Babar Latif wrote:
-------------------------------
How about making multitones as of DTMF from 8051.

- - - - - - - - - - - - - - - -
MICRO GENERATED DTMF:
====================
We did that at Tandy 15 years ago. We designed for such high volumes that every penny we saved in the design, was worth $50K of profit over the sales lifetime of the product. DTMF chips were therefore outlawed and considered a waste of money. Instead we used 4 pins of the micro connected to a resistive ladder to generate 16 level DAC output. An analog engineer converted that into the telco line through a few cheap components.

The 16 bit patterns at the 4 output pins gave us the ability to synthesize by firmware, two blended sinusoidal waves (DTMF) with a little external filtering.

The trick is the generation of the data tables. As I recall, all tones are made by a matrix of 4 by 4 frequencies. By making independent tables for each 8 frequencies, you can create a blended tone for the 16 patterns represented by the intersections of the 4x4 matrix. Most people find the proper creation of those 8 tables non-intuitive and are usually extremely convinced that their mistaken approach is correct. If you've got good mathematic instincts, you might see the true solution.

AN EXAMPLE OF THE PITFALLS:
==========================
When I used this at Tandy, I was told to consult someone that had used the technique the previous year and was their expert on the subject.

He was no expert but was instead what I secretly termed "an engi-NEARLY" which at Tandy was either a technician promoted to engineer title or a engineering school 2nd year drop-out. Tandy allowed them to design products through what appeared to be a mixture of borrowing ideas from ap-notes and trial&error; an example of foolharty Tandy management.

The guy told me he used the technique the year before in a 1200 baud modem design and he spent 4 months of tracking down problems in their tones. They had to keep adding wave shaping circuitry and never really liked the results. He advised me to expected at least 4 months of very frustrating work.

He gave me a diskette containing Motorola's BASIC program for generating sinusoidal data tables for their micro (pre-Spreadsheet method). I converted the BASIC program for my TI micro characteristics and spent he rest of the afternoon trying out different combinations of all eight frequency table lengths to get them relatively prime.

The next day the enginearly stopped by my office to tell more stories of his horrible experiences with DTMF and I told him I was done.

Of course he didn't believe it and thought I was a fool. As he challenged my statement and looked at my work, I understood why he had wasted 4 months of effort chasing an elusive problem - he didn't understand the principle or advantage in making your independent wave table length relatively prime. I gave him a quick refresher on least common denominators and why relatively prime table lengths for any two DTMF component frequencies would better synthesize the required tone without so prominent a beat frequency of interference that he experienced.

He was skeptical but finally quiet. The analog enigineer assigned to my project had suffered through the four months with him and didn't have any reason to think I could make those mysterious microprocessors do things someone else couldn't. The both of them went to talk to management and warn them of the disaster brewing in my lab. :)

We arranged a quick test circuit and the following day we ran my tables with a chorus of non-believers huddled in his office looking at the telco analysis equipment and oscilloscopes.

Of course it ran perfectly. The tones were very pure and the analog engineer even started pulling components out of his filtering circuit to see how much he could reduce it. We never had any trouble with DTMF thereafter.

This typifies a problem with Tandy Management back then. Tandy paid enginearlies back in 1986 from $14K to $26K while they were paying me $55K. Instead of saving money on cheap labor, they generally lost money on the deal. In fact there were so many such silly cases of wasted effort around that design lab that I was motivated to do what I stated I'd never do....

Go back to college for an MBA degree to replace the fools running that department. :)

Unfortunately, Tandy sold off their design company while I was in university and fired almost everyone. When I graduated I tried to contact Tandy to send a resume and I couldn't find anyone I knew still working there. :)

Well I guess Tandy found one way to solve their design problems... heheheheeh. And you wouldn't believe me if I told you who in that lab is the only one they kept. :)

DTMF TABLE ASSISTANCE:
=====================
If you choose to use tables to synthesize your DTMF waves, I can assist. I'm not an analog engineer though and can't help you with the filtering or telco stuff.

However this is a really busy and wonderful time in my company and I have a few 8051 emails backed up awaiting my replies. Please be patient as it may take a few days for me to get back to anyone in email.

-aka Jay C. Box