DDS Evaluation Board from Analog Devices

on:

on

on:

That chip seems to have some sort of differential current mode output, would that be useful to you? I'd try to put some sort of buffer on there to get the usual voltage output into 50 ohms. It would also protect the chip from potential damage. I note the crunky parallel port on there, I would prefer USB, but what can you do? Kind of weird for a chip that has a serial interface.

To me it seems like you'd learn a lot and end up with a quite handy device.

I find the price too high though. AD usually includes the layout in their datasheets. Just toss the file to Olimex and get a board made for 35$ and buy the chip for 10$.

on:

on

on:

I noticed that. I probably would make an output section suitable for the work I do. Op amp otta do.

I suspect the software programmer found parallel port interfacing easier than the USB interface in Windoze XP.

I don't think the artwork drawing at the bottom of

Yup...I wish it was cheaper..like $50.00! :) A stable accurate PC controlled 37Mhz sine/triangle sweep generator for $50.00 would be wonderful. :)

D from BC

...

I think your earlier idea of actually rolling one makes far more sense. Here you are definitely in the ballpark of "as much money spent as a standalone", it's not a complete unit, (slap an op amp on may not be quite enough for quality output drive over the full frequency range) and you are at the mercy of keeping a computer with centronics port running

- not a big deal now, but in 5-10 years? They are dropping off the newer computers.

Design something with USB or Ethernet interface, and make it good enough to sell - then you should be money ahead. Otherwise, your $158.61 Canadian will go a long ways towards getting you a fully functional, fully built unit from either Asia or *B*y (or from Asia via *B*y).

Personally, I'd say more trouble than they're worth. You buy the board, then still have to add on a fast differential amplifier, then (most importantly) a decent low pass filter and then probably another amp/buffer. End up with another PCB just to get the sine out and usable. What's never mentioned in the datasheets is how impossibly useless the output waveform is at maximum operating frequency. That "37.5MHz" is theoretical only. For good useable sines, something like 1/8th the clock frequency should be considered a max (eg 10MHz). Triangle, what is that used for?. Forget the square output as well. The basic DDS generating mechanism becomes so visible on squares that a 'scope can't even lock on to some particular frequencies. (observations based on a slower AD9833 I though would be OK as a function gen') Would suggest getting any serial type (clock data) DDS chip and wiring it yourself. Only needs 3 wires from the LPT and the frequency programming you can do yourself. Software will probably be more reliable than what AD supplies!.

On the + side.... I have a computer here with a useless parallel port. So, it can be dedicated to interfacing with the DDS.

On the - side... Yup...that $158 is in the ballpark of most EEbbaayy generators which are complete. This is a spoiler. :(

It's now a decision of make my own PCB + add what's missing or get something off Eebaayy.

D from BC

Get something off eBay. My strategy is to have the patience of a spider. Spin a web of email warnings and wait for a good price. Don't wait to actually *need* the device. If you buy when you don't need it, you

I have a lab of stuff just waiting for me to actually build stuff again.

Yup.. I'll still be making a board for an output amplitude control section. So..I might as well do the PCB for the DDS chip anyways. I don't even need a connector on the PCB. I can hack off some old parallel cable and solder it direct.

I was thinking of using the triangle wave to make a variable duty square wave generator section.

I am thinking about the filtering.

D from BC

I believe this is in the chip already. There is a comparator. (going from memory here).

I think this will be the most work...

Ahh yah..it's nice to play on Eebayy without desperation :)

Well...the AD eval board was tempting...

AD Eval Board |Ebbaayy

---------------------------+------------------------- I know what I getting. |It could be broken. Negligible shipping cost(1)|Around $65.00 by USPS Missing electronics |Complete Needs R&D time |Ready to use Needs benchtop PC |No PC required New electronics |Possibly old electronics (1) It wouldn't change my shipping cost if the eval board was included in my usual Digikey parts order.

D from BC

Any news on your DDS project? I'm building something similar, see my first try:

You can get inexpensive FPGA development boards, like this for about $60:

RS232 is already integrated.

I've used another board, because I've used it for other projects already and it has a DAC integrated:

The advantage compared to the Analog Devices chip is that you can do more with it, e.g. generating digital patterns, controlled by the PC over the RS232 port etc., without the need for additional chips.

Adding something like the AD5445 provides a 20 MHz, 12 bit, analog output (but a fast opamp after the DAC would be a good idea, like described in the application note). I think all parts, including some housing, would cost less than $100 and you can program any special functions you need in the FPGA, which is impossible if you buy something ready-made (or at least costs much more).

I put the DDS project on hold until I finish my other projects.

I was interested in comparing: A DIY DDS poorman generator versus spending ~$300.00 on some possibly broken E*bay generator made in ~1992.

D from BC

I just read in a German newsgroup about this project:

A really inexpensive, but good DDS generator, if you don't need frequencies higher than 200kHz.

You should look into DSP filtering. There are a lot of better ways to upsample your sparse sin table than linear interpolation.

Yes, but I wonder if I need it. I could do cubic interpolation, but if I have 12 bits output, only, it will be better just to store as many samples with 12 bit width as needed to avoid more than one LSB change in adjacent values. This would give perfect results without any interpolation at all.

But if I want 24 bit output, it might be more accurate to use something like CORDIC to calculate the values, or polynomial approximation, if I have some hardware multiplier or if I use a fast DSP, because interpolation produces not mathematically exact sine values.

Here is the new version of my DIY generator (ok, you'll need the Spartan3E starter kit, too, but at least all the VHDL and C code is DIY)

Now you can compare it to the $300 generator :-)

Interesting... Originally, I was thinking of just downloading the control app from analog devices and building my own DDS board based on an Analog devices DDS chip.

But source code is better to customize and augment.

D from BC