the job (see:
http://www.smallnetbuilder.com/content/view/30238/96/) for under $200.
And you probably want his spartan FPGA to talk ethernet as well.
Yeah... slapping down some gigabit Ethernet reference design is one heck of a
lot easier than slapping down a high-speed wireless router design. If you
look at the reviews, you can see that what often makes the difference between
"solid and fast" vs. "flaky and slow" routers isn't the actual hardware --
which is often very similar, if not identical -- but how much tweaking
occurred in the radio controller firmware. Unless I'm building a radio from
scratch, I'd prefer to let someone else worry about that. :-)
Sure, if we are looking for a $200 solutions, we can build multi-TX,RX
radios. We don't have to deal with channels and/or routings as well.
Yep. Too bad the O.P. here seems to have disappeared!
So, I need customized hardware to do this. Are you guys sure that
there are no chips available to do what I want to do? Any suggestions
or pointers that how should I proceed. I have FPGA generated clock and
data stream . Data is 48 bits wide and I am serially outing the data
So you are dealing with 1.5Mbps. In theory, 5.8GHz 802.11n can
handle it. But with all the overheads of channellings and routings,
it will not guarantee the data rate. Your FPGA would have to talk
ethernet/TCP/IP in this case. You can bump it up with Multi-
Transceivers as another poster suggested, but it will not be cheap.
Your best bet is to do a simple modulation/protocol at 5.8GHz.
What's the actual required throughput? Could you bring the 48 bits
into a FIFO and send them out at a slower (cheaper, easier to find)
Okay , What is the slower , achievable rate? I can try to slow the
data rate down to that value
6 feet? if you can use line of sight, consider optical. You can
serialize your 48 bits (I guess that would be 72 Mbps) and modulate
your optical transmitter with that, de-serialize at the other side.
Why can't you use a wire again?
My understanding is that the internal data path is 48 bits, but the
serial clock rate is 1.5MHz or 1.5Mbps. Either way, he need a custom
solution. If I am right, it would be a cheap custom solution. If you
are right, it would be an expensive custom solution.
Yep, it looks like 1.5 Mbps as you said. I wonder why no wires? 6 feet
looks like such a short range for wireless, especially RF wireless.
Since it's apparently only 1.5 Mbps, I'd consider something simple,
like FM carrier at 2.4 or 5.8 GHz, Manchester coded. That would double
the BW, but improve the SNR. It's still cheap and simple. I'd publish
a 1.5 Msps demodulator in Verilog for the right kind of beer ;-). The
up/downconverter is up to the OP.
He says he is "serial outing the data ay 1.5 MHz" . He needs to explain
better what he is doing or intends to do. Otherwise this thread is a
"waste of bandwidth".
Joe Leikhim K4SAT
"Treason doth never prosper: what's the reason?
For if it prosper, none dare call it treason."
"Follow The Money" ;-P
We need the OP back for sure. If he is really from Wayne State
University Medical Center, perhaps isolation issues from the operating
I am looking for a digital design or technique that I can use to
transmit the 48bit of data at 1.5MHz frequncy wirelessly. So, what I
gathered so far is that a digital frequncy modulator can be made using
a FPGA and the demodulator can also be made using FPGA. Am I right ,
if yes then i need more guidance , some literature to understand
exaclty how can I do this? The thing that I do not still understand
that If I am right about the FPGA based frequency modulator and
domodulator then how would I design the wireless part of the system?
So, is it 1.6Mbps or 48x1.5=72Mbps?
> So, what I
> gathered so far is that a digital frequncy modulator can be made using
> a FPGA and the demodulator can also be made using FPGA.
The modulator/demodulator can be configurated/controlled by the FPGA,
but you need analog RF circuits. You need to be very careful with PCB
layouts. RF PCB layouts are very tricky.
>Am I right ,
> if yes then i need more guidance , some literature to understand
> exaclty how can I do this? The thing that I do not still understand
> that If I am right about the FPGA based frequency modulator and
> domodulator then how would I design the wireless part of the system?
First thing is to pick your encoding scheme. Manchester FM sounds
good, as suggested by other posters.
Haven't you bothered to read any of the dozens of answers you've already
This is why people don't like googlies. Do you expect the answers to
just show up on your desk or something? If you ask on USENET, you get
answers on USENET.
Send 48bits of data at 1.5Mhz is either:
48bits at a time 1.5M times a second.
a 1.5Mhz carrier frequency to send 48bits.
Which is it ??
As i come at this reading the whole thread OP is looking for about 1
mile range and three data sources.
Each bit of the 48 bit data gets out of the FPGA at the rising edge of
the 1.5MHz of the Clock. I found following two transciever chips
What do guys think? Has anybody worked with these chips before?
Get a (better) news client. Even Outhouse Express (Outlook Express) is
ok by comparison. You need to have proper quoting.
Now that is off my chest, I suspect that neither of these chips will
do what you need directly.
The Nordic Semi device seems to be designed around a 100 ms sample
The fastest data transfer rate of the Micrel is 128 kb/s. Far less
than the speed it comes out of the FPGA.
What is the timing between 48-bit packets from your source again?
I still don't know the OP's requirement for data rate, it's still
ambiguous; however, I meant to use the FPGA as the 2nd stage down-
conversion and the demodulator both. The whole receiver (all COTS):
1) 5.8 GHz BPF
2) Mini-circuits mixer
3) Fixed L.O. (with amp)
4) IF amp (assume a 1st I.F of about 1.2 GHz)
5) IF filter (e.g. EPCOS) - don't use SAW unless triple-transit ok.
6) ~500 MSPS converter (with at least 1200 MHz BW)
7) FPGA: Cyclone 3, Spartan, Virtex5, etc ...
The above is for 72Mbits/s. If it's only 1.5 Mbps, then it can be a
lot simpler. Step 4,5,6 could be adapted for direct conversion, which
would be cheaper, but lower performance.
Not enough details (especially the data rate) on your first link.
Can't access your second link.
If you email them to me, I will take a look.
contact me at www dot linnix dot com
What inofmation do you require? Please let me know.
For the first link. I don't think you need the 8051, if you already
have an FPGA.
I can't read the second link, so I can't tell if it's the right chip
or not. If you have the datasheet somewhere, just email me.
I believe your original problem statement was 48bits per word,
transmitting continuously at 1.5 Mb/s. Neither chip will help you.
Your data rate is beyond what can handle. For the Micrel transmitter
you would have to cut the data rate down to 100kb/s. It may be
possible to use the WiMAX Chip but you will now eat batteries.
You should look for 1.5 to 2Mb/s chips for ISM bands, and narrow beam
John wrote on 5/12/2008
> 48 bits at 1.5Mhz rate?
Clear as mud.
After re-reading some of your posts, it seems clear that your data
rate is 72 Mbps. It would help to know if this is a one-off
(feasibility) or a design project ... Since you already have a Spartan
on the transmit side, you can use it as a modulator (e.g QAM),
followed by an up-converter, and use another FPGA after down-
conversion on the receive side. I think questions about what "chips"
to use for the RF (or optical) section might be more productive after
we know more about the cost / size and power constraints.
1.5MHz is the serial clock rate. So, the time period of this clock
will be approx. 66 ns. Now, at each rising edge of the clock, FPGA
puts a bit out and keep it at its output for one clock cycle. So, FPGA
uses 48 clock cycles to output 48 bits. So, according to my guess the
data rate to transmit one 48 bit frame is (66ns x 48 = 32usec)....The
time interval between two consective 48 bit frames can be anything...
I can transmit one after another or can have some delay between them.
Now , I need a wireless system that can transmit and recieve this
data, clock and a synch. I heard that I can use FSK technique to do
this by programming FPGA as modulator and demodulator. or I can
interface FPGA with the some RF chips that can handle this data rate.
So, I am looking into chips. I can lower the data clock to 600kHz.
The chips links that I found are as follows
This is your internal data clock. You have to buffer the data and
shift out at a lower rate.
> Now , I need a wireless system that can transmit and recieve this
> data, clock and a synch. I heard that I can use FSK technique to do
> this by programming FPGA as modulator and demodulator. or I can
> interface FPGA with the some RF chips that can handle this data rate.
> So, I am looking into chips. I can lower the data clock to 600kHz.
You have to lower it to 128khz for the Micrel chip
By 128khz, you meant the Data clock frequency. Am I right?
First you say 1.5M clock, 48 bits at 1.5 MHz, then 66nS, It's
difficult to know what you mean.... If you want to call me, I'll help
you sort it out. I can help you if you want (I have the day off), but
I think it will go 100x faster by phone.
Sorry if you are unable or unwilling to read the whole thread. It
might make better sense to you then. Not that it is all that clear in
the first place. OP has been given several offers of professional
advice (probably paid for). I have no idea if OP has pursued them.
You got it. OP does not seem to understand what the actual
Looking at this differently...
Is there perhaps a *smarter* system level way for you to achieve the
same end result?
i.e. do you have to actually output that amount of data continuously
to the DAC via the wireless interface?
Does the DAC output anything repetitive that you can automate on the
receiver side? e.g. if it's say for example a sine wave bust, can you
send a short "generate sine wave burst" "command" instead of the huge
amount of actual DAC data itself?