[GreenKeys] Arduino Code/Speed Converter

Mon Jan 27 00:56:53 EST 2020

I had some scope captures on this computer, but I can't find them now.  I'm
jammed up for time for the next 10 days or so, so I won't be able to get
them on this machine for posting for a while.

But attached is a PDF of the schematic sheet that shows the business end of
the mag driver.  Some of my notes on the sheet are old, so they may not
make sense.  The snubber shown here is directly from the ST-6000, except I
show it across the magnets rather than across the switching device.  I
think  the values of the 470 ohms 100nF cap can be jiggled to make the
magnet current drop faster without much voltage increase.  I think the 1nF
to limit the initial dV/dt for the first nano seconds after the switch
opens is a good value.  The snubber is to limit EMI from the magnets as the
flux collapses, as well as limit voltage spikes seen by the switching
FETs.  DV1 is to protect the switching FETs if the snubbers are not working
right (DV1 should never trigger if the snubber is working correctly).

The circuit operates as follows: a space to mark transition sets the NAND
latch, which turns on the FET Q1 (with no series loop resistance other than
magnet coil resistance and sense resistor 10 ohms) and Q2 (which doesn't
contribute much current at this point).  Current rises straight slope at about
15mA/ms until there is 600 mV across the current sense resistor (60mA
magnet current).  At that point the latch is cleared and then the only
source of magnet current is through Q2.  Strap the loop resistors for the
sustained magnet current you want.  I like 30mA as the sustained current once
the magnet has reached 60mA and the mechanicals have settled, but other
people may want other values.  The 100mA fuse is in case something goes
wrong with the magnet driver.  It hasn't blow yet in my prototypes, but,
well, fuses have saved my bacon in the past on other projects.  I can't
afford to burn out selector magnets.  Do NOT omit the fuse.  The timer
formed by U1B is to extend the mark time by a value similar to the delay time
caused by the 15mA/ms slope at the start of the mark time.  I use op-amps
as comparators here because they are slow, slow, slow.  I never like to use
components that are faster than necessary, if I can avoid it.  The op-amp
as a comparator switches in about a millisecond, and that's plenty fast
enough for this application.

For power I use a 12V wall wart transformer (for 12 and 5 volt rails) and
then another 24V wall wart is series connected on top of the 12V to get
rail for 36V for the magnet driver.  I can put them in series because they
are isolated from each other.  The 24V only needs to put out about 100mA,
but 1A is the smallest wart I've been able to find at that voltage.  Two
wall warts is the cheapest way I've found to get 36V, 12V and 5V.  You
don't see the 12V used on this sheet, but I use it for the motor control
relay and motor current sensor (that's all on another sheet, and has not
been tested yet).

73, Paul, ad7i

On Sun, Jan 26, 2020 at 11:53 PM Gil Smith <gil at baudot.net> wrote:

> That is very interesting Paul.  Yes indeed, in a classic loop it is the
> desired R you want for the exponential, and the resulting loop voltage is
> merely the needed means to that end.
>
> That is a very interesting multi-mode scheme you have there -- any scope
> shots of the switching?
>
> What are you doing for snubbing?  I played with typical RC snubbing but
> found that a clamping schottky was all that I needed.
>
> gil
>
>
> gil smith, AF7EZ
> greenkeys moderator
> gil at baudot.net
>
>
> -------- Original Message --------
> Subject: Re: [GreenKeys] Arduino Code/Speed Converter
> From: ad7i <ad7i at ad7i.net>
> Date: Fri, January 24, 2020 10:36 pm
> To: Jim Haynes <jhhaynes at earthlink.net>
> Cc: Gil Smith <gil at baudot.net>, Ralph Mowery
> <rmowery28146 at earthlink.net>, greenkeys <greenkeys at mailman.qth.net>
>
> I have found (sample of 1) that the M15 is much more tolerant of low
> voltage, low resistance, drive than is the M28.  Bill Henry of HAL told me
> years ago that the inductance of the M28 magnets is much larger than the
> 15, with the 28 being 2 Henrys.  My measurements for my one M28 show the
> magnet inductance at 1.7 Henry, which is still a big number.
>
> A model 28 selector magnet will pull in with as little as 6 volts.  The
> reason that classic loop supplies are ~150V 2500 ohms is that the time
> constant of an R-L circuit is in proportion to L/R.  If L gets bigger the
> time constant gets longer.  To shorten the time constant, to make the
> magnet action fast and "snappy", then R must get larger.  As R gets larger
> then the driving voltage needs to get larger to maintain the same loop
> current.  To get the time constant of the selector magnet to be under 1 mS,
> the circuit resistance needs to be at least about 1.5K ohms.
>
> For comparison purposes, when using the classic 170 VDC 2800 ohm loop
> circuit, the selector magnet current takes about 600uS to go from zero to
> 30 mA (final value is of course is 60 mA).  Also when tested in a static
> situation (sample of one, model 28) the pull in current for the selector
> magnet is about 24 mA and the release current is about 12 mA.  I also found
> that my M28 ran well with only a 30 mA loop (but that 30mA had to come from
> a high voltage high resistance loop -- 120V 4000 ohms in this case), but I
> didn't try adjusting the range control with the 30mA loop to see if I had
> much latitude at that 30mA loop current.
>
> In my 36 volt electronic magnet driver the system has three modes, (mode
> 0) off, (mode 1) 36 volts with zero ohm loop and then (mode 2) 36 volts
> with 1200 ohm loop.  During mode 1 the resistance isn't really zero, it's
> the sum of the magnet resistance (66 ohms), a 100 mA fuse (5 ohms) and a 10
> ohm current sense resistor (the MOSFET switch is less than one ohm when ON
> so I ignore that) for a total of 81 ohms.  When the serial input to the
> magnet driver goes from SPACE to MARK the magnet driver enters mode 1.
> When in mode 1 the magnet current has the classic exponential decay toward
> it's final current value (36V/81ohms or 444mA), but from 0 mA to 60 mA we
> are so low on the exponential curve that the curve looks like a
> straight-line sawtooth rising current, rising at a slope of 15mA per
> miillisecond.  As the loop current rises so does the voltage across the 10
> ohm current sensor resistor.  When voltage on that sense resistor reaches
> 600 mV we then have 60mA through the sense resistor (600mV/10 ohms equals
> 60 mA), and thus also 60 mA through the selector magnets) the device
> switches to mode 2 and the magnet current then decreases at an exponential
> decay rate to it's final value of 30 mA (36V/1200ohms) and stays at 30 mA
> until the serial input goes SPACING.  When the input goes from MARK to
> SPACE a timer extends mode 2 from zero to 5 extra milliseconds (controlled
> by a pot) and then enters mode 0 which causes the magnet driver to
> disengage and the magnet current goes to zero at a rate determined by the
> snubber components (usually about 1 mS).  The purpose of the timer at the
> end of Mode 2 is to compensate for the time it took at the start of Mode 1
> to get the magnet current from 0 to 45 mA (about 3 mS).  I could probably
> replace that timer pot with a fixed resistor that provided a fixed 3 mS
> delay and call it close enough.
>
> This setup seems to work fine with my model 28 at 100 WPM as well as my
> model 15 at 60 WPM.  The useful adjustment range of the "range" control on
> the M28 is about the same with the 36 V electronic magnet driver as it is
> with a 120V  2000 ohm loop.
>
> Paul, ad7i
>
>
>
>
>
> On Fri, Jan 24, 2020 at 9:52 PM Jim Haynes <jhhaynes at earthlink.net> wrote:
>
>> On Fri, 24 Jan 2020, Gil Smith wrote:
>>
>> > 36 volts is probably a good choice also.  The purists may say it needs
>> to be
>> > over 100V to burn oil off contacts and such, but if a lower voltage
>> loop is
>> > working for you go for it.
>> >
>> A reference on this topic is Western Union Technical Review 15:4, October
>> 1961, p. 149.
>>
>> Also Western Union Technical Review 5:1, January 1951, p. 32.
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.qth.net/pipermail/greenkeys/attachments/20200127/ef65bca9/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: TTY Loop Supply Build 191219-01 sheet 3.pdf
Type: application/pdf
Size: 38870 bytes
Desc: not available
URL: <http://mailman.qth.net/pipermail/greenkeys/attachments/20200127/ef65bca9/attachment-0001.pdf>