bewildered by protection relay circuit

[samgraysound]

Hi,

I'm trying to fix an Ashdown MK500 bass amplifier head. When I got it, the speaker protect relay was stuck open. I cannot figure out why. The rest of the amp works fine. At this point I have completely disconnected everything else from the power supply and protect circuit and the relay still wont close under normal conditions. Attached is a schematic.

I have replaced the JFET, the relay, and all diodes. I have checked the other two transistors with a transistor tester, and checked the electrolytics with a ESR meter. I have checked all resistor values.

When I bring it up on my variac to a mains voltage of about 130 the relay closes.

My understanding of the circuit is the JFET gets reversed biased, causing a positive bias to appear at the base of TR17, which causes a positive bias to appear at the base of TR15 allowing current to flow through the relay to ground. It seems like the JFET is not cutting off. But the voltage at the gate of the JFET is -2.5vdc, and it should cutoff below -1vdc according to the datasheet. When I raise the mains voltage and the relay closes the gate voltage only changes to -2.6vdc

TR17 voltages are:
B - 110mv
C - 42.5v
E - 640mv

TR18 voltages are
B - 640mv
C - 42.5v
E - 0v (grnd)

Tell me the obvious thing I'm missing people, I've been fiddling with this thing for months.

 

[JohnRoberts]

Hi,

I'm trying to fix an Ashdown MK500 bass amplifier head. When I got it, the speaker protect relay was stuck open. I cannot figure out why. The rest of the amp works fine. At this point I have completely disconnected everything else from the power supply and protect circuit and the relay still wont close under normal conditions. Attached is a schematic.

I have replaced the JFET, the relay, and all diodes. I have checked the other two transistors with a transistor tester, and checked the electrolytics with a ESR meter. I have checked all resistor values.

good

When I bring it up on my variac to a mains voltage of about 130 the relay closes.

huh? The relay should close at lower voltage if I understand?

My understanding of the circuit is the JFET gets reversed biased,

Initial conditions when mains power is first applied, the gate of the JFET is 0V so JFET is on hard. This turns off TR15 and TR17 so relay gets no power and stays open.

Whan mains voltage is applied  D4 and D3 pull current from C16 and C13 driving the gate negative. When TR18 is cut off, R26 turns on TR17 and TR15 (a darlington connection).

causing a positive bias to appear at the base of TR17, which causes a positive bias to appear at the base of TR15 allowing current to flow through the relay to ground. It seems like the JFET is not cutting off. But the voltage at the gate of the JFET is -2.5vdc, and it should cutoff below -1vdc according to the datasheet. When I raise the mains voltage and the relay closes the gate voltage only changes to -2.6vdc

TR17 voltages are:
B - 110mv
C - 42.5v
E - 640mv

TR18 voltages are
B - 640mv
C - 42.5v
E - 0v (grnd)

Tell me the obvious thing I'm missing people, I've been fiddling with this thing for months.

well for one TR18 does not have a base,emitter,and collector its the JFET.

Some simple tests...  If TR18 is removed from the circuit... does relay turn on and stay on?

If not look at TR15 /17 that should easily turn on with TR18 out of the picture.

JR

 

[samgraysound]

huh? The relay should close at lower voltage if I understand?

well for one TR18 does not have a base,emitter,and collector its the JFET.

Some simple tests...  If TR18 is removed from the circuit... does relay turn on and stay on?

JR

Yes this is what is weird. At a normal mains voltage the relay is not energized. When I raise the mains voltage on the variac the relay energizes.

The voltages I gave for TR18 were for TR15, typo on my part.

When I remove TR18 from the circuit. The relay does energize immediately once the amp is turned on. The voltage at the cathode of D5 (what would have been at the gate of TR18) is -2.4vdc.

I suppose  I could replace the JFET again, and check for trace damage between the drain and ground...what else?

 

[Gene Pink]

The voltage at the cathode of D5 (what would have been at the gate of TR18) is -2.4vdc.

-2.4V isn't right, unless a 400-500W amp has less than 30 volt DC rails. Measure the voltage at the point where D3 and D4 anodes are connected together (should be the same as the neg rail), the -2.4V should be 1 / 11 (one-eleventh) of that, from the voltage divider R19 and R23.You mentioned checking cap ESR, but did you check leakage? A leaky C13 or C16 would do this, dragging the voltage down to below a threshold to work properly, and a bit of overvoltage with a PowerStat (*) gets it above that threshold to work.

(*) Why is "Variac" the generic name for these things? Who decided that? :

Gene

 

[JohnRoberts]

Gate voltage will be -unregulated divided by 10.

-2.5V is not a lot, but J112 should cut off at -1V Vgs, so -2.5V should switch it completely open circuit. 

Yes, try another JFET.

JR

 

[mjrippe]

(*) Why is "Variac" the generic name for these things? Who decided that? :

Because they Vary the AC voltage?  Where as PowerStat might provide some statistics about the power (via metering)?  Actually my Variac (R) has meters too...but really General Radio sold a huge number of these so it became genericized if that can be a word.

 

[Gold]

It’s a little easier to say than variable auto transformer.

 

[JohnRoberts]

It’s a little easier to say than variable auto transformer.

Yup, technically it's a variable or tapped "autoformer"....  generally only a single winding.

JR 

 

[samgraysound]

-2.4V isn't right, unless a 400-500W amp has less than 30 volt DC rails. Measure the voltage at the point where D3 and D4 anodes are connected together (should be the same as the neg rail), the -2.4V should be 1 / 11 (one-eleventh) of that, from the voltage divider R19 and R23.You mentioned checking cap ESR, but did you check leakage? A leaky C13 or C16 would do this, dragging the voltage down to below a threshold to work properly, and a bit of overvoltage with a PowerStat (*) gets it above that threshold to work.

V- at the output of DR1 is -49.5
At the anodes of D3 and D4 it is -25 (this seems  odd?)
I replaced C13 and C16, no effect
I double checked the values of R19, R22, and R23. They are all spot on.

 

[samgraysound]

Gate voltage will be -unregulated divided by 10.

-2.5V is not a lot, but J112 should cut off at -1V Vgs, so -2.5V should switch it completely open circuit. 

Yes, try another JFET.

JR

Today I socketed the position of TR18 then tried a number of different J112s, no effect. I also checked connectivity for all the signal grounds and they are all connected.

 

[JohnRoberts]

Today I socketed the position of TR18 then tried a number of different J112s, no effect. I also checked connectivity for all the signal grounds and they are all connected.

Looking at a data sheet today, I am seeing a different story so disregard my earlier post.

Vgs (off) for j112 varies from min of -1V to max of -5V so -2.5V may not be enough to pinch it off, and this is consistent with your evidence of higher mains voltage tripping it.  data sheet http://www.farnell.com/datasheets/2287770.pdf?_ga=2.72393875.120755367.1521658079-360060878.1521658079&_gac=1.48548754.1521658079.EAIaIQobChMIyNfyrqT82QIVgbrACh0nEQQnEAAYASAAEgLqQPD_BwE

Before we change the design (divider ratio) what is the unregulated voltage  feeding D3 and D4 ?  To work in the worst case for J112 that voltage would need to be like -50V. Note: if C13 or C16 are polar and installed backwards their leakage current could make it harder to reach jfet pinch off voltage.

1- check polarity of C13/C16 (+lead should connect to ground).
2- check peak voltage at D3/D4. This may be hard to measure without a scope...  A small film cap (say 0.1uF mylar) attached from anodes of D3/D4 to ground might help with that measurement.
3- as a last resort we can change the ratio (R23) , but it is rarely a good idea to assume a published design is blatantly wrong.  :

You can perhaps parse out unusual leakage caused by C13/16 by measuring DC voltage drops across  R19, R22, and R23...

JR

 

[samgraysound]

Looking at a data sheet today, I am seeing a different story so disregard my earlier post.

Vgs (off) for j112 varies from min of -1V to max of -5V so -2.5V may not be enough to pinch it off, and this is consistent with your evidence of higher mains voltage tripping it.  data sheet http://www.farnell.com/datasheets/2287770.pdf?_ga=2.72393875.120755367.1521658079-360060878.1521658079&_gac=1.48548754.1521658079.EAIaIQobChMIyNfyrqT82QIVgbrACh0nEQQnEAAYASAAEgLqQPD_BwE

Before we change the design (divider ratio) what is the unregulated voltage  feeding D3 and D4 ?  To work in the worst case for J112 that voltage would need to be like -50V. Note: if C13 or C16 are polar and installed backwards their leakage current could make it harder to reach jfet pinch off voltage.

1- check polarity of C13/C16 (+lead should connect to ground).
2- check peak voltage at D3/D4. This may be hard to measure without a scope...  A small film cap (say 0.1uF mylar) attached from anodes of D3/D4 to ground might help with that measurement.
3- as a last resort we can change the ratio (R23) , but it is rarely a good idea to assume a published design is blatantly wrong.  :

You can perhaps parse out unusual leakage caused by C13/16 by measuring DC voltage drops across  R19, R22, and R23...

JR

The voltage off the secondary of the transformer is 65vac rms.

The caps are definitely oriented correctly.

At the anodes of of D3 and D4, it measures -27vdc and 13vac on my Fluke. When I connect the .1uf cap to ground, it becomes -37dc and 5vac.

Voltage drops:
R19 - 25.7
R22 - 0
R23 - 2.46

 

[JohnRoberts]

The voltage off the secondary of the transformer is 65vac rms.

65 ac rms is like 90V peak

The caps are definitely oriented correctly.

good

At the anodes of of D3 and D4, it measures -27vdc and 13vac on my Fluke. When I connect the .1uf cap to ground, it becomes -37dc and 5vac.

this seems a little low if transformer is putting out 95V peak. Do you have access to a scope?

Voltage drops:
R19 - 25.7

This voltage seems a little low

R22 - 0

this is cool... no current flowing into or out of hi-z gate.

R23 - 2.46

consistent with R19 measurement.. so divider is working and JFET is OK (as far as we can tell)
So problemo appears to be voltage at anode of D3/D4...  Probably should be 2x that.

Check the integrity of wires and solder connections in that area... I would expect a bunch more negative voltage than -37V at that junction. 

Vp-p is 2x1.4x rms V  so that voltage is way way low, and obviously the problem.

Good luck...

JR

 

[samgraysound]

Do you have access to a scope?

So problemo appears to be voltage at anode of D3/D4...  Probably should be 2x that.

Check the integrity of wires and solder connections in that area... I would expect a bunch more negative voltage than -37V at that junction. 

Vp-p is 2x1.4x rms V  so that voltage is way way low, and obviously the problem.

Good luck...

JR

Thanks for all your help JR. I do have a scope, but it isn't calibrated so I can't get an accurate p2p reading from it at the moment

I replaced D3 and D4 again, and checked all the connections to and from them and for shorts to ground. connections are good and still -27v.

I pulled DR1 and c7 off the pcb so that the only connections to the transformer secondaries are D3 and D4 . still -27v

I hooked up 2 1n4002s directly to the transformer secondary, completely isolated from the pcb. -29v referenced to the transformer ct.

Think I'll try changing the value of R23 just to see if it will work.

 

[JohnRoberts]

Thanks for all your help JR. I do have a scope, but it isn't calibrated so I can't get an accurate p2p reading from it at the moment

I replaced D3 and D4 again, and checked all the connections to and from them and for shorts to ground. connections are good and still -27v.

I pulled DR1 and c7 off the pcb so that the only connections to the transformer secondaries are D3 and D4 . still -27v

I hooked up 2 1n4002s directly to the transformer secondary, completely isolated from the pcb. -29v referenced to the transformer ct.

Think I'll try changing the value of R23 just to see if it will work.

For the record I DO NOT APPROVE changing the design because something else is misbehaving.

If you actually have 65V RMS on the transformer, you should get more like -90V peak, not -35V DC

If you bandaid it, the problem can continue to get worse and the bandaid won't work forever.

JR

 

[samgraysound]

For the record I DO NOT APPROVE changing the design because something else is misbehaving.

If you actually have 65V RMS on the transformer, you should get more like -90V peak, not -35V DC

If you bandaid it, the problem can continue to get worse and the bandaid won't work forever.

JR

I absolutely agree. That's why I haven't band-aided it for the last six months.

But I have literally tested or replaced every component, many twice or more. I have literally done a continuity check on every trace. I'm completely out of ideas. I've tried every suggestion from this thread and am no closer to figuring it out. I suppose I could breadboard it to completely eliminate the possibility of a problem with the pcb. Although, it seems more likely there is something I'm not understanding about the circuit or some sort of testing error on my part.

I just sunk another 2 work days into this, and I need it off my bench. Maybe after a load test the problem will really become obvious.

Oh I also modeled it in Spice, which put the voltage at the gate of JFET at -1.8v. so wtf is even going on. I have no idea any more.

 

[JohnRoberts]

I absolutely agree. That's why I haven't band-aided it for the last six months.

But I have literally tested or replaced every component, many twice or more. I have literally done a continuity check on every trace. I'm completely out of ideas. I've tried every suggestion from this thread and am no closer to figuring it out. I suppose I could breadboard it to completely eliminate the possibility of a problem with the pcb. Although, it seems more likely there is something I'm not understanding about the circuit or some sort of testing error on my part.

I just sunk another 2 work days into this, and I need it off my bench. Maybe after a load test the problem will really become obvious.

Oh I also modeled it in Spice, which put the voltage at the gate of JFET at -1.8v. so wtf is even going on. I have no idea any more.

Sorry I use spice(s) only when cooking in the kitchen...

We have a logical paradox... You tell me you measure 65V RMS across the transformer, but do not get the expected as much as 90V peak driving the gate circuit.

It could be something obscure like the transformer CT  J1 is not the same 0V "ground" as as the JFET source 0V ground.

This is physics not magic so the answer is something simple and logical, while it will seem a lot more logical after you find it.

Probably  a bad connection somewhere, as I suggested earlier confirm the integrity of associated connections (like transformer CT).

It's always something.

JR

 

[PRR]

> Why is "Variac" the generic name for these things? Who decided that?

General Radio developed the product for internal uses, then promoted the HECK out of it. Perhaps to recoup the substantial (for GR) tooling costs to make those ring-cores. This is Variac. Like aspirin and heroin, it got genericized.

PowerStat is a me-too product from a big transformer company. Despite having some cost leverage, and making a wider range of sizes with various frills, I think Variac had the bulk of the market until GR's business got fuzzy. Their trademark lapsed(?) in 2002, another company registered the word "for use" in 2004.

 

[Audio1Man]

Here is a quick circuit description: V+ = 50 volts (Measured), AC1 or AC3 to Common is about 35vac( 50vdc x 0.707), D3 & D4 voltage is negative 50 volts, C16 voltage is about negative 5 volts (R19 & R23 divider), TR18 is a JFET and is ON (Low ohms) until negative voltage @ the gate turns it off (High ohms) RC time, R26 then goes high about 1.2 volts this bias turn TR17 & TR15 and this turns on the relay. D5 helps the turn off when power is turned off. If the heat sink gets to 100 degrees the relay opens.

The JFET may need more bias to turn it off. By lowering the value of R19 by 30%,  the negative voltage will increase and the relay will energize sooner.
Duke

 

[Gene Pink]

> Why is "Variac" the generic name for these things? Who decided that?

General Radio developed the product for internal uses, then promoted the HECK out of it. Perhaps to recoup the substantial (for GR) tooling costs to make those ring-cores. This is Variac. Like aspirin and heroin, it got genericized.

Jeez, guys, I meant that as a rhetorical  question, not a topic swerve, but I do appreciate the input.

Being curious, I took inventory today around here.

1) tiny General Electric "Volt Pac" 2.5 amp
1) Radio Shack 5 amp, dunno what is actually in it
1) Variac 20 amp
2) PowerStat 5 amp
1) PowerStat 240V / 5 amp
1) PowerStat 20A
3) PowerStat 50 amp with separate servo motors, in a working CPU based Staco three phase voltage regulator 8)

Just sayin', only one actual Variac in the menagerie here.

Now, back to our regularly scheduled topic.

@  Sam: When you say the VAC out of the transformer is 65VAC, you mean end-to-end, correct? End-to-center tap is more important here, should be half of that, 33.5VAC. This means peaks at D3,4 anode junction of about -45V. With no filtering cap, it will be full wave humps, and most DC meters will average that out in some way. Regardless, eventually the both caps should charge up to 1:11 of that to over -4V, in an "as time approaches infinity squared" way, eventually.

A scope would really help here, as either the hump peaks aren't getting to -45V, or something is leaking to ground. It doesn't take much to bleed down that cap connected to the gate, as it is getting charged through 200K total, in tiny pulses. Maybe try cleaning the circuit board well in the cap/jfet area?

Gene