Picking apart a couple old schematics

[metertech58761]

Years ago, I used to repair some electronics that my work produced for a company that supplied equipment to chip foundries.

Among these circuit boards and assemblies was a driver assembly for a turntable motor. The original incarnation of of this assembly was somewhat underpowered, as it had a history of blowing out the two TO-220 MOSFETs (MGP-20N45).

The design was improved by adding a brake daughterboard (which added a third MOSFET and a 10 ohm/50 watt resistor), but it was still failure prone - and very unforgiving of certain failures (On one occasion, I managed to make a 1' tall flame shoot out of that brake resistor after forgetting to replace a 1N4005, which was nearly hidden between two MOVs). Eventually, however, the MOSFET in question was discontinued, and after supplies (and spare boards) dried up, the company finally did a respin of the board to use a pair of beefy IGBTs, and that spot on the repair shelf soon cleared out.

Some time back, long after that customer went away, I found my working copies of the speed control schematics and before tossing them, I digitized them into KiCAD. I've never understood why this board has four AND gates connected in parallel. Can anyone enlighten me?

[metertech58761]

And here's the accompanying Logic board.

While the PWM was prone to blowing up, the Logic board rarely ever gave trouble - even if all the ICs on the earlier units were in those junk Scanbe brand sockets.

It was also interesting in that it was practically self-contained in terms of how it was controlled, as all the inputs were activated by grounding them (the test box that plugged into P2 consisted of a thumbwheel switch and two toggle switches). So, I never had much need to study this board more closely, but I still wonder about things like the "mode" jumper setup.

If you can see a better way to arrange parts of this schematic, don't hesitate to pick on me and I'll see what I can do.

[Hawaka]

I've never understood why this board has four AND gates connected in parallel. Can anyone enlighten me?

On a first look I would say for added drive strength. But with 1k gate resistor not sure that’s really doing something…

[bdunham7]

I've never understood why this board has four AND gates connected in parallel. Can anyone enlighten me?

Probably because the 4081 refers to a CD4081 CMOS AND gate that has an output current (source or sink) on the order of 4mA and there are 4 gates per package anyway, so wiring them all in parallel gives you 16mA or so.  You need 12mA to drive the optocoupler and you need another 24mA for a brief period to charge up the MOSFET gates.  That gate charge time may have been part of the reason they blew up.

[jwet]

I don't know why I enjoy this kind of stuff so much but I do.  I am an old Maxim FAE and am used to looking at bad designs- at least I don't have to be quite so diplomatic on line in a forum since you're not the original designer.

First of all the four AND gates in parallel are to make a low impedance driver for the FET's Q3 and Q4.  They use 4000 series logic because they can take 15v and the 4 in parallel increases drive.  4000 series CMOS could source and sink about 5 mA, so four of them could source or sink 20 mA- CMOS can be paralleled this way- sort of unique. This is a poor man's mosfet driver but not a very good one.  Using a hex inverter would have made more sense since they're burning the whole 4081, this would have given 6x vs 4x- maybe they had a bunch of 4081's somehow?  The 1k resistors on the outputs to the FET's aren't a great decision, this kind of limits the max current to about 15 mA short circuit- something about 1/4 of this value would make more sense.  The rise time won't be real fast.

The FET's are not really FET's, they're insulated gate bipolars, IGBT's, in an ancient Motorola technology called TMOS- very early trench FET's.  IGBT's are still used but only for very high voltage apps.

The electronic brake stuff uses an inverted signal through an opto of the main motor PWM that discharges the motor current when the PWM is off- this is kind of goofy.  Electronic braking is fine but you don't want to do it on every cycle only when you're trying to stop the motor.  The driver for the brake IGBT is another kluge mosfet drive, it uses two 5 ohm FET's to drive the brake FET. 

I don't know why there are optos in the circuit- there is really no need- this are no floating high voltage or anything- they just slow things down.  If you have  bad design, sometime optos can help as they isolate circuits and keep spikes out of upstream circuits.  There are usually better ways to accomplish this.

The LM339 stuff at the bottom is really kind of bizarre.  The LM339 is an open collector comparator.  The first one U2C takes the input through an opto (for no good reason) and then has a kind of low pass filter.  A high input from the opto will charge C8 with 4.7K, in about .5 mS, when it goes low, it will discharge in about 1 mS.  This sort of rejects high frequencies on the input and will let low frequencies pass to trip the next comparator U2A.  Where it gets bizarre is on the output.  The .1 uF is discharged when the comparator goes low, with an RC of 47K and .1uF- about 5 mS/200 Hz.  This is another kind of low pass, if the input rate goes too high, C7 never charges and disables U2A which drive the FET's though paralleled AND gates modulated by the oscillator on its inverting input- U2B.   I guess the idea is to reject noise on the input but this is strange?   U2B is an oscillator that sets the basic drive frequency for the motor- this is RC of R9xC6- 470K and 2200 pF, about 1 KHz.  This is a strange circuit which might be getting sort of sawtooth across the input of U2B as a sort of PWM.  There is a bunch of supression stuff across the motor, tranzorbs, diodes and RC's. 

Overall, a kind of horror show but no unusual for the time.  Thanks for sharing.  I haven't looked at the logic board yet.

What are your plans- are you redesigning/updating it?  Throw this schematic in the trash- terrible.

[jwet]

After a little more study, I see that that the motor is driven directly from recified line volts- the optos make more sense now.  I originally thought it was all 12v.

[metertech58761]

So I actually have one of these squirreled away and dug it out for another look.
(I found it on Ebay and not only does it have the repair stamp I used to use, the S/N makes me actually wonder if it was that very same unit where I blew the brake resistor after finishing an overhaul!)

I take it that one needed to let the unit stabilize for a few seconds before starting the turntable?

Anyway, I opened up the unit to see the exact part number on the 4081 and it is a MC14081BCP.

I shoulda mentioned the optos and the fact the PWM board had +170V DC on it.

I did go back and check KiCAD, it has IGBT symbols and I went and updated Q3, Q4, and Q7 accordingly.

I still remember the final testing process for these - the controller was set to drive the turntable at 200 RPM (during which we calibrated the trimpot on the Logic to get it within +/- 2 RPM), then checking it at 300, 400, 999, and finally 1499 RPM, then let it coast to a stop (I presume this is where the brake board comes in), then a check of the 20 RPM maintenance mode.

I THINK, but I cannot be sure, that the redesigned PWM board had different sized gate resistors, but the rest of the circuitry was unchanged from the old design - the brake daughterboard circuitry was even incorporated into the PWM board (and for some reason, they went with a 4-layer PCB as well).

If I remember right, the date on the schematics was from the very early 1980s (and this was a slightly redesigned version, so I guess the original design was from 1980 or so).

Also, looking at the Logic, I got the frequency wrong (it wasn't written on the schematic I transcribed) - it is actually 3.58MHz.

No plans to do anything with the unit or design - mainly keeping it around for nostalgia.

[jwet]

That's pretty cool.  I have found stuff that I designed or worked on in junk stores in Silicon Valley- always a wisp of nostalgia, once in a while on ebay too.  I have an old cigar box that I keep boards and bits and pieces of old products in for nostalgia- most of the stuff I designed was handheld so its not so difficult.

Motorola put that MC1 in front of their 4000 series part numbers- all the same though.  RCA, the original inventors of the series, used a CD prefix.  Signetics prefixed theirs HE.  Everyone seemed to do this.  Just quirks of old chip companies making commodity parts and trying to differentiate or confuse.

Its funny how I can remember precisely how some old piece of equipment worked in arcane detail but can't remember my anniversary...

The crystal frequency of 3.58 Mhz is the US color burst frequency in NTSC TV's.  Each TV set had one of these in it and they were the most abundant and cheapest crystal around, the 5369 chip took advantage of that- they could use any crystal they wanted.  One verson of the National MM5369 made a precise 60 Hz (there were other versions too), it was used to keep line operated real time clocks on time though power failures or in DC only apps.

Have fun