Re: New Monitor: #2 CIRCUIT FOR REVIEW

It is I, the mighty Circuit Shredder, returned for a rematch. So, let's see
where the nasty surprises are this time.

First of all, I know you all want it, but stay away from switching power
supplies for a vector monitor. Switchers work great with stable loads but have
one hell of a time when the loads keep changing (exactly what a vector monitor
does). With very careful design a high-frequency switcher could be made to
work, but design of one is not for the faint of heart.

Ignore sense resistor inductance at your own peril. Or better yet, look at a
game where somebody has put in normal wirewound resistors in something like a
Tempest. See all the little jaggies in the lines? There is nothing that says
you can't use 10 10 ohm 1 watt non-wirewound resistors in parallel you know.
You even spread the heat out better that way as a bonus.

I know the spec sheet says you don't need a fuse, but put one there anyway. You
don't need a shorted IC (it's not a matter of IF, it's a matter of WHEN) burning
up a yoke. Or at least getting the yoke wires hot enough to shatter the CRT. I
know it's far-fetched, but PLEASE keep the fuse.

Mute pin (pin 8) can't be tied directly to the negative supply, there must be
resistance there. Say 22 k ohms or so. This pin is a current (not voltage)
driven pin. 22 K ohms will give you a touch over 1 mA. Current has to be over
.5 mA, but the graphs all stop at 10 mA (all bets are off when you leave the
graph).

The power supply .1 uF capacitors will need at least some heavy electrolytics
across them (100 uF or more). All of these will need to be VERY close to the
amplifier pins.

R1 and R7 need to be VERY close, preferably the same pad group. If the power
supply caps and IC ground connect to the same group it's better yet.

For the real purists out there, the resistance on each of the input pins of the
amp should be close to the same. Practice is just too picky for words, but has
to do with thermal stability (amongst other things). And get the value of R2
down. This beast driving an inductive load is going to be hard enough to
stabilize without unexpected phase shifts from high value resistors.

That's enough of that, I guess. The parts you have marked as "none" are going
to be real interesting to pick. You might want to use the original monitor
board design as a guide for at least a starting point since these are very
dependent on yoke inductance. OR find an engineer that know his poles from his
zeroes to calculate them for you. I never was any good at the pole/zero stuff,
but without it it will be hard to get things stable. I suppose you could dump
it all into a Spice simulation and see what flies. At least some good starting
numbers could be arrived at that way.

James Nelson wrote:

> Hi everybody!
>
> This deflection circuit design is very simple, and that's the beauty of it
> :-)
>
> The LM3886 has all kinds of protection built in to it, so many of our past
> concerns are finally addressed, such as overvoltage, undervoltage,
> overloads, shorts, thermal runaway & instantaneous temperature peaks!
>
> I will probably need to design an on-board switching power supply and a
> decent spot killer (I'm not sure if it justifies a microprocessor at this
> point) as well as connectors for hookup. There may be a special oscillator
> for Sync inputs on the conversion monitor. This would stabilize the High
> Voltage section.
>
> http://www.angelfire.com/nh/northamericantelecom/images/deflectionamp02.gif
>
> Don't mind the wrong symbol for the yoke.
> Fuses are not required due to the protection inherent to the 3886.
> I believe the sense resistor inductance, in this design, is unimportant due
> to yoke inductance being orders of magnitude more significant. 1-10 uH -vs-
> 1 mH.
>
> Please keep in mind, I'm trying to keep the parts cost to an absolute
> minimum so the max people can afford it. It also means, hopefully, that I
> won't have to invest $5000 in a parts stock just to have 20 boards on the
> shelf.
>
> Design comments, complaints and criticisms, as usual, are requested.
>
> Thanks everyone!
>
> James
>
> http://www.angelfire.com/nh/northamericantelecom
> low cost regular phone service click links to 6.9c or 5.9c no monthly fees!
Received on Mon Nov 8 01:43:17 1999