# Need design for 120VDC 60mA constant current supply

L

#### [email protected]

Jan 1, 1970
0
If the 120V was only on a charged capacitor, how large would the
capacitor have to be to accomplish a brisk switching?

1uF @ 120V and 4H @ 60mA store the same energy, 7.2mJ

that's a place to start

-Lasse

W

#### whit3rd

Jan 1, 1970
0
John Larkin wrote:
[ use two power supplies, and a class G type amplifier that
takes current from both]
But how to regulate all this?
The selector coil is floating, nicht wahr? A sense resistor on
the (-) terminal to ground is all it really needs, your low
voltage supply will power op amps and a grounded-base
transistor (base to +12V in this case) is the level
translation to the HV pass transistor.

J

#### James Arthur

Jan 1, 1970
0
James said:
Nobody said:
Agreed. The question is, given that the capacitor feeds the coil, and the
HV feeds the capacitor, how to prevent the HV feeding the coil? The
resistor can't be any higher if the capacitor is to recharge fully.

You could switch the HV off with e.g. a FET during the on state, but that
requires a transformer or a high-side driver. If you're going to that
level, it's getting to the point where you may as well go the whole hog
and use an H-bridge, which lets you recover the energy stored in the coil
on switch-off to use for the next switch-on.

I like the simplicity of the previous circuit charging the cap
via R1.

FIG. 1
======
R1
4.7K / 5W
+120v >---\/\/\-----o-----------o-----o-------.
| | | |_
D1 | [R2] [C2] _) L1
+4v >----|>|------o [10k] [1uF] _)
| | | _) (4H, 55ohms)
| '--o--' |
--- C1 | D2 |
--- 1uF '----|<|---o
| |
| | |--'
=== | |<-. Q1
_____| |--+
|
===
GND

The only way to reduce that dissipation is to replace
R1 with a switching element.

I gave such a scheme in block form up above--it's a lot
busier. But, if you're dead-set on saving dissipation,
that's what needs doing.

Here's one possible sketch:

FIG. 2
======
D1
.--------------|<|---------------------.
| Q1 |
+120v >-----o-----o----. .-----o-------. |
| e \ / c | | |
[R1] ----- - | |
| | b ^ D2 | |
'-------o - _) L1 |
| | _) 4H / |
DATA+ >--. | === _) 55r |
| [R2] GND | |
| | | |
.-----o----o-----[R3]---|--------. | |
| | | | | | |
| | | |/ Q2 | o-------'
| | | +5v---| MPSA42 | |
| [R4] | |>. | |
| | | | | |
| | | __ [R5] | |
| | '--| \ | | |
| | | )O----o | |
| o-------|__/ | | |
| | 74HC132 | | |
| | | | |
[R6] \| Q3 | | |/ Q4
| |---------------|--------o-----| MPSA42
| .<| | |>.
| | | |
o-----o------[R8]-------' [R9]
| |
[R7] ===
| GND
===
GND

This is a hysteretic buck. Q1 is the switch, current is
sensed across R9, and Q3 is the comparator. R6-7 sets
the current limit as a fraction of the DATA+ signal input,
while R8 provides hysteresis.

When the current in R9 is below setpoint, Q3 is biased off
by R6-8, both 'HC132 inputs are high, driving the output
low, and switch Q1 is on. When i(L1) reaches the setpoint,
Q3's base rises until it conducts, 'HC132 output goes high,
Q1 cuts off, and L1 freewheels efficiently through D2 and Q4.

When DATA+ goes low, Q4 turns off and the inductor flies
back, returning its energy to the +120v supply.

The currents are small, so D1-2 are signal diodes and
snubbing is probably unnecessary.

The DATA+ signal could also serve as the +5V supply
powering the logic gate.

The whole should draw roughly 3 or 4 mA from the +120v.

Cheers,
James Arthur

Gee, no bites. Doesn't anyone like talking circuits?

Here's another hysteretic bucker, from scratch:

+120v
-+-
| D2
.-----------------------------o-----------|<----------------.
| Q1 | |
| 2n5401 | |
o------------. .-----o------)----o------------------. |
| e \ / c | | | | |
| --- | | - L1 _) |
| ___ | b .-. | ^ D1 4H _) |
0-----|___|----o | | R3 | | 55r _) |
| R1 | | | 1.8M| === | |
. | . 2.2k .-. '-' | GND | |
\ . | | | | | |
\ . SSR | | R2 | | o-----'
. | . '-' 5.6k | | |
| | | | |
.-. .----)-------' | |
| | R11 | | C1 | |
| | 680k | | || 470p | ___ |
'-' | o-----||-------)---|___|--. |
| | | || | R8 | ||--+
| ___ | |/ \| 47k | ||<-.
o--|___|--o--| Q2 Q3 |--------o .-----||--+ Q4
| R4 | |>. .<| | | |
| 39k .-. | | | | ___ |
| | | '------o-------' '--|--|___|--o-------.
| | | R5 | | R9 | |
| '-' 10k .-. | 15k .-. |
| | | | R7 Q2-3: | | | R10 --- C2
| === | | 470r 2n5550 | | | 33r --- 100n
| GND '-' MPSA42 | '-' |
| | | | |
| === | === ===
| GND | GND GND
'-------------------------------------------'

(created by AACircuit v1.28.4 beta 13/12/04 www.tech-chat.de)

LTSpice says it draws about 3mA average, 5mA worst case.

This version is meant to be driven by the solid state relay (SSR)
IC1 shown in the OP's circuit here:

http://www.animats.com/nagle/aetheric/connecting.html

C2's kind of optional--it simplified analysis.

I don't think it gets much simpler than this. I did another
version using an IC switcher, but you still need a level shifter,
high-side switch, and Q4, so using an IC doesn't save much more
that a transistor and some discretes.

Cheers,
James Arthur

~~~~~~~~~~~
(Oh, I just used whatever diodes and such I had models for.
The 1n4148 isn't suitable; use a higher voltage diode.)

Version 4
SHEET 1 1520 680
WIRE 304 -192 -144 -192
WIRE 448 -192 304 -192
WIRE 832 -192 512 -192
WIRE -144 -96 -144 -192
WIRE -144 -96 -272 -96
WIRE 48 -96 -144 -96
WIRE 208 -96 144 -96
WIRE 624 -96 208 -96
WIRE 736 -96 624 -96
WIRE -272 -64 -272 -96
WIRE 624 -48 624 -96
WIRE 208 -32 208 -96
WIRE -144 -16 -144 -96
WIRE -96 -16 -144 -16
WIRE 96 -16 96 -32
WIRE 96 -16 -16 -16
WIRE 96 0 96 -16
WIRE 736 0 736 -96
WIRE -272 48 -272 16
WIRE 624 64 624 16
WIRE 736 112 736 80
WIRE 832 112 832 -192
WIRE 832 112 736 112
WIRE 208 128 208 48
WIRE 208 128 -32 128
WIRE 96 208 96 80
WIRE 176 208 96 208
WIRE 368 208 240 208
WIRE 480 208 448 208
WIRE 96 272 96 208
WIRE 304 272 304 -192
WIRE 736 288 736 112
WIRE -368 320 -464 320
WIRE -208 320 -288 320
WIRE -160 320 -208 320
WIRE -32 320 -32 128
WIRE -32 320 -80 320
WIRE 32 320 -32 320
WIRE 480 320 480 208
WIRE 480 320 368 320
WIRE -32 352 -32 320
WIRE -464 368 -464 320
WIRE 208 368 96 368
WIRE 304 368 208 368
WIRE 688 368 624 368
WIRE 208 384 208 368
WIRE 480 432 480 320
WIRE 512 432 480 432
WIRE 736 432 736 384
WIRE 736 432 592 432
WIRE 816 432 736 432
WIRE -32 464 -32 432
WIRE -464 480 -464 448
WIRE 736 480 736 432
WIRE 208 496 208 464
WIRE 816 496 816 432
WIRE -208 544 -208 320
WIRE 624 544 624 368
WIRE 624 544 -208 544
WIRE 736 576 736 560
WIRE 816 576 816 560
FLAG 624 64 0
FLAG 736 576 0
FLAG -272 48 0
FLAG -32 464 0
FLAG -464 480 0
FLAG 208 496 0
FLAG 816 576 0
SYMBOL pnp 144 -32 M270
WINDOW 0 111 66 VLeft 0
WINDOW 3 80 84 VLeft 0
SYMATTR InstName Q1
SYMATTR Value 2N5401
SYMBOL diode 640 16 R180
WINDOW 0 24 72 Left 0
WINDOW 3 24 0 Left 0
SYMATTR InstName D1
SYMATTR Value 1N4148
SYMBOL ind 720 -16 R0
SYMATTR InstName L1
SYMATTR Value 4H
SYMATTR SpiceLine Rser=55 Rpar=220k Cpar=1nF
SYMBOL res 720 464 R0
SYMATTR InstName R10
SYMATTR Value 33
SYMBOL npn 32 272 R0
SYMATTR InstName Q2
SYMATTR Value 2N5550
SYMBOL voltage -272 -80 R0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR InstName V1
SYMATTR Value 120
SYMBOL res -176 336 R270
WINDOW 0 32 56 VTop 0
WINDOW 3 0 56 VBottom 0
SYMATTR InstName R4
SYMATTR Value 39k
SYMBOL res -48 336 R0
SYMATTR InstName R5
SYMATTR Value 10k
SYMBOL voltage -464 352 R0
WINDOW 3 24 104 Invisible 0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR Value PULSE(0 120 10mS 100uS 100uS 20mS 40mS)
SYMATTR InstName V2
SYMBOL res 0 -32 R90
WINDOW 0 0 56 VBottom 0
WINDOW 3 32 56 VTop 0
SYMATTR InstName R1
SYMATTR Value 2.2k
SYMBOL diode 512 -208 R90
WINDOW 0 0 32 VBottom 0
WINDOW 3 32 32 VTop 0
SYMATTR InstName D2
SYMATTR Value 1N4148
SYMBOL res 80 -16 R0
SYMATTR InstName R2
SYMATTR Value 5.6k
SYMBOL npn 368 272 M0
SYMATTR InstName Q3
SYMATTR Value 2N5550
SYMBOL res 192 368 R0
SYMATTR InstName R7
SYMATTR Value 470
SYMBOL res 608 416 R90
WINDOW 0 0 56 VBottom 0
WINDOW 3 32 56 VTop 0
SYMATTR InstName R9
SYMATTR Value 15k
SYMBOL res 192 -48 R0
SYMATTR InstName R3
SYMATTR Value 1.8meg
SYMBOL cap 240 192 R90
WINDOW 0 0 32 VBottom 0
WINDOW 3 32 32 VTop 0
SYMATTR InstName C1
SYMATTR Value 470pF
SYMBOL res -384 336 R270
WINDOW 0 32 56 VTop 0
WINDOW 3 0 56 VBottom 0
SYMATTR InstName R11
SYMATTR Value 680k
SYMBOL nmos 688 288 R0
SYMATTR InstName M1
SYMATTR Value Si3440DV
SYMBOL res 464 192 R90
WINDOW 0 0 56 VBottom 0
WINDOW 3 32 56 VTop 0
SYMATTR InstName R6
SYMATTR Value 47k
SYMBOL cap 800 496 R0
SYMATTR InstName C2
SYMATTR Value 100nF
TEXT 848 232 Left 0 !.tran 0 .2s 100uS 2uS
TEXT 888 -32 Left 0 ;Driver for a Teletype Model 15 selector\nmagnet
(L1).\n \n12-Mar-2009 by James Arthur

J

#### James Arthur

Jan 1, 1970
0
Jim said:
On Fri, 13 Mar 2009 01:35:24 GMT, James Arthur
Gee, no bites. Doesn't anyone like talking circuits?

Here's another hysteretic bucker, from scratch:

+120v
-+-
| D2
.-----------------------------o-----------|<----------------.
| Q1 | |
| 2n5401 | |
o------------. .-----o------)----o------------------. |
| e \ / c | | | | |
| --- | | - L1 _) |
| ___ | b .-. | ^ D1 4H _) |
0-----|___|----o | | R3 | | 55r _) |
| R1 | | | 1.8M| === | |
. | . 2.2k .-. '-' | GND | |
\ . | | | | | |
\ . SSR | | R2 | | o-----'
. | . '-' 5.6k | | |
| | | | |
.-. .----)-------' | |
| | R11 | | C1 | |
| | 680k | | || 470p | ___ |
'-' | o-----||-------)---|___|--. |
| | | || | R8 | ||--+
| ___ | |/ \| 47k | ||<-.
o--|___|--o--| Q2 Q3 |--------o .-----||--+ Q4
| R4 | |>. .<| | | |
| 39k .-. | | | | ___ |
| | | '------o-------' '--|--|___|--o-------.
| | | R5 | | R9 | |
| '-' 10k .-. | 15k .-. |
| | | | R7 Q2-3: | | | R10 --- C2
| === | | 470r 2n5550 | | | 33r --- 100n
| GND '-' MPSA42 | '-' |
| | | | |
| === | === ===
| GND | GND GND
'-------------------------------------------'

(created by AACircuit v1.28.4 beta 13/12/04 www.tech-chat.de)

LTSpice says it draws about 3mA average, 5mA worst case.

This version is meant to be driven by the solid state relay (SSR)
IC1 shown in the OP's circuit here:

http://www.animats.com/nagle/aetheric/connecting.html

C2's kind of optional--it simplified analysis.

I don't think it gets much simpler than this. I did another
version using an IC switcher, but you still need a level shifter,
high-side switch, and Q4, so using an IC doesn't save much more
that a transistor and some discretes.

Cheers,
James Arthur
[snip]

I can think of some far simpler configurations. I'm just waiting on
an answer to this previously posted question...

"If the 120V was only on a charged capacitor, how large would the
capacitor have to be to accomplish a brisk switching?"

Also, since I haven't seen a flexowriter since I was a kid, what's the
duty cycle and rep-rate?

...Jim Thompson

The energy needed to charge to inductor is 60mA into 4H.
1/2 * L * i^2 = 0.5*4H*(0.06)^2 = 7.2mJ

1uF at 120v has the same, so that's ballpark. That's what
Nobody used. Whether you need more or less cap depends on
how you plan on using it.

The problem is charging that cap efficiently. I figured
to just skip that step and charge the inductor directly.

As to duty cycle and rep rate, IHNI--I was depending on
the OP. It seems they run at about 50 baud:

http://www.baudot.net/teletype/M15.htm

Cheers,
James Arthur

J

#### John Nagle

Jan 1, 1970
0
1uF @ 120V and 4H @ 60mA store the same energy, 7.2mJ

that's a place to start

Right.

It looks like we have some good designs for an efficient driver
which requires an external 120VDC supply. The next step is to get
rid of the high voltage supply, and use some kind of up-converter
to charge up a capacitor with the 7.2mJ needed to pull in the
selector magnet. At 45.45 baud, one bit time is 22ms. Even
better, there can only be one 0 to 1 (SPACE to MARK) transition
every two bit times, so the charging system has maybe 1.8 bit times
or so to charge the capacitor. 75 baud is as fast as the mechanical
Teletypes ever got, so a 13ms bit time is the worst case, giving a
charging time of 23ms. So design for a 20ms charging time and a
1-2ms discharge time.

This is something like a "hammer driver" for a daisy wheel printer.

It would definitely be convenient to run the whole thing off
+5VDC. Right now, 90% of the hardware volume is the open
frame 120VDC 200mA supply.

John Nagle

J

#### John Nagle

Jan 1, 1970
0
James said:
Version 4
SHEET 1 1520 680 ....
TEXT 888 -32 Left 0 ;Driver for a Teletype Model 15 selector\nmagnet
(L1).\n \n12-Mar-2009 by James Arthur

Tried that in SPICE. During ON periods, the current through the selector
magnet is usually 60mA, but there are 1uS spikes to 200mA every
3ms. Not clear what's happening.

John Nagle

N

#### Nobody

Jan 1, 1970
0
Tried that in SPICE. During ON periods, the current through the selector
magnet is usually 60mA, but there are 1uS spikes to 200mA every
3ms. Not clear what's happening.

That's the 1nF parallel capacitance specified for the inductor.

The 120V voltage pulses have a rise time of 1uS, with a peak slew rate of
~140V/uS, giving 140mA through the parallel capacitance, on top of the
60mA inductor current.

D

#### [email protected]

Jan 1, 1970
0
That's the 1nF parallel capacitance specified for the inductor.

The 120V voltage pulses have a rise time of 1uS, with a peak slew rate of
~140V/uS, giving 140mA through the parallel capacitance, on top of the
60mA inductor current.

Yep, that's right. The real selector magnet will have considerable
capacitance. I didn't know the value, so I took a wild guess.

As far as getting rid of the 120v supply entirely, boost topology
off say +12v on the "hot" end of the coil, FET to GND on the cold end,
open the FET to dump the coil quickly?

That is, you might initially load up the boost coil (L1) to the
desired total energy, then let it fly back (single impulse)
through a rectifier, charging an appropriate capacitor to a
desired voltage, e.g. 1uF to 120v. The capacitor then decays
into and charges the selector magnet to the desired current.

In steady-state, maintain desired current with a low voltage supply
through a diode and resistor, as before, or switchmode, via SW1.

Open FET to dump the coil (L2).

Like this:

L1
.-.-.-. D1
+12v>----| | | |---o--->|-----o------.
| | |
SW1 \ C1 --- _) L2
| --- _) selector
=== | _) magnet
GND === | coil
GND |
||--'
||<-.
------||--+
|
===
GND
Cheers,
James Arthur

N

#### Nobody

Jan 1, 1970
0
As far as getting rid of the 120v supply entirely, boost topology
off say +12v on the "hot" end of the coil, FET to GND on the cold end,
open the FET to dump the coil quickly?

That is, you might initially load up the boost coil (L1) to the
desired total energy, then let it fly back (single impulse)
through a rectifier, charging an appropriate capacitor to a
desired voltage, e.g. 1uF to 120v. The capacitor then decays
into and charges the selector magnet to the desired current.

LTSpice version:

Version 4
SHEET 1 940 680
WIRE 16 -144 -96 -144
WIRE 192 -144 96 -144
WIRE 288 -144 192 -144
WIRE 432 -144 352 -144
WIRE 592 -144 432 -144
WIRE 752 -144 592 -144
WIRE 592 -80 592 -144
WIRE 592 -80 544 -80
WIRE 640 -80 592 -80
WIRE 752 -64 752 -144
WIRE 544 -48 544 -80
WIRE 640 -48 640 -80
WIRE -96 64 -96 -144
WIRE 192 64 192 -144
WIRE 544 64 544 32
WIRE 592 64 544 64
WIRE 640 64 640 16
WIRE 640 64 592 64
WIRE 320 80 240 80
WIRE 432 96 432 -144
WIRE 592 128 592 64
WIRE 640 128 592 128
WIRE 752 128 752 16
WIRE 752 128 704 128
WIRE 320 144 320 80
WIRE 192 208 192 144
WIRE 240 208 240 128
WIRE 240 208 192 208
WIRE 752 208 752 128
WIRE 880 224 800 224
WIRE 880 272 880 224
WIRE 752 352 752 288
WIRE 800 352 800 272
WIRE 800 352 752 352
WIRE -96 384 -96 144
WIRE 192 384 192 208
WIRE 192 384 -96 384
WIRE 320 384 320 224
WIRE 320 384 192 384
WIRE 432 384 432 160
WIRE 432 384 320 384
WIRE 752 384 752 352
WIRE 752 384 432 384
WIRE 880 384 880 352
WIRE 880 384 752 384
WIRE -96 400 -96 384
FLAG -96 400 0
SYMBOL voltage -96 48 R0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR InstName V1
SYMATTR Value 5V
SYMBOL ind 0 -128 R270
WINDOW 0 32 56 VTop 0
WINDOW 3 5 56 VBottom 0
SYMATTR InstName L1
SYMATTR Value 630mH
SYMBOL sw 192 160 R180
SYMATTR InstName S1
SYMBOL voltage 320 128 R0
WINDOW 3 24 104 Invisible 0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR InstName V2
SYMATTR Value PULSE(-1V 1V 20ms 1us 1us 20ms 100ms 5)
SYMBOL cap 416 96 R0
SYMATTR InstName C1
SYMATTR Value 0.47ÂµF
SYMBOL ind 736 -80 R0
SYMATTR InstName L2
SYMATTR Value 4H
SYMATTR SpiceLine Rser=55R
SYMBOL res 528 -64 R0
SYMATTR InstName R1
SYMATTR Value 10K
SYMBOL cap 624 -48 R0
SYMATTR InstName C2
SYMATTR Value 1ÂµF
SYMBOL sw 752 304 R180
SYMATTR InstName S2
SYMBOL voltage 880 256 R0
WINDOW 3 24 104 Invisible 0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR InstName V3
SYMATTR Value PULSE(-1V 1V 39ms 1us 1us 20ms 100ms 5)
SYMBOL schottky 288 -128 R270
WINDOW 0 32 32 VTop 0
WINDOW 3 0 32 VBottom 0
SYMATTR InstName D1
SYMBOL schottky 704 112 R90
WINDOW 0 0 32 VBottom 0
WINDOW 3 32 32 VTop 0
SYMATTR InstName D2
TEXT -40 432 Left 0 !.model SW SW()
TEXT -130 506 Left 0 !.tran 500ms

J

#### James Arthur

Jan 1, 1970
0
Nobody said:
LTSpice version:

<snip>

I envision S1 being triggered on the rising edge of data,
with its ON-time controlled either by time (a 290uS one-shot)
or by a current sense limiter.

S2 would be controlled directly by the data level (high
= ON).

Cheers,
James Arthur

Version 4
SHEET 1 940 680
WIRE 16 -144 -96 -144
WIRE 192 -144 96 -144
WIRE 288 -144 192 -144
WIRE 432 -144 352 -144
WIRE 464 -144 432 -144
WIRE 592 -144 544 -144
WIRE 752 -144 592 -144
WIRE 592 -80 592 -144
WIRE 592 -80 544 -80
WIRE 640 -80 592 -80
WIRE 752 -64 752 -144
WIRE 544 -48 544 -80
WIRE 640 -48 640 -80
WIRE -96 64 -96 -144
WIRE 192 64 192 -144
WIRE 544 64 544 32
WIRE 592 64 544 64
WIRE 640 64 640 16
WIRE 640 64 592 64
WIRE 320 80 240 80
WIRE 432 96 432 -144
WIRE 592 128 592 64
WIRE 640 128 592 128
WIRE 752 128 752 16
WIRE 752 128 704 128
WIRE 320 144 320 80
WIRE 192 208 192 144
WIRE 240 208 240 128
WIRE 240 208 192 208
WIRE 752 208 752 128
WIRE 880 224 800 224
WIRE 880 272 880 224
WIRE 752 352 752 288
WIRE 800 352 800 272
WIRE 800 352 752 352
WIRE -96 384 -96 144
WIRE 192 384 192 208
WIRE 192 384 -96 384
WIRE 320 384 320 224
WIRE 320 384 192 384
WIRE 432 384 432 160
WIRE 432 384 320 384
WIRE 752 384 752 352
WIRE 752 384 432 384
WIRE 880 384 880 352
WIRE 880 384 752 384
WIRE -96 400 -96 384
FLAG -96 400 0
SYMBOL voltage -96 48 R0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR InstName V1
SYMATTR Value 12v
SYMBOL ind 0 -128 R270
WINDOW 0 32 56 VTop 0
WINDOW 3 5 56 VBottom 0
SYMATTR InstName L1
SYMATTR Value 470ÂµH
SYMATTR SpiceLine Rpar=22k
SYMBOL sw 192 160 R180
SYMATTR InstName S1
SYMBOL voltage 320 128 R0
WINDOW 3 24 104 Invisible 0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR Value PULSE(-1V 1V 20ms 1us 1us 290uS 100ms 5)
SYMATTR InstName V2
SYMBOL cap 416 96 R0
SYMATTR InstName C1
SYMATTR Value 1ÂµF
SYMBOL ind 736 -80 R0
SYMATTR InstName L2
SYMATTR Value 4H
SYMATTR SpiceLine Rser=55R
SYMBOL res 528 -64 R0
SYMATTR InstName R1
SYMATTR Value 10K
SYMBOL cap 624 -48 R0
SYMATTR InstName C2
SYMATTR Value 1ÂµF
SYMBOL sw 752 304 R180
SYMATTR InstName S2
SYMBOL voltage 880 256 R0
WINDOW 3 24 104 Invisible 0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR Value PULSE(-1V 1V 20ms 1us 1us 20ms 100ms 5)
SYMATTR InstName V3
SYMBOL diode 288 -128 R270
WINDOW 0 32 32 VTop 0
WINDOW 3 0 32 VBottom 0
SYMATTR InstName D1
SYMATTR Value MURS120
SYMATTR Description Schottky diode
SYMATTR Type schottky
SYMBOL schottky 704 112 R90
WINDOW 0 0 32 VBottom 0
WINDOW 3 32 32 VTop 0
SYMATTR InstName D2
SYMBOL res 560 -160 R90
WINDOW 0 0 56 VBottom 0
WINDOW 3 32 56 VTop 0
SYMATTR InstName R2
SYMATTR Value 130
TEXT -40 432 Left 0 !.model SW SW()
TEXT -128 504 Left 0 !.tran 500ms

J

#### James Arthur

Jan 1, 1970
0
John said:
Right.

It looks like we have some good designs for an efficient driver
which requires an external 120VDC supply. The next step is to get
rid of the high voltage supply, and use some kind of up-converter
to charge up a capacitor with the 7.2mJ needed to pull in the
selector magnet. At 45.45 baud, one bit time is 22ms. Even
better, there can only be one 0 to 1 (SPACE to MARK) transition
every two bit times, so the charging system has maybe 1.8 bit times
or so to charge the capacitor. 75 baud is as fast as the mechanical
Teletypes ever got, so a 13ms bit time is the worst case, giving a
charging time of 23ms. So design for a 20ms charging time and a
1-2ms discharge time.

This is something like a "hammer driver" for a daisy wheel printer.

It would definitely be convenient to run the whole thing off
+5VDC. Right now, 90% of the hardware volume is the open
frame 120VDC 200mA supply.

John Nagle

+5v is a pain--you can't charge the boost inductor quickly, and then
you need to stagger the switch timing. +12v wastes a little power,
but it makes the control circuitry simpler and faster.

Cheers,
James Arthur

J

Jan 1, 1970
0
J

#### James Arthur

Jan 1, 1970
0
Jim said:
For DC in chokes I prefer gapped cores... pot cores or E-I.

...Jim Thompson

Me too. Gapped is the way to go. Here I just searched for
whatever Digikey had handy--saved me the calcs for a home-rolled.

Mouser's search is impossible for this -- slow, and won't permit
even a small subset of the range of Idc to search 3-7A chokes ...
grrr.

Cheers,
James Arthur

N

#### Nobody

Jan 1, 1970
0
<snip>

Yeah, that makes more sense. I was thinking 1:1 current; don't know why.

I don't know how much of an issue the peak current will be, but you can
just increase the pulse width and inductor size if that's an issue. E.g.
2ms and 7.2mH gives 1.3A peak, 146mA RMS.

Also, I'm using 5V, which ISTR is what the OP said.

Regarding the current spike, the capacitor is also worth worrying about.
In the event that the capacitor fails open, the next pulse is going to
produce a rather nasty voltage spike.
I envision S1 being triggered on the rising edge of data,
with its ON-time controlled either by time (a 290uS one-shot)
or by a current sense limiter.

S2 would be controlled directly by the data level (high
= ON).

That certainly makes the timing simpler.

J

#### James Arthur

Jan 1, 1970
0
Jim said:
Jim said:
On Wed, 18 Mar 2009 00:46:43 GMT, James Arthur
[snip]
It's low duty, low frequency. This toroid might be pushed
to 5.7A (we'd have to review the datasheet sat curve, but
20% reduced inductance at high current wouldn't hurt a
thing here):

http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=M8842-ND

Cheers,
James Arthur
For DC in chokes I prefer gapped cores... pot cores or E-I.

...Jim Thompson
Me too. Gapped is the way to go. Here I just searched for
whatever Digikey had handy--saved me the calcs for a home-rolled.

Mouser's search is impossible for this -- slow, and won't permit
even a small subset of the range of Idc to search 3-7A chokes ...
grrr.

Cheers,
James Arthur

Back in my capacitive discharge ignition system days, I'd roll my own
on old filament transformer E-I bobbins, shim them up to what I
wanted. When they worked the way I wanted I'd hand them off to
Arizona Coil to knock me out a few hundred for demos.

My favorite size was 5mH at 5A... 62.5mJ... just the right kick for a
spark ;-)

There were times when all the police cars in Detroit were running on
one or another model of my ignition systems (Motorola and Philco-Ford
days).

...Jim Thompson

Good point--this <50Hz app is pretty similar. I wonder how laminated
iron cores would stack up for this inductor. Surely spark coils
use 'em for good reason...

Peak current here could be reduced by raising the inductance,
but that's just as demanding on the inductor.

Everything else being the same, doubling turns gives:
o 4x inductance,
o 2x amp-turns per amp, cutting Isat rating in half

0.5 (4L) * (i/2)^2
------------------ = 1, so, no add'tl energy is stored by
0.5 (L) * (i)^2 just adding windings on a given core.

Cheers,
James Arthur

J

#### James Arthur

Jan 1, 1970
0
Jim Thompson wrote:

[snip]
I like iron for slow stuff like this.

I'm up to my ears in a project right now, commenting here as
simulations sit there and churn.

My thoughts are that this hammer driver could be attacked very much
like a CD-ignition (see my web site)... generate the energy just as
needed... whack the bottom of the coil to -120V, top of coil tied to
+5V.

...Jim Thompson

That's pretty much what Nobody and I've been up to,
with a few variations. If John Nagle (the OP) doesn't
mind scrounging or winding an inductor, our work is done.

I Googled 'lightly' for laminated core inductors and learned
that those wound with hard-drawn copper _sound_ better too.
So that's nice to know.

Cheers,
James Arthur

J

#### James Arthur

Jan 1, 1970
0
Nobody said:
On Tue, 17 Mar 2009 23:57:51 +0000, James Arthur wrote:
I don't know how much of an issue the peak current will be, but you can
just increase the pulse width and inductor size if that's an issue. E.g.
2ms and 7.2mH gives 1.3A peak, 146mA RMS.

Also, I'm using 5V, which ISTR is what the OP said.

Yep, and it's more power efficient during the steady-state.
I used +12v for speed and convenience.
Regarding the current spike,
a)

the capacitor is also worth worrying about.
b)

In the event that the capacitor fails open, the next pulse is going to
produce a rather nasty voltage spike.

It a) is, and b) will. A nice film cap should do well and last forever.

Cheers,
James Arthur

J

#### James Arthur

Jan 1, 1970
0
John said:
It looks like we have some good designs for an efficient driver
which requires an external 120VDC supply. The next step is to get
rid of the high voltage supply, and use some kind of up-converter
to charge up a capacitor with the 7.2mJ needed to pull in the
selector magnet. At 45.45 baud, one bit time is 22ms. Even
better, there can only be one 0 to 1 (SPACE to MARK) transition
every two bit times, so the charging system has maybe 1.8 bit times
or so to charge the capacitor. 75 baud is as fast as the mechanical
Teletypes ever got, so a 13ms bit time is the worst case, giving a
charging time of 23ms. So design for a 20ms charging time and a
1-2ms discharge time.

This is something like a "hammer driver" for a daisy wheel printer.

It would definitely be convenient to run the whole thing off
+5VDC. Right now, 90% of the hardware volume is the open
frame 120VDC 200mA supply.

John Nagle

As one option your large +120v supply could be replaced with
a much smaller unit at this point.

If the selector coil can take it you could even cobble a 1:1
isolation transformer out of back-to-back wall warts or some
such, and use +170v. That'd be easy, with parts that are
widely available. A 1:1 audio transformer might even do for
just a few mA (check the ratings carefully first!). The switcher
uses jelly-bean parts as well--nothing that's hard to get.

The boost-inductor circuits are simpler and slick, but you might
have to scrounge for or make the boost inductor. JT was right
that--8mJ inductors are available, but not quite as dirt-common
as I'd thought.

Cheers,
James Arthur

J

Jan 1, 1970
0

Replies
0
Views
565
Replies
2
Views
1K
Replies
6
Views
2K
Replies
45
Views
9K
Replies
4
Views
895