Replacing mechanical switch with relays, relays damaged, considering MOSFETs. Need advice.

Thread Starter

Agent24

Joined Jun 14, 2010
25
I wish to repair an old 80s\90s Sanyo tape\radio 'boombox'. The unit uses some custom (or at least, very obscure and obsolete) end-actuated slide switches in a 6P3T configuration, one for mode selection and one for radio band selection.

The mode selector switch switches, among other things, the main power rail(s). This first section of the switch (which I have called SW1A) has burnt out after some years and no longer functions. The second section of the switch, (now called SW1B), has only one function - to pull low some part of the tuner circuitry when switching from AM to FM mode.

See attached schematic I have drawn up for how the original design was intended to operate.

To clarify, the 9.6v rail comes from the tape deck mechanism, as an output from the supply to the motor, switched on only when a function button on the deck is pressed. (ie: the 9.6v rail is at 0v when the tape is doing nothing)
The 12v rail comes straight from the DC\AC selector switch, which switches between the battery compartment and the mains transformer\rectifier diodes.

SW1A allows either the 9.6v or 12v rails to continue through to the main supply filter capacitor, a 2200uF 16v part, which then supplies the rest of the circuitry.


As a new switch was impossible to find, my repair idea consisted of taking the good contact from SW1B and putting it in SW1A, then using SW1A to activate some relays to perform its (and SW1B's) functions, with the idea being that SW1A will now last much longer as it is only switching a small current to some relay coils, and SW1B gets replaced with a relay as well since I stole its contact.

I used a 7805 to tap from the main 12v rail and power three 5v reed relays, RL1, RL2 and RL3.
The circuit I designed worked as intended during testing but failed immediately in practical use - both RL1 and RL2 failed with their switching contacts shorted together. They were rated for 1A which seemed plenty given the size of the transformer, but now I think the surge current to charge up the 2200uF capacitor may have been too high for them.


Is it possible to use my relay circuit with the addition of some kind of protection to prevent the relays from being damaged or do I just need heavier relays?
The problem with larger relays is that space and mounting options are limited, and their cost may make the entire thing uneconomical.


I also thought about using MOSFETs instead, but would this fall down another rabbit hole? - if RL1 and RL2 were N-channel MOSFETs (with NPN BJTs for inverting the gate drive), then whichever one was off would get its Source pin pulled high by the output of the other. I don't know if this also has the potential for damage.
 

Attachments

crutschow

Joined Mar 14, 2008
34,285
You might try a power P-MOSFET as a switch. The plus power is connected to the transistor source and the drain is the output.
You connect a 10k resistor between gate and source to bias the transistor normally off.
You ground the gate through the control switch to turn on the transistor.
One rated for a few amps should be sufficient.
 
Last edited:

Thread Starter

Agent24

Joined Jun 14, 2010
25
Unfortunately I don't think MOSFETs of either type would work, there are two problems.

1) For your idea of a P-Channel MOSFET there is the issue that the 9.6v rail is not present when the tape deck is not operating. In fact it is literally disconnected by a switch on the deck. So the Gate and Source of the MOSFET for that rail would be sometimes floating (not sure if that matters though)

2) Worse though is if someone operated the tape deck while the tuner was running (say, to record radio to tape), then the MOSFET for the 9.6v rail would have 9.6v on Source and 12v on Drain. (Or vice-versa for an N-Channel MOSFET) In either case, would this not cause a current to flow backwards through the body diode and cause problems?

Also, a third thing - could the tape deck motor and its switch introduce noise\spikes into the power rail and cause damage to the MOSFET in some way?
 

crutschow

Joined Mar 14, 2008
34,285
1) Floating the gate and source is not a problem in this situation.

2) If you need to block in both directions then you can use two P-MOSFETs in a back-to-back (sources connected together) configuration. That connects the body diodes back-to-back.
Connect the 10k resistor from the common source connection to the two gates in parallel.
That configuration will block in both directions when the gate connection is left open.
Both MOSFETs will then turn on when the gates are grounded (since a MOSFET conducts equally well in either direction when ON).
Edit: The main disadvantage is that is the two ON resistances of the MOSFETs are in series, but the total should still be sufficiently low with reasonably high current MOSFETs.

If you are worried about spikes you can add a zener to ground from the line to the motor next to the MOSFET (anode to ground).
It's voltage rating should be a few volts above the highest supply voltage (say 15V for a 12V supply).
 
Last edited:

Thread Starter

Agent24

Joined Jun 14, 2010
25
Thanks for your help so far. Now I have some more questions...

I always read that floating inputs\gates etc with MOS devices was never a good idea, so that's why I asked. Why is it OK in this scenario?

What kind of Rds(on) is low enough? Can I get away without heatsinking?
I was looking at these: http://nz.element14.com/infineon/ipd90p03p4l-04/mosfet-p-ch-30v-90a-to-252-3/dp/2432731

I chose them because a 12v audio amplifier kit I have been building uses an IRF1405 for reverse polarity protection, which has to handle a few amps and doesn't have a heatsink. It has an Rds(on) of 0.0053 Ohms. I found the IPD90P03P4L-04 (With 0.0033 Rds(on)) by searching for P-Channel MOSFETs of 0.0053 Ohm Rds(on) or lower and then ... sorting by price. I figure these are probably OK, but I know there is always more than meets the eye at first glance!


Another question too... I used the 1N4148 diode to provide the logic function of switching RL2 and RL3 on in FM mode, but only RL2 in AM mode. Will the voltage drop on this diode cause problems when switching the MOSFETs instead?
(If I understand right, it would mean the gate will not be pulled completely to ground?)
 

GopherT

Joined Nov 23, 2012
8,009
@Agent24
Have you tried to open up the defective switches to clean and polish the contacts? Switches like this tend to oxidize - copper turns to black Copper(I) oxide and stop conducting. Use a good quality pencil eraser. If that is not enough, various sprays exist. You could also use a relay hone to polish the contacts.

It would be much easier than any new circuitry.
 

crutschow

Joined Mar 14, 2008
34,285
I always read that floating inputs\gates etc with MOS devices was never a good idea, so that's why I asked. Why is it OK in this scenario?
The floating gate is only a problem if you have a voltage connected across the source-drain terminals because the gate input level is then indeterminate as well as the drain-source current.
In this case the gate is not floating, it is connected to the source by the resistor.
What kind of Rds(on) is low enough? Can I get away without heatsinking?
I was looking at these: http://nz.element14.com/infineon/ipd90p03p4l-04/mosfet-p-ch-30v-90a-to-252-3/dp/2432731
The Rds(on) needed depends upon the current it carries as that determines the voltage drop and the transistor dissipation.
If the Rds(on) is .01 ohms, for example, then a 5A current would drop the voltage 50mV (negligible) and dissipate 5A * .05V = 0.25W (requires no heat sinking).

The NFINEON IPD90P03P4L-04 should work fine in your application. The Vgs won't quite reach 10V specified for full-on in the data sheet, when operating from the 9.6V rail (and if you also have a diode in the gate circuit), but it should still be close enough to turn on sufficiently for your lower current application (much lower than the max transistor rating).

Another question too... I used the 1N4148 diode to provide the logic function of switching RL2 and RL3 on in FM mode, but only RL2 in AM mode. Will the voltage drop on this diode cause problems when switching the MOSFETs instead?
(If I understand right, it would mean the gate will not be pulled completely to ground?)
Using diodes to control the P-MOSFETs gate is fine (anodes towards the gate).
The voltage doesn't have to go completely to ground, just low enough so that the gate-source voltage is sufficient to fully turn on the MOSFET (typically 5V for a logic-level type MOSFET and 10V for a standard MOSFET).
 
Last edited:

Thread Starter

Agent24

Joined Jun 14, 2010
25
@Agent24
Have you tried to open up the defective switches to clean and polish the contacts? Switches like this tend to oxidize - copper turns to black Copper(I) oxide and stop conducting. Use a good quality pencil eraser. If that is not enough, various sprays exist. You could also use a relay hone to polish the contacts.

It would be much easier than any new circuitry.
It sure oxidised alright... into nothing. I did originally open the switch with that intention, but the offending contact was not dirty, it had burnt away. This is after probably 20 years of use.
 

Thread Starter

Agent24

Joined Jun 14, 2010
25
The floating gate is only a problem if you have a voltage connected across the source-drain terminals because the gate input level is then indeterminate as well as the drain-source current.
In this case the gate is not floating, it is connected to the source by the resistor.

Using diodes to control the P-MOSFETs gate is fine (anodes towards the gate).
The voltage doesn't have to go completely to ground, just low enough so that the gate-source voltage is sufficient to fully turn on the MOSFET (typically 5V for a logic-level type MOSFET and 10V for a standard MOSFET).
Thanks for the explanation and information. I'll draw up a new schematic and post it here to be sure I haven't missed anything silly.

The Rds(on) needed depends upon the current it carries as that determines the voltage drop and the transistor dissipation.
If the Rds(on) is .01 ohms, for example, then a 5A current would drop the voltage 50mV (negligible) and dissipate 5A * .05V = 0.25W (requires no heat sinking).

The NFINEON IPD90P03P4L-04 should work fine in your application. The Vgs won't quite reach 10V specified for full-on in the data sheet, when operating from the 9.6V rail (and if you also have a diode in the gate circuit), but it should still be close enough to turn on sufficiently for your lower current application (much lower than the max transistor rating).
That's what I was thinking, the Rds(on) is so low the dissipation should be very low indeed for this low power circuit. But having never actually done this before I wasn't sure in case I had missed something about MOSFET specifications.
 

Thread Starter

Agent24

Joined Jun 14, 2010
25
Finally got around to giving this a try. With the MOSFETs it seems to work and they haven't failed like the reed relays did.

What's left of the switch still seems a bit wonky though. Sometimes only one speaker works, or it doesn't switch properly into radio mode, and I have to give it a wiggle. Spraying contact cleaner into it didn't make any difference.

I do notice a bit more hiss in the FM band than I used to, which changes in amplitude when I poke the wiring with my finger. I suspect I may just need shorter wires. Hopefully it is not the MOSFETs picking up noise somewhere. I can't remember if this was different with the relays.

Also had to re-string the dial after accidentally melting the original cord with my soldering iron...
 

Attachments

Top