Resistance of copper traces, or a mil2 bypass pcb prob

[Auke Haarsma]

Situation: I've 'designed' my own millenium 2 bypass, based on RG's schematic and pcb. Mine includes traces to an onboard 2pdt footswitch. I've etched the pcb myself. Checked connections of the traces. They are all ok.

It's a simple schematic. 1 Led, 1 BS170 mosfet, 1 current limiting resistor, 1 Switch, 1 High leakage diode. (the protection diode is omitted, it plays no part in this situation)

Problem: The LED does work. However, when I connect an FX to my pcb and put the footswitch in the 'OFF' position, the LED stays on. (it should turn off)

Trouble Shooting steps taken:To test, I build the schematic on a breadboard. It works flawlessly on the breadboard. I swapped the mosfet (it was socketed on my pcb), so I know for sure the mosfet is ok.

I connect the breadboard-bypass to the FX-out. The LED is turned off. Next, I connect the pcb-bypass to the FX-OUT. The LED stays on.

Now I connect the breadboard-bypass directly to GND -> LED is turned off. I do the same with the pcb-bypass, LED is also turned off. So the bypass does work!

I rechecked the PCB a million times, I'm positive the PCB is correct.

Question: Could it be that the PCB traces have a too high internal resistance so that the PCB when connected to the FX-OUT has too high a resistance to ground? What other causes could there be for this problem? Why does the pcb-bypass work when I connect it directly to ground and not when I connect it to the FX OUT?



/edit: added the highleakage diode to 'situation', which ofcourse is also on my pcb/breadboard

[jlullo]

isn't this because you haven't wired jacks yet?  i could be wrong, but i think the circuit will function as you mentioned until you are breaking the -9v connection on your input jack ring

[Auke Haarsma]

Thanks for your reply.

I should have mentioned the jacks, input and output of the effecft are connected to the pcb. They are not soldered but I use testwires with claws. These connections are ok. I doulbe check them with a DMM before testing the switch. So I don't think that is the problem.

[jlullo]

are you sure that the negative lead of the 9v is going to the ring and not the shield?

could you post a copy of your layout so we can take a look at it?

[Auke Haarsma]

are you sure that the negative lead of the 9v is going to the ring and not the shield?

I'm at work now, but I'm 99.9% positive jacks are wired correclty (these are my 'test-jacks' which I have used for all my breadboard expiriments.

could you post a copy of your layout so we can take a look at it?

Not at this time (at work atm), but it is the same as:


With the 2pdt added. I've checked this many a time and I'm sure it is ok (ofcourse I'll recheck when I get home )

[Paul Perry (Frostwave)]

Whatever the problem is, it isn't the PCB trace resistance.

[Auke Haarsma]

Whatever the problem is, it isn't the PCB trace resistance.

That's exactly what I would think. It must be something else, a mistake I overlooked. But I've been staring at the pcb and breadboard for quite some time...puzzled what the difference between the two is.

Paul, do you have any idea in what direction I should look?

[Auke Haarsma]

Here's an image of the pcb I made.



The order of the pads at the bottom is: FX OUT, FX IN, SIGNAL IN, SIGNAL OUT.

FX goes to FX in/out, Signal goes to jacks/signal line.

The switching works. The LED just stays on, and it's close to driving me crazy...

Any clues?

[R.G.]

Let's think.

The LED always stays on. That means the MOSFET is always conducting.

There are only three ways that can happen.
(1) the MOSFET is dead; we will discount that as the most unlikely
(2) the MOSFET drain/source leads are reversed in the PCB holes and the body diode is conducting
(3) a defect of some sort in the circuit is pulling the MOSFET gate higher than about 2.5V

So let's eliminate (3) and test (2). Disconnect the PCB from the effect entirely. Temporarily solder a 1M resistor between FX_OUT and ground on the PCB. Attach a battery to the +9V and GRND locations on the PCB. If the circuit is correct, the LED will turn on and off. (By the way, I think the PCB layout does correspond to the schematic.)

Now test. Does the LED go on and off on alternate switch pressings? If not, the problem is entirely on the PCB.

So now we eliminate a defect on the PCB. Using your DMM, measure the voltage between the anode of D2 and ground as you press the switch. If it stays at or near 9V, D2 is shorted somehow. If it goes to 0V one way and either stays at 0V or rises to a middle voltage the other way, that would be normal because your meter is loading the circuit down.

Finally, test the orientation of your BS170's. There is at least one manufacturer who makes backwards pinout BS170s. Use your DMM set to ohms, and measure between the drain and source both directions on a MOSFET that you don't care if you kill. If you don't kill it, you will see open one way and a forward biased diode the other way. Or just try it the "right" way AND the "wrong" way.

One of these should fix the thing.



(it's worth noticing that if you test the switching PCB all by itself, the LED will always be on; there must be a DC resistance to ground from FX_OUT to tu

[Auke Haarsma]

R.G. thanks! Your structured way of pointing out possible causes really helps to understand what's going on.

I followed the steps right away.... first to no avail. Swapped around with BS170's...all kept the LED on. Checked with the 1M in place to FX_OUT...still no go.

I did tests both on the PCB and on the breadboard... same results in both situations.

I measured the anode of D2. This seemed ok.  However...still all the LEDs stayed on. Checked the BS170's as RG described, but didn't got varying results. Some appeared ok, some didn't, but I wasn't sure what I should get as a result on my DMM.

Sometimes it's better to just let it go for a while, so I grabbed a movie and got a good nights rest.

This morning I started over again. Cleared my breadboard. Put two milbypass2's on the board. Put in a fresh BS170...and...IT WORKED. Phew... I then used the breadboard to test the BS170's. I tested all BS170's I had (about 25). I took the working one out, put a new one in. If the LED turned ON and OFF when (dis)connecting with the 1M to ground I knew the BS170 was ok. Turns out *all* BS170's I used last night where defective. Every single on off them.

Next I moved to the PCB's. After fiddling around with the components, resoldering bits and swapping in working BS170's I got one to work! The PCB is correct, the components are correct. Phew!

I tried another PCB...it works too!! I tried a third...it's not working. But I know it is something on the PCB.

One question left. I tried the mil2bypass with a 25k b pot between the +9v and the 9v input on the board. This allows me to lower the voltage on the board and thus the brightness of the LED. I tried this the PCB's and the breadboard and didn't experience any problem at all. However the question is: does the mil2bypass require 9v to work or could it work with like 5v? (it does work with 9v, I checked that, but could I expect problems in the future when dimming the leds by lowering the voltage on the bypass?)

RG, thanks again!

[R.G.]

I feel a little silly. The actual problem you had - multiple bad parts that were new - was the one I discounted. I'm glad you kept trying and found that.

The Mill2 will work down to about 3-4V depending on the specific MOSFET and the LED. The BS170 needs something like 0.5 to 2.5V on its gate to turn on. If you get a high one, at 2.5V, then a bit more than that would be needed to make it work. Call it 3V. The LED needs 1.5V for old-style red ones, up to 2.5V for new green ones, even as much as 4V for high brightness blue, maybe more. You need some voltage across the resistor in series with the LED to limit its current.

So a practical limit is about 3.5-4V with a 2V LED. That lets you put a resistor with 2V across it in series with the LED, and also ensures that the MOSFET should toggle. Some combinations of components will work lower if you hand select them.

Congratulations! Good work.

[jlullo]

i'm happy you figured this one out  

[Auke Haarsma]

I feel a little silly. The actual problem you had - multiple bad parts that were new - was the one I discounted. I'm glad you kept trying and found that.

Well, I'm pretty sure I killed those BS170's myself when messing around with the problem ... I'm glad that I now know for sure the parts and pcb are good.

So a practical limit is about 3.5-4V with a 2V LED. That lets you put a resistor with 2V across it in series with the LED, and also ensures that the MOSFET should toggle. Some combinations of components will work lower if you hand select them.

I'm actually using the bright blue leds for the switches. I learned yesterday that these are among the Leds that require quite some voltage to work... They seemed to get dimmer quite fast, so I'm pretty sure I can keep the mosfet working and making the let dimmer at the same time.

Congratulations! Good work.

What can I say... but 

i'm happy you figured this one out  

Thanks I'm very very happy myself too. This switch is part of my multi-fx project...so I have like 17 of these... my project stalled at these switches for 2-3 weeks already (I have most time for the project during weekends only...). Not understanding why the switches didn't work was really close to driving me mad. RG saved the day again

[igor12]

Don't feel bad RG, I woulda though the mosfet is the least likely thing to fail! Always use you ant-static wrist straps!