DOUT chip Substitution

[John_W._Couvillon]

Is it possible to substitute a MIC5841 8bit Serial-input latched Driver  for the 74HC696/ULN2803 pair in the typical combo for driving magnets, excluding the current DOUT pcb design..  The 5841 will allow 4 outputs on (200ma @12vdc) at 70 deg. c ambient for an 86% duty cycle, , or 100% duty cycle for 3 outputs on at 70 deg. c ambient.  In an application using one 8 bit driver package to serve 4 SAMS ( 4 on magnets and 4 off magnets)  where  max condition would be  4 outputs at capacity per 8 bit driver. As i understand the ULN2803 spec, only one of the 8 outputs can be at capacity at the same time.  Seems a better optiion.

Johnc

 

[Duggle]

 As i understand the ULN2803 spec, only one of the 8 outputs can be at capacity at the same time.

 

I just looked at the datasheet of the ULN2803.

 

It is simply an array of darlington drivers.

 

Each output can switch 500mA.

 

The maximum voltage across the switched load is 50V.

[John_W._Couvillon]

Duggle,

The 5841 appears to  contain the equivalent of both the 74hc595 and the uln2803. From an application point of view, using the 5841 the DOUT4X would need only 4 ic's on the pcb rather then 8.  Actually the 5841 is used on a popular brand of  SAMS driver. 

Look closely at the datasheet,  only one of the 8 outputs can carry the 500ma at a time.  The ic cannot carry all 8 outputs at capacity.

[Duggle]

The data sheet I looked at was the TI one, which returned near the top of the google search.

It does say that the total substrate terminal current is -2.5A

This means with all outputs "on" each terminal could carry 2.5/8=312mA.

There is an "on" voltage drop of about 0.7V so 8*0.312*0.7= 1.75Watts which would make the chip warm, even hot, to the touch.

 

So it would seem the driver outputs of MIC5841 and ULN2803 are similar (could argue that the 2803 has a bit more current).

Whether it makes sense to substitute 2 devices for 1 depends on the application and situation. (btw, I haven't looked at the Shift register within the MIC5841, so I can't say it's compatible)

[John_W._Couvillon]

Duggle,

Ok, i see your point.

i am probably miss interpreting the notes on the max ratings table  (motorola Datasheet)  it says

MAXImUM RATING  (Ta=25deg, C,  and rating apply to any on device in the package, unless otherwise noted)

 

Below the table it says:  Do not exceed maximum current limit per driver. 

 

If you are running  SAMS with 30 -40 ohm coils, thats 300 - 400ma each,  connecting 4 sams to one 2803 means there is a possibility that a gen cancel will cause 4 off coils to be energized at the same time, for .5 sec.  if i understand you accurately,  it is allowable to have all 8 outputs on at 312 ma with out exceeding the max. temp rise over ambient.  since  only 4 of the 8 will fire, then i should have no problem. 

Am I reading you correctly.

johnc

[Duggle]

If only 4/8 loads are ever active at the same time then I'm sure 300 to 400 mA would be fine.

 

Another approach worth mentioning is to have the DOUT pin drive the gate of an N channel MOSFET. Although they are discrete devices, they don't require any other components (except protection diodes!) The main benefit is their extremely low "on" resistance. It means they run stone cold, which is good for long term reliability. This means 8 mosfets and 8 diodes (which would be painful though, if you need to replicate this circuit many times!)  

[John_W._Couvillon]

Duggle,

That sounds good, but would require a total new PCB design with new PCB's for 64 sams, not to mention the cost of  the nchannel mosfets, even in a quantity of 100.  There is one commercial line of driver boards using the 74HC595 which drives simple 2n222 single transistor drivers.  thats 128 transistors, but they are cheap.

for the time being, SMASH TV's pcbs are the best value going. 

My current problem (different thread) deals with  grounding  issues when using a 12vdc  PS for the SAMS magnets.  The 28ga ribbon cable ground conductor that loops through the DOUT4X pcbs was never intended to pass high amps produced by the SAMS aps, nor was the ground trace on the PCB's.  In an attempt to resolve the issue,  KPete recommended attaching a high capacity (#18ga) ground lead to the ground plane on all Dout4X  pcbs  and terminating them at one poiint on the -12vdc supply.  Assuming the new grounding path is sufficiently low impedence to keep the current through the ribbon cable in limits.  In addition,  The LPC uses USB power, and the 8bit core uses a wall wart.  The wall wart -5vdc can be connected to the -12vdc common point, but not the USB power.  Again the recommended fix is to solder a good ground on the the grounded back plane of the LPC and affix it to the -12vdc common point.  That creates ground loops big time.

To bad that technology doesn't provide an optical connection inbetween  the 74HC595's and the ULN2803, whereby the magnet power could be isolated from  all other supplies. probably could be done, with some optical isolation.  All you need is a way to switch 5vdc on the input to the ULN to turn it on or off. Any thoughts.   Take a look at "midification"  midio128 ver. 3.  there is a block diagram of my organ system.  Clearly, IMHO, there is a need for a high current solution when dealing with DOUTs and low imped. magnets such as SAMS.  Driving organ pipe electromechanical pallet valves works just fine, much lower duty cycle, however, grounding and high current return path to the negative PS term is still a problem..

 

Please take a look at the datasheet for the MIC5841/5842,  http:/www.micrel.com.  Note in the General Description, Page.1, "The drivers can be operated with a split supply where the negative supply is down to -20V".   The split supply is what caught my eye, in view of the grounding situation described above.  would this design aid in keeping the 12vdc power loop isolated from the 5VDC logic supply?    using the MIC5841 would allow more space on the pcb for larger ground traces, and output traces from the drivers without increasing the overall PCB size .  your comments please.

Johnc

[Duggle]

If the power supplies (5V, 12V)  are floating then you can have common ground connection without a forming a "ground loop".  Even if the 5V is coming from a grounded USB device (computer) it won't be a ground loop as such.

 

Without knowing much about the application, if you identify the current path of the coils:

+12V terminal of p.s, 

to "+" of coils,

to "-" of coils,

to o/p of driver (say uln2803),

to gnd pin of driver,

to gnd plane of PCB,

to gnd of 12V supply.

 

These are the wires that need to be kept low impedance, particularly the gnd plane of the PCB(s) to gnd of 12V supply, and the +12V from the supply as these nodes carry current for all coils.

Now if the 12V power supply is floating (gnd not connected to mains ground) then you will not have an earth loop, even though the gnd plane of the driver boards is connected to the core.

This is very often the case (floating) and can be checked with a continuity tester mode of a multimeter.

 

There is a substrate pin on the micrel chips which allows the driver to use a different potential but this does not isolate the supply (there still need to be a ground connection somewhere).

 

Just to repeat: you wont have earth loop problems if you use a 12V supply that is isolated from it's mains earth pin.

 

[John_W._Couvillon]

Duggle.

The current path as you describe it is accurate!

Looking at the two main devices, the 8bit core and LPC --  The 8bit core uses a wall wart, but the core has its own rectifier and regulator, the -5v side being grounded to the core ground plane.  The LPC accepts 5vdc fromt he USB connectiion, but also has its own power supply, with grounded negative.  Each  train of I/O is powered through the 5/c i/o cable interconnecting the DINS snd DOUTs.

if there were no ULN2803 drivers there would be no need for additional 12vdc PSU, or additional common ground point.

Adding the ULN Drivers on the DOUT pcbs and creates the need for an additional 12vdc PSU, which by design, must have a  common connection point which turns out to be the backplane of the DOUT pcb.  IMHO, The back plane of the DOUT PCB  is  the common with regard to the two  supplies.  Current will flow as you described in the 12vdc loop, and current will flow through the 5vdc loop, however the common backplane of the DOUT pcb is the only tie. So i agree, there is no ground loop.  The ground plane of the core 8 and LPC should remain floating.

Johnc

[Duggle]

if there were no ULN2803 drivers there would be no need for additional 12vdc PSU, 

But the coils require 12V, and you'll need something that switches this and can carry 400mA, right? 

I'd suggest a standard DOUT module with ULN2803 fitted should work fine. You may have discovered that with this configuration you need to make sure GND (pin 9, I think) makes a (good) connection with the ground plane. 

[kpete]

Hi John and Duggle,

I must agree with your elimination of ground loops by not grounding the 12volt common point to the frame ground and not putting a common ground on the LPC board. My original suggestion to put a frame ground at the 12v return point was for safety reasons only and not ground loop elimination reasons.

 

Pete

Edited March 2, 2013 by kpete

[John_W._Couvillon]

Duggle, Pete,

Take a look at the attached sketch.

As long as the only point of connection between the two loops is the new ground lug that i show on the DOUT pcb, the high current loop involving the 12vdc PS, magnet coil and driver on the ULN and the 5vdc loop though the I/o cabling should remain separate.  When the ULN conducts the voltage from pin 9 (gnd) to any of the output pins drops to less then 1 volt and the current from pin 9 to any one of the output pin goes up to 200 - 300ma.  Pin 9 on the ULN  must be connected to pin 8 on the 595 which is at ground plane potential, however the current flow will be normal.

Adding  a high current jumper from pin 9 on the ULN to the new grond lug, ot a beefier trace will do the job. and minimize voltage drop in the loop, so that the voltage across the magnet is as high as possible.

Johnc

[Duggle]

The drawing looks right, to me.

[John_W._Couvillon]

Duggle,

Back to our original discussiion of the position of ULN outputs and initialization of the shift registers on startup,  If stops are on when the organ is turned on,  they are immediately cancelled when jOrgan come on.    Boot sequence is midibox cores, ubuntu,  the soundengine,  then jOrgan, and lastly the disposition.  jOrgan sends out a note off  message to all off magnets, whether they are on or not. 

My 12 volt power supply is a converted computer PS which is only rated for 10 amperes at 12vdc, and if it is on during boot up,  it immediately trips out when the jOrgan disposition boots.  The off magnets on 32 SAMS is sufficient to  trip it out.  I need a higher capacity PS.  simply unplugging it for 10 seconds allow it to reset.  

Johnc

[Duggle]

32*0.2=6.4A  (?)

32*0.3=9.6A

 

Make sure the rating is for the 12V not the 5V output of the PS.

 

Of course you could use 2 or more separate 12V supplies.  Naturally, the +12V from each supply is *not* connected to each other at all.

They have a common ground connection, but the +12V goes to separate groups of coils. The "groups" is defined by those coils connected to the same ULN2803 because they share a common protection diode common cathode pin on the driver that terminates to the +12V side of those coils in the group.

[John_W._Couvillon]

Good idea!

Thanks

[John_W._Couvillon]

Duggle,

you mentioned the  protection diode in your last post.  one of my keyboard matrix encoders is doing strange things.  when a key is pressed it sends out the appropriate note on/ note off message, rite channel, rite note,  but in addition it sends out several other note on/note off messages on different channels with different notes.  This phenonmon was viewed with midiox.

If a protection diode is open, or shorted, can it cause such things?

Johnc

[Duggle]

The "protection" diode I mention is the one in the driver(s) to protect it from back emf's generated in the load (coils). They come into play when the + common pin of the ULN is connected to the load positive. 

 

What keyboard matrix circuit are you using?

Edited March 7, 2013 by Duggle

[John_W._Couvillon]

Duggle,

i am using 3 KB's from a "liberated"  Saville organ.  they are configured with a full length PCB forming an 8x8 matrix, with diodes.  8 terminals for rows, 8 terminals for columns.

Current problem involves unwanted messages.  Depressing a key produces the correct noteon/noteoff messages, but also some extras with note value and channel not included in the .mio file.

I have checked everything, even replaced all the IC's on the 2 KB pcbs, to no avail. Take a look at the midio128 ver.3 thread in "midification".

johnc