Randall Century 1000 II Low Output

[DckTech]

Hi,

I have finally had some time to check this amp out some more.

Roly:
To be more specific:
I removed Q3 & Q4 and jumpered the wiper of P1 to R5 = good sound @ Pin 7 IC2
(should there not be audio at Pin 5 IC2? = no audio)
I can hear the effects of Treble, Bass, Middle & Reverb on main speaker, although signal is low.

I have good audio at Reverb In and no audio at Reverb Out Pin 3 IC2.
Why no audio on my probe at this point, yet the audio is making it to main speaker?
No matter the position of the Reverb pot.

The voltages on IC1 Pin 7 (still have Q3 & Q4 out at this time) are:
Normal = -9.5v
O/D     = 0V

Probing wiper of P3 = loud burst & crackle
Probing ends of R5 = softer bursts & crackles

With regards to the Reverb, I have:
swapped the tank with a known good one = no change
made new set of RCA patch cables for In & Out = no change


Let me know what you think, or if you have any other ideas.

I feel we are getting close to figuring this out, thanks to all the help.

Thanks,
Dan

[DrGonz78]

I suspect that something has failed in IC3 which is loading down the switching circuit and preventing the other FET's from switching (despite the correct LED indication).

After reading through this thread I get stuck on Roly's thought here... So if IC1 is working as the switching FET that leads over to IC3 TL604, hence leg 2 AD/DA is working on IC3. Then shouldn't there be signal that is routed through S1 and S2 through that IC? Which leads us through to Q6 and Q7 and that circuit for the distortion channel. It all leads us back after that to the R5 connection. I am not sure what I am alluding to here, but really just have questions. I want to understand this circuit better as others here can explain where I might getting off track with my questions.  8|

[Roly]

I removed Q3 & Q4 and jumpered the wiper of P1 to R5 = good sound @ Pin 7 IC2
(should there not be audio at Pin 5 IC2? = no audio)

Audio at Pin 7 IC2 is real good.  Yes, there must be audio at pin 5 for there to be audio at pin 7 (it's coming out, so it must be going in, right?), but there is quite a lot of gain between those points (roughly R19/R18, 1meg/2k2 or 1000/2.2 ~=500 times).

In this situation you should also have some audio at pin 1 IC1 (or Fx Send), but again the level may be a bit low for your signal tracer.

I have good audio at Reverb In and no audio at Reverb Out Pin 3 IC2.

This suggests that there may be an open somewhere through your reverb tank.  A better place to probe would be the output, pin 1 IC2, or just on the hot end of the Reverb level control, BUT at this point I would put your reverb problem to one side for later and concentrate on the channel switching which is ahead of that point, since a problem with the reverb tank doesn't deal with why there is no signal getting past IC3, and until you can be sure of the input to the reverb chain is working correctly it is better to concentrate on that first.

Before we do however you can try two tests; turn the Reverb control up to half way and "crash" the reverb springs.  This should produce a loud spring crash from the speaker.  No crash?  Then disconnect the reverb tank output to pin3 IC2 and try the crackle/blurt test at pin3 IC2 which will show if the recovery amp (pin 1,2,3 IC2) is working.  If it is check the tank coils for continuity, and look carefully where the tank wires go internally to it's sockets and see if one of them are broken off.

Why no audio on my probe at this point, yet the audio is making it to main speaker?
No matter the position of the Reverb pot.

Because the audio is going directly via C14 and not via IC2 and the Reverb control - IC2 and the reverb tank are a "sidechain".

The voltages on IC1 Pin 7 (still have Q3 & Q4 out at this time) are:
Normal = -9.5v
O/D     = 0V

Okay, this is good and indicates the audio switching control circuit is working correctly, but the audio switches it controls, Q3 and Q4 in the Normal path, and IC3 in the Overdrive path, are still suspect.

We always look for a single failure first, and what is still possible is that a failure in IC3 is preventing the line that feeds the Gate of Q4 from going -ve and turning off this shunt switch, and if it can't turn off, despite its mate Q3 turning on, it will still be sinking most of the signal to ground.

Probing wiper of P3 = loud burst & crackle
Probing ends of R5 = softer bursts & crackles

These are very good signs.  The loud noises from the wiper of P3 tells us that what follows, IC1 (pins 1,2,3) and the main amp are apparently healthy.  We expect there to be less response from R5 because the tonestack (R16, C11,12,12, P3,4,5) is lossy and reduces the signal level quite a bit, so this is all good.

You can check that when pin 7 IC1 is around -9V that the Gate pad for Q4 is half that, around -4.5V.  This is required to switch Q4 off so it stops shorting the signal path in the Normal direction.  Similarly pin 2 of IC3, the TL604, should also be at -9V (and at the moment I'm suspecting IC3 has failed internally and you will find a more positive voltage here, maybe 0V but also maybe higher, and this is what is preventing Q4 from turning off).

If you restore Q3 and Q4, taking care to get their leads correctly oriented, and lift one end of R15 you should get switching of the Normal channel off and on (but by lifting R15 will have isolated IC3 and won't get switching of the Overdrive path).  If this works correctly then the finger is pointing at IC3.  This is a little difficult since TL604's aren't made any more, however there are small sub-assemblies available that will substitute.


Note that in the drawing I have (that you added component numbers to) the input of Q5 is shown going up to the ground rail under Jx1 - this cannot be right and it must actually be coming from the hot side of Jx1, i.e. at C1 (or it simply wouldn't work at all).

HTH


@DrGonz - the TL604 is an analogue switch which controls the path pin 7 to pins 4 and 5, doing the same job as Q3 and Q4.  Why they chose to do one channel with FET's and the other with an IC beats me, but basically they are identical in function while opposite in action.

[DckTech]

Hi,
So I did some more troubleshooting ( Q3 & Q4 back in)

Roly:
Test #1 = reverb crash = yes I get the "crash"
Test #2 = IC2 pin3 = feeding audio in at this point works great = good sound

IC1 pin 7 = -9.5volts = good
Q4 gate = -4.5volts = good
IC3 pin 3 =  -.7 volts = bad (suspect IC3 is bad)

I unsoldered & lifted one side of R15, isolating IC3 = no change

I noticed lots of comments about the schematic, I emailed Randall directly and this is the best they had, i guess.  I know some areas look questionable.

I sent this link & schematic to a college professor at the local college, when I once attended, this was his response:

Hi Dan,

I had a look at your schematic and posts and I have the following comments/questions
-when in normal mode the audio signal is amplified by Q1, buffered by Q2 and then sent to IC2 for more amplification
-when in overdrive the audio is amped by Q5, Q6, buffed by Q7 and then sent to IC2 (schematic with designators is not the same as schematic without see connection left of C1)
Is there a component missing on the schematic? I see a node to the right of Q4 but nothing connected to it.
Can you trace the audio signal through the overdrive stage?
I see the last comment on the post has you checking Q3, Q4, etc... I am very interested in the outcome of that test and will wait for your results
Have you tried injecting a signal on the right of R5? How about injecting on wiper of P3 (same idea that is in the last post)
I have included the expected gate voltages for Q3, Q4 to help you out (I assume the opamps are powered with +/- 12V)
PNL switch closed (normal mode)
Q3 gate approx. -0.2V
Q4 gate approx. -6V
PNL switch open (overdrive)
Q3 gate approx. -12V
Q4 gate approx. 0V
I don't expect Q3 or Q4 to be a problem because it should then work fine in overdrive mode

Hope this helps you out

So, I also went through the above and found:
- as I mentioned above I think the schematic is a bit flawed ( node by Q4 )
- yes I can trace the signal thru the OD section (possible my audio probe is not that sensitive) may no hear the audio at some areas
Normal:
- Q3 gate = -.2v
- Q4 gate = -4.5v

OD:
Q3 gate = -12v
Q4 gate = 0v

So I started to inject the audio signal in various areas:

- right side of R5 = good all the way through
- wiper of P3 = good all the way through

Then I started injecting audio more & more to the left of the circuit, I got:
"good all the way through" everywhere, EXCEPT to the left of Q1 (gate) = no & Q5 (gate) = no
I'm not really sure if that is a good way to test - but I was really curious.

I replaced Q1,C5,
The input jack - a problem -really?

Lots to take in, let me know what you all think.
Much appreciated,
Dan

[DckTech]

Hi,

Update,
I have replaced the input jack and for some reason now that has improved things a little. 
After doing some more accurate sound injecting and testing it appears i am getting about 1/2 the total output volume when I inject the audio anywhere in the pre amp stage. Normal & OD = same results.
When I inject sound into the return, I get full volume.
I guess not really much has changed.


Thanks for any help,
Dan 

[Roly]

Test #1 = reverb crash = yes I get the "crash"

Okay, well that means that everything from the springline to the speaker is working.

- as I mentioned above I think the schematic is a bit flawed ( node by Q4 )

It's a rule of nature that all circuits have errors (and it's Murphy's Law that the error will be right in the bit you are interested in  ).

Then I started injecting audio more & more to the left of the circuit, I got:
"good all the way through" everywhere, EXCEPT to the left of Q1 (gate) = no & Q5 (gate) = no

Ho-kay.  Now this implies that the two channel switching circuits are working okay.  You should be able to confirm this by the Normal "Volume" (only) working in one state, and the OD "Gain" (only) working in the other state.

The latter part could be for two different reasons; input sockets are normally arranged to short to ground if nothing is plugged in to them to silence the amp.  If you didn't have anything plugged into the input this might explain the lack of signal at that point.

It could also be that either (or both) Q1 or Q5 have failed, perhaps due to an excessive signal voltage being fed in - it happens, and either or both are damaged and now shunting the input signal.

Check the Drain and Source voltages on Q1 and Q5.  With a 24V supply I'd expect somewhere around half that on the Drain, and somewhere around a couple of volts on the Source.

I'm surprised that changing the input jack has made any difference, but it could be that you have some sort of short from another cause (outer of screened cable melted into inner insulation) that you have disturbed in the process of replacing the jack.

If the voltages around Q1 and Q5 look reasonable I'd be lifting the input ends of C1 and C6 and trying injecting a signal into each to see if the respective preamps work okay disconnected from the input wiring.

It's all a process of progressive elimination.

The main reason for the difference in sensitivity between the preamps and the Return socket will be the tonestack which is by its nature fairly lossy.

[DckTech]

Hi,

Roly:

Just to confirm:
Normal "volume" = works
OD "gain" = works

I understand the grounding on the input jack - that makes sense.

Voltage readings:
Q1 & Q5 drain = 1.92 volts (I'm assuming not good)
Q1 & Q5 source = 1.70 volts

I measured R3 & R7 = 33k
I compared Q6 drain = 22volts (R22 = 47k not 33k as stated)
The =24 volts is actually 28 volts, a bit high but I'm sure it's within spec.

Just because I had another FET, I replaced Q1 = no change

I did remove C1 & C6, and injected audio into the gates = no change.
I went over all the connecting wire & made sure no wires were exposed or touching anything.

This is all a serious "process of progressive elimination".

Thanks again for your help, any thoughts?

Much appreciated,
Dan

[DrGonz78]

Well I may have played the fool in regards to how the TL604 chip functioned in this amp. However, I think that Q5 is not getting the correct voltage there on the drain. I did look at the schematic and start to question Q5 and Q6. Q5 would be my bet for testing and/or replacing.

[Enzo]

Going back a page.   As someone said, if you have signal coming OUT of an op amp, it must have gotten to its input.   But in general you cannot measure the signal right at the input pins because they are at "virtual ground".  Look that up.   

If the IC is in a socket, you can remove the IC and verify signal to the socket pins.  If there is a resistor leading to the input pin, you can go to the far end of the resistor and see what is there, and if the resistor is good itself, we can infer the signal passes on to the IC.

[Enzo]

And not so sure the Q4 thing is a mistake.  There is a dot on the line, and one nearby to the right on the ground bus.  I suspect they removed a part and neglected to clean up the dots.

[Roly]

And not so sure the Q4 thing is a mistake.  There is a dot on the line, and one nearby to the right on the ground bus.  I suspect they removed a part and neglected to clean up the dots.

Agreed, that's what I assumed.

Q1 & Q5 drain = 1.92 volts (I'm assuming not good)

Correct, that is very "not good".

You answered my next question which would have been about the supply getting to the first stages...

The =24 volts is actually 28 volts, a bit high but I'm sure it's within spec.

...agreed, but now you confound me with...

Just because I had another FET, I replaced Q1 = no change

That is not what I would expect, because 2V on Q1 Drain suggests that it's boofed.

At this point I'm still inclined to the view that both Q1 and Q5 have been damaged by excessive input signal at some point (since they are both unprotected and it is not unknown for an amp output to get connected to another amp input, either by accident or as an an experiment)

I did remove C1 & C6, and injected audio into the gates = no change.

This is to be expected as the DC conditions are so wrong (both Q1 and Q5 saturated on).

STOP THE PRESSES: I've just done an LTSpice sim and discovered that with MPF102's these stages won't bias properly.  When I tried 2N5484's (as used by Randall in the RG80) bingo, the stage biassed correctly! (see attach)

Now, either Q1 and Q5 have to be 2N5484's, OR the values of the Source resistors have to be increased to correctly bias MPF102's.  The sim says that R2 and R8 need to be increased from 2k2 to 5k6. (or perhaps a couple of 10k trimpots so that they can be rebiased for whatever FET you can get, J201's etc).

[DckTech]

Hi,

Thanks to everyone for your replies in trying to solve this problem.

I am using 2N5486 as replacement FET's in this amp, the local supplier did not have the 2N5484, but according to the attached spec, the 2N5486 should be fine.

I could try replacing Q5 again and maybe Q6 & Q7.  That would be all the FET's in the preamp.

Roly:
Thanks for running the sim, so what it's saying is the voltages I'm getting are not correct, should be 7.5v on the drain. 

Not sure what the next move is?

Thanks,
Dan

[Roly]

Not sure what the next move is?

Correctly DC bias both preamp stages.

1. Put in known good FET's at Q1, 2 and 5.
2. Confirm the Drain voltage at Q1 and Q5 is way low (saturated).
3. Tweek the value of the Source resistor so the Drain voltage is ~half supply.(*).
4. Profit!

(* I just repeated the sim with 2N5486 and found it needs 8k2 as the Source resistor)

The easy way is to lash in a 10k-50k pot as the Source bias resistor; set this for 1/2 supply on the Drain, remove and measure value and fit nearest value fixed resistor.  The same value should also work for the other signal path (i.e. Q1 and Q5).

The slightly longer method is to fit a 8k2 and measure the voltage at the Drain; if it's above 1/2 V supply reduce R, if below, increase R - i.e. tune the Drain voltage to roughly half the supply.  {If you want to be picky about it with Q1 and Q2, you can split the difference between the Drain voltage of Q1 and the Source voltage of Q2 so that they both have similar voltages to ground and supply, i.e. maximise signal headroom, but I don't think that will actually make any material difference.}

Take home lessons;

- any active device that is saturated or cut off has negligible gain.

- in any Class-A amplifier stage the idle Drain/Collector/Anode voltage needs to be about half the supply to get maximum available output voltage swing (and stage gain), that is, it must be biased into its linear operating region.

[DckTech]

Well Roly - you are the MAN.

After my weekend of other duties, I finally had time to get back to the Randall.

I put in good FET's and replaced R2 & R8 with 8k2's, which brought the drain voltage to 12v.
And in quoting Roly "PROFIT".
The amp is working great, lots of volume.

Recap:
Looking back, I'm assuming there was a few FET failures.  Brought on by the "loose/broken" leg of the 3300uF in the supply.  It was probably arcing, spiking the voltage or???
That cap was the first thing I replaced.
So replacing the FET's with a substitute, may have caused my agony also.

I have lots of "take home lessons" from this forum - thanks to Roly & everyone who helped.

Thanks,
Dan

[Roly]

Great!  Pleasing result.   :dbtu:

It was probably arcing, spiking the voltage or???

I'll go with the "or???" and suspect that the loose cap was co-incidental. 

If you can find the two FET's that originally came out of Q1 and Q5 and test them I wouldn't be surprised to find they were both shorted Gate-to-Source.

They were playing one night, then suddenly this problem occurred.

Given that this is a true, accurate, and complete account of what happened(*) it's still possible that an excessive voltage spike found its way into the inputs.

(* and people often don't accurately relate what happened, mostly because they are lousy  observers, but sometimes because they now realise that connecting the output of one amp to the input of another was, perhaps, in retrospect, a pretty stupid thing to do.  The main problem is that in attempting to save face they don't tell you things you really need to know to fix it, or to prevent it happening again.  You have to quiz clients about what happened and listen very carefully, but you also have to take what you are told with a fairly large grain of salt.)


To repeat ('coz it's important) a linear stage must be biased somewhere between cutoff and saturation, between no conduction and full conduction, so the output has some "room to move" up and down in voltage.

If the Drain/Collector/Anode is either at the supply (no current) or ground (full current) then the gain of the stage will be nothing; i.e. a change in the input produces no change in the output.

So in general we are looking for the Drain of a FET/Collector of a BJT/Anode of a valve to rest somewhere around half the supply voltage so that it has roughly equal room to move up and down from the resting voltage, maximum "headroom".


After a bit of initial diagnosis we often test our conclusions by substituting a suspect component with another.  We normally make a couple of assumptions; that the substitute is in good working order, and if not identical is near enough - and sometimes we get trapped because these assumptions are invalid for one reason or another.  In this case we all assumed that the substitute FET you had used was "near enough", and as it always is, it isn't until someone gets suspicious and tests that assumption that we find out what is really going on.   :tu: