Please review this Matrix Circuit

Thread Starter

Register

Joined May 31, 2010
20
Please check and corrrect this Matrix circuit.
The major challenge has been getting over the Voltage leakage which has been tried to encounter with lots of semiconductor diode!
Was this the correct approach or exists a better way to handle the leakage?

thanks for your help

 

Attachments

Last edited:

beenthere

Joined Apr 20, 2004
15,819
There are some some symbols that do not seem to be found anywhere but in the schematic (they are not standard symbols). It is nearly impossible to follow the schematic because the ground and voltage points are not conventionally placed.

Perhaps you could add reference designators to all the unlabeled components and give a description of what the circuit is supposed to do?
 

Thread Starter

Register

Joined May 31, 2010
20
Hello beenthere,
Thank you for your response. Please note this circuit was created with Yenka Sim. and attached the schem. with notes. Please advise if further details are needed.
Generally it is an inductor matrix that works fine except for the voltage leaking through the inductors which are not active and causing a short with the other side of the circuit. I have included the Diodes to make one way flow but somehow I feel this is a lame and messy way of doing it and there must be better ways that I need to find out about. All help appreciated.

thanks a lot

 

Attachments

SgtWookie

Joined Jul 17, 2007
22,230
Well, you can select column A and/or column B via switch A and B, respectively - and rows 1 through 3 via the unlabeled switches on the left.

One big problem you have is the ratio of the resistors on the MOSFET gates. Try using 2k resistors from the optocoupler outputs to the gates, and 20k for the pull-up/pull-down resistors.

I don't know what optocouplers you are using. The current transfer ratio generally isn't very good with optos, unless you are using Darlington optos. I suggest using 360 Ohm resistors instead of your 620 Ohm resistors for the IR emitter side. That should give you around 10mA current through the IR diode.
 

Thread Starter

Register

Joined May 31, 2010
20
SgtWookie,
Thanks for the very helpful guidance. Can you please also advise on how to make the voltage stop passing through to the inactive inductors?
Is my usage of the Diodes correct in this respect or is there a better way?
Presently just one inductor is used at a time but some of the others keep getting voltage too, how can this be stopped so just the desired inductor gets the voltage and none of the others?
Would using an ICl7667 be a better way to assure only 1 Solenoid gets all the applied voltage?

thanks for your great help
 

Thread Starter

Register

Joined May 31, 2010
20
Also,why are Solenoids A1 & A2 are showing negative voltage? is that important?

p.s.
changed the Opto Res. to 360 and worked fine but changing the Pull up/downs to 20k from 3.3k blew Fets in the simulator and did not activate the fets so are back to 3.3k. Opto to Gate is/was at 2k as advised.
 
Last edited:

SgtWookie

Joined Jul 17, 2007
22,230
Are you using Depletion-mode MOSFETs? If so, you need to be using Enhanced mode MOSFETs, like IRF520 (N-ch), IRF9530 (P-ch).
 

SgtWookie

Joined Jul 17, 2007
22,230
SgtWookie,
Thanks for the very helpful guidance. Can you please also advise on how to make the voltage stop passing through to the inactive inductors?
The current shouldn't pass through, if you are using enhanced-mode MOSFETs..
Is my usage of the Diodes correct in this respect or is there a better way?
You have an absurd number of diodes in the schematic. The only place you really need them is across the solenoid coils, to take care of the reverse-EMF pulse when the MOSFETs turn off.
Presently just one inductor is used at a time but some of the others keep getting voltage too, how can this be stopped so just the desired inductor gets the voltage and none of the others?
If you have on only switch A or switch B, and only one of the row switches on, the others should not have current flow through them. Their might be some minor leakage current, but that should be of little consequence; it won't be enough to actuate the solenoid.
Would using an ICl7667 be a better way to assure only 1 Solenoid gets all the applied voltage?
No. They are very low-power DC-DC converters.
 

SgtWookie

Joined Jul 17, 2007
22,230
Also,why are Solenoids A1 & A2 are showing negative voltage? is that important?
Your schematic is so spaghetti-like, it's hard to see WHAT is going on.

changed the Opto Res. to 360 and worked fine but changing the Pull up/downs to 20k from 3.3k blew Fets in the simulator and did not activate the fets so are back to 3.3k. Opto to Gate is/was at 2k as advised.
The Vgs won't be high enough to turn on a standard-level enhanced MOSFET.
 

Thread Starter

Register

Joined May 31, 2010
20
SgtWookie,
Your generous help and contribution above should resolve my problems. I'll be looking into Enhanced mode MOSFETs now and report back the outcome.
BTW, before the leakage as you have indicated I was only using 1 Diode per Solenoid to take care of the reverse-EMF pulse and it was after the problem that I added all the rest of the Diodes to stop passage of Voltage to the inactive Solenoids.
BTW, is there a possibility that the reverse voltages are turning the Solenoids active without going through the Fets??? Because I believe that is what's happening. Please advise.

many thanks
 

SgtWookie

Joined Jul 17, 2007
22,230
If the P-ch MOSFET for the respective column is not turned on, there is no current source for the solenoid coils.

If the N-ch MOSFET for the respective row is not turned on, there is no current path to ground for the solenoid coils.

I think that perhaps you have simply complicated the schematic to the point where you no longer understand it.

Try starting fresh, with a bit more simple of a design.

Try just using one N-ch MOSFET to sink current from one solenoid. Use one diode across the solenoid for the reverse-EMF.

Use a 2k resistor from the optocoupler to the gate. Use a 20k from the gate to the source terminal. Ground the source terminal. Use ground symbols instead of running wires everywhere.

Once you get that single circuit working properly, then you can duplicate it to create a column.
 
Last edited:

SgtWookie

Joined Jul 17, 2007
22,230
Please note I amusing "Advanced HEXFET® Power MOSFETs" like irfz44n, is this enhanced MOSFE?
Yes.

Is that what you are using in the simulation?

What is your solenoid current requirement? IRFZ44 MOSFETs have a large gate charge. It will take quite a while to charge/discharge them through large resistors. You may need to use a gate driver.
 

Thread Starter

Register

Joined May 31, 2010
20
sgtWookie,
I did in fact first did as you have mentioned in making first a single Solenoid and then adding on. In fact a single column produces no problem and even a 3x3 matrix as long as no solenoids are connected the voltages at the pins where solenoids get connected are perfect. The whole problem starts after populating the board with more than on column which is what made me believe the problem is not Fet related rather because when activating just 1 solenoid a whole row and column get voltage and the inactive solenoids are acting like shorting the pins and flowing the current all over the place since after all they are all interconnected?!
BTW, aren't the pull up/down resistors acting as gate drivers? if not what gate driver would you please recommend? Someone had recommended the ICL7667 as a gate driver?!
Appreciate all your help and advise...
 

Thread Starter

Register

Joined May 31, 2010
20
Solenoid current is under 3 Amps.
I just changed to IRF520 in the simulator and had the sameproblem.
I'll try to simplify the circuit and make it more clear and re-post it.
 

SgtWookie

Joined Jul 17, 2007
22,230
OK, I didn't read the part number correctly; I thought you meant an ICL7660.

You could use ICL7667's, except you won't have the isolation that the optos provide.

The resistors are very poor gate drivers.
 

Thread Starter

Register

Joined May 31, 2010
20
You could use ICL7667's, except you won't have the isolation that the optos provide.

The resistors are very poor gate drivers.
Well, this could help a lot now. Do you mean if I switch to Gate drivers(at the cost of losing the Opto isolation) then it should resolve the reverse voltage problem?
Could my belief that the reverse voltage has nothing to do with the Fets be true?If yes, then the only solution would be to use as many Diodes as necessary to avoid the stray reverse voltage?
Sorry for so many questions but I'm getting close to a solution I think.

thanks
 

John P

Joined Oct 14, 2008
2,025
I think that in this circuit, you can't use freewheeling diodes across the solenoids. What you need is a diode in series (not parallel) with each solenoid. Then you apply +V to one row, and Gnd to one column, and there will be only one conduction path through which current can flow.

It seems as if a freewheeling diode per row is sufficient to protect all the transistors. You just have to be sure to turn off the row (NPN or N-FET) first. Well, I suppose you could have diodes to Gnd on the columns as well, and then it wouldn't matter. If both transistors turned off at once, you'd just have conduction through 2 diodes instead of one.
 

SgtWookie

Joined Jul 17, 2007
22,230
I think that in this circuit, you can't use freewheeling diodes across the solenoids. What you need is a diode in series (not parallel) with each solenoid. Then you apply +V to one row, and Gnd to one column, and there will be only one conduction path through which current can flow.
This isn't correct. Each solenoid should have it's own "flywheel"/reverse-EMF diode.

There are intrinsic body diodes in all of the MOSFETs. However, it is not good practice to rely on them. In our OP's circuit, the intrinsic body diodes would not provide a discharge path for the solenoid current. Without the flywheel/reverse-EMF diodes, the voltage at the drain would reach very high instantaneous voltages when the N-ch MOSFET turned off.
It seems as if a freewheeling diode per row is sufficient to protect all the transistors.
This is also not correct. There would not be a complete current path.

You just have to be sure to turn off the row (NPN or N-FET) first. Well, I suppose you could have diodes to Gnd on the columns as well, and then it wouldn't matter. If both transistors turned off at once, you'd just have conduction through 2 diodes instead of one.
Also not correct.
 

Thread Starter

Register

Joined May 31, 2010
20
Thank you all with help provided so far.
Below is a less spaghetti version of the circuit

The problem seem to be caused by stray voltage and not the fets. as I will later explain.

 

Attachments

Top